Book - Contributions to Embryology Carnegie Institution No.14: Difference between revisions

From Embryology
(Created page with " http://www.archive.org/details/contributionstoe04carn CONTRIBUTIONS TO EMBRYOLOGY Volume IV, Xos. 10, 11. 12, 13 Published by the Carnegie Ixstitltion of Washingt...")
 
mNo edit summary
 
(15 intermediate revisions by the same user not shown)
Line 1: Line 1:
{{Header}}
{{Ref-Weed1917}}
{| class="wikitable mw-collapsible mw-collapsed"
! Online Editor  
|-
| [[File:Mark_Hill.jpg|90px|left]] This historic 1917 paper by Weed describes the development  of the cerebro-spinal spaces.
<br>
<br>
'''Modern Notes:'''
<br>
{{Neural Links}}
|}
{{Historic Disclaimer}}


http://www.archive.org/details/contributionstoe04carn
=The Development of the Cerebro-Spinal Spaces in Pig and in Man=


[[Book_-_Contributions_to_Embryology|Contributions to Embryology]]




CONTRIBUTIONS TO EMBRYOLOGY
By Lewis H. Weed


Volume V, No. 14


==I. Introductory==


Volume IV, Xos. 10, 11. 12, 13


Probably no field in embryology has been less explored than that relating to the meninges. Our knowledge of the transformation of the perimedulla




Published by the Carnegie Ixstitltion of Washington Washington, 1916
ry mesenchyme into the three fully developed membranes about the cerebro-spinal axis has been largely of a crude sort, with gross generalities based on inexact or incomplete evidence. The present work was undertaken in the hope that by a study of the various stages in the development of the cerebro-spinal spaces there might be gained some knowledge which would afford a basis for a conception of this dynamic metamorphosis.




Many of the problems centering around the development of the meningeal spaces have recently been expounded by Cushing^^) . * Not only do we lack knowledge as to the method of differentiation of the primitive mesenchyme, but we know little about the establishment of the circulation of the cerebro-spinal fluid. When do the chorioid plexuses begin to secrete? When does the venous absorption of the fluid take place? When does the perivascular system begin to remove waste products from the cerebral tissue? And also, what factors play a part in the formation of the subarachnoid and subdural spaces?
These questions, some of which it is hoped the present study will answer, relate to the field of physiological anatomy. Consideration of the subject, however, serves to convince one that they must be investigated coincidently with the stages of morphological differentiation; for it may readily be conceived that the physiological use of the meningeal spaces may precede any morphological differentiation of the three membranes, nor indeed is it unlikely that one of the active causative factors in the metamorphosis concerns this filling of the mesenchyme about the nervous system with fluid.


CARNEGIE INSTITUTION OF WASHINGTON
Publication No. 224


This study, therefore, has been anatomical, but with a broader scope than purely morphological studies would have afforded. Not only has it dealt with the morphological differentiations about the nervous system, but throughout the investigation the relationship of these structures to the possible presence of cerebrospinal fluid has been considered. As the problem developed it was projected more and more into the difficult realm of "tissue spaces." Interest in these spaces largely concerned their physiology, but many points of correspondence between structure and function were found.




niKHS or OIDBON DROTUEnS WASIIINOTON
In some measure this work is a development of an earher study of some of the anatomical and physiological problems of the cerebro-spinal fluid, carried out in the laboratory of Dr. Harvej^ Cushing at the Harvard Medical School.




* The figures in parentheses refer to the bibliography at the end of this paper.


CONTENTS.




PAGE.
==II. Review of Literature==




No. 10. The human magma reticule in normal and in pathological development. By Franklin P. Mall (3 plates) 5-26
In order fully to understand the problems which confront one in the study of the embr3'onic cerebro-spinal spaces, a comprehension of the stage to which investigations have brought our knowledge of these fluid-pathways in the adult is necessar}'. It is with this purpose that the adult relationships are here considered. The inclusion of this material may be pardoned, for it will be seen that unanimity of opinion has bj' no means existed in regard to any of the problems concerned in the circulation of the cerebro-spinal fluid.
11. The structure of chromophile cells of the nervous system. By E. Y. Cowdry
(1 plate) 27-43
12. On the development of the lymphatics of the lungs in the embryo pig. By R. S.




CuNNiNGH.\M (5 plates) 4.S-68
Modern anatomical knowledge of the meninges dates from the work of Axel Key and Gustav Retzius'29). These Swedish investigators, in their excellent monograph published in 1875, first conclusively demonstrated the anatomical continuity of the spinal and cerebral subarachnoid spaces. But for years after their publication appeared, a physiological continuity between the subdural and subarachnoid spaces was argued for bj' many observers, notably by HilK^*). Gradually, however, workers in this field have reached the opinion that the subarachnoid spaces (the interrupted but continuous channels between arachnoidea and pia) are functionally the channels for the cerebro-spinal fluid. Between the intra-leptomeningeal and the subdural spaces no anatomical connection exists; physiologically there may be some mode of fluid-passage. Thus Hill' 24) states that either by filtration or through actual foramina fluid passes readilj- from one space to the other. Quincke '*^, from observations on animals, somewhat similarly premised a connection between the two spaces, but only in the direction from subdural to subarachnoid. His experiments, based on the results of the injection of cinnabar granules, are open to criticism as indicating a normal passage-way for the fluid; for, as he has recorded, an intense phagocytosis of practically all of his granules occurred. More modern conceptions of the subdural space treat it as a space anatomically closed, lined externally by a polygonal mesothelium. Less error is introduced if it be regarded as analogous in many respects to well-known serous cavities rather than as an essential portion of the pathway for the cerebro-spinal fluid.
13. Binucleate cells in tissue cultures. By Charles C. Macklin (4 plates, containing .




70 figures) 69-106
The question of the absorption or escape of cerebro-spinal fluid from the subarachnoid space has claimed the attention of many workers. Since the original conception that the meningeal coverings were actually serous cavities, anatomical investigations have furnished many new views. Key and Retzius, by spinal subarachnoid injection of gelatine masses colored with Berlin blue, demonstrated an apparent passage of the injection fluid into the great cerebral venous sinuses through the Pacchionian granulations (die Arachnoidzotten). Their observations were made on a cadaver and the injections carried out under fairly low pressures (about 60 mm. of mercury). A lesser drainage of the fluid into the lymphatics was also shown.
3




Since the view advanced by Key and Retzius of the absorption of cerebrospinal fluid, the general trend has been away from the idea of an absorption into the venous sinuses. Quincke's observations, made on lower animals after the subarachnoid introduction of cinnabar granules, really offer some substantiation of this theory, but the failure to find the great Pacchionian granulations in infants and in the lower animals caused many workers to reject utterly the conception of the Pacchionian granulations as the functionally active mechanism for the fluid escape.


CONTRIBUTIONS TO EMBRYOLOGY, No. 10.


Physiological evidence, however, advanced by Hill^24) fj-Q^i intraspinous injection of methylene blue, indicated that the major escape of the cerebro-spinal fluid was into the venous sinuses of the dura, while a slow and minor absorption took place along the lymphatic channels. Ziegler(57j^ with potassium ferrocyanide introduced into the cerebro-spinal space, Ukewise found that the venous absorption was much greater and more rapid than the lymphatic. Reiner and Schnitzler with the same agent detected the ferrocyanide in the jugular blood-stream after injection. With oUve oil these investigators found a similar venous absorption, but with a slowing of the venous blood-stream. Lewandowsky also using ferrocyanide. found this salt in the urine within 30 minutes after its subarachnoid injection. Spina 52'^ from observations on freshly killed and hving animals, presented somewhat similar evidence of a major venous and lesser lymphatic absorption. Gushing suggested a valve-Uke mechanism of escape of the fluid, his hypothesis being based on the findings after the introduction of mercurj' into the meningeal spaces.




Several theories concerning the absorption of cerebro-spinal fluid into the bloodvascular system have more recenth' been offered. Mott '**', from a studj' of dilated perivascular and permeuronal spaces, has advanced the idea of fluid-escape by way of the perivascular system into the cerebral capillaries. Dandy and Blackfan^'"^, from an analysis of their evidence, consider that the chief drainage of the fluid is into the capillaries of the pia-arachnoid. Opposed to this conception of a major drainage of cerebro-spinal fluid into the l)lood-vascular system is the view championed by Cathelin'^', that the Ij'mphatic drainage is the chief method of fluid-escape. Cathelin's contention of a veritable circulation of the fluid has not received support from other workers.




THE HUMAN MAGMA KETICULE IN NORMAL AND IN PATHOLOGICAL DEVELOPMENT.
Thus it will be seen that since the work of Key and Retzius the trend of opinion has been away from the view that the Pacchionian granulations carry the cerebrospmal fluid into the venous sinuses.




By Fkanklin P. Mall.
In the earlier investigation carried out in the Harvard Medical School the problems of this fluid absorption were attacked in a somewhat different manner than by previous workers. True solutions of potassium ferrocyanide and ironammonium citrate, such as have been used in the present investigation, were injected into the spinal subarachnoid space under pressures but shghtly above the normal. The animals (dogs, cats, and monkej-s) were kept imder anesthesia during the period of injection, which was usually  continued for several hours. Complete filling of the subarachnoid channels was secured by this technique, provided the injections were continued for a sufficient length of time. At the conclusion of the experiment the foreign solution was precipitated in situ and blocks were carried through for histological purposes.




Many of the anatomical findings in this work carried out as outlined are of interest in the present problem. The complete correspondence of the spinal and cerebral subarachnoid spaces as demonstrated by Key and Retzius was amply verified. The normal return of the cerebro-spinal fluid to the general circulation by way of the arachnoidal villi into the great dural sinuses was demonstrated. These viUi are projections of the arachnoidea through the dural wall, prolonged directly beneath the vascular endothelium of the venous sinuses. Furthermore, columns of arachnoid cells were found, normally affording fluid channels in the dura. In addition to the major escape of cerebro-spinal fluid into the sinuses a lesser drainage was alsi) demonstrated, slower than the primary drainage, out along certain of the emergent nerves into the lymjihatic system. No evidence whatsoever was obtained in support of any of the theories of a drainage of cerebro-spinal fluid into either the leptomeningeal or cerebral capillaries, nor could an anatomical valve-like mechanism along the great sagittal sinus be demonstrated. The process of escape of cerebrospinal fluid from the arachnoid villus unto the great sinus appeared to be a simple one of filtration or of diffusion.




Another of the problems concerning cerebro-spinal fluid, which has been of interest to anatomists and phj'siologists, is the source of the fluid. Haller^^D and Magendie'^^ to whom the greatest credit for work on this subject must be given, believed it to be the product of the leptomeninges. Faivre^^^} {^ 1853 and Luschka^''*) in 1855 were the first to suggest the chorioid plexuses as the elaborators of this circumambient medium. Since then the view has been generall}- accepted that these villous structures do give origin to the fluid, but the early evidence was based wholly on the glandular character of the plexus. Cappelletti^^) and Pettit and Girardi) offered more definite proof of this relationship by the introduction of pharmacological agents which affected the rate of production of the fluid. These latter authors recorded definite histological changes in the cells of the plexus when influenced by these drugs, indicating, in conjunction with the changed rate of production of the fluid, an undoubted relationship of the chorioid plexus to the fluid elaboration. Since these early investigations many observers — Findlay, Meek(37)^ Mott^'*^', Pellizzi''*2), Hworostuchin'26)^ and others — have studied the histology of the chorioid plexus with reference to its function as an elaborator of the cerebro-spinal fluid.


With three i)lates.


In addition to the elaboration of the fluid by the chorioid plexuses, increments are furnished bj' the nervous tissue itself. This elimination from the nervous system occurs bj' waj' of the j^erivascular spaces. In the previous work referred to'55j it was found possible to inject the entire perivascular system by continuing a physiological injection of the spinal subarachnoid space, and subsequently causing an extreme cerebral anemia. By this procedure an injection of the system to its termination about the cerebral capillaries and nerve-cells could be secured. From this and other evidence the view was advanced that the cerebro-spinal fluid was derived from a dual source — in part from the perivascular system and in greater part from the chorioid plexuses. This view had already been advanced, but on rather insufficient grounds, by Mestrezat'^s* and by Plaut, Rehm, and Schottmuller (, Recently Frazier) has signified his acceptance of this conception of the source of the fluid.




Such, then, is the basis for our present understanding of the meninges, in regard to their characteristic morphology and particularly their functional relationship to the cerebro-spinal fluid. Without a consideration of the circumambient fluid morphological studies of these membranes would be incomplete, for in order to understand the meninges knowledge concerning the cerebro-spinal fluid is necessary.




CONTENTS.
===The Comparative Anatomy of the Meninges===




PAGE.
Sterzi's has pubUshed a comprehensive report of the comparative anatomy of the spinal meninges. From his studies he has advanced hypotheses, supported by observations on a Hmited number of fetuses, regarding the development of the human meninges. On account of the interest of this subject in relation to the present discussion a brief summarj' of Sterzi's work will be here included.


e
In the acrania there is no special envelope of the central nervous system, but rather a fibrous sheath corresponding to the meninges of higher forms. This fibrous sheath is largely made up of circular fibers, except in the median ventral line, where there occurs a ventral hgament of longitudinal fibers. In cyclostomes, however, there is found a single "primitive meninx" — vascular and composed of white and elastic fibrils coursing in a longitudinal direction. Some of these fibrils traverse the perimeningeal spaces (filled with star-like cells, with some fatty tissue) and are attached to the inner surface of the vertebrae. This same general plan of a single "primitive meninx" is hkewise found in fishes (elasmobranchs, teleosts, etc.); the membrane here is often pigmented and follows closely the external architecture of the spinal cord. The perimeningeal space is filled by mucus in elasmobranchs, but in teleosts this is replaced by fat. For the most part there are found dorsal and ventral ligaments and two lateral ligaments.


1 lit roduct ion 7-8
The maftina in normal development 8-17
The magma in pathological ova 17-23
( ■ontiusion 23-24
Hililiography 25
Explanation of plates 2G


The next stage in the development of a more complete form of spinal covering is found in the urodele amphibia. A "primitive meninx," formed of two layers, often artificially separated from each other, replaces the simpler meninx of C5'clostomes and fishes. Of the two layers in this membrane the external is thin and free from pigment; the inner, strongly pigmented, adheres to the spinal cord. The meninx is perforated by the denticulate Ugaments.




THE IILMA.N MAliMA KETirilLE IN NORMAL A\\) IN rATllULUGlCAL
In amphibia (Anura) Sterzi found the first evidence of a "secondary meninx," corresponding to the pia-arachnoid. Surrounding this membrane, but separated from it, is the dura, thin and transparent; between the two meninges is the intradural (subdural) space. The dura lies in the peridural space. The spinal prolongations of the endohonphatic canals he in the dorsal part of the peridural space. Both the dura and the "secondarj^ menmx" continue outward along the roots of the spinal nerves and along the filum terminale. Embryologically the perimedullary mesenchyme is differentiated into these two meninges in the Anura.
DEVELOPMENT.




This arrangement of the two meninges in Anura is followed out in reptiles. The dura, thin as in the amphibia, is covered by endothehum and is vascular. The "secondary meninx" possesses laterally the denticulated hgaments and ventrally the ventral hgament. Both the peridural and intradural spaces are very small.






By Franklin P. Mall.
Likewise in birds Sterzi was able to differentiate only two meninges — the dura and the "secondary meuinx." These membranes are quite similar to those of reptiles. The " secondary meninx " has acquired three layers — an outer endothelial covering, a middle vascular layer, and an inner membrane closely adhering to the cord. This is a distinct approach to the three meninges of mammals. An intradural (subdural) space covered by endothelium can be easily made out. The development of these avian meninges concerns a differentiation of the perimedullary mesenchyme.




The arachnoid, according to Sterzi, first appears as a definite membrane in mammals (marsupials and placentals). In marsupials this arachnoid has become well differentiated and the pia mater possesses denticulated and ventral Ugaments. A transformation of the extradural portion of the denticulated ligaments unites the dura to the endorachis. In perissodactyla the differentiation of the three meninges (particularly of the arachnoid) is incomplete. The arachnoid is separated from the pia mater by a pecuUar tissue wliich contains numerous lymphatic lakes, forming the intra-arachnoid spaces. No intradural (subdural) space is apparent, due to the approximation of dura and arachnoid. The subdural space is clothed by endothelial cells; these can not be made out in the intra-arachnoid spaces. The dura is surrounded by a fatty pad.




According to Sterzi the augmentation of the intra-arachnoid (subarachnoid) space is the distinguishing characteristic of the meninges of carnivora. This increase takes place at the expense of the peridural space.


INTRODUCTION.


As Sterzi developed the knowledge of the comparative anatomy of the lower forms — of the transition from the primitive meninx of the cyclostomes to the three membranes of mammals — the possible correlation of this analogj'^ to the embryological development in mammals became apparent. He extended his observations to human beings and to human fetuses. His findings will be detailed in the following section.


Students of embryologj' are familiar with the jellj'-like substance found in the human exoccelom, which varies much in appearance in different specimens. Sometimes this substance is gelatinous, with, delicate fibers; at other times it is mixed with granules; and, in extreme cases, it forms quite a solid body. I think it was Giacomini who pointed out definiteh' that the morphological appearance of the magma determines, with considerable certainty, whether or not the contained embryo is normal or pathological. We are indebted to him for about a dozen papers on pathological embryologj^, a summary of which he published in ^lerkel-Bonnet's Ergebnisse. In this summary the following statement is made:
" In the early stages of development we can determine by the extent of the exo^tElom and its contained magma whether or not the embryo under consideration is normal. A large coelom, containing a rich magma, with its meshes sufficiently filled with a flaky precipitate to mask the embrj^o, is a certain sign of pathological development."
It is well known that the magma is least conspicuous in fresh specimens and becomes more pronounced after being hardened in alcohol or other preservative fluids. In recent years it is found that magma shows to the greatest advantage in specimens hardened in formalin; the fibrils are somewhat tougher, but the magma has usually the same appearance as in the fresh state. However, the experience of embryologists has been that the magma is more pronounced in pathological specimens, and for this reason it has been suspected that it does not exist in normal development. In fact, the illustration of the magma given bj^ Velpeau in his monograph — ■ in which he first uses the term "magma reticule" — is undoubtedly of a pathological specimen. A glance at the other plate which accompanies this handsome monograph shows clearh^ that most of the specimens he describes are decidedly pathological. During the 80 years which have elapsed since his time, embryo


Farrar'i, in a short discussion of the development of the meninges of the chick, finds in early stages three laminse about the spinal cord, "the middle one of which alone still presents the primitive features of the mesoblastic-sheath." The inner layer, close to the medullary tissue, is highly vascular; in the outer zone "the connective-tissue elements arc assuming elongated forms and crowding together with long axes parallel, giving a very close mesh with long but extremely narrow spaces, in contradistinction to the loose irregular reticulum of the pia-arachnoid." The outer lamina becomes dura mater, while the inner two zones are considered together as the embryonic pia-arachnoid. Farrar defines the pia-arachnoid as developmentally a single membrane consisting of a loose reticulum, at the outer and inner borders of which limiting membranes are formed.


logists, through comparative study, have been able to separate normal from pathological embryos with considerable precision; and in the abortion material, as collected in various laboratories, far over one-half of the specimens of the first 2 months of pregnane}^ are pathological, and in them we usuallj'^ find a highly differentiated magma. However, if normal specimens are studied with care, we find that thej^ too, contain some magma; therefore, magma must be viewed as a normal constituent of the human ovum.


===Literature On The Development Of The Mammalian Meningeal Spaces===


It has been shown by Keibel that there is marked magma within the exocoelom of monkey embryos. In specimens containing embryos 1.3 mm. and 5 mm. in length, he describes it as a flaky, reticular mass outside the amnion, and speaks of it  
The development of the meningeal spaces in mammals has not been studied extensivel}', and the literature in regard to it is quite meager. Only a very few workers have touched upon the subject except casually. Reford"*^', working in the Anatomical Laboratory of the Johns Hopkins University, studied the development of these spaces bj^ the method of injection with india ink. His work unfortunatelj' has never been published, but it has been rather extensively referred to bj' Sabin *3' in 1912 and by Cushing'^^ in 1914. Their sununaries of this work are here included. Miss Sabin thus speaks of it:
7
"In a study of the arachnoid made bj' the injection method in the Anatomical Laboratory of the Johns Hopkins University by L. L. Reford, and as j^et unpublished, it has been shown that the thinning out of the mesenchjine around the central nervous sj-stem is not haphazard, but that injections of the same stage give the same pattern, and that the form of the arachnoid space changes as the brain develops. That is to say, the arachnoid space has as definite a form as the coelom, and it never connects with the IjTnphatics."


Cushing() gives the following summary:
:"It was thought that an investigation of the cerebro-spinal spaces in the embryo would most hkely shed Ught on the subject, and some unpublished studies in this direction were undertaken in 1904 and 1905 by Lewis L. Reford in ^Mall's laboratory' in Baltimore. In living pig embryos of various stages low spiaal india-ink injections were made either into the wide central canal or into the subarachnoid space, and the embryos were subsequently cleared. It appeared from the course taken by the injection mass that the full development of the spinal arachnoid preceded that of the intracranial spaces, the impression being gained that the separation of the primitive meninx into its layers occurred later over the cerebral vertex than in the basilar portion of the chamber. Still, I never felt quit« convinced that the failure of injection of the meninges over the surface of the hemispheres in many of Reford's specimens was not due to the floating up of the brain against its envelopes by the introduction of the injection mass from below. Howe\'er this may be, it was nevertheless apparent that a venous injection of the body of the embryo was often produced, and the impression was gained that a communication existed between the basal subarachnoid spaces and the precursors of the sinusoidal veins of the cranial chamber which empty into the jugulars. If due to an artifact from a vascular ruptm-e, at all events the conununication always occurred at the same point. Reford, moreover, in agreement with Cruveilhier, Reichert, and KolUker, came to doubt the existence of the foraminal opening described by Magendie, beUeving that the opening was an artifact and that the fluid escaped by seepage through a persistent membrane."
It is regrettable that Reford's study has not been pubUshed, as it represents the only attempt to solve the problenas of the development of the cerebro-spinal space by the method of injection. As stated in subsequent sections of this communication, his apparent failure to control pressures of injection and to use only granular suspensions is unfortunate.




8 IIIMAN MACMA UfiTIClI-l': IN NDUMAL AND PATHOLOGICAL DEVELOPMENT.
In a study of the development of the blood-vessels of the human brain, (Mall) noted the ease with which an extravasation into the embryonic arachnoid spaces could be brought about by increasing the pressure in a venous injection. In a specimen of 46 mm. an arterial injection with aqueous prussian-blue resulted in a complete subarachnoid spread, due to rupture of the vessels as they perforated the nervous tissue. In general, it was found that this rule held: an arterial extravasation always took place from the perforating capillaries, while a similar venous rupture occurred in the veins themselves.




as coaguluni composed of reticular magma which had to be removed before the embryo could be seen. Undoubtedly he was dealing with normal specimens, thus showing quite conclusively that a delicate magma must be viewed as a normal constituent of the exocoelom. According to Keibel's figure 7, the magma appears to he ilenser in monkeys than is usually the case in normal human specimens. However, a very dense magma in human specimens invariably indicates, as was first demonstrated l>y Giacomini, that the ovum is pathological.
Mall made similar observations on living pig embryos from 30 to 80 mm. in length, with analogous results. But when, in these embryos, the arachnoid spaces were completely filled by an intraventricular injection of india ink, no passage of the granular injection into the veins or sinuses occurred. The ventricular injection flowed into the extraventricular spaces "through the medial opening of the fourth ventricle." From the spinal cord the ink extended for a short distance along the main trunks of the spinal nerves. In the larger embryos (above 50 mm.) the ink usually gushed from the mouth, reaching it by way of the Eustachian tube. Using, in the pig embryo, the heart as the mechanism for injecting the ink, extravasation from the cerebral vessels in the arachnoid spaces occurred.




The best account of magma reticule is given bj- Retzius, who brought the subject up to 1890, and left it with the conclusion that magma reticule is a normal constituent of the human ovum. The statements of the earlier embryologists, from the time of Haller, are mainly of historical interest; but these investigators were at times inclined to view magma as a "middle embryonic layer" of the ovum, and, again, they believed it to represent the allantois of lower animals. Retzius reinvestigated the subject, taking into consideration normal as well as pathological embryos, and his conclusion is that magma is present in both kinds. His own words are as follows:
In one human specimen of 90 mm., Mall found both the arachnoid spaces and the cerebral ventricles filled with india ink after an arterial injection of that suspension. He states: "The injection passes through the medial opening into the fourth ventricle (Magendie), and apparently the ventricles are injected through this opening from the arachnoid."
" Bei (ler Oeffnung des Chorionsackes der Eier des ersten und zweiten Monats sah ich, wie in der Einlcitung crwalint wuitlo, und dies oft, am bcsten nach kurzer Erhartung in Ueberosmivimsiiure (von 3.1 Proc.) odor in Miillcr.sclRT Lusunfi ip'wcihiilich nach doiipeltor ^'ol•dunIunlg) , in dcni scliloiniigcn Irihalt, wclclicr zwisi-hcii dcni Chorion unil dcm Amnion, also im sulK'horionischfn Kaume vorhandcn war, diiiuioro odor dickere Fiidcn und J^triingc, (He mehr oder wcnigor dicht von der aus.*<orcn Fliiche des Amnion zur inncren Fliichc des Chorion hinul)orIicfen, um sicli dort mit ihrcn Enden an den beiden Hautcn zu licfcstigcn, indcm sie sich oft an ihnen vcrbreitcrtcn und in ihre U-kleidende Schicht iibcrgingen. Diese Faden und Striingo, wolehc im frischen Priiparate kaum sichtbar waren, traten nach der Behandlung mit den erwahnten Fliissigkciten deutUch hcrvor. In der Fig. 15 der Taf. X^'III habe ich ein seiches Ei abgebildet. Das stark zottige Chorion (ch) ist geoffnet, und man sieht im subchorionischen Raume den Amnion.sack (a) an weisslichcn Strangen (m) aufgehiingt liegen.
To His'25) and to Kolliker belongs the credit of first having established on a firm basis the development of all the meninges in man from mesenchyme. This perimedullay' layer of mesenchyme Salvi'^o^ called the "primitive meninx" — a term now used extensively in comparative anatomy. The primitive meninx divides into two layers, the outer forming the dura and the inner the pia-arachnoid. Sterzi(53), working on the development of the human spinal meninges, advanced a view similar to that of KoUiker. The perimedullary mesenchjTne (the "primitive meninx") divides into two portions, one hugging the inner surfaces of the vertebra? and the other adhering to the cord. This inner layer of the perimedullary mesenchyme, according to Sterzi, should properly be termed the "primitive meninx," as it divides subsequently into dura and the "secondary meninx," which in turn forms both arachnoid and pia. The denticulate hgaments develop in the "primitive meninx." The dura and arachnoid in human embryos are modeled up to a certain point on the cord; then, with the augmentation of the subarachnoid space, they follow the outline of the vertebral canal.




" Die Strange, welche im unerhiirtotcnZustandecinc scliIcimigfasorigoConsistcnzhaben,sichaber ohne Schwierigkeit Stiickweise au.ssclineiden iiussen und dann in Mikroskope eine ileutlich f;iserige !>truc(ur darbieteii, zeigeu nach der JCrhiirtuug einen ausgcjjriigt fibrillaren Habitus. In einer homogen, structurlosen (iruud.substanz tretcn Zlige echtcr bindcgcwebiger Fibrillen hcrvor, welche oft eine Hauptrichtung ein.schlagen, also ziemlich parallel verlaufen. Jedoch konimen auch viele sich kreuzende Fasern vor. Hier und da bemerkt man dickere Biindel verschiedenen Calibers welche aus dicht gedriingten Fibrillen besteiieii. Es sind also fibrillar-bindegewel)igo Balken, welche (lurch cine homogene, zahlreiche einzelne Fibrillen enthaltende Intercellularsubstanz ziehen. Zwischen den Balken und Fibrillenzugeii sieht man recht zahlreiche Zellen, welche theils und am moisten rutuUich oder oval, theils auch s])in(lelf()rmig sind und in ihrem oft reichlichten Proto]ilasma gr().s.serc Klanzendc Korner enthalteii. Diese Zellen liegen in der (Irundsubstanz ohne Ix-soiidere anordnung zerstreut, bilden al.so keine Scheiden o. d. um die Filjiillenbiindel.
His'25) has given information regarding the development of the meninges, with particular reference to the formation of the subarachnoid space. He affirms the mesenchymal origin of all of the cerebro-spinal membranes. His describes the first differentiation of mesenchyme to form the meninges as consisting of two zones of condensation, the outer being closely associated with the developing perichondrium of the vertebral column and the inner facing upon the cord. Between these two zones of condensation the subarachnoid space develops, posterior and anterior spaces first appearing, with later fusion laterally. These appearances were met with in chicks of 10 to 12 days' incubation. Quite soon after this process of spacedevelopment a separation occurs which gives rise to a complete subarachnoid space. Later the splitting-off of dura from the vertebral periosteum takes place.




"Man hat es hier ofTenbar mit einem iiurcifcn B i itdcgcwcbe zu ihun, inncm cmbryoiuilcn mucosen liindcgnvebf, irclchcg imlesmn in drr Kitlwickluiuj zum jlbrilUircn liindcycH'cbc schon tccit vorgcschritcn ist."
==III. Methods Of Investigation==
THE MAGMA IN NORMAL DEVELOPMENT.




\\v have now in the litcTaturc a detailed description of a number of young litunan ova, smd, according to their clinical histories, some of them, at least, must be normal. The da.ssic specimen is that described by Peters, which came from a woman who had committed suicide. The specimen was hardened in situ in an approv('<l manner, and was worked up and de.'^cribed under the best ])ossible conditions. In it the co'loni is filled with a gelatinous substance, through which radiate
In the study of any problem dealing with the development of fluid-spaces within the body, the method of investigation must of necessity be such as to offer exceptional opportunities for control. In the present work several well-known and generally accepted anatomical procedures were naturally suggested, such as injection of the spaces about the central nervous system, reconstruction from serial sections, or merely study of the various stages by means of serial sections.




It was ascertained early in the investigation that by injection and serial sections without reconstruction the necessary stages in the process of meningeal differentiation could be estabUshed. In regard actually to the physiological aspects of the problem more reUance was placed on the results of injection than on any histological differentiation, for, as explained above, considerations of the pathway and of the flow of the cerebro-spinal fluid were deemed most important. No method of injection, however, holds out much promise in such a problem unless it can be applied, under conditions approximating the normal, wnthin the spaces about the nervous system. The greatest objection to reliance upon injections in this problem is in relation to pressures. From the very nature of the case it wall be reahzed that any ordinary injection into the embryonic central canal or perispinal space must result in an extraordinary increase in the normal tension of the fluid. This objection applies to any method employed, whether that of a simple syringe and needle, the glass tube and bulb devised by Knower, or a glass capUlary-tube contrivance.


HTMAX MAGMA RfiTICULfi IN NORMAL AND PATHOLOGICAL DE\-ELOPMENT. 9
delicate bands of fibrils, among which appear scattered nuclei. Near the embryo there is a small si)ace, the interi)retation of which was verj^ difficult at the time the specimen was described.


The erroneous conclusions drawTi by investigators from the emploj^ment of excessive pressures of injection are nowhere more strikingly illustrated than in studies of the circulation of the cerebro-spinal fluid. Many such examples were recently brought forward in a critical review ''5' published in connection with a study of the fluid. In the embryo, with structures and membranes still of very little tensile strength, the consequences of a disregard for the pressures of injection are even more disastrous.


Since Peters studied this specimen, the sections have been carefully reworked and discussed in a critical way by Grosser, who gives a new interpretation in two figures and states that the cavity of the ovum contains reticular magma which is partly made up of heavier strands of tissue accompanied by nuclei. In the neighborhood of the embryo there are two large spaces, hned with cells, which appear to be the primitive body-cavities. In his work on the comparative development of the embryonic membranes, Grosser describes this space in great detail and also gives us two new illustrations of the embryo in his plates 3 and 4. According to this authority these two body-cavities communicate b.y means of a slit-like canal just behind the umbihcal vesicle (Grosser's figure 31, plate 4). This interpretation of the Peters specimen shows that the cavit}' of the ovum is first filled with a free mass of reticular magma, after which the coelom begins to form near the body of the embryo. As this cavity expands subsequently, it probably first destroys the more delicate strands of magma, leaving the heavier ones; thus in a short time the cavity of the ovum is lined b}- the endotheUum of the coelom, which also must cover the stronger bands of magma radiating as trabecula> throughout this cavity (Grosser, pp. 78 and 79).


A second criterion for the study of fluid-pathways in the body is necessarily the type of injection mass. Not only should attention be paid to the pressures involved, but the peculiarities of the particular body-fluid concerned must be considered. Adopting for this work on the embryo the same standards followed in the previous investigation on the adult, true solutions were used in place of the customary granular suspensions. Emulsions and viscous solutions were not emploj'ed because of their obvious disadvantages in studying the passage through membranes. India ink and process black (in which carbon granules are the particulate matter) were also used, but only for comparison with the standard true solution, as the likelihood of the insoluble granules being phagocyted wathin the period of experimentation or of being caught mechanicalh'^ in tissue meshes appeared a priori to be too great.


Keibel explains the formation of the human coelom as follows:
■ It is, however, not quite clear how the cavity traversed by scattered strands of mesoblast and lying between the yolk-sac and the chorion in the Peters ovum is to be interpreted. It may be supposed to represent the extraembryonic coelom; but it may also be imagined that it has arisen from an extensive loosening up of ths tissus, and not by a splitting of the meso Jarm, and that the triangular space between the caudal extremity of the embryo, which is lined with flat cells having an epithelial arrangement, is the first jmrnordium of the coelom."
A condition similar to that found in the Peters specimen has been observed by Lewis in the Herzog specimen, which is of about the same stage of development. Lewis says (see his paper, p. 300) that there are occasional clefts in the mesoderm of the chorion of the Herzog embryo, but that they are of doubtful significance. His reconstruction shows a strand of mesoderm, more pronounced than m the Peters ovum, extending from the yolk-sac to the chorion and circumscribing a space on the ventral side of the embryo


In any study of fluid-pathways in the body, not only must the injection fluid be a true solution, but it must also be one which is not attracted to particular cells (as with many stains). Likewise, colloid stains (such as the benzidene group) could not be employed, because of the fact that certain cells (macrophages, as described by Evans’s phagocyte the small colloidal particles. In addition, the true solution must be readily precipitated as an insoluble salt, capable of remainhig unchanged in histological technique. After trying many salts in long-continued injections into the adult cerebro-spmal spaces, it was found that solutions of potassium ferrocyanide and iron-ammonium citrate in equal parts were admirably adaj^ted to the purposes of the experiment. By the addition of a mineral acid (preferably hydrochloric) ferric ferrocyanide could be precipitated. This prussian-blue is insoluble in the routine technique and is readily identified in sections. After mounting in damar or balsam the blue granules can be observed unchanged for several months, but after a year there is some deterioration in the specimen, due to a conversion of the blue into indefinite greens.


.




Eternod has written several papers in which he describes the formation of the exocoelom and the fate of the magma reticule. He sa3-s that it first fills the entire space between the primordium of the embryo
Text-figure 1. — .Schematic sketch of mechanism used (or replacing ventricular and spinal fluid of an embno with a foreien solution or suspension. The system is here shown in balance, the difference in fluid-level in reservoirs and needles representing the hydrostatic pressure necessary' to overcome the capillary resistance of tubes and needles. The stands holding the injecting needles may be moved about without altering the balance of the system. As one reserv'oir is raised, the other is lowered in a corresponding degree.




and the chorionic wall. Later, larger spaces appear within the substance of the magma, leaving denser strands of magma fibrils to support the embrj-o wdthin the gradually expanding chorion. In general this coincides with the opinions just cited.
In regard to these two major factors in the employment of injections (pressure and true solution) it was found necessary to devise a method of experimentation which would satisfy the requirements of the problem. Solutions of the ferrocyanide and of the citrate were non-toxic within the central nervous sy.stem and afforded an excellent histological means for following the fluid-pathways. It was hoped at first that a simple "replacement" type of injection might be employed, as in the adult animals. In this procedure a given amount of fluid was withdrawn from the subarachnoid spaces and immediately replaced by an equal quantity of the injection fluid. The method was successfully tried on fetal cats of consideralile size, but was impracticable on small embryos. After such a replacement the animals were allowed to Uve for varying periods of time (up to 3 hours) and then killed.




' The relation of the exocoelom to the magma Ls strikinglj- shown by Waterston in a section of a small embryo in situ. The space between the embryo and the chorion is fiUed with a dense mass of fibrils, into which the exocoelom is burrowing. Waterston's figure 1 shows the relation of this cavity to the magma, and only near the embryo is the exocoelom lined with a layer of cells. "When this figure is compared
It was soon ascertained that the essential circulation of the cerebro-spinal fluid was established in pig cml^ryos of less than 30 mm. in crown-rump measurement. Hence the ordinary method of replacement had to be discarded for some more delicate system. With the realization that a simultaneous withdrawal and introduction in a living embryo would be far preferable to a two-stage procedure, the extremely simple apparatus pictured in text-figures 1 and 2 was employed. This device consists of two glass tubes of uniform and like bores, suspended from above by a string running over a pulley. To the tapering lower ends of these reservoirs are attached rubber tubes which connect the reservoirs to two needles. These needles are held at the same level by two metal brackets which can be moved at will on a level glass plate.






10
Text-figure 2. — Diagrammatic representation of the method of rcplacinR the cerebro-spinal fluid in a living embryo. The spinal needle is inserted into the central canal of the spinal cord, while the cerebral needle is introduced into one of the cerebral ventricles. The canal of the spinal cord and the cerebral ventricles are represented by the interrupted lines. The foreign fluids are introduced by the spinal needle and withdrawn by the cranial.






HIM.VN M.\GM.\ Hfn-ICl'Lf; IN NOKM.Xl. .\ND PATUOLOGIC.\L DEVELOPMENT.
The apparatus is employed as follows : Both tubular reservoirs are filled up to the point where the fluid is just ready to fall from the needle in a drop. This point is easily obtained by fiUing the reservoirs slightly in excess and allowing this excess fluid to run off from the needle. With the system thus in balance the needles he in the same horizontal plane and can be moved without altering the balance of the solutions. The injection is made by inserting one needle into the central canal of the spinal cord and the other into one of the lateral ventricles; then as the reservoir connecting with the spinal needle is raised the other is lowered, so that an amount of fluid equal to that introduced into the spinal canal is withdrawn from the cerebral ventricles. In this way the whole contents of the cerebral ventricles and central canal of the spinal cord can be slowlj' withdrawn without increasing the pressure in the central nervous sj^stem. The initial pressure necessary to secure this flow is only that required to overcome the capillary resistance of the medullarj'-canal system. In practically all cases this can be accomplished by using a positive pressure of less than 60 mm. of water (associated with a negative pressure of the same degree) .






In the present study the above procedure was the routine method of injection employed. Pig embryos, brought from the abattoir, contained in the uterus, were found to be wholly satisfactory material. If not permitted to cool excessively in transit the embryos lived for at least two hours in a 38° incubator. On being received at the laboratory a section of the uterine wall contaming the placenta was excised, with the embryo left connected by the umbilical structures. As soon as the technical preparations for injection were completed the amnion was opened and the embryo placed upon a padded block at the proper level. The first needle was then inserted into the easily discernible central canal of the spinal cord and the second into the left cerebral ventricle or into the mesencephalic ventricle. By elevation of the reservoir connected with the first needle the cerebro-spmal fluid was replaced by the injection solution. As soon as the replacement was complete the needles were withdrawal and the embryo and its uterine portion replaced in the incubator. The heart of the embryo could be easily obser\^ed in the smaller forms and served as the index of a continued circulation.




with C.rosscr's figure of the Peters ovum, it becomes clear that the two si)aces in the latter are in reality the beginning of the exocoelom.
The incubation of the embryos was continued for varying periods of time, but it was soon ascertained that a period of over 30 minutes generally resulted in a complete spread of the injection solution. For comparison the period of incubation was lengthened and shortened, but the best results were usually obtained with a 45-minute incubation after the replacement.




Tlie .>;tuclies referred to above indicate that the space near the embryo i.s the primitive exoccvlom and that the remainder of the so-called cavity of the chorion is .•dimply the young normal ovum filled with delicate fibrils which communicate freely with the fibrils of the chorionic membrane. We have in our collection a young normal specimen, No. 763, containing an embryo anlage 0.2 mm. in length, which in general confirms the observations in the Peters ovum. A hst of the normal specimens in our collection discussed in this paper is given in table 1. Table 1. — List of normal embryos.
Injections of the necessary true solutions were made, in the routine experiment, with a 1 per cent concentration of potassium ferrocyanide and iron-ammonium citrate in distilled water. By a 1 per cent solution is meant a salt concentration of this amount (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm.; water, 100 c.c). The resultant true solution should be practically isotonic with the body-fluids. In this way any injurious consequences due to hypertonic or hypotonic solutions were apparently overcome. The factors of osmosis and diffusion also had to be considered in this connection.




Other concentrations of the so-called "ferrocyanide mixture" were used, but only for the sake of comparison or for the purpose of investigating some particular phase of the problem. The results obtained by the use of these concentrations were not relied upon as affording standards for the normal pathway of the cerebro-spinal fluid.




In addition to the replacement type of injection, many observations were carried out on pig embryos, with a simple syringe-injection of the ferrocyanide solution into the central canal of the spinal cord or into the cerebral ventricles. It proved to be a very simple matter to regulate the pressures by this method, and three arbitrary standards (mild, moderate, and strong) were found to be of value in a comparison of the extent of the spread obtained by replacement and by injection.


Cat. No.


The prussian-blue reaction (formation of ferric ferrocyanide) was obtained in these experiments by fixing the whole embryo in an agent containing hydrochloric acid. For histological study the best results were obtained by immersing the specimen from 1 to 10 minutes in a 10 per cent formaldehyde solution containing 1 per cent hydrochloric acid. After this primary procedure, during which the ferrocyanide was precipitated, the embryo was transferred to Bouin's fluid (saturated aqueous picric acid, 75; formaldehyde 40 per cent, 20; glacial acetic acid, 5). The specimens were allowed to fix over night and were then dehydrated in graded alcohols. From 30 per cent alcohol, use was made of 4 per cent changes up to 60 per cent; and from this point to absolute the changes were by 2 per cent gradations.




Length
In addition to the technique outlined above, Carnoy's solution and 10 per cent formol were employed. The Carnoy fluid, containing acid (absolute alcohol, 60; chloroform, 30; glacial acetic acid, 10; hydrochloric acid, 1) proved to be of particular service in the study of specimens cleared by the Spalteholz method; histologically, however, it has not been as valuable as Bouin's fluid.
of embryo




Besides the ferrocyanide solution, two other injection masses were constantly employed. Solutions of silver nitrate in concentrations of 0.5 per cent were injected into the central canal of the spinal cord and into the cerebral ventricles. This method, with reduction of the silver salt in the sunlight, gives very pleasing preparations. It is, however, subject to obvious limitations. The intraspinous toxicity of the silver, together with its action as a precipitant of albuminous substances, renders its use unsatisfactory in replacement experiments. Furthermore, it reacts apparently with any protein tissue, irrespective of the true function of that tissue (as, for example, its coagulation of the lining ependyma of the ventricles).




Dimensions of chorion.
India ink, the other substance employed, is of extreme value in anatomical studies. Because of the suspension of carbon granules it possesses the disadvantages already commented upon for the study of any true pathway of fluid. It has been of service, however, in the present work in showing marked differences in spread from that of true solutions and in furnishing information in regard to fluid passage through a membrane.




This investigation has been carried out on the basic idea of correlating the physiological spread of the embryonic cerebro-spinal fluid with the gradual transformation of the perimedullary mesenchyme into the three fully formed meninges. This has necessitated a histological study of the embryo. Pigs for the most part were the animals used, but the findings have all been verified by a study of the same regions in the human embryos in possession of the [[Carnegie Collection|Carnegie Institution of Washington]]. In addition, certain structural characters have likewise been identified in sections of chick, rabbit, and cat embryos.


Menstrual age in days.


It was early apparent that the material to be of value must be free from any great shrinkage about the central nervous system. Comparative freedom from this artifact was obtained by fixing the embryo alive in Bouin's fluid and dehydrating by 2 and 4 per cent gradations of alcohol. The material was chiefly cut in paraffin after being embedded by means of xjdol.




Condition of magma.
The methods of investigation outhned in the foregoing paragraphs have been followed throughout the major portion of the work. In many minor instances other procedures not commented upon have been employed ; these wiU be detailed in appropriate subdivisions of this paper.




==IV. Injections And Replacements In The Cerebro-Spinal System==


Cat.


Results Of Replacements In The Ventricular System Of True Solutions.


No.


The results of experiments carried out on embryo pigs by the technical procedures outlined in the previous section will be detailed here. The study was made on this animal because of the facility with which it could be obtained living and in good condition and also because it exhibits the characteristic meningeal anatomy of all mammals. The chick could not be used in this investigation on account of the dissimilarity between the avian and the mammalian menmges.




Length
The chief problem concerned here was the actual physiological extent of the cerebro-spinal spaces. This apparently could be ascertained by the replacement of cerebro-spinal fluid by the ferrocyanide mass. But there was also to be considered the passage of fluid from the ventricles out into the periaxial* spaces, corresponding exactly to a similar passage in the adult.
of embryo.




If into the central canal of the spinal cord of a hving pig embryo of 9 mm., crown to rump measurement, an injection of the ferrocyanide solution be made under very mild syringe-pressure, the ventricles can be fairly well filled without rupture of any element. Incubation of this experimental embryo with its circulation continuing almost unabated for an hour should cause a further spread of the fluid throughout the normal canals. If at the end of this time the whole embryo is fixed in an acid medium the ferrocyanide will be precipitated in situ.


Dimensions of chorion.


Such a specimen, subsequently cleared by the Spalteholz method, is represented in figure l.f In this drawing the spread of the injection solution is clearly shown. Running upward from the point of introduction, wholly within the central canal of the spinal cord, it reaches the bulbar region and extends outward into the large fourth ventricle, appearing as a dense collection of the prussian-blue. Cephalad from this region it spreads in diminishing intensity until it is finally lacking in the diencephalon.




Menstrual age in days.
The injected solution, then, in spite of the unavoidable increase in the normal intramedullary pressure, is contained only within the medullary-canal system (central canal of spinal cord and cerebral ventricles). There is no evidence of any spread outwards, either from the third or fourth ventricle.




In the next stage of meningeal development the replacement method can be used, as the embryo is no longer too small for its employment. In figure 2 is represented an embryo of 13 mm., in which the circulation continued for 90 minutes after the replacement. The same general picture shown in figure 1 results. The whole medullary-canal system is filled with the precipitated prussian-blue, which is densest in the region of the fourth ventricle. The roof of the ventricle, however, shows a striking difference from that of the ventricle in the embryo of 9 mm. Just posterior to the cerebellar lip is a regular oval, which is covered from within by a dense collection of prussian-blue granules, causing it to stand out in clear contrast to the thinner and more evenly distributed blue lining of the remainder of the roof. This oval area is comparatively large and comprises a portion of the superior or anterior half of the ventricular roof. This area, differentiated from the remainder of the rhombencephalic roof, is clearly shown in figure 2, a drawing of a cleared specimen of this stage.


1 Condition of magma.


* Throughout this paper the term "periaxial " has been used in the seii90 of "around the central nervous system " or "around the ccrebro-spinal axis."
* Throughout this work the reference "figure " 1, etc., refers to plate illustrations; the word "text-figure" refers to the illustrations inserted in the text.






With the exception of this strikingly dense area in the rhombic roof, the injection spread in an embryo of 13 mm., subjected to replacement of the cerebro-spinal fluid by the ferrocyanide, differs in no way from that in the embryo of 9 mm. Careful inspection of figure 2 is convincing that the spread still remains within the medullary canals, with no extension of the fluid into the spaces outside of the cerebrospinal axis. It seems justifiable, then, to speak of the cerebro-spinal spaces at this stage of development as being only mtramedullary in type, with no indication as yet of a meningeal fluid cushion (corresponding to the adult subaraclmoid space).




mm.
With the use of larger embryos, however, for the medullary replacement with ferrocyanide and citrate, the picture gradually changes. The first indication of a more advanced stage of development is obtained in embryos whose length exceeds 14 mm. Figure 3, of a pig embryo of 14.5 mm., is included here as representing this further extension of the injection fluid. The cerebro-spinal fluid of this specimen was replaced, by the compensating mechanism, by a solution of potassium ferrocyanide and iron-ammonium citrate. The embryo was then kept alive (as judged by the heart-beat) for a period of one hour. At the end of this time it was fixed in an acid medium and subsequently cleared in oil of wintergreen after careful dehydration.




The essential differences between an embryo of this stage and one of the stage represented in figure 2 concerns the spread of the injection fluid from the roof of the fourth ventricle. Both specimens show a complete filling of the intramedullary system (cerebral ventricles and central canal of the spmal cord) with the precipitated prussian-blue granules. The specimen of 13 mm. (fig. 2) is characterized bj' a dense oval collection of the prussian-blue on the upper and inner surface of the rhombic roof. In the specimen shown in figure 3, in contradistinction to this localized aggregation of granular matter, there is a deUcate extension of the injection fluid caudalwards from the roof of the fourth ventricle. This fusiform projection is here readily made out, lying beneath the skin over the ventricular roof and separated quite distinctly from the easily discernible line of the roof. This outward extension of the fluid has a fairly wide and deep origin from the upper portion of the roof, but tapers caudally to a sharp point with considerable rapidity.


mm.


At the stage of 14 mm. the roof of the fourth ventricle shows the small depression which marks the formation of the chorioid plexuses. With this depression occurrmg transversely the relation of the external surface of the embryo to the ventricular roof necessarily alters somewhat in this region. The chorioidal depression of the roof graduallj' becomes separated from the skin; and it is into this area between the skin and the ventricular ependyma that the first spread from the cerebral ventricles occurs. At this stage, illustrated in figure 3, the injection is intramedullary in type, with but sUght extension into the pericerebral tissues.




The pericerebral spread may be made out in nearly all replacements in embryos of 14 mm., but in a few cases the injection has remained intramedullary in type. In embryos of 16 mm. the spread into the pericerebral tissues is invariably found. Often, with this extension of the replacement solution outside the ventricles, the oval area noted in the stage of 13 mm. persists. (This phenomenon is especially well shown in a simple injection of silver nitrate, illustrated in figure 11.)
The next stage of importance in the development of the cerebro-spinal spaces is represented in figure 4, a drawing of a pig embryo of 18 mm. in which a typical intramedullary replacement of the cerebro-spinal fluid with a solution of potassium ferrocyanide and iron-ammonium citrate had been made. Here, with the exception of the region of the roof of the fourth ventricle, the replaced fluid is contained solely within the central canal of the spinal cord and within the cerebral ventricles. The roof region, however, exhibits a new phenomenon, which distinguishes it from the stage shown in figure 3. The chorioid plexus invagination has become strongly developed, dividing the roof into two parts. These roof divisions have been termed superior and inferior, the former lying anteriorly and orally from the chorioid fold. The general surface outline is but little changed, due to the mesenchyme filling up the area between roof and skin. From two areas in the entire roof of the fourth ventricle the foreign fluid has escaped into the pericerebral tissue. These points of fluid passage he in the two divisions of the ventricular roof. The superior area of escape corresponds to the oval outlined by the prussian-blue in figure 2 and to the point of emergence of fluid shown in figure 3. The lower area of fluid escape is in the inferior half of the ventricular roof, where the ependymal lining and its supporting tissue are developing into a well-marked dorsal distension. This area corresponds to Blake's'3' caudal protrusion, though, as Heuser'23) has pointed out, the shape of the structure in the pig in no way resembles the "finger of a glove."
The extraventricular spread of the injection fluid in this specimen is considerably greater than in the pig embryo of 14 mm. (fig. 3). On the whole, however, the distribution of the replaced fluid is not extensive as compared with the adult relationship, where the central nervous axis is entirely surrounded by its subarachnoid cushion of cerebro-spinal fluid. From the superior area of fluid passage the replaced solution (as shown by the resultant precipitation of the prussian-blue) has passed both superiorly and inferiorly. In the median line, and extending laterally but slightly, a projection of the blue may be seen occu])ying a large portion of the extraventricular area formed from the chorioidal invagination. This area of fluid passage occupies at this stage about one-third of the total transverse diameter of the ventricular roof. From it the blue tapers caudally, diminishing in all directions. Above, the precipitate may be made out extending superiorly over the cerebellar lip. Its extension into the pericerebellar tissue is not marked; here again it tapers from the area of fluid passage, its midline prolongation stretching farthest anteriorly. This relationship is easily made out in figure 4, a frank lateral view of such an experimental rei)lacement. The granules which result from the introduced ferrocyanide solution are found only in the central canal of the spinal cord and not in any perispinal arrangement.




In the pig embryo of 18 mm., shown in figure 4, the replaced solution has been carried somewhat farther than in the embryo of 14 mm. (fig. 3). The chief point of differentiation lies in the fact that in the latter stages two areas have apparently become permeable to the intraventricular fluid, so that a larger periaxial spread has resulted. Then, too, the extension of the ferrocyanide solution from the superior area is considerablj^ greater, overlapping the cerebellar Up and filling in some degree the pericerebral tissue in the chorioidal invagination.




With a definite periaxial spread established for the cerebro-spinal fluid in pig embryos of 14 to 18 mm., it seemed not unreasonable to expect a gradual increase in the extent of the future subarachnoid distribution in more advanced stages. The earliest extension of the fluid into the peribulbar tissues occurred with the inception of the infolding of the ventricular roof to form the chorioid plexuses of the fourth ventricle. Its further extension, particularly its passages through a second area, occurred with the greater development of the chorioidal invagination (i. e., 18 mm. stage). A still more extensive pericerebral flow of the ferrocyanide and citrate is illustrated in figure 5. Here the cerebro-spinal fluid in a hving pig embryo of 19 mm. was replaced by the ferrocyanide solution. The embryo was kept alive for about an hour after the replacement and was then fixed in toto in an acid fixing medium, which caused the precipitation of the prussian-blue. On clearing subsequently by the Spalteholz method the spread of the solution was found to be somewhat more extensive than in the stage of 18 mm. (cf. figs. 4 and 5). In figure 5 the whole periaxial area over the roof of the fourth ventricle is shown to be completely filled by a dense aggregation of the prussian-blue granules. The separation of the two areas of fluid passage can not be made out in such a specimen. This dense periaxial extension ahnost completely covers the cerebellar Up, not onlj^ in the medial region but laterally to the limit of the ventricular roof. The injection precipitate lies directly beneath the skin in this area, but more posteriorly its separation from the skin becomes more marked. Tracing this dense periaxial injection posteriorly, it is seen (fig. 5) to end somewhat abruptly in the region of the cephaUc flexure. The Une of termination of the denser mass, to the ventral surface of the medulla, tapers somewhat anteriorly. This extraventricular spread is medial to the otic vesicle, but extends peripheraUy along the caudal cerebral nerves, reaching outward as far as the peripheral gangUa. The periaxial spread also closely covers the ventral surface of the medulla and extends in this plane around the pontine flexure for a short distance upwards along the basilar surface of the mid-brain.




Examined from its dorsal aspects, the superior portion of the spinal cord is found to be covered (in a perispinal relation) by a fine deposit of the prussian-blue. This is shown in figure 5. Caudally from the higher cervical region there is no exidence indicating a further spread in the perispinal tissues. Such a spread from above downward is wholly at variance wdth Reford's^^' conception of a development of the spinal meningeal spaces before the cerebral. The complete filling here of the central canal of the spinal cord and of the cerebral ventricles with the replaced fluid, with no evidence of a periaxial spread except in the region of the fourth ventricle, indicates that in the pig cinbrj'o the adult human relationship between the cerebral ventricles and the subarachnoid spaces endures. There is apparently in this embryo no evidence of the foramina of Bichat and of Mierzejewsky, a findmg in accord with the observations of Dandy and Blackfan'i*''.




In the slightly larger embryos the further extension of the embryonic extraventricular spaces progresses rapidly. Figure 6 represents such an extension in a pig embrj-o of 21 mm., in which the normal cerebro-spinal fluid was rei)laced by a dilute solution of potassium ferrocyanide and iron-ammonium citrate. In this specimen the central canal of the spinal cord and the cerebral ventricles are completely filled with the precipitated prussian-blue. But m addition there is almost a total filling of the periaxial spaces. Viewed laterally the densest aggregation of the blue granules is again in the region of the roof of the fourth ventricle. As in the embryo of 19 mm. (fig. 5), the whole extraventricular tissue posterior to this ventricular roof is filled with the granules precipitated from the foreign solution. The spread from this region is similar to that in the previous specimen, except in its far greater extent. The granules may be traced caudalwards in the perispinal spaces to the point of injection. The arrangement of the precipitated material, both withm the central canal of the spinal cord and surrounding it in the perispinal relationship, is well shown in figure 7, a frank dorsal view of the same specimen represented m figure 6. The greater density of the perispinal granules in the upper region of the cord, as contrasted with the granules in the thoracic region, is probably of importance in indicating the direction of the flow from above dowaiwards. The increased amount of the injection fluid in the region about the pomt of insertion of the spinal needle.is in all likelihood due to a local spread from the needle, such as frequently occurs in a very limited area. The phenomenon may, however, be due to an actual increase in the size of the potential perispinal space, though observations upon other embiyos of the same stage of development argue against this view. The segmental outlming of the caudal portion of the perispinal space is to be noted in this figure.


mm.


The cephalic regions in the specimen of 21 mm. show a quite extensive spread (fig. 6), and there is the same general distribution of the granules about the medulla, as in the specimen shown in figure 5. The rhombencephalon is completely surrounded by the blue, the ventral sheet inclosing it tightly. Laterally the prussianblue is shown in a dense mass, in intimate relation to the cranial nerves as they join the brain-stem. The cerebellum is practically completely covered by the precipitate; from the ventral portion of the pericerebellar granules the replaced solution (as evidenced by the granules of prussian-blue) spreads forward and surrounds a portion of the mid-brain. Only the ventral surface of the posterior half of the mid-brain is circumscribed by the granules; anteriorly it is wholly surrounded by the i)eriaxial injection; more anteriorly tlic extension is iimited to the mesial structures, leaving unsurrounded the cerebral hemispheres, althougli creejnng between the hemispheres and the mid-brain.




mm.
The pecuhar avoidance by the replacement fluid of the extreme dorsal half of the mid-brain is also to be made out in the dorsal view of the specimen (fig. 7).




The two lateral extensions from the ventral sheet of the injection granules approach on either side this mesencephalic eminence. The peculiar appearance of the injection spread caused by the chorioidal invagination of the roof of the fourth ventricle is also here illustrated.




In this specimen, then, of a pig embryo 21 mm. the periaxial spread is almost complete, the only areas not entirely surrounded being the aiiterior mesencephalon and the cerebral hemispheres. In an embryo but a few milhmeters larger this periaxial exten.sion of the solution is complete. The mesencephalon first becomes entireW covered by the jjrussian-blue precipitate, with later extension over the hemispheres. This complete periaxial injection occurs usually in replacements in embryos varying in length from 24 to 28 mm.




A specimen exhibiting a complete extension of the replaced solution around the central nervous system is shown in figure 8. This specimen was prepared by replacing the cerebro-spinal fluid in a Uving embryo of 26 mm. and then keeping the embryo alive for an hour. After fixation in an acid medium, dehydration, and clearing, the uijection was found to occupy the whole medullarj-canal system and also to surround completely the cerebro-spinal axis, as shown in the lateral view. The striking features of this stage are similar to those observed in the younger specimens — the dense accumulation of granular material in the region of the roof of the fourth ventricle, the surrounding of the central portion of the caudal cranial nerves, and the thin pericerebral covering by the replacement mass. In addition the specimen exhibits in the thoracic region an extension of the granular material laterally along each spinal nerve. An observation of this peculiarity reveals the prussian-blue extending outwards only as far as the ganglia on the posterior roots.




The relationships, then, observed in an embryo pig of 26 mm. are those which exist in the adult; the cerebro-spinal axis contains cerebro-spinal fluid within its cerebral ventricles and within the central canal of the spinal cord, while in turn it is cornpletely surrounded by cerebro-spinal fluid within the subarachnoid space. Communication between the ventricles or intra-medullary sj'stem and the perispinal spaces occurs only in the region of the fourth ventricle. Here again the adult human relationship holds. The evidence, therefore, from a study of the fluid spread in a replacement experiment with the use of true solutions, indicates that in pig embryos of about 26 ram. an adult distribution of cerebro-spmal fluid occurs.


763


===The Results of Injections of True Solutions===


0.2
In the preceding section there have been detailed the results of experiments on living pig embrj-os in which the cerebro-spinal fluid of both the central canal of the spmal cord and the cerebral ventricles has been replaced by a dilute solution of potassium ferrocyanide and iron-ammonium citrate. After the replacement, carried out so as to avoid any increase in the normal tension, the embryos were incubated for varj^ing periods of time so that the normal current of the fluid might cause an extension of the loreign solution. In the experiments which will be recorded in this section the same true solution was injected from an ordinary syringe and the salts immediately precipitated as prussian-blue. The purpose of these observations was solely to ascertain the effect of injections at pressures above the normal tension, so that the conclusions drawn from the replacement method might be more fully substantiated.




4X 2.2
It was soon ascertained that the pressures caused by injections with a simple syringe could be fairly well controlled and that several degrees of tension might be employed. Thus it was found to be simple and serviceable to designate the injections as those made with mild, moderate, or strong syringe-pressure. Most of these injections were made into the central canal of the spinal cord, but occasionally into the perispinal spaces or cerebral ventricles. Injections under equivalent pressures in the central canal of the spinal cord or into the cerebral ventricles always gave corresponding results. It is necessary to record that the injections, even under strong pressure, were not carried to the point of macroscopic rupture.




60
The so-called mild syrmge-pressure, making use of solutions of potassium ferrocyanide and iron-ammonium citrate, resulted in extensions of the prussian-blue wholly similar to those obtained in the replacement experiments which were carried on for 30 minutes and over. This similarity indicates a complete filhng of the available cerebro-spinal system in the replacement method, for certainly (even in the mildest syringe injections) the intraventricular pressure must be excessively increased. Figure 1 shows a specimen under such conditions, with a marked thinning of the injection mass in the region of the fore-brain. This finding is customarily present in the injections under mild pressure, due to the pushing upwards of an existent ventricular fluid.




Some reticular.
"When moderate pressures are employed with the syringe the picture gradually changes. The essential difference in the results obtained by moderate syringe injection and by the replacement method lies in the greater extension of the foreign solution in the smaller embryos. Thus in figure 9 the spread of the injection precipitate in a pig embryo 16 mm. is shown to be about as extensive as that obtained by the replacement method in an embryo of 19 mm. (fig. 5). The extra ventricular distribution of the injected solution around the medulla, the extension (even more marked here) along the central roots of the caudal cranial nerves, and the localized perispinal spread are easily made out in this specimen of 16 mm.




This general rule applies to all of the results obtained with the use of syringepressures above the mildest. Dependent upon the degree of syringe-tension, the spread extends in simple ratio. Thus, by the use of moderate pressures of injection into the central canal of the spinal cord, a complete intramedullary and periaxial spread was secured in a pig embryo of 22 mm. somewhat earlier than the equivalent stage was obtained by the use of the replacement method.


588


With the highest syringe-pressures (insufficient, however, to cause macroscopic rupture) the same general type of injoctiou spread was obtained, bringing the more complete stages down into smaller and smaller embryos. Most of these embryos, however, on microscopic section showed obvious rupture of some part of the central nervous system.


4


The most important feature of these findings in the embryo pig injected with true sohitions under moderate pressures from a syringe concerns the fact that the extension of the injection coincides, except as to the size of the embryo, in every instance with that obtained by the replacement method. Thus similar and analogous spaces are filled by injections under syringe-pressures in small embryos and by the solution under normal tension in larger embryos. It must be assumed, then, that the pressure of injection is sufficient to dilate potential cerebro-spinal spaces which normally would not be concerned in the pathway taken bj' the cerebro-spinal fluid. No evidence of new or abnormal pathways for the fluid is afforded by the observations made with the increased pressure; these phenomena indicate great potential strength in the tissues which limit the immature cerebro-spinal spaces.


19X15X 8


Injections with a simple syringe may he made with such a degree of pressure that gross rupture of the tissues becomes apparent. In such an injection into the central canal of the spinal cord the infundibular region ordinarily ruptures in the smaller embryos (under 15 mm.), while in larger embryos rupture usually occurs into the subcutaneous tissues of the back of the neck over the fourth ventricle.


49


In discussing the effects of the introduction of solutions of ferrocyanide under pressures higher than normal into the central canal of the spinal cord, it may be appropriate to record observations made in the attempt to inject the cerebro-spinal spaces from the perispinal space. In embryos under 15 mm. in length it is quite difficult to make a perispinal injection. As the embryos exceed this measurement the injection becomes increasinglj' easy, but not until a length of 20 mm. is attained can it be made under the mild pressure advisable. These observations tend to substantiate the findings alreadj' recorded in both the intramedullary replacements and the injections under mild pressure.


Strands of magma.


===Results Of Injections Of Nitrate Of Silver===




391
In a number of experiments a dilute solution (0.5 per cent) of nitrate of silver was injected into the central canal of the spinal cord and the salt then reduced in the sunhght. This solution, although a true one, is wholh- unsuited for the replacement type of injection, on account of its great toxicity and its power to coagulate protein. It was employed here onh' for the simple type of injection.




2
The results obtained by this intraspinous injection of solutions of nitrate of silver were of but little value in the determination of a pathway for the cerebrospinal fluid, but they vividly present certain aspects of the problem. Thus, in figure 11, a drawing of a specimen (pig) of 16 mm., the area through which fluid passes in the superior portion of the roof of the fourth ventricle is clearly outlined by a denser deposition of the silver. This specimen was prepared by introducing the solution of nitrate of silver into the central canal of the spinal cord under the so-called moderate sj-ringe-pressure. The drawing shows a shght, cone-shaped extraventricular spread of the injection fluid. This spread takes place solely from the superior area of fluid passage, a result in accord with the finding that the solution of potassium ferrocyanide and iron-ammonium citrate passed first through the superior area. Of course it is realized that the precipitant action of the silver may have exerted a more potent action on the structures constituting the lower area of fluid passage.




16X14X12
Another interesting phenomenon of the injections of silver nitrate is shown in figure 12. The embryo of 13 mm. here represented was injected under strong syringe-pressure with a sohition of silver nitrate into the central canal of the spinal cord. On subsequent reduction and clearing it was found that the excessive pressure had resulted in a comjilete intramedullary injection with a localized pedunculate spread into the tissues from the roof of the fourth ventricle. This bulbous extravasation into the extravcntricular tissue has not been observed in any specimens except those into which the solution of silver nitrate was injected. Such a spread is probably to be accounted for by an immediate coagulation of the surrounding tissue.




14(?)  
The extensive use of solutions of sUver nitrate as a means of demonstrating vascular channels naturally suggests a careful comparison of the results obtained from its use and those obtained from the employment of other available true solutions, in regard to the evidence afforded by the two methods of intraspinous injections. The chief objection to the use of silver nitrate, as has aheady been mentioned, is its power to coagulate protein. This is illustrated by many features of the specimen shown in figure 11 — by the sharp outhning of the area of fluid passage, the markings on the caudal process of the fourth ventricle, and the delimitation of the cerebellar hp. But much more marked are the evidences of this coagulative power as shown in figure 12, the pedunculated extraventricular spread, the transverse corrugation of the cerebellar hp (amountmg to circumscribed indentations), and the peculiar outlining of the roof attachment to the bulb. These phenomena obtained by the intraspinous injection of solutions of silver nitrate must be classed as artifacts. The different degrees of this corrosive action of the silver probably result from the varj^ing rates of reduction of the salt to the metal, a factor which is not easily regulated. The findings, therefore, with this method are worthless unless controlled.




Do.
Many embryos of varying sizes were injected with the silver nitrate. In the main these observations followed the course of development of the cerebro-spinal spaces as evidenced by the replacement experiments with the ferroc3'^anide. The injections required moderate pressure in the syringe in order to secure more than a local extension from the roof of the fourth ventricle, and to secure the same extent of spread it was generally necessary to use embryos a few millimeters larger than those required in the replacement experiments; but this is to be expected, m view of the i^robability of a constant precipitation of the albuminous tissues by the injection fluid.




Specimens prepared by the intraspinous injection of silver nitrate, then, afford but little reliable evidence in this j)roblem except of a substantiative sort. The findings by this method indicate that the perispinal and pericerebral spaces, in pig embryos of 25 mm. and upward, could be filled by an injection of silver nitrate under moderate pressures into the central canal of the spinal cord. The point of passage of the fluid from the intramedullary to the periaxial system was in the region of the roof of the fourth ventricle.


136


===The Injection Of India Ink===


4


The objections to the use of any fluid of insoluble particles in suspension have already been discussed in considering the methods of injection which were possible for use in this study; but for comjiarisou Avith results obtained by more promising methods and to ascertain to what extent injections with India ink are reliable they will be further considered here. No granular substance other than India ink (carbon granules) was employed in this investigation. In every way this suspension possesses advantages over other possible masses — in its ease of preparation, in the small size of the granules, and the insolubihty in the reagents used for microscopic technique.


14X11X 6


Suspensions of nidia ink (diluted from 4 to 10 times) were introduced first into the medullary-canal system of living pig embryos by the replacement method. In no case, however, even though the circulation of the embryo may have continued for 90 minutes, was there any evidence of an extension of the replaced mass outward into the periaxial spaces. The carbon granules remained wholly within the ventricles, a striking difference from the results obtained by the ferrocyanide replacements. It would appear, then, without the further evidence afforded by microscopic section of the specimens, that there is an existing mechanism which prevents the passage of the carbon granules from the fourth ventricle into the periaxial spaces. This finding was found to be constant ui all the living embryos subjected to the cerebro-spinal replacement.


56


Quite similar to these results by the replacement method are those from the injection of a suspension of india ink under mild syringe-pressure. In no instance, provided the pressure was maintained at a low enough degree, was there any passage of the granular material into the periaxial tissue. In embryos of over 30 mm., however, even with the lowest pressure, it becomes increasingly difficult to prevent a sudden spread into the periaxial spaces. The type of spread indicates a sudden release of some restraining agent and suggests a rupture of a membrane. This spread is usually local and takes place from the roof of the fourth ventricle.


Reticular exce.ssive.


With moderate and strong syringe-pressures, however, it is possible to secure a periaxial spread, but this is quite different from the distribution of the uijections by the use of ferrocyanide solutions. Figure 10 illustrates a specimen of a pig embryo of 21 mm. into whose central spinal canal india ink was injected under strong syringe-pressure. The resultant spread of the injection is easily discerned; the cerebral ventricles are quite filled with the carbon, while from the superior portion of the roof of the fourth ventricle a dense but localized periaxial spread is made out. This extraventricular extension of the ink is well defined; it stretches caudalwards for a slight distance, curving about the bulbous caudal portion of the ventricle and extending lateralwards but a short distance. The median portion of the cerebellar lip is covered by the granules. E\adeuces of the excessive pressure at which the injection was made are shown by the lines of mvasion of the spinal cord and mid-brain. A comparison of the spread of this injection mass with the extension of a ferrocyanide replacement in an embryo of the same size (21 mm.) is afforded by figures 10 and 6. With such a divergence in the results obtained by the two methods of approach it is not surprising that observations such as Reford's) fail to coincide with these findings. The unsuitabiHty of suspensions of granular material in the investigation of the cerebro-spinal sjiaces has been many times verified in this work.




779
In the further study of the course of the spread with injections of india mk it was found that, in pig embryos of approximately 22 mm. and over, a partial periaxial injection could be secured by plunging the syringe-needle into the perispinal spaces. The carbon granules could subsequently be seen filling the perispinal spaces and also mounting upwards in jjartial pericerebral relationships, particularly around the medulla. This result was obtained by the use of strong syringe-pressures. Apparently the resistance to the spread of the ink in injections or replacements in the medullary-canal system occurs in the passage of the fluid from the roof of the fourth ventricle into the periaxial spaces. So far as is known, Reford^"*^) did not control his injection pressures. These results with the injection of india mk under strong pressures coincide with the idea of his observations afforded by the abstracts given by 8abin(*3) and Cushing^^'. Suspensions of india ink. then, injected under mild syringe-pressure or by the replacement method, offer no evidence, in the pig embryo, of a passage of the cerebro-spinal fluid into the j^eriaxial spaces. Only by employing pressures much above the normal tension can such evidence be obtained.


==V. Undescribed Structures in Roof of the Fourth Ventricle==


2.75
The results of the replacement of the existing cerebro-spinal fluid by a true solution of potassium ferrocyanide and iron-ammonium citrate in a living pig embryo indicated, as detailed in the foregoing section, that the fluid passed from the ventricular system into the periaxial tissues in the region of the roof of the fourth ventricle. This important transit of the fluid, agreeing with the established conception of the relationship in the adult, was first observed in an embryo pig of 14 mm. (fig. 3). At this stage the exudation of the replaced fluid occurred in one defined area, seemingly corresponding to the dense oval in a smaller embryo shown in figure 2.




16X14X12
Such a passage of fluid from ventricle to i)eriaxial tissue is necessarily a ])hysiological phenomenon, and it was in the hope of finding an anatomic basis for this phenomenon that the roof of the fourth ventricle was studied histologically. It was reahzed that failure to demonstrate anatomically differentiated structures would not vitiate the physiological observations, but that a correspondence between function and structure was most desirable. Hence observations were undertaken to determine, if pn.ssible, an area of histological differentiation in the roof of the fourth ventricle which might be concerned in the primary passage of fluid from the cerebral ventricles into the periaxial tissues. The investigation concerned first the examination of this region in pig embryos of 14 to 15 mm., at which stages the fluid passes from a single area. Subsequently, similar studies were undertaken in regard to the second, more inferior area (shown in figure 4). The results of these studies will be given here.




40
AN UNDESCRIBED AREA IN THE SUPERIOR PORTION OF THE ROOF OF THE FOURTH VENTRICLE.




None.
===The Area Membranacea Superior In The Pig Embryo===




Examination of the roof of the fourth ventricle in a pig embryo of 14 mm. revealed a peculiarly differentiated area in the superior portion. The general topography of this area is show7i in the rectangular area marked off in figure 32 — a median sagittal section from a pig embryo of this critical stage. In figure 33 this rectangular area is enlarged to show the morphologA' in greater detail.


1 836


In this figure the densely staining ependj'ma lining the fourth ventricle approaches from both sides. The superior portion of the ependyma ends abruptly, while the inferior Hne of the layer tapor.s more slowly. Between these two jjoints is an area having none of the characteristics of the ventricular lining at all other points. The comparatively smooth contour of the ependymal cells is replaced by an irregular cell-border. The pyknotic nuclei of the cells have been replaced by less densely staining, elongated, spindle-like nuclear bodies. The cell-layer lining the ventricle is here really only of a single cell in thickness, although blood-capillaries closely applied to it suggest a greater thickness. The mesenchj-nie lietween this layer and the peripheral epidermis is quite thin, but resembles in everj' way the mesenchyme in the immediate neighborhood.


4


There is, therefore, as pictured hi figiues 32 and 33, an area in the roof of the fourth ventricle which is morphologically dissimilar to the characteristic ependyma lining the cavity. Is this the result of some distortion in fixation or in the routine histological technique? Is it a constant finding and, if so, what is its historj'? Does it arise at a definite period and persist throughout intra-uterine life onlj' or through adult life also?


22X18X11
The question of the actual existence of this area, or of its being caused by technical manipulations, is one which must be answered. That this differentiated portion of the roof of the fourth ventricle is not an artifact is verified bj* the general history of its formation, by its invariable occurrence (not onh' in the pig but in other animals), and by its general histological appearance. Moreover, the physiological importance of this area undoubtedh' incUnes one completely from the possible explanation that it is due to an artifact. Xo single finding wholly excludes such a possibility; rather is one convinced, by many features, of its actual occurrence.




36(?)
Considering the fact, then, that this differentiated structure in the roof of the fourth ventricle may be found in all embryo pigs at the stage of 14 mm., it becomes necessary to ascertain at what time in the development of the embryo it first appears and how it is formed. Obviously the most satisfactory' method is to trace the area through the lower stages and also through the older embryos. For the sake of greater clearness, however, a description of the area \s-ili be given from its first differentiation through its maximum transformation to its final disappearance — for the structure is only temporary.




Delicate reticular.
In pig embryos of 8 mm. and less in crown-rump measurement, the roof of the fourth ventricle is fcfrmed of cells morphologically and tinctorially different from those lining other parts of the ventricular cavities. These cells are quite unlike the deeply staining ependjTiial cells, which can be so readily identified as the lining cells in older embryos. In this yoiuiger stage of 8 mm., the entire ventricular roof is composed of several layers of cells with round or somewhat oval nuclei and fairly abundant cytoplasm. The cell-boundaries are not well defined. The nuclei are not deeplj' tinged with hematoxylin. The chromatin material is sparse and irregularly distributed. Nucleoli are prominent. The cytoplasmic border hning the ventricular cavity is rough and ragged at times, often blending with the coaguUvted albumen of the cerebro-spinal fluid. Altogether, these lining cells bear a much greater resemblance to the epithelial cells than to the ependymal.




These characteristics of the lining cells of the roof of the fourth ventricle are shown in figures 24 and 25, from a pig embryo of 8 mm. The close association of the roof cells to the surface epithelium is easily made out in figure 25, as well as the general character of the lining cells.


164


At the stages of 8 mm. and under, in the pig embryo, the roof of the fourth ventricle is relatively'' quite large. In its whole extent it is formed of the peculiar lining cells described above. With the growth of the embryonic nervous sj^stem, the roof of the fourth ventricle is subjected to alterations in form and position; to some extent these changes influence the cells which line the cavity in the early stages.


3.5


In pig embryos between 8 and 12 mm. in length the roof of the fourth ventricle undergoes a change. The ependyma, which from comparison with later stages is regarded as typical, begins to encroach upon the epithelial-like cells which are so numerous in the 8 mm. stage (fig. 25). The area occupied by these cells diminishes, not only relatively but absolutely. It becomes smaller and the cells gradually change their character. These changes are shown in figures 26 and 27, from a pig embryo of 11 mm. Figure 26 gives the location, in a sagittal section near the midline of the area in figure 27, taken at a higher magnification.


17X17X10


In figure 27 the densely staining lips of ependymal and nerve cells are seen approaching each other. For a considerable space in the central portion of the photograph there is an area similar to that shown in figure 33. But considered in connection with figure 25 this area represents the epithelial-like cells of the roof of the fourth ventricle. This relationship is more clearly shown in figures 28 and 29, taken in a more lateral plane from the same embryo (11 mm.). Examination, however, of th(; area in figure 29 shows the epithelial-like cells again a])i)arent in the roof of the fourth ventricle.




The i^rocess of transformation, then, as shown in these photographs from an embryo i)ig of 11 mm., concerns a gradual encroachment upon the area of epithelial-like cells by the more densely staining and more closely packed ependymal cells. Gradually the epithelial-like cells in the central portion of the area lose their former character (fig. 27), while around the periphery, especially on the lateral sides, the epith('lial-lik(! ajjpearance persists (fig. 29).




Few strands
On the lateral side of this area, just as the tyincal ependymal lining is al)out to become isolated (fig. 29), the epithelial-like lining cells form a several-celled layer.




148
The nuclei are poor in chromatin material and the cytoplasm somewhat small in amount. The inner cytoplasmic border hning the ventricle is in contrast, by its ragged outline, with adjacent smoother ependj^ma on both sides. At this stage of the pig embryo the characteristics of the epithehal-like cells are still to be made out, but a gradual transformation is becoming evident.




4.3
The metamorphosis becomes much more marked in the central portion of the area, as shown in figures 26 and 27. In these figures the whole central area seems to have lost some of its former character as an intact cell-laj'er. Closer examination, however, under higher power demonstrates that it still possesses an intact surface as a hning for the ventricle. Delicate cytoplasmic strands stretch in a continuous line across the whole area between the Ups of denser tj-pical ependjina. The nuclei in this difi'erentiated area are seeminglj' altered from their rounded form and have elongated almost into spindles. The inner cytoplasmic border is characteristically rough, with, small amounts of coagulated albumen adhering to the processes. The area, then, in its central portion, at the stage of 11 mm., has assumed the character of the stage of 14 mm. (fig. 32). On the periphery, however, the cells stiU resemble those of smaller stages (8 mm.).




17X14X10
From the pictures presented by the intermediate stages (figs. 27, 28, and 29) the differentiation goes on verj' rapidly, so that in the pig embryo of 13 mm. there is rarely any evidence of the epithelial-like cells. Figures 30 and 31 are photomicrographs of a sagittal section of an embryo pig of 13 mm.; here there are no evidences of the epitheUal-like cells. The whole area, pictured in figure 31 as sharply delimited from the tongues of tjijical ependyma above and below, has become well differentiated. The cell-character observed in figures 27 and 33 (elongated nuclei and scanty strands of protoplasm) has become very obvious. The ragged and roughened intraventricular border, the coagulated albumen, and the abrupt transition from the neighboring tj-pical ependyma are well shown in the photomicrographs of this specimen.




34
The differentiation of this area in the roof of the fourth ventricle of the pig embryo proceeds at a very rapid rate, so that within the growth of a few millimeters (from 8 to 13 or 14) a great histological change occurs. Figures 32 and 33, already described, show the extent of this metamorphosis in a pig embrj^o of 14 mm. The process, however, continues, modified possibh' bj- the changing of the roof of the fourth ventricle. For this roof structure is subjected to marked alteration in stages of 14 mm. and upwards, both by the lateral development of the chorioid ple.xuses and b}^ the readjustment of the cervical and pontine flexures. Its maximal differentiation may be said to appear at a stage of 18 mm.; this is maintained through several millimeters, until undergoing final retrogression.




Small amount of magma
This maximal change in the roof of the fourth ventricle is shown in figures 34, 35, 36, and 37. Several points of interest are brought out in these photomicrographs. Figure 35 represents an enlargement of the rectangular area in figure 34, taken from transverse sections of an embryo pig of 18 mm. The area is particularly well shown in this figure, in which, from the right, the typical ependyma, in a fairly smooth single-cell layer, approaches the differentiated cells in the central portion. On the left, too, similar typical ependyma is shown. In the central area, which has been repeatedly described, the elongated nuclei, the strands of protoplasm, and the ragged, iiTegular intraventricular surface are well presented. The photomicrograph has been reproduced to show the relation of this differentiated area to the various blood-channels in the supporting mesenchyme. Apparently the whole ventricular roof is, at this stage, a site for an extensive capillary plexus; from both sides, as shown in figure 35, vessels (one of great caliber) approach the central area of differentiation. Directly beneath this area smaller capillary channels can be made out, from which, apparently, a sUght extravasation of red blood-cells has occurred. Here, as in the greater part of the basilar pericerebral region, extravasation of the blood-cells is very frequent. This phenomenon has already been pointed out by INIaU'^"".




463
The large extent and the great differentiation of this pecuUar area in the roof of the fourth ventricle are well shown in figures 36 and 37, taken from a transverse section of a pig embryo of 18 mm. In the photomicrograph of higher magnification the two sharp tongues of typical ependyma are quite striking. Their abrupt termination in the wide, differentiated area has nowhere been more convincingly shown. The resemblance of these lining cells in the central area to the mesenchymal elements adjoining is here also seen. The most interesting of all the phenomena exhibited in this reproduction, however, is the attachment, apparently by precipitation, of the coagulated albumen of the cerebro-spinal fluid. This coagulation, in this specimen, deUmits the differentiated area in the roof of the fourth ventricle. The phenomenon is seemingly only an ampUfication of a similar attachment of small fragments of the albuminous precipitate shown in other figures.




3 9
Beyond the stage of 18 mm., which may be termed the maximal stage, the differentiated area in the roof of the fourth ventricle undergoes a regression. This is apparently due to the morphological alterations in this rhombic roof. The chorioid plexuses in embryos over 18 mm. long deeply invaginate the fourth ventricle, possibly drawing some of the true roof with them, but surely encroaching upon the mid-hne with their lateral tuftings. This growth tends to decrease the available extent of the differentiated area, but an even more potent factor is the rapid development of the cerebellum. The caudal growth of the cerebellar lip soon largely occupies or replaces the superior half of the roof. These two factors, the cerebellar growth and the enlargement of the chorioid plexuses, render the persistence of the differentiated area impossible, so that a regression or disappearance is to be expected.




17X12X 7
With these considerations before us, the study of sectioned pig embryos of a greater length than 18 mm. becomes important. The process of disappearance, however, does not occur at once. Thus, in an embryo pig of 19 mm. (figs. 42 and 43) the differentiated area is as large and as characteristic as in the stage of 18 mm. This same aijpoarance and maintenance of size may be observed through the next several millimeters' growth, but in pig embryos of 23 mm. the chorioid plexus has usually developed to such an extent that a continuation of the former size becomes impossible. This is shown in figures 44 and 45. Figure 45, the enlarged squared area from figure 44, is a photomicrograph from a pig embryo of 23 mm. The differentiated area, duo to the factors favoring its regression, now appears in close proximit}^ to the chorioid plexus. It has more the appearance of a degenerating area at this stage than in any of the younger embryos, but it still shows a characteristic delimitation of both edges — on the one from the typical ventricular ependyma, and on the other from the differentiated ependj'ma of the chorioid plexus. The cytoplasmic strands of the area which forms the ventricular border do not show to advantage in the photomicrograph, but the .same ragged character with the covering of coagulum may be made out. The process of regression, mechanical as it perhaps is, has begun at this stage in the pig, and in the course of the next few millimeters' growth will become even more active.




48
With the encroaclmaent of the chorioid plexuses and the downward growth of the cerebellar lip, the superior portion of the ventricular roof soon disappears, and is practically non-existent in embryos of 30 mm. and more in length. The differentiated area thus encroached upon from the sides and above becomes a mere vestige of its former size. Thus in a pig embryo of 32 mm. (figs. 46 and 47) it appears as a very small break in the lining continuity of the ventricular ependyma. Without the intermediate stages such a picture would undoubtedly be considered as an artificial erosion of the ependjinal lining of the ventricle, but when studied in connection with figure 45 the true vestigial character of the area becomes established.




Much reticular.
The final fate of this differentiated area in the roof of the fourth ventricle is a complete disappearance, with the occupation of the region by chorioidal epithelium and cerebellum. In this study it was impossible to find traces of the differentiated areas in pig embryos of over 33 mm. in length; vestiges maj' persist, but so small as to present difficulties of decision. The persistence of such a differentiated vestige in rare instances would not be surprising; the transitory character of the area and the method of disappearance make this seem not unlikely.




This transitory area of differentiation in the roof of the fourth ventricle of the pig has not, so far as can be determined, been noted or described bj- any p^e^•ious author. His'25\ in a retouched photomicrograph of a sagittal section of a human embryo of 17 mm., reproduced the area as differentiated from the roof, but he has made no comment upon it. I have called this differentiated area in the superior portion of the rhombic roof ventricle the "area membranacoa superior ventriculi quarti." This terminology is based on the anatomical character of the area as a continuous membrane, but chiefly on its physiological significance. For, as will be shown in the succeeding section of this paper, the transit of embryonic cerebrospinal fluid from ventricle to periaxial tissue occurs in this area, which functions apparenth- as a physiological membrane. With such a physiological conception of the area, the ttim "area membranacea" seems most suitable, inasmuch as it also meets the anatomical requirements.






===The Area Membranacea Superior in the Human Embryo===




The finding of the diflferentiated area in the superior portion of the roof of the fourth ventricle in the embryo pig suggested the value of a study of the same region in the human embryo in the further solution of the problems underlying its occurrence. Hence this region in the roof of the fourth ventricle has been examined in the sectioned human embryos of the Department of Embryology of the Carnegie Institution of Washington. It was found that a similar area occurred in the human embryo of approximately the same age.




The study of the roof of the fourth ventricle is usually more difficult in the human embrj^o than in the pig. This is due to the fact that the roof of the fourth ventricle quickly suffers from poor fixation and dehydration — collapse or inversion of the whole structure being commonly met with. It is rarely possible, in the younger embryos, to secure the most satisfactory fixation, whereas in the pig these factors may be controlled as desired. Furthermore, the undue pressures to which the human OAaim is frequently subjected in abortion may cause crushing of the more delicate parts of the nervous sj^stem.




It is probably best, in the human embryo as in the pig, to trace the formation of the area membranacea superior ventriculi quarti from its beginning, through the various differentiations.




In a human embryo of 4 mm. (No. {{CE836}} of the Collection of the Carnegie Institution of Washington) the entire roof of the fourth ventricle is composed of cells with round or slightly oval nuclei and palely staining cytoplasm. The nuclei of the cells are poor in chromatin material as contrasted with the pyknotic character of the typical ependymal cells. The lining tissue is of the thickness of several cells. The ventricular cytoplasmic border is fairly smooth at this stage. This characteristic ventricular lining is shown in figures 40 and 41, both taken from embryo No. 836. The whole picture is similar to that exhibited by the pig embryo of 8 mm. (figs. 24 and 25).




around cord.
A similar accumulation of epithelial-like cells is found in a human embryo of 7 mm. (No. {{CE617}} of the Carnegie collection). This is pictured in figures 48 and 49. The photomicrograph of higher magnification shows these poorly staining cells heaped up in a rather localized part of the ventricle, fairly sharply dolimitod from the adjoining ventricular lining. This accumulation of cells in the roof of the ventricle invariably occurs, and it must not be considered as being due to the distortion of the ventricular roof. The reason for the asymmetry of the rhombic roof shown in these figures hes in the fact that in this embryo, as in practically all the embryos of similar stages in this collection, some degree of distortion of the roof of the fourth ventricle is present. Photomicrographs (figs. 50 and 51) taken more posteriorly (from embryo No. {{CE617}}) give strong evidence of this distortion. They are reproduced not only to show the possible distortion, but also to give a further picture of the lining of the ventricle, with its epithelial-like cells in several layers (fig. 51).




Similar accumulations of these epithelial-like cells are to be found in human embryos of 9 mm. Reproductions of a much fragmented specimen of this size (No. {{CE721}}) are given in figures 52 and 53. In the latter figure the complete occupation of the ventricular roof by these cells is well illustrated. Moreover, the specimen shows the many-layered stage to a degree but seldom found. It is unfortunate that such a degree of fragmentation and distortion is found throughout this specimen.


486


Thus far, in human embryos up to and including 9 mm. in length, the roof of the fourth ventricle has shown the same architecture as appears in the pig. As will be recalled, the first evidence of a further differentiation of these cells in the pig embryo was found at a stage of 11 mm. (figs. 26 and 27). In one human embryo of this stage (No. 544) a distinct break in the roof of the fourth ventricle can be made out. This is shown in two photomicrographs (figs. 54 and 55). The picture in this cas3 is somewhat obscured bj' the shrinkage and distortion of the ventricular roof, but a distinct differentiation of the lining epithelium can be made out. On the caudal side of figure 55 considerable nervous tissue is seen. Just superior to this (toward the left) the lining tissue is almost lacking, a few nuclei, only, preserving the contour of the ventricle. Above this area appears again the ventricular lining of many layers of cells. It has been quite difficult to interpret these findings. The area under discussion shows a rather typical adherence to the coagulated albumen ; there is evidence of its extension also into the adjacent mesenchyme, a finding observed in no other similar stage. The caudal position of the opening, the character of the tissue approximating the ventricular cavity, and the presence of the albumen in large amount in the adjacent mesenchyme — all indicate that in great measure the pictures presented in this specimen are largely artifacts. It seems most likely, though, that some differentiation of the tissue in this area has occurred.


4


In a human embryo of 14 mm., as in the pig of the same stage, the area membranacea superior has attained a great degree of differentiation. This is particularly well shown in figures 56 and 57, the latter being an enlargement of the squared area in the former. These photomicrographs are from embryo No. {{CE144}} of the collection of the Carnegie Institution of Washington. Figure 57 shows a characteristic which distinguishes the area membranacea from that of the pig, although in the later stages of the pig embryo (figs. 45 and 47) this feature is present. This concerns the marked decrease of cellular tissue in the membranous area. In figure 57 the deeply stainmg typical ependyma is shown approaching from below. These cells end abruptly at the border of the area membranacea; the ventricle in this area is lined by cells possessing small elongated nuclei and long cytoplasmic processes, which unite to form a ventricular lining. The oval nuclei along the ventricular border become more closely massed together in the superior portion of the area, but nowhere is there the same architecture as in the equivalent stage in the pig (fig. 33). A feature of the histological appearance of the membranous area in the pig embryo is also shown in figure 57; this is the marked adherence of the coagulated albumen of the cerebro-spinal fluid to the area membranacea superior.


22 X22 X22


The roof of the fourth ventricle in the human embryo is subjected to the same factors causing changes in the form and relationships which were commented upon in the pig; but these play little part until the chorioid plexuses become of sufficient size to divide the ventricle into a superior and inferior portion. In the human embryo, as in the pig, the superior half of the ventricular roof is sacrificed to the greater growth of the cerebellum.


44


* Measured on the slide after mounting.


Do.
In human embryos of 17 mm., however, these factors have not begun to influence the membranous area. This is shown in figures 58 and 59, photomicrographs from embryo No. 57G. The section is somewhat to the side of the midline, but in the superior portion of the roof of the fourth ventricle the differentiated membranous area can be made out. The sharp delimitation of this area from the denser t3T5ical ependyma on both sides is quite apparent. The ragged character of the ventricular border, with its few elongated spindles, seems wholly in keeping with the transverse view of this area afforded by figure 37.




Embryo No. 576 exhibits one characteristic of the area membranacea superior very frequently seen in human embryos, but almost invariably absent in these stages in the pig. Along the lateral margins of the superior membranous area are dense borders of the many-layered epithelial-like cells which lined the ventricular roof in younger stages. This feature is well shown in figures 60 and 61, the latter figure being a higher magnification of the former. The cellular border of the superior area reaches transversely only through a few 15-micron sections, but it extends throughout the whole cephalo-caudal diameter of the area. It seems likely that this represents purely a survival of the epithelial-like cells in the younger embryos. In rarer instances the whole area membranacea superior may be surrounded bj^ such a border of many-layered cells, but even in these cases the superior and inferior margins are quite thin.


576


No apparent agencies favoring the disappearance of the superior membranous area m the roof of the fourth ventricle of the human embryo are apparent in stages up to the fetus. Thus, in human embryos of 18 mm. this differentiated area in the roof has reached its maximal differentiation. A section from an embryo of this size (embryo No. 409) is reproduced to show the distortion and its influence upon the topography of the area membranacea. The two photomicrographs (figs. 62 and 63) show the extreme collapse and distortion of the roof of the fourth ventricle. In the figure of higher power (No. 63) the membranous area appears facing posteriorly, due to the shrinkage; the proper leader runs to this area. It shows the differentiation from the adjoining tj'pical ei)endyma which is characteristic of the full}' developed area membranacea superior.


17


In a beautifully preserved and sectioned human embryo of 21 mm. (No. {{CE460}}) in the collection of the Carnegie Institution of Washington the area membranacea superior appears as a sharply delimited area (figs. 64 and 65). These figures give a very good idea of the definiteness of this area when the fixation and dehydration approach the perfect. The tissue of this membranous area lining the ventricle here appears to be wholly lacking in an epithelial covering; the mesenchyme seems to serve as the ependymal lining. Study of this area, however, through different stages argues most strongly against such a view.


30X30X25


he process of regression of the area memhranacea sujjerior in the human embryo differs somewhat from that described in the pig. This alteration in the mode of disappearance is largely due to the fact that in the period of growth from 20 to 35 mm. the superior portion of the roof of the fourth ventricle in the human embryo is not sacrificed to the cerebellar lips; for in the human the cerebellum grows largely into the fourth ventricle, enlarging beneath the superior part of its roof. Thus, the attachment of this part of the roof is not greatly interfered with by the rapid development of the cerebellum. The total extent, then, of the superior portion of the roof is hardly altered in these stages in the human, while in the pig embryo the roof is shortened by its attachment to the inferior portion of the cerebellar hp, which retains its earUer characters. These differences in the relationship of the superior portion of the ventricular roof in human and pig embryos may be seen by comparison of figures 74 and 89.




Another factor which renders the mode of disappearance different in the two embryos concerns the greater tufting and development of the chorioid plexuses of the fourth ventricle in the pig. This greater size and complexity of the plexus causes an encroachment upon the roof structures which, in the pig embryo, seems of considerable importance in the t'nal closure.




Small amount of mayma.
In the human embryo, however, it has been found very difficult to explain the final disappearance of the superior membranous area on the same mechanical factors w^hich seemed so well to account for its transitory characters in the pig; but at approximately the same stage of growth the process of regression occurs in the human fetus. The area maintains a fair size in stages up to a length of 23 mm. Thus, in figures 89 and 90 (No. {{CE453}} of the Carnegie collection) a sagittal section from a human fetus of this size is illustrated. In the higher power (fig. 90) the superior membranous area is short, rather sharply delimited on its superior border by the typical, dense ventricular ependyma. Below, its edge is irregularly formed by the deeply staining ependyma over the invagination of the chorioid plexus. The cell-character of this area resembles that shown in the photomicrographs from the specimen of 21 mm. (figs. 64 and 65). There is left in the area no indication of the cellular architecture which characterized the original ventricular ependj-ma; the cells with their elongated cj'toplasmic processes here have the oval nuclei which are found almost invariabh' in this membranous area.




In the human fetus of 26 mm. (No. {{CE1008}} of the collection of the Carnegie Institution of Washington) there is but sHght evidence of a superior membranous area in the upper portion of the roof of the fourth ventricle. The evidence present in this specimen consists in a localized thickening of the lining cells of the ventricle in the situation of the area in other stages. This thickening is illustrated in figures 91 and 92; it consists of several layers of epithelial-hke cells, similar in all respects to the many-layered border shown in figure 83. The picture is somewhat obscured by the vascular pl-^xus directly beneath the ventricular lining.


470


There is difficulty in determining exactly when the last evidences of the superior membranous area in the roof of the fourth ventricle may be found. This is due to the likelihood of artifacts disturbing the character of the ventricular lining in human material, where the freshness and fixation of the specimen may not be ideal. In the larger specimens in the collection of the Carnegie Institution, which are well fixed and sectioned, the existence of the area membranacea superior could not be wholly verified. Thus, in specimen 405 (26 mm.) the presence of the area seemed probable though not definite. In another embiyo of this same size (So. 782) the existence of this area was still more questionable. In a larger embryo (30 mm. No. {{CE75}}) the presence or absence of the area could not be assured; many indications suggested its existence, but the resemblance to an artificially separated ependyma was strong. In all specimens of human embryos of over 30 mm. examined, no evidence of the area membranacea superior could be found. It appears likely, then, that the final disappearance of this differentiated area in the roof of the fourth ventricle occurs at a slightly earlier stage in the human embryo than in the pig.


4


The final disappearance of the area membranacea superior in the human embryo is not accompanied by the same ingrowth of typical ependyma that characterizes the process in the pig. There is a great tendency, in the human, as indicated in figure 92, for a replacement of the area by the same type of epithelial-like cell which comprised the whole ventricular roof in the earlier stages (fig. 41) and later formed lateral borders for the superior membranous area (fig. 83). Thus, in a human embryo of 24 mm. (No. G32 of the Carnegie collection) there is evidence of a very small membranous area surrounded by a border of epithelial-like cells. In a slightly larger specimen (No. {{CE840}}, 24.8 mm.) the whole membranous area is occupied by the epithelial-like cells. The frequent association of these cells with the area indicates that in disappearing the area membranacea is probably replaced first by these cells, which in turn disappear, so that the whole roof is finally composed of the typical, densely staining ependyma.


20X13


===The Area Membranacea Superior in Other Animals===


34


In order to ascertain whether the area membranacea superior existed in other animals examinations of serial sections of the rabbit, cat, sheep, and chick of suitable stages were made. All of these animals were found to possess a differentiated area in the roof of the fourth ventricle.


Very few fibrils.


Opportunity was afforded for the study of serial sections of the head of a chick* of 121 hours' incubation. The head was carefully dehydrated and embedded by Dr. E. R. Clark, and was subsequently sectioned bj' Dr. C. R. Essick. The material was beautifully fixed and dehydrated, showing practically no evidence of shrinkage. Typical portions of the superior membranous area are reproduced in figures 66, 67, 68, and 69. Figure 67, taken near the crown of the embr3-o and representing the squared area in figure 66, shows the two dense masses of ependyma separated by the more lightly staining area membranacea. The cellular character of this differentiated zone resembles more the histological features of the similar afea in the pig than those of the human embryo. This resemblance is also to be seen in figure 69, taken more posteriorly than the two preceding figures. The dense ependynia approaching on both sides is sharply delimited at the edge of the hroad membranous area. This is composed of cells having elongated, chromarin-poor nuclei, and long cytoplasmic processes, which form the ventricular roof. The adlierence of the albuminous coagulum occurs here also.




!
* The chick measured 14 mm. in 40 percent alcohol.




In the rabbit the occurrence of the superior membranous area was verified as in the other species studied. In a rabbit embryo of 13 mm. (series x in the embryological collection of this laboratory the area was well differentiated from the surrounding typical ependyma. The cells of the area resembled those of the adjacent mesenchyme. The ventricular surface was roughened by the projection of numerous protoplasmic processes. An albuminous coagulum was attached to the cells of the membranous zone.




One sheep embryo from the collection of this laboratory was also studied. The sections, although labeled as an embryo of 10 mm., resembled in every way a pig embryo of 18 mm. The area membranacea was easily identified in the roof of the fourth ventricle; it is similar in every respect to the same area in the pig and the human embrjo.




In a cat embryo of 10 mm. a smiJl but highly differentiated area membranacea superior was made out. The most striking feature in this specimen is the great adherence of the coagulated albumen to the cells of the area and the resemblance of these cells to the mesenchymal elements adjacent. The edges of this differentiated area are sharp and clear-cut.




No attempt was made to identify the area membranacea superior in other animals — as further suitable material was not immediateh- available. The chief study has been made on pig embryos and on human embryos. The occurrence of the area in the cat, sheep, and rabbit probably indicates its existence in aU mammals. The finding of such an area in the chick is also suggestive.




===General Consideration of the Area Membranacea Superior===




The occurrence of a definite area of differentiation in the superior portion of the roof of the fourth ventricle has been pointed out in preceding subdivisions of this paper. It has been described in detail in the pig embryo and in the human embryo; it has been identified also in cat, sheep, rabbit, and chick embryos. It remains here to discuss the general characteristics of this area.




No description of such an area of differentiation in the ventricular roof has been found in the literature. It may be that the distortion of this structure in the course of the usual embryological technique has rendered its discovery Iess likely. His'-='', m his description of the ventricular roof, has not commented upon the occurrence of this membranous area, even though in a retouched photomicrograph of his fetus C-1 (a human specimen, of the beginnmg of the third month) the area membranacea superior can be made out. Likewise in his description of the pUca chorioidea he faJs to mention any differentiated areas in the roof, although plate I, in his "Die Entwickelung des menschUchen Rautenhims, von Ende des ersten bis zum Beginn des dritten Monats," shows a slight irregularity in the roof. Practically all of the contributions to the anatomy of the roof of the fourth ventricle deal with the lower half of the structure, with particular reference to the occurrence of the foramen of Magendie.


Specimen No. 763 was removed from a woman who was the mother of 6 children, the oldest being 10 years old. She had had one miscarriage. During the year before the operation she suffered much from headache and backache, but otherwise her health appeared to be normal. When she was admitted to the hospital she complained of abdominal enlargement and there was some urinary disturbance. At tlie operatif)n for rui)ture of the perineum the uterus w'as scraped out ; subsequently the ovum was found in one of these scrapings. The fragments both of the mucous membrane and ovum appear to be normal.


The general biological process involved in the formation of the area membranacea superior concerns a dilTerentiation of the epidermal elements which Une the ventricular cavity. This differentiation, both in human and in pig embryos, first begins with the occurrence in the ventricular roof of an area of epithelial-hke cells. These, in the course of enlargement of the roof, become more or less isolated in the superior portion of the structure, and then undergo a metamorphosis into the typical cells of the membranous area. They are characterized by oval or elongated nuclei (rather poor in chromatin as compared with the nuclei of the typical ependjonal elements) and by cytoplasmic strands (in which the cell-boundaries are very poorly marked) which compose the ventricular border. The ventricular surface in the area membranacea is more ragged and u-regular than where lined by typical ependjana. In many instances, as in figure 57, from a human embryo of 14 mm., this transformation has proceeded to such an extent that the epithehal character of the lining cells is almost wholly lost, and the ventricle seems, in this area, to be lined by mesenchyme. Study of the membranous area in many stages convinces one that such an hypothesis is untenable; in every case the ventricle must be considered as being lined by epidermal elements, no matter to what extent the process of differentiation has proceeded. There is no real evidence to support the view that the ependymal lining of the ventricle has been replaced by mesenchymal elements to form the area membranacea superior.


Unfortunately we have only a few of the sections of this valuable specimen, but these show that we are undoubtedly dealing with a normal ovum of the same stage of development as that described bj' Peters. The chorionic cavity is ])artly filled with mother's blond, ])ut there are some strands of reticular magma, with nuclei and protoi)lasm radiating through the blood. The specimen has been staimnl in hematoxyhii and eosin, which is not esjiecially favorable for defining magma fi])rils.


In general the area membranacea superior is a rounded oval. Its measurement is quite difficult except when fixation and dehydration have been excellent, because of the highly abnormal distortion of the ventricular roof which frequently occurs in the technically poor specimens. Measurements have been made in a considerable number of favorable specimens, both of human and pig embryos. With the history of this area in mind, it will be realized that the size of the structure necessarily varies with the length of the embryo, attaining its greatest dimensions at about the length of 18 or 20 mm. Herewith is a short table of the measurements taken.


The specimen described by Ilerzog is also undoubtedly normal, as it was obtained from a woman who was killed by a stab-wound through the heart. The large colored plate i)ubli.shed by Herzog shows the specimen to be quite identical with that of Peters. It shows free cells in the coclom, which contains no other foreign substance, but a i)hotograph (figure 24, published by Ilerzog) shows that the cu'lom is filled by a very pronounced substance, reminding one very much of reticular magma. The same is true of a specimen recently described by Johnstone. A colored photograph which he j)ublished shows quite distinctly a ]M-onounced magma throughout the coclom. (See, for instance, his figure 3.) This establishes definitely the i)resence of reticular magma in ova the size of the si)ecimen of Peters. AVe have, however, the valuable specimen of Bryce and Teacher, which also shows the condi


Dimensions of area membranacea superior.


Species.


HUMAN MAGMA UKTICULE IX NORMAL AND PATHOLOGICAL DEVELOPMENT. 11
No. of specimen.
tion of the magma in an earlier stage. In this specimen the chorionic cavity is filled with a dense mass of fibrils, throughout which are scattered numerous nuclei, as shown in their plates 3 and 4. The specimen was not perfectlj^ hardened and there is a small cleft between the chorionic wall and the mass of magma. As .vet there is no exocoelom, showing that it is younger than the Peters specimen.


Length of embryo.


jMore advanced stages of the condition of the magma are represented in the specimens described by Jung and by Strahl and Beneke. In the Jung spccunen the cavity of the ovum is filled with a very pronounced magma, running together in stronger bands, as in our own normal specimen. No. 836, to be described later. The larger cavity Jung marks "exocoelom," but it is not clear that this is lined with endothelium. From his large illustration one gains the impression that the specimen is somewhat pathological, for it is of the same type as numerous specimens in our coll?ction with embryos that are usually found to be pathological. Taking the illustrations given in Jung's plates 1 and 2, the specimen again appears to be pathological, and I should be inclined to pronounce it such did not his plate 6, figure 17, show this same section on an enlarged scale, which gives a very sharp outUne of different embryo structures and scattered through them are numerous cells undergoing division. It would be impossible, with our present knowledge, to accept such sections as coming from a pathological embryo. The specimen described by Strahl and Beneke is of about the same stage as the Jung specimen, although the magma does not seem to be so well pronounced. It is unequal m nia.ss and has scattered through it deUcate strands, as shown in their figure 63. In fact, the above-described specunen underlies also the diagram on the form of the ccelom given by Strahl and Beneke on page 18 of their monograph.
Width of area.


Length of area.


]\Iagma of uniform consistency, as seen in the Bryce and Teacher specunen, soon arranges itself in bands, which gradually become more and more pronounced in older specimens. Between these bands are spaces filled with fluid, and those spaces near the embrj^o become lined with endothelium to form the exocoelom. There are other spaces between the exocoelom and the chorionic wall. The sharper bands of magma fibrils — well shown in our embryo No. 836 (plate 1, figs. 3 and 4) — apparently support the embryo and the wall of the exocoelom within the chorion.
Species.


No. of specimen.


We have in our collection an excellent embryo. No. 391. which is a Uttle larger than that described by Strahl and Beneke. This specimen came to us in formalin and was opened with great care. It was found that the embryo and appendages were suspended bj^ means of numerous delicate fibrils which radiated from them to the chorionic wall. As the sections were stained with cochineal, the fibrils do not show in them, so that this description is based entirely upon the apearance of the uncut specimen. In general the specimen appears to be normal.
Length of embryo


Width of area.


Our specimen No. 779, somewhat older than the one just mentioned, apparently contams no magma. It also was hardened in formalin. The ovum is entirety covered with vilU, which branch twice, are of uniform size, and appear to be normal. In the main chorionic wall there is a pronounced fold. The specimen was bent along the line of the fold, but the chorion was gradually dissected away with the aid of direct sunlight. The chorion is entirely Imed by a smooth membrane, and contains a cavity wliich is filled with a clear fluid and which apparently contains
Length of area.


Pig


98 107 144 119 106


1L» lUMAN MAt;MA UftTICULft IN NoHMAI, AND PATHOLOGICAL DEVELOPMENT.
mm.




no magma, ^\■ithm there is a clear, worm-like body, which is bent upon itself, with another body ari-sing from the middle of the bend. Apparently this is a flexed emI)ryo with the umbilical vesicle attached to it. The body is of uniform diameter, mea.^^uring less than a millimeter. We are probably dealing here with a normal ombryo. In opening this siiecimen great care was taken not to touch the embryo, so as to avoid injuring it. The embryo was taken out and cut into serial sections. It contains 14 somites and is without limb-buds. The sections give the impression that the embryo is pathological. There are no data in the history of the case which bear upon this point; therefore, for the present we may view it as a normal specimen without magma — or, if the embryo is taken into consideration, as a pathological specimen with dis.solution of the magma. Usually in pathological specimens the magma is greatly inerea.sed in quantity.
12
13
14*
14
14  




No. 104 is a somewhat older specimen. It came to us from an autojjsj^, with the entire uterus, and the sections of it indicate that the embryo is undoubtedly normal. The only record of the magma which we now have is given by several photographs which were taken at the time we received the specimen. These show a few strands of reticular magma, without any granular magma, radiating from the embrj^o. The i)hotograi)hs were taken while the sjiecimen was in formalin.
mm. 0.37 0.95 1.25 0.45 0.65




The next s]iecimen, Xo. 463, is somewhat more advanced in development and contains a flexed embryo, 3.9 mm. in length. The ovum is covered completely on one side, and partlj' on the other, with villi 1.75 to 2.75 mm. long. On the partlj^ co^•ered side the villi leave relativelj^ bare one area, centrally situated, measuring 8 by 4.5 mm. Over it the villi occur onh' here and there, about 2 mm. apart, and are branched and apparently normal. On opening the ovum the reticular magma is f(»und to (ill the exocoelom. By carefully exploring with fine tweezers, an ai)i)arently ncjrmal embryo is seen with a j'olk-sac measuring 3.5 bj- 4 mm. The embryo has anterior limb-buds and at least three gill-slits which are visible externally. No note was taken at the time regarding the condition of the magma, but sections of the entire chorion show that there is a very decided reticular magma between the embryo and the chorionic wall. There is no granular magma. The magma is composed mostly of fibrils, of much the same appearance as tho,sc of mesenchyme. Ik'tween the network of magma fibrils are denser strands accomi)anied by cells. In the fresh state imdoubtedly the denser strands would appear as filn-ils, while the rest would be transparent and jelly-like. The specimen came from a woman who was i)erfectly healthy and had given birth to 2 children durmg the last 4 years. This was her first miscarriage, and there was no indication of uterine disease.
7iim. 0.5 0.4 1.1 0.6 0.85




Specimen No. 486, of the same degree of development as the one described above, is in a perfect state of i;)re.servation, but there is no historj'^ which would indicate whether or not the specimen is normal. However, the chorion is covered with vilh about 3 nun. long, with a bare spot on one side about 4 nun. in diameter. The sections of the embryo do not show any attached fibrils of magma, but the chorionic wall, after hardening in alcohol, shows a decided layer of magma attached to it.
Pig




No. 470 is an interesting specimen, as it was found floating in a ma.s.s of bloodclotfi, which were sent to the laboratory in formalin. The ovum is covered with
1 mm.




121 1 16
576 1 17
108 18 (?)
45 18
84 1 22


HUMAN MAGMA RKTICUI.K IX NORMAL AND PATHOLOGICAL DEVELOPMENT. 13
normal villi and contains a well-formed embryo


mm. 0.6 1.5 0.8 0.9 0.8


within the amnion. It is apparently normal in every respect. No magma could be seen at the time, but drawings of the embryo subsequently made show delicate strands of fibrils forming a fuzzy laj'er around the umbilical cord and extending over the umbilical vesicle: undoubtedlj^ these are magma fibrils. This seems to be the normal condition for this stage and is verified in specimen No. 836, to be described later. Sections through the mass and the chorion, stained with carmine, show the magma as a granular mass; only at points is there any indication of fibrils. However, this mass resolves itself into the most definite fibrils when colored with Van Gieson stain, in ^lallorj'^'s stain, in hematoxylin, aurantia and orange G., or in iron hematoxylin. With Van Gieson stain the fibrils take on fuchsin color about as intenselj' as do the fibrils of the chorionic wall, with which thej' are continuous. The contrast obtained with ^lallory 's stain is quite marked, as the endoplasm of the mesenchj^me of the chorionic wall stains shghtly blue, while the exoplasm and the fibrils of the magma reticule remain unstained. This difference is not shown in sections stained in iron hematoxylin, as all fibrils are colored intensel}^ black. However, it does not come out with the Oppels-Biondi method or with hematoxylin and eosin or aurantia. As the fibrils of the magma are continuous with those of the exoplasm of the chorionic wall, which do not stain in jNIallory's connective-tissue uiLxture, they can not be considered as white fibers, and from their failure to stain in Weigert's elastic-tissue mixture thej' are not elastic. As will be shown subsequently, they give the reactions of embryonic connective-tissue s}Tic5'tium; and this is Retzius's opinion regarding their character. In specimen No. 486 the fibrils of the magma are not accompanied by any nuclei; so they must be viewed as belonging to the cells of the chorionic wall, from which they extend to bind the chorion with the primordium of the embrjo.


mm.


Specimen No. 588 came from a woman who had 2 children living, aged 14 and 20 years respectively. Since the last birth she had aborted 11 times, and in the opinion of her phj'sician all the abortions were due to mechanical means. This indicates that the specimen is normal. A figure of this embryo with strands of magma radiating from the umbilical cord and vesicle is shown in plate 3, figure 2.


0.48
0.9
0.8
0.4
0.7


Specimen No. 136 is of about the same stage of development as No. 588, although the chorion is covered with poorly defined viUi. For an embryo of this stage it is unusually small, and I have therefore Usted it with the pathological specimens in my paper on monsters. A photograph of the ovum after it had been cut open shows that the chorion is completely filled with reticular magma, so that the embryo


Rabbit
Human
Pig


is practically obscured. A block of the whole ovum encircling the embryo was cut in serial sections. These show that there are strands of tissue accompanied by cells which form partitions in the exocoelom. The quantity of the magma appears to be somewhat excessive for a normal o^'um of this stage of development.


Human
Sheep
Pig


No. 836, a perfect specimen containing an embrj-o 4 mm. ia length, settles definitely the condition of the magma at this stage of development (plate 1, figiu'es 3 and 4). In this ovum the exocoelom, measuring 9 by 4 mm., contains a dehcate spiderweb-like reticular magma, several of -the strands being considerably larger than the others. ^lost of this magma occurs between the yolk-sac and the amnion


Chick




14 III >tA\ MAC.MA UKTICn.K IN NORMAI. AND PATHoLOf.K'AI, nrAKI.Ol'MKNT.
Pig




Miul the adjacent ehoiionic \v;ill wlicic tlu' fibrils are unusually abuiulant. This speeiinen was ohtaiiieil from a hystereetoniy upon a woman, 2o years old, for a fibrous tumor of the uterus. She had been married 4 years, this being her first pregnaney. There were no special symptoms bearing upon the case, excepting the discomfort which accompanied the tumor of the uterus. Her last menstrual jx'riod had been delayed, and as it had been more jjrofuse than usual she believed that she h:id had a miscarriage; otherwise, everything ai^peared normal. This w-as confirmed by a careful examination of the specimen, which showed it to be normal in every respect. The uterus was opened by the surgeon at the time of the operation, but fortunately the site of the ovum was not injured. The specimen was sent to the laboratory innnediatcly. where it was fixed In- Dr. Evans, who niadc the following record:
In a rough way, then, we may consider the area membranacea as an oval; in some cases the longitudinal diameter exceeds the lateral, and in others the reverse holds. The measurements given above were taken from mounted sections and are probably somewhat disturbed by the histological tcchnifjue which was followed.
"Tfic specimen consists of a myomatous uterus which has l)ccu opened (apparently in a midhne anterior incision) so as to disclose an abundant deciduous endometrium thrown into large folds. At the upper posterior surface of the uterus an oval mass, about 251)y 20 by 20 mm., projects. It is a sjic and is covered with a rather smooth membrane (decidua reflexa), beneath which tortuous vessels are apparent. On one side the sac (tlie implanted chorion) is adherent to the uterine mucosa (decitlua vera). With a sharp .«cali)el the entire mass was dissected away fiom the uterus and brought under a iiinocular microscope in warm salt solution. The middle poitionof the free surface was o])cned carefully, beautiful villi being found, and then the delicate wall of the chorion was divided. Within, a transparent young embryo and its umbilical vesicle were seen, the embryo ajiiiearing to be about 5 mm. in length. Through this opening in the chorion, warm (40° C.) saturated aciueous solution of HgCl;, containing o per cent glacial acetic acid, was gently introduced and the entire mass placed in .500 c.c. of this fixation fluid. The main body of the uterus was dissected from the mj-omatous nodule andfixe<l in 10 per cent formalin, the site of the imjilanted ovum being marked by a short wooden rod."
The fixed and hardened specimen had undergone a readily appreciable shrinkage from the condition seen in warm salt solution. All of the tis.sues were beautifully preserved. The implanted ovum, covered with the decidua capsularis, measures approximately 22 by 18 by 11 mm. The adjacent decidua parietalis is thrown into large folds, which are themselves marked by numerous tiny elongated crack-like depressions, as well as by more circular pit-like apertures. The relatively smooth but irregular surface of the decidua capsularis is marked here and there by very consi)icuous, small, oval pits, which may attain 0.5 mm. in diameter. The four f1a|)s f»f this coat at its highest point, where it was oi)ened dir(>ctly over the middle of the ovum, an; rather smooth on their iimer surface an'H stand ai)art from the subjacent chorionic villi (intervillous space) to which they were originally adherent. Tlie villi are about 2.5 mm. in length and possess one or two large branches and many ".stub-like" tiny bulbous ones on the main stem. The villi are imiformly distributed in the small area exposed. With a slender .scalpel the ovum was carefully divided under the dissecting micro.scope, the embryo and yolk-sac being visible. TIk' yolk-sac appears to be almost 2 cm. in diameter and the embryo is surroimded by its amnion, its head (visible from above) being about 3 cm. in length and showing the fourth ventricle covered by a transparent ependyma. Two gill-arches are visible. The yolk-sac surface presents an exquisite picture of irregidar. clear va.scular channels tuid a uniform p.itfern of small, opa(|ue. white blood-islands. The ])re.servation seenis perfect.




* MuisurecI on alidc uftpr iiiouDting.


The borders of this oval area membranacea are usually fairly regular and smooth, but in some instances they are irregular, due to the fact that small extensions of the area run into the bordering ependyma. These extensions are more commonly met with at the stage when the area has reached its maximum size, as in figures 38 and 39, photomicrographs from an embryo pig of 19 mm. The higher power of these two photographs shows two areas in the smoother ependymal wall. These are extensions of the area membranacea, and within a section or two directly connect with the differentiated area. Both of these small spots on the circumference resemble technical errors; their ragged appearance, the relative excavation of their surface, and the intact ependymal borders w'ould seem to encourage such a view; but when considered in connection with the character of the whole area membranacea they assume a definite relationship in this regard. Other similar areas, rather rare in occurrence, are found separated entirely from the main area membranacea. These isolated areas are of the same size as those shown in figure 39. In significance and character they are probably identical with the larger area membranacea superior.




HUMAN MAGMA RKTICULK IX NORMAL AND PATHOLOGICAL DEVELOPMENT. 15
Most of the general features of the area membranacea superior have been commented upon in descriptions of the various stages of differentiation in both pig and human embryos. The characteristics most commonly observed concern the differentiated character of the cells of the area, the sharp borders of the typical ependyma, the ragged ventricular surface throughout the whole extent, and the peculiar adhesion of the albuminous coagulum from the embryonic cerebro-spinal fluid to the lining cells. The area membranacea superior should be considered, then, as a transitory focus of differentiation of the typical ependymal hning of the roof of the fourth ventricle.
After the embryo had been carefully removed, the ovum was cut into blocks which included its implantation. A block 1 mm. thick, which included the largest circumference of the embryo




, was embedded in celloidin, the sections being stained in various waj^s. A photograph of this block is represented in plate 1, figure 4, which shows strikingly the extent of the magma. Sections which have been stained in hematoxylin and aurantia show the magma much as it appears in the other embryos that have just been considered. There is a denser magma just under the chorionic wall, and heavy strands radiate in every direction, with a fine network resembling spider-web, among the main strands. A number of loose nuclei accompany these strands, but they do not have the appearance of the nuclei of the main wall of the chorion. They are mostly round and are of unecjual thickness, simulating very much the blood-cells. Occasionally there is a large nucleus. Sections which have been treated by the Weigert fibrin method do not show these fibrils. This confirms a previous experience which I have published elsewhere in my paper on monsters, namelj'^, that magma fibrils do not give the reaction of fibrin, nor do these fibrils stain well in Van Gieson's mixture; however, they take on color similar to the mesenchj'me of the chorion.
===An Undescribed Area In The Inferior Portion Of The Roof Of The Fourth Ventricle===




At points it appears as though these fibrils arise dhectl}' from the chorionic wall. Thej^ stain intensely blue by the ^lallorj^ method, and in sections treated in this way the nuclei of the mesenchyme of the vilU look much like the accompanj'^ing nuclei of the magma fibrils. On one side of the ovum a denser mass of the magma is directly continuous with the mesenchyme of the chorionic wall. However, just in this region the magma contams no nuclei. It, therefore, appears that the magma fibrils must be associated, at least partly, with the nuclei of the chorionic wall. Exceedmgly good histological pictures were obtained from sections stained by Heidenhain's method, which show all the transition stages between magma containing no nuclei and magma very rich in nuclei. It would seem that there is quite a free wandering of the nuclei along the magma fibrils, and whenever they come in contact with the chorionic wall the fibrils enter it, showing direct continuity. The most instructive specimens are obtained by the Weigert elastic-tissue stain, which gives a sUght blue-black tinge to the mesenchj^me fibrils of the chorionic wall, as well as to those of the centers of some of the villi. The magma itself takes on a very light stain, but where it is in contact with the chorionic wall it grades over into its blue network. It appears, then, that the centers of the villi, which represent their older portion, stain somewhat with elastic-tissue stain; and, if we view the chorionic waU as the more differentiated portion of the chorion, we must conclude that the older mesenchyme fibrils behave more like elastic-tissue fibrils than do the j^ounger. At any rate, the magma fibrils do not take on elastic-tissue stain.
With success attending the effort to find in the superior portion of the rhombic roof an anatomically differentiated area which would furnish a morphological basis for the jihysiological phenomenon of the extraventricular passage of the cerebrospinal fluid, attention was necessarily directed to the inferior portion of this roof (considering the whole roof structure to be divided by the chorioid plexuses). The spread of the replaced injection fluid (fig. 4) into the periaxial tissues through two points in the roof of the ventricle suggested a study of this stage (pig embryo of 18 mm.) as the basis of the investigation. As a histologically differentiated area in this inferior portion of the roof is easily made out, the complete history of the area will be given chronologically. It has been termed the "area membranacea inferior ventricuU quarti," the terminology being based on the same physiological and anatomical features which led to its adoption in the case of the analogous area in the upper portion of the roof.




From all that has been said it is clear that the mesenchyme of the chorionic wall and the magma fibrUs are continuous and, as I have pointed out elsewhere, they together form a common syncytium. I have already demonstrated that verj' young connective tissue arises directlj' from the mesenchyme, the earlier stages of which I have designated r^s the connective-tissue syncj'tium. Towards digestive reagents the connective-tissue syncj'tium gives somewhat the reaction of j-ellow elastic tissue, just as do the mesenchyme and the magma of No. 836 when treated with Weigert's
===The Area Membranacea Inferior In The Pig Embryo===




The inferior portion of the fourth ventricle shows no evidence of a differentiation from the typical lining ependyma until the length of 15 mm. is reached. In this development consideration must be given to the factors concerned in the process. It will be recalled that in the younger embryos, both pig and human, up to and including a length of 9 mm. the whole roof of the ventricle is occupied by the epithehal-like cells. With rapid growth of the medulla and corresponding enlargement of the fourth ventricle the roof becomes elongated and widened. This process results in the isolation of the area composed originally of the ejiithelial-like cells and the subsequent formation of the superior membranous area. The epithehal-like cells remain in the superior portion of the enlarged ventricular roof, while the whole inferior half is composed of the densely staining, typical ependyma. The division of the roof by the laterally developing chorioid plexuses becomes evident in pig embryos of 14 mm. At this stage the whole inferior portion shows a ventricular lining composed of the typical ependyma.


1() urMAN MAdMA nfnicn.f: ix normal and pathoi.ooicat. devklopment.


The first indication of a differentiation in this inferior half of the roof was found in a pig embryo of 15 mm. This is illustrated in figures 70 and 71. The sagittal section from which these photomicrographs were taken is near the mid-line of the embryo, as is indicated by the partial section of the central canal of the spinal cord (fig. 70). The division of the ventricular roof into two parts is also indicated in figure 70 by the invagination of the chorioid plexus. The squared area in the lower half is reproduced in figure 71 under higher magnification; here the first evidence of an ependjonal differentiation is observed. The dense Une of the typical ependjona appears from both sides, but in the center of this ventricular lining a small area of differentiation is seen. This area, isolated by the abruptly terminating pyknotic ependymal elements, is composed of two or three layers of less deeply staining cells. The nuclei are round, rather larger than those of the adjacent mesenchyme, and contain httle chromatin. The cytoplasm stains fairly well with eosin and is not scanty in amount. The cells resemble those epithelial-hke elements which so largely make up the ventricular roof in the earlier stages. No albumen is found near this point of differentiation, although the whole ventricular cavity is filled with the normal amount. In figure 70 the marked zone of the area membranacea superior may easily be seen.


I'la.stic-t issue stain. 1 have also shown that tho younger the connective-tissue syncytium is. tlie more ilifficult it is to digest it in jx'jjsin. Frozen sections shrink Injt little when treated with acetic acid, while white fibers become transparent. The syncytium itself is somewhat elastic, as shown by pressure upon the coverglass over a frozen section. If treated for 2-i hours with pepsin, the fibrils disintegrate. They are tlierefore much more resistant to the action of pepsin than are white fibrils.


After this initial indication of a differentiation in pig embryos, the further differentiation of the tissue proceeds but slowly until the length of 18 mm. is attained. Thus, in a similar specimen from an embryo pig of 18 mm. the area of differentiation is not greatly hicreased in size. This is shown in figures 72 and 73. In the higherpower figure (fig. 73) both the superior and inferior membranous areas can be made out by the attachment to these areas of the protein coagulum of the ventricular cerebro-spinal fluid.


The action of pancreatin is, in a measure, the opposite of that of pepsin. \Mien the main mass of syncytium is formed by exoplasm, it digests readily in pancreatin. The more the syncytium is developed, the more resistant it is towards pancreatin. \"ery young syncytium fibrils, therefore, react towards pancreatin and jiejisin much like elastic fibers and this is confirmed in a measure, by tinctorial methods, when applied to sections of the chorion and magma, in specimen No. 836.


In the higher-power figure (fig. 73) of the squared area from figure 72, the area membranacea inferior shows the same character as exhibited by the specimen of 15 mm. (fig. 71). The opening maintains the same approximation to the lateral lip of the medulla, but the area is larger and the histological character more nearly approaches the permanent feature of the tissue. The nuclei in this zone are paler than those of the adjoining ependymal elements and contain less chromatin. The cytoplasm is not scanty, nor is it very abundant in amount. The area is also characterized by the occurrence of the cells in a layer, two or three cells in thickness.


I have discussed the denser strands of tissue within the main mass of the magma. In the fresh state it appears that these are distinct fibrils, as shown in l)late 3, figure 2. They are, also, observed in plate 1, figure 3. It is not quite so dear that there are fibrils in the magma as shown on plate 1, figure 4. In fact, it ai)pears as though we have compartments separated by membranes, and that at the junction of several of these membranes the fibrils become denser, and therefore often appear as distinct fibers. It would be more appropriate than to state that the exoccelum is broken up into compartments the walls of which arc composed of membranes, and that where .several of the membranes come together the increased amount of tissue gives the point of juncture the appearance of fibers to the naked eyi' and under the enlarging lens.


In view of the very slow differentiation of the area membranacea inferior in the growth of the embryo from 15 to 18 mm., the enormous enlargement of the region within the next few millimeters growth is very astonishing. This period, as has been jjointed out, is a critical one in the extension of the embryonic cerebro-spinal fluid from a ventricular to a periaxial relationship. Apjjarently, in the course of the embryo's growth during these next few millimeters the whole inferior roof of the ventricle undergoes a transformation and enlargement, so that the differentiated area membranacea comes to occupy practically the whole inferior half of the roof. This portion of the roof, persisting, enlarging, and suffering no extension of nervous t'ssue upon it, becomes the tela chorioidea inferior.


I liavc taken great pains to follow the cells which mark the stronger bands of magma, and it is difficult to arrive at any conclusion, for, in a measure, they seem to be related to the endothelial lining of the exoccelom. In the Peters ovum the spaces near the embryo are lined by a distinct laj'er of cells, but otherwise there is no indication of endothelial lining in any other portion of the chorionic cavity, nor is there any indication of such a lining in the figures given bj' Herzog, Johnstone, Jung, or Strahl and Beneke. It would seem that what corresponds to the exocoelom of the chorion in the later stages is represented by a diffuse mass in the specimen of lirycc and Teacher where the nuclei are scattered through it. The mode of the destruction of the mesenchyme is well indicated in figures on page 18 of a monograph by Strahl and Beneke. These irregular cells are first of all attached to the heavier strands of magma, and they must, therefore, correspond to the endothelial lining of (he exoca'lom. For the present, however, it appears as if the exoccrlom of the human chorion is lined only in part by a layer of endothelium; these cells al.so nccomi)any the magma fibers and line the inner side of the chorion near the embrj'^o.


The rapid differentiation of the whole iiiferior half of the roof of the fourth ventricle is a very interesting process. Apparently the typical ependymal elements, visible on both sides of the membranous area in figure 73, undergo a very rapid alteration, so that in the course of a few millimeters' growth the cubical lining of the ventricle is replaced by a low- type cell, with round or oval nuclei, staining much less densely than do the ependymal elements. The whole area membranacea rapidly becomes a membrane in the true sense of the word; it is a continuous, intact laj'^er of cells, generally only one cell in thickness, closing in the fourth ventricle from the chorioid plexus above and the bulbar lips on the sides.


As tlie amnion expands, it naturallj^ j)ushes these strands of magma up against the chorion, and in a short time we can recognize only a few fibrils in the exocoelom whicli encircle the uml>ilical cord. The.se are well seen in si)ecimen Xo. 14S, and their remnants are .shown in Xo.'jTO, of which I give an illustration on plate 2, figure 2. Xo. 14S is vmdoubtedly normal, for it was obtained by mechanical nieans. and Xo. .')7fi is also a normal spccinu-n obtained from a tub.al pregnancy.


The general characteristics of this transformation are seen in figures 74 and 75. These photomicrographs are taken from a sagittal section of a pig embryo of 23 mm. On one side of the sharply deUmited membrane shown in figure 75 is a tongue of nervous tissue of the medulla; on the other is the differentiated ependjina of the chorioid plexus; between these two structures stretches uninterruptedly the area membranacea inferior. The flattened cells of the membrane, with their oval nuclei and almost continuous cytoplasm, effectually close the whole ventricle. The photomicrograph also shows an interesting characteristic of this membranous area which is universally present in the larger forms; this is the relatively unsupported character of the membrane. The highly vascular mesenchyme posterior to the area has gradually developed, during growth, larger and larger interstices between the cytoplasmic processes. The phenomenon is not due to shrinkage, but is intimately connected with the formation of the future cisterna cerebello-medullaris. This phase of the mesenchymal differentiation will be more fully considered in an appropriate section of this paper. It will suffice here merely to record the lack of support of the membrane.




Another phenomenon of unportauce in the cerebro-spinal fluid relationships of this stage is shown in figure 75. In the mesenchjanal spaces directly beneath the membranous area there is a large amount of albuminous coagulum. This phenomenon does not occur to any appreciable extent in earlier stages or in other parts of the mesenchyme, except about the nervous system. The c1o.se association of the coagulum from the ventricular cerebro-spinal fluid with the inner border of the area membranacea (shown in figure 75 as a slight roughening of the border) is of very great significance in this connection. In one i)oint in the membranous area (fig. 75) the albumen can be traced almost without interruption from the ventricle into the wide spaces of the mesenchyme (cf. fig. 8). This observation strongly suggests that the embryonic cerebro-spinal fluid, which is rich in protein material, is passing, in this stage of embryonic growth, from the ventricle into the periaxial mesenchyme; and such an interpretation becomes established by the comparative findings in the embryo of the same stage in which a replacement of the cerebro-spinal fluid by the ferrocyanide solution had been effected. These comparable findings are surelj^ of the utmost importance for the final solution of the problems centering about the embryonic cerebro-spinal fluid.




lUMAN MAGMA RETICILK IX NORMAL AND PATHOLOGICAL DEVELOPMENT. 17
In the later stages of development of the area membranacea inferior in the pig embryo the same structural relationships persist that are shown in figure 75. Figures 76 and 77 are photomicrographs taken from a sagittal section of a specimen of 32 mm. In the enlargement of the squared area, from the first of these figures, the continuity and completeness of the membrane are well established. The photograph shows well the flattened character of the cells comprising the membrane and its sharp differentiation from the nervous tissue and ependyma below and from the ependj'ma and chorioid plexus above. Most important in this case is the distribution of the albuminous coagulum. Within the ventricular cavity this appears in considerable amount, and in several places it is in close adhesion to the lining area membranacea. This albuminous precipitate may likewise be traced in some places apparently through the cellular membrane into the periaxial spaces. For here, as indicated in figure 75, the clotted albumen from the cerebro-spinal fluid apjjarently exists in large amounts in the space just posterior to the membrane — the future cisterna cerebello-medullaris. Delicate strands of mesenchyme are still observed running through the wide space, but in general the whole tissue has returned to the line of the future arachnoid. The relative lack of substantial support of the membrane is well brought out in figure 77. A characteristic feature of this membrane, which Blake'3^ has championed, and which is indicated in figures 76 and 77, is the posterior bulging of the roof — "the caudal process like the finger of a glove."
The conclusion regarding the condition of the magma of normal development is that the cavitj^ of the ovum is filled with delicate fibrils which are intorpersed with numerous nuclei and which form one continuous network, extending from the embryo to the chorionic wall, and blending with its connective-tissue network. It forms one continuous syncytium, and as the ovum grows the magma reticule differentiates .somewhat. Stronger bands of membranes soon form, breaking the cavity of the chorion into compartments. This process continues until the amnion begins to expand, and then these fibrils are pushed up against the chorionic wall. The exococlom begins as two larger spaces near the embryo, and in this portion of the ovum its cavity is lined with a layer of endothelium. It is quite certain that this sharply defined cavity does not extend to include the whole cavity of the ovum, but the cells lining it arise in common with those which accompany the magma fibrils. The exact extent and the fate of the two small spaces near the embryo in the Peters specimen is still undetermined, but Waterston's specimen indicates that they do not extend to fill the entire chorionic cavity. The examination of numerous specimens, however, indicates very definitely that the exocoelom of the ovum at 2 months does not contain a complete endothelial lining.




THE MAGMA IN PATHOLOGICAL OVA.
Another section from the same pig embryo, taken more laterally, is represented in figures 78 and 79. In the photomicrograph of higher power the flattened character of the lining cells, the intactness of the membrane in isolating the ventricular cavity, the unsupported freedom of the membrane, and the relation to the albumen coagulum on both sides are of |)articular interest.




Since the publications by Giacomini it has become well known that an increased quantity of magma within the coelom indicates with certaintj' that the embryo
The ultimate fate of the area membranacea inferior will not be more fully entered into until the early history of the similar area in the human embryo has been detailed. For in this connection the occurrence of the foramen of Magendie requires discussion, and it seems best to delay the further consideration of the present topic until the whole question can be reviewed.




is pathological. When the magma is pictured or described, it is quite easj'- to determine whether or not the embryos and ova published in the literature are normal or pathological. This is demonstrated in the plates accompanying Yelpeau's work. His was able to separate most of the normal from the pathological embryos, but he relied mainly upon the external form of the specimens, which he compared with other mammalian embryos. Unless an embryo
THE AREA MEMBRANACEA INFERIOR IN THE HUHMN EMBRYO.




appeared much like those of other mammals and was not transparent and sharply defined, he decided that it was not normal but pathological. The work of Hochstetter, who Kmited his studj' to embryos obtained through hysterectomy, has been used to advantage by Keibel and Else in the preparation of their Normentafel, so that now we have adequate tables and plates which enable us to recognize with considerable certainty whether or not an embryo is normal, without paying much attention to the magma or the chorion. However, embrj^ologists are well aware that they can predict whether a specimen is normal or pathological by the quantity of the magma which masks the embryo when the ovum is opened.
The same process in the formation of an area of differentiation in the inferior portion of the roof of the fourth ventricle may also be followed in the human embryo. Unfortunately, however, human omliryological material can rarely be subjected to the immediate fixation and preservation which j'ield excellent histological results in the more plentiful specimens. It does not seem strange, therefore, that the determination of the exact stage at which an area of differentiation can be made out in the ventricular roof should be practicall}' impossible; for, in poor technical procedures, the roof of the fourth ventricle suffers almost more than does any other portion of the specimen.




By the contents of the exocoelom it is quite easj' to classify pathological ova into three chief groups. In the first group, which includes most pathological specimens, the magma is changed into an organized mass of reticular fibrils, intermingled more or less with granular substance.
In a human embryo of 13 mm. (No. {{CE695}} in the collection of the Carnegie Institution of Washington) there is slight evidence of a differentiation in the lower portion of the rhombic roof. The changing character of cells in this specimen is not marked, but as the central portion of this inferior roof is reached the ependymal cells seem to assume gradually a more cubical morphology. Associated with this change in shape, there is also a slight loss of the deeply staining character of their nuclei. The whole differentiation, however, is slight and would be commented upon only from the conception of this area in the pig embryo.




To the second group belong specimens in which the exoccelom is large and contains only a fluid mass — that is, a liquid substance which does not coagulate in either formalin or alcohol. I have pictured a number of specimens of this sort in my paper on monsters. Specimen No. 512, of which I give an illustration on plate 2. figure 1, belongs to this group. The embryo is atrophic, and it is questionable
The first definite evidence of differentiation in the inferior portion of the ventricular roof was found (specimen {{CE390}} in the Carnegie collection) in a human embryo of 15.5 mm. This initial differentiation occurs, then, in the human embryo of approximately the same length as in the pig. The specimen showed the same change in character of the hning ependj-ma as was found in the pig. The deeply staining ependymal elements are replaced in a limited central area in the inferior portion of the roof by cells with more elongated nuclei, poorer in chromatin, and resembling somewhat the epithehal-like cells which early filled the ventricular roof. These cells tend to compose a layer of more than one cell in thickness — a feature particularly noticeable in the peripheral portions.




The size of the area membranacea inferior observed in specimen {{CE390}} suggested that the earUest evidence was probabh' to be observed in somewhat smaller specimens. This could not, with the material at my disposal, be verified, but it is probablj^ safe to assume that the first signs of an ei)end\Tnal differentiation will be found in human embryos of about 15 mm. This time of appearance of the area in the human would coincide with its time of primarj* differentiation in the pig embryo. In this limitation of the first appearance of the area membranacea inferior, the standard has been an unmistakable differentiation of ependyma and not an isolated change of a lining-cell or two which might have been the result of the technical procedure. Such a criterion was necessitated by the verj' marked changes in the ventricular borders observed in specimens in which distortion of the chorioidal roof had occurred.


18


The area membranacea inferior very rapidly increases in extent after the onset of the process of ependymal differentiation. This was hkewise observed in the pig embryo, although perhaps more stages could be made out. In a human embryo of 16 mm. (No. {{CE406}} of the collection of the Carnegie Institution) the area nierabranacea inferior is quite extensive, as is shown in figures 80 and 81. In the photomicrograph under higher power (fig. 81) the denselj' stained ependyma approaches the membranous area (ami) as tongue-hke processes from above and below. These tips gradually lose their dense character and are prolonged as a delicate membrane, lining, in this localized area, the ventricular cavity. The nuclei of the cells here are not heavily laden with chromatin; they are oval and somewhat larger than the more densely packed nuclei of the typical ependj-mal element. Unfortunately, the middle portions of the membranous area in this specimen are surrounded by extravasated red blood-cells obscuring somewhat the structure (fig. 81). The process, though, of the differentiation of these ependymal elements into paler and larger epithelial-like ceUs is quite apparent.




iiiM AN M.\f;M.\ nf'.Ticn.ft ix ndhm.vi. .\xn patholooicxi, dicvelopment.
As in the pig, the tendency of the differentiated ependymal cells forming the area membranacea inferior to lose in some degree their distinctive appearance and to approach in character the undifferentiated mesenchymal element is apparent in the human embryo very shortly after the original steps in the process of differentiation have occurred. Photomicrographs from two human embryos of 17 mm. have been included to show this phenomenon. Thus, in figure 88, an enlargement of the blocked area from figure 58, the area membranacea inferior (ami) is well defined. The sagittal section from which this photomicrograph was taken is from embryo No. {{CE576}}, in the Carnegie collection. Above and below the dense fine of ependj-ma may be made out; this tapers quite abrupt!}', to be succeeded bj' the cells of the area membranacea inferior. These cells, products of ependymal differentiation, have lost much of their epithelial-like appearance; they now show rather small, oval or rounded nuclei, poor in chromatin. The cytoplasm of the cells is small in amount, but not disproportionate for the size of the nucleus. The ventricular border of these cells (fig. 88) exhibits a rather characteristic phenomenon, the adherence of a shght albuminous coaguluni. The fine processes of this coagulum fuse^ith the cytoplasmic borders of the cells and render these borders vague and indefinite. Beneath the cells of this inferior area small vascular channels may be made out. These tend to make the membrane appear denser than its cellular character warrants.




In another section from this same embryo (No. {{CE576}}) the inferior membranous area is shown in relation to the tufted chorioid plexuses (figs. 82 and 83). In the reproduction under higher magnification (fig. 83) the ependymal lining may be traced caudalwards to a gradual fusion into the area membranacea inferior. From the rather high cubical cells in the immediate proximity to the plexuses the ependymal elements become reduced in size and in height, and then rather abruptly the pyknotic character of the ventricular lining is lost. This loss of the deeply staining character coincides with the superior border of the area membranacea inferior {ami). The membrane of this area shows the same cell-character as already desorilied for this embryo. On the superior side of the plexuses (fig. 83) the lateral border of the area membranacea superior (ams) is shown composed of epithelial-like cells.




The apparent tendency of the cells composing the inferior membranous area to lose the epithelial-like character, as shown in the figures from embryo Xo. 576, is not an invariable phenomenon. Rather is an aggregation of epithelial-like cells met with in human embryos very commonly in this area, not onlj- in embryos of small size, but also in small fetuses. This phenomenon is illustrated in figures 84 and 85, reproductions of photomicrographs from a human embryo of 18 mm. (No. {{CE409}} in the collection of the Carnegie Institution). In figure 85 the total transverse extent of the area membranacea inferior (ami) is illustrated, with the villous chorioid plexuses appearing to the left. Although this membranous portion of the embryo has been distorted somewhat by the technical procedures to which the specimen was subjected, the cellular character of the membranous area is well indicated. The most striking feature, apart from the characteristic tinctorial differentiation from the typical ependymal elements, consists in the marked clumping of the cells in certain parts of the membrane. On one lateral extent the membrane is thickened into a bulbous swelling several cells in thickness. These cells have palely staining nuclei, poor in chromatin, with an oval or round form. In other places in the membrane smaller but no less characteristic clumps of similar cells maj' be made out. Between these cellular aggregations the membrane stretches in a continuous line with but few nuclei.


whether or not if is oiicircled l)y tlic amnion. In these specimens the ocrlom is usually enlarged and sometimes it is greatly distended. Often there is a small granular precipitate in older specimens, but this is not of sufficient (juantity or density to form a continuous mass. The histories of these specimens show that they are consideral)ly older than their sizes indicate, and I am inclined to view them as having once had a dense mass of magma within the coelom, which subsequently underwent dissolution, leaving a more or less flaky deposit that finally disajiijearcd altogether.


Analogous clumps of cells, with pale, rounded or oval nuclei, may be made out in figures 86 and 87, taken from a human embryo of 19 mm., No. {{CE431}} in the collection of the Carnegie Institution. Only a small portion of the membrane is reproduced in the figure under higher magnification, but a characteristic clump of epithelial-like cells (epc) is shown. These cells of the differentiated ependyma here again have oval and rounded nuclei, poor in chromatin, similar to those which have been pointed out manj^ times in the foregoing pages. A second broadened area in the inferior membrane is also showm in figure 87.


In the third group, the coelom is greatly distended, the ainiiioii is usually absent, aiul the ovum is tilled with a gelatinous substance. This is well illustrated by specimen No. 604, plate 1, figures 1 and 2.


The further development of the area membranacea inferior proceeds in the human embryo in a manner very similar to that described for the pig. In the stages but shghtly above those already described the differentiation goes on slowly, but withui a few millimeters the cellular pictures resemble those given for the embryo of 17 mm. (figs. 82, S3, and 88). The cellular clumps which appeared quite frequently in the embryos under 20 mm. have not been found in the larger forms. Thus, in an embryo of 23 mm. (No. {{CE453}} in the collection of the Carnegie Institution) the inferior membranous area {ami) appears as an extensive membrane comprising almost wholly the inferior portion of the chorioidal roof. The membrane is here of a single cell in thickness; these cells are rather small, with oval nuclei, simulating in some measure those of the surrounding mesenchj-me. The most mteresting phase of the membranous area at this stage of 23 mm. concerns its completed cellular differentiation and its rather slow increase in size.




Wholly similar pictures of the inferior membranous area of the roof of the fourth ventricle are afforded by a human fetus of 26 mm. (figs. 91 and 92). These photomicrographs were taken from embryo No. {{CE1008}} in the collection of the Carnegie Institution. In this specimen (fig. 92) the fourth ventricle seems almost to lack a lining of oj^endymal (epidermal) elements in the area membranacea inferior (ami). The cells of this area are small, inconspicuous ua their distinctions from the underlying mesenchjme. The whole character resembles that of the superior area membranacea shown in figure 57.




The appearances exhibited by the inferior membranous area in the stages above 26 mm. are modified in great part by the development of the great cisterna cerebellomedullaris. As in the pig, the breaking-down of mesenchyme to form this cistern results finally in the almost total isolation of the inferior membranous area. The cistern is fairly rapidly formed when once the process begins, and so in an embryo of 35 mm. (No. {{CE199}} in the Carnegie collection) the isolated character of the area membranacea inferior (a?ni) may be easily made out. This is shown in figure 94, an enlargement of the blocked area in figure 93. The general architecture of the membrane, particularly its intact character, appears in this photomicrograph, but its finer structure is obscured by the albuminous coagula which adhere on both surfaces. The cell structure of the area membranacea resembles closely that described in the embryos already pictured.




Discussion of the final disposition of the area membranacea inferior will be undertaken in the following subdivision of this paper, in order that the findings in the pig and in the human embryo may be correlated.


===General Consideration of the Area Membranacea Inferior===




The ependymal lining of the caudal portion of the roof of the fourth ventricle undergoes a process of differentiation which results in the formation of the area membranacea inferior. This transformation has been observed in pig and human embryo


T.\BLE 2.— List of specimens containing pathological magma.


s; in both, the first definite evidence of the cellular change has been observed in specimens of 15 mm. The essential phases of the process are identical in the two embryos. The tendency of the deeply staining typical ependymal elements is to lose their highly pyknotic character; the nuclei become poorer in chromatin and the cytoplasm somewhat more abundant. In the first stages of the metamorphosis the lining cells come to assume epithelial-like appearances, but in the final change the nuclei become small oval bodies, poor in chromatin, resembling to some degree the nuclei of the adjoining undiff'erentiated mesenchyme. In the human embryo, a tendency for the epitheUal-like characters to persist in isolated cellular aggregations is apparent.




After the initial process of differentiation has begun, the area membranacea inferior increases rapidly in extent and the differentiated cells which characterize it come to occupy the greater portion of the caudal part of the chorioidal roof. In the somewhat later stages the area membranacea is almost wholly unsupported by other tissues, due to the development of the cisterna cerebello-medullaris. As soon as the cistern forms, the area membranacea serves as practically the sole dividing membrane between the ventricular system and the future subarachnoid spaces.






Cat No.
The ultimate fate of this area membranacea inferior is necessarily involved in the distribution of the tela chorioidea inferior. Likewise it necessitates a discussion of the possible formation of the so-called foramen of Magendic and its mode of origin from the "caudal process" of Blake. It is proposed to discuss briefly some of these questions in the hope that some phases of the problem may be brought forth.




It must be clearly understood that the questions of the ultimate fate of this area membranacea inferior probably differ considerably in the different species of mammals. In the horse and in the pig the absence of the medial foramen (]\Iagendie) is fairly well established, but in man its existence .seems to rest on equally firm grounds. While, ))rimarily, this investigation has not been concerned with the possible existence of the foramen of IMagendie, the question has been presented manj' times in regard to the pig and human embryos examined.


Lcneth
of ciiibrj'o.


As far as can be determined, no descriptive study of the development and differentiation of the inferior portion of the rhombic roof has been published. Heuser's'23' studies on the form of the cerebral ventricles of the pig have afforded a very good conception of the gradually changing relationships in this region. Hess(22j has devoted attention to the histological appearances of the inferior roof in the embryo. One of his interesting obseivations concerns the caudal portion of the rhombic roof in a fetal cat of 10 cm., where he noticed a very sudden interruption in the epithelial lining of the ventricle, with a complete closing by a fibrous net. This description by Hess is the only comment upon the histological appearance of the ventricular roof that has been found. His^^s^ pictures, without comment, in a retouched jihotomicrograj^h, a differentiated area in the proper situation in his fetus C-1 (beginning of the third month) .




Dimensions of chorion.
The many writers in embryology have commented upon the roof of the fourth ventricle. Minot, in 1892, stated regarding it:
:"Several writers have thought that the membrane was broken through at several points, but it probably is really continuous throughout life. The fourth ventricle is to be regarded, then, as an expansion of the central canal permanently bounded by the original medullary walls."


Kollman (32) on the other hand, advances the view that during the third month the rhombic roof is broken down to form the foramen of Magendie and the two foramina of Luschka. Streeter^S'*), in his chapter on the development of the nervous system in the Keibel-i\Iall Handbook of Embryology, advances a similar view. The majority of investigators to-daj' incline to the beUef that the roof of the fourth ventricle in man is perforated to form the median foramen of ]Magendie.




Menstrual age.
jjpgg (22) has advanced a conception of the foramen of Magendie that is supported by numerous observations. To test KolUker's statement that the fourth ventricle remained closed during human embryonic life. Hess sectioned the region in human fetuses, new-born infants, and in adults. The lengths of the fetuses cut were as follows: 7, 12.5, 15, 16, and 17 cm. In the 47 cases the roof showed a medial opening (IMagendie), except in one case, in which it was closed by a "thin pial membrane." Hess's conception of the process of formation of this membrane was that in earh' embn'ological life the rhombic roof was bordered by a regular, meshed tissue. Later the small meshes in this tissue fused to form the larger foramen of Magendie.




Blake's (2) hypothesis of the formation of the medial foramen has been quite extensively quoted in the more recent publications on this subject. In a study of the chorioidal roof Blake found a caudal bulging of the inferior velum; this outpouching became more and more extensive in the older embryos.. In man this pouch became sheared off at its neck, leaving the foramen of IMagendie.


Contents of osTim. '


In addition to the few studies referred to above, there have been in the past 25 years a great mmiber of articles (notablj' those of Wilder^^s) and Cannieu^^^) offering evidence that this median foramen of the fourth ventricle is an existent, functional opening. Into this literature it is not proposed to go in the present communication; it may be stated that in the larger part the views presented have been in favor of the consideration of the true occurrence of the foramen of IMagendie.


Cat. No.


The material on which this study is based has been purely embryological in type, so that no relialile data regarding the foramen of IMagendie could be obtained. But even in the largest fetuses examined, there was no evidence which indicated a breaking-down or a shearing-off of the inferior roof of the fourth ventricle. In the largest human fetus at my disposal, in which the histological material was good enough to permit an accurate examination of the chorioidal roof (embryo No. 448, 52 mm. in the Carnegie collection) the area membranacea inferior appeared as an intact membrane supported only be a few pial cells. In the pig the material at hand has been such that accurate study of the roof could be made in specimens up to 20 cm. ; in all of these later fetal pigs the roof has been wholly without foramina. If, however, in these larger stages the histological procedures have not been of the best, ruptures and other artificial separations are very frequently found.




Length
The area membranacea inferior, then, may be regarded as a region of ependymal differentiation. Whether it persists as an intact membrane or undergoes, in certain animals, a perforation to form a foramen of Magendie can not be here answered; this study has been concerned solely with the embrj^ology of the cerebro-spinal spaces, and it affords no evidence in favor of or against the existence of such a foramen. Nor has any study been made of the two foramina of Luschka, the two openings from the lateral recesses of the fourth ventricle into the subarachnoid spaces. It can Ije stated, however, that these foramina arc not in existence at the time of estal)lishment of the circulation of the ccrebro-si)inal fluid. This phenomenon, as recorded in the previous section, occurs in pig embryos of 26 mm.; at this time the lateral reces.ses are anatomically and physiologically closed.
of embryo.


==VI. Passage Of Fluid Through Roof Of The Fourth Ventricle==




Dimensions of chorion.
On pages 20 to 'SO is a description of the passage of a true solution, substituted without increase in pressure for the embryonic cerebro-spinal fluid, through the roof of the fourth ventricle into the extravcntricular or periaxial spaces. This extension of fluid occurred in two localized areas, one in the superior half and the other in the inferior half of the rhombic roof. Histological study of these regions revealed a localized differentiation of the ependyraa, both in the upper and lower halves of the ventricular roof. It becomes necessary, then, to correlate, if possible, the areas of this fluid-passage to the anatomical differentiations pointed out.




===The Accumulation of Injection-Masses in the Superior Membranous Area===


Menstrual age.


It has already been recorded that the first evidence of a change in the reaction to a replacement injection occurred in an embryo about 13 mm. long (fig. 2). This stage was characterized by a dense collection of the precipitated granules in a definite area in the roof of the fourth ventricle. At this stage also the area membranacea superior is well differentiated (fig. 31). That the site of the granular accumulation is this membranous area is easily proved by an inspection of figure 117, which represents an enlargement of the squared area in figure 116. In the low power photomicrograph the prussian-blue granules are not represented, but are found scattered through the ventricles, with a definite collection in the posterior region of the fourth ventricle. Under a higher magnification (fig. 117) the blue can be traced in but small quantity along the normal ependj'mal lining (shown to the left in the figure), but as soon as the differentiated area (area membranacea superior) is reached the granular material is heaped up in a dense mass, which extends as a thickened pad into the ventricle.




Contents of ovum.
The same phenomenon of the accumulation of the injection fluid in the superior membranous area is shown in figures 112 and 113, the second photomicrograph representing the area outlined in the first, but reproduced under much higher magnification. In this specimen (an embryo pig) a dilute solution of silver nitrate was injected into the central canal of the spinal cord. On histological examination the accumulation of the silver also shown in figure 11 was found. Thus, in figure 113 the ventricular epithelium can be made out in the upper right-hand corner, while below (in the area membranacea superior) the silver is densely accumulated.




The explanation of this phenomenon of accumulation in the superior membranous area is not whoUj' clear. It occurs only in stages in which the histological differentiation of the ventricular roof has proceeded to some degree and in stages where the fluid-passage into the periaxial tissues is not wholly unobstructed. This aggregation of the precipitated granules of prussian-blue and of the reduced silver in a locaUzed area certainly suggests a phj'sical exi^lanation, as in these cases the physical laws of precipitation and reduction must hold. The many figures of the superior membranous area of the ventricular roof show that in the stage under consideration the cell-outlines projecting into the ventricles are rough and ragged as contrasted with the smoother and more regular surface of the adjoining ependyma.


278 660 813
78 531 250
12 318 543
Ooia
244 402 122 533 545 21 560 135


Could not these roughened, irregular cell-surfaces become the site of the first and most extreme precipitation of the prussian-bluc and of the reduction of the silver? Certainly they would serve much more efficiently as the foreign substances about which precipitation would occur in greatest amount. This phj^sical explanation finds man}' arguments for its suj^ijort in these studies.


mm.


Another explanation of the phenomenon concerns the normal flow of the fluid and the relation of the direction of this flow to the roof of the fourth ventricle. As has already been emphasized, it is difficult to assume that there is any marked production of cerebro-spinal fluid before the periaxial spread occurs. Such an assumption would argue against the development of any special current toward the roof of the fourth ventricle in any stage smaller than that represented in figure 3, and would vitiate the explanation of the occurrence of the granular accumulation shown in figure 2 (a pig embryo of 13 mm.). In the later stages (16 mm., cj. fig. 11) this explanation would probably suffice for the phenomenon exhibited.


1 1.5
2.1 2 5 3 3 4
4 5 5 5 5
5.5 • 7 9


===The Sites of Fluid Passage Through the Roof of the Fourth Ventricle===


mm.


With consideration of the evidence presented as to the accumulation of the precipitates of the injected fluid about the area membranacea superior during certain stages in the development of the cerebro-spinal spaces, it would seem that the same area must be concerned in the passage of fluid from the ventricular cavities into the periaxial tissues. This view receives support from the reproduction of a cleared specimen (fig. 11) in which an injection of silver nitrate had been made into the central canal of the spinal cord. The pressure employed was great enough to force the fluid into the periaxial spaces, but the resultant picture clearly showed the oval outline of the area membranacea superior.




days.
The study of the passage of fluid from the ventricular to the extraventricular spaces can best be made by simple histological serial sections. In these observations pig embryos in which the cerebro-spinal fluid had been replaced by the compensating device, supplying a true solution of potassium ferrocyanide and iron-ammonium citrate, were sectioned and examined with reference to the sites of fluid passage. The results of these studies are given here in order that the whole question of the connection of the cerebral ventricles with the subarachnoid spaces may be discussed.




67
In the stage represented by figure 3 (in which fluid passes from one area in the roof of the fourth ventricle into the extraventricular tissues) histological sections show that the point of fluid ])assage is localized and concerns solely the area membranacea superior. The replaced fluid (as demonstrated by the subsequent precipitation of the prussian-blue) passes through this entire membranous area into the adjoining mesenchyme. The process is wholly confined to this area; the adjoining ependyma is entirely imj)ervious to the ferrocyanide. This phenomenon of passage of the replaced fluid through the sujjerior membranous urea is well shown in figures 14, 18, and 23.
200
87
45
41 42
42 65 50 53
49




. 1
The distriltution of the minute granules of ])russian-l)lue in the cells of the superior membranous area is of iini)ortance in any discussion of the passage of fluid through a membrane; for this area (in the superior portion of the roof of the embryonic fourth ventricle) must be considered as a memljrane permeable in certain degrees to the fluids bathing it. That the area mcml:)ranacea is intact and does not contain stomata or other minute foramina has been demonstrated histologically. Further evidence of the entire lack of intercellular stomata is afforded by the distribution of the prussian-blue granules precipitated in situ after the replacement of the cerebro-spinal fluid by the ferrocyanide solution.




512
Figure 14 is a reproduction of the superior area from a transverse section of a pig embrj^o in which the routine replacement had been made. The position of the area is shown by the squared outline in figure 13. On both sides the impermeable ependyma is seen, with granules of the blue adhering to the ventricular border of the cells, but not penetrating them at all. To the left of the drawing the few ependymal cells possess, beneath their central border, a chain of the granules which have entered from the abrupt edge of the area membranacea. In the cellular border between the two hps of the ependyma, the area membranacea superior, the passage of the replaced fluid is easily made out by the resultant blue granules. The area is roughly deUmited by a ventricular collection of the blue granules. Examination of these cells shows that the prussian-blue is present within the cytoplasm, avoiding the nuclei with perfect precision. Some of the cells are rounded and almost free from the granules; others, particularly those whose cj'^toplasm is elongated, are completely filled with the granules, the nuclei standing out in a blue granular cytoplasm.




mm.
The question of the passage of the fluid between the cells must also be answered by the histological evidence. In the same drawing (fig. 14) in one or two places there are indications of a slight stream of granules between the cells of the area membranacea superior. This apparent transit of the fluid through intercellular passages is particularly clear in the small areas where the cellular cytoplasm is relatively free from the granular deposits. But upon careful examination of these areas under oil immersion it is always apparent that the adjoining cytoplasm is also mvolved in the granular precipitation, indicating that the cells, although almost free from the deposit, are also engaged in the process of the fluid passage. Compared to the whole area of fluid transit, the points indicative of a passage through possible intercellular stigmata are almost negUgible. It seems not unlikely that the outlining of canals between cells may be a physical phenomenon, as in most cases no cellular borders (as demonstrated bj' the precipitated granules) can be made out. These pecuUarities of fluid passage may be seen in figures 14, 18, and 23.




10
Consideration of all the evidence afforded by histological examinations of the essential character of the area membranacea superior and of the passage of fluid through it incUnes one inevitably to the belief that this area functionates as a cellular membrane. The fluid passes through it as through any permeable liv-ing membrane. Histologically the passage is for the most part through the cj-toplasm of the cells, but occasionally an intercellular course is suggested. Both processes are wholly compatible with the accepted view of a cellular membrane de\dsed for the passage of fluid through it.
10 [10.5]
12
12 14 14
17 17 20 22 25 33 57 90




mm. 30X27X18 28X28X22 45 X40 X25 35X35X15
The same phenomenon of the passage of fluid from the fourth ventricle into the periaxial spaces is beautifullj' illustrated in figure 23. This drawing is from a transverse section of a pig embryo (23 mm. in length) in a stage when the superior membranous area is rapidly being encroached upon by the developing cerebellum and by the caudal chorioid plexuses. Between the deeplj' staining epondj-mal cells on cither side the membranous area is densely outlined by the deposition of the granules of prussian-blue in the cytoplasm of the cells of the area membranacea superior. The avoidance of the nuclei of these cells by the ferrocyanide is well demonstrated in this reproduction, as is also the impenetrability of the ependymal cells. In a specimen of this nature the question of the passage of the injection fluid through possible intercellular foramina loses its significance; for the drawing shows clearly the importance of considering the entire area membranacea as a functioning whole — a permeable, Uvmg, cellular membrane.




days. 56 35
It has been shown in a foregoing section of this memoir that histologically the area membranacea superior decreases to an almost neghgible remains in specimens of embryo pigs over 30 mm. long. This same rule apparently holds for its functional importance, as determined by the relative and absolute amount of prussian-blue granules deposited in the cells of the superior membrane. This decrease in the functional imi;)ortance may be inferred from figure 47, a photomicrograph from a pig embryo of 32 mm. Apparently the size of the membrane determines in large measure the amount of the replaced fluid which passes through it.
73
14 91




No magma.
Thus far we have been concerned solely with the passage of fluid through the area membranacea superior. In the earUer stages of from 14 to 23 mm. the importance of the superior membrane functionally is great, but in the later stages the inferior membrane assumes far greater significance. This is demonstrated not only by the structural history of the two areas, but by the functional index afforded by the replacement of the cerebro-spinal fluid by a foreign solution.




Granular-hyaline ; also no magma.
In the foregoing section the first evidence of any histological differentiation in the inferior portion of the roof of the fourth ventricle was shown to occur in pig embryos of 15 mm. in length. From this stage upwards (figs. 4, 5, etc.) a portion of the inferior roof allows fluid to pass through it. The exact point of fluid iiassage is the localized ependymal differentiation forming the area membranacea inferior. This relationship is easily verified by reference to figure 18. In this drawing of a median sagittal section of a pig embryo the two localized pomts of fluid passage into the periaxial tissue are readily identified; they are quite Umited in comparison to the extent of the periaxial spread.




Hyaline magma. Granular; hyaline also.
Figure IG represents the inferior membranous area of the roof of the fourth ventricle from a i)ig embryo of .similar size (18 mm.). The histological character of the inferior area is well shown in this drawing. It will be seen that, except in small areas, the histological differentiation of the ependyma has not proceeded to any great extent; the fluid from the ventricular cavity (as traced by the precipitated granules) closely follows the points of greatest cellular differentiation. There is no possibility of an intfri)r('tation of the findings concerned with the existence of intercellular stomata; the passage of fluid is here again to be looked upon as a transit through a cellular membrane.




Hyaline magma.
The same general i)heiioinena of the pa.ssagc of fluid through a localized area (the area membranacea inferior, hi the caudal portion of the roof of the fourth ventricle) that have been observed in the superior portion of the roof are shown in figure 18. Cliief among these phenomena is the careful avoidance bj- the precipitated granules of the ependymal lining of the ventricles and the adherence of the granules to the Uning walls at the points of fluid passage. The ependymal lining, except in the two areas of differentiation, is everywhere impenetrable to the solution of the ferrocyanide.




Granular; hyaline.
As the size of the embryo increases the functional importance of this more caudal area becomes much greater (c/. figs. 3, 4, 5, and 76). The whole caudal half of the fourth ventricle becomes an area of ependj-mal differentiation and of fluid passage. It serves everywhere as a complete diff'using membrane, un))roken by the occurrence of stomata. Through this whole membrane the replaced solutions of potassium ferrocyanide and iron-ammonium citrate pass with apparent ease, as demonstrated by the precipitated granules of prussian-blue (fig. 18). From stages of 24 mm. and over, the lower membranous area is the onlj' one of significance in the total fluid passage.




Do. Do.
The areas, therefore, through which the replaced solution of potassium ferrocyanide and iron-ammonium citrate jiassed, in the experimental pig embryos, are the two areas of histological differentiation in the roof of the fourth ventricle — the areae membranacese superior et inferior. There is no evidence whatsoever of any other point of escape of the fluid from the ventricular system into the periaxial spaces. The precipitated prussian-blue does not penetrate any of the lining cells of the ventricle except in the two areas under consideration. Nor is any evidence afforded by histological study of the escape of ventricular fluid through the described foramina of Bichat and of ^Mierzejewsky.




===Factors Concerned in the Experimental Fluid Passage===


40 X35 X30 80 X50 X25 36X33X13 19X19X19 10 X 9X 8 18X18X 8 20X18X11 70 X30 X25 35 X30 X30 25X15X15 40 X25 X20 20X16X 6 35X30X30 12 X 9X 9 12X 9X 5


It becomes necessary to discuss the question of the passage of the replaced fluid through the two cellular membranes in order to ascertain to what extent the results obtamed by the method may be relied upon. Naturally in such questions the factors concerned in the normal transit of body-fluid through such structures must be considered.


636 Hyaline magnia. 005 Fluid. 1 104 Flaky.


Probably the most essential element in obtaining rehable results in any injection is the control of the pressure at which the foreign fluid or mass is introduced. This matter has been fully discussed in the resume of the methods employed; it is suflRcient to reaffirm here that, in these observations, the normal cerebro-spinal tension has not been disturbed because of the use of a compensatory replacement. Other experiments, carried out under increasing pressures of injection, have been made, in order to compare the results with those furnished by the replacementmethod.


1 211 1117


Consideration must next be given to the factors of diffusion, filtration, and osmosis in the passage of fluid through the roof of the fourth ventricle. The third factor, however, may be largely excluded, owing to the fact that the solutions of potassium ferrocyanide and iron-ammonium citrate employed were for the most part practically isotonic with the body-fluids. Furthermore, the use of hypertonic solutions apparently gave no difi"crent results (except in the increased density of the resultant precipitate) from those obtained by the employment of the isotonic solutions. Finally, it was found to be of service to use hypotonic replacement solutions in order to obtain very sUght precipitates; in these experiments also the spread of the replaced ferrocyanide solution was similar to the standard result afforded by the isotonic solution. These observations with varying concentrations of the foreign solutions replacing the cerebro-spinal fluid serve to indicate that osmosis plays but little part in the passage of fluid through the roof structures of the fourth ventricle. Undoubtedly the factor of osmosis can not be ignored in any consideration of the passage of fluid through a cellular membrane, but it seems unhkely that with solutions of practically the same salt-content it should be of great importance.




The influence of diffusion in this passage of the solution of the ferrocyanide and citrate from the cerebral ventricles into the extraventricular space is probably great. The whole plan of the experiment concerns the introduction of salts foreign to the body-fluids, even though in analogous concentrations. It seems not unhkely that as soon as the replacement of the existent cerebro-spinal fluid is effected the ferrocyanide and citrate must immediately begin to diffuse out into the periaxial tissues and the normal salts return to the ventricles. Probably, however, this same phenomenon plays a normal role in the human body. Jacobson's^^?) extensive and important studies on the chemistry of cerebro-spinal fluid have shown that the ventricular cerebro-spinal fluid is not identical with the subarachnoid fluid. The differences in the two fluids are probably to be accounted for by the fact that the ventricular fluid represents the pure elaboration of the chorioid plexuses, whereas the lumbar subarachnoid fluid is composed not only of the products of the chorioid plexuses but also of the fluids from the perivascular system. In this transference of the ventricular fluid to the subarachnoid space diffusion may play some part, the relative importance of which can hardly be estimated.




1
But will diffusion alone accoimt for the passage of the experimental fluid in the ventricle through two well-defined areas into the periaxial tissues? Will diffusion account for the varying extent of the injection in different stages of embryonic develojjment? There are several arguments against according diffusion a maximal imjjortance in the process. In the first i)lace, an injection of the solution of the ferrocyanide under mild syringe-pressure will give a spread similar in every re.spect to thos(; obtained by the replacement method. This indicates that the course taken by the two solutions is not necessarily the result of diffusion, but rather of the capabilities of the tissues for fluid-spread; and similarly the jiassage of this true solution through the roof areas need not be solely a diffusion process, but may be accounted for by the true flow of the fluid in this direction. Again, in the stages represented in figure 2 one would expect as extensive a spread of the replacing solution into the periaxial tissue were diffusion the active force in the movement of the fluid. Instead of such a periaxial spread the injection fluid remains wholly within the ventricular system, indicating that other forces than that of diffusion play an active role, in the more advanced stages, in the movement of the fluid. Finally, if diffusion is to be considered the sole agent in the distribution of the replacing fluid, why does not the ferrocyanide penetrate all the cellular structures lining the ventricular cavity? Surely it would he expected that diffusion between the body-fluids and the ferrocyanide solution would occur in each ependymal cell — a phenomenon observed only in the cells comprising the ventricular surfaces of the membranous areas of the rhombic roof.
Granular. Fluid.




Do.
While acknowledging that diffusion and osmosis may play important parts in the process of the passage of fluid from the fourth ventricle into the periaxial tissues, it seems apparent that some other factor or factors must be the determining agent or agents. It is not unlikely that the formation of cerebro-spinal fluid by the cells of the chorioid j)lexus may cause, in the replacement experiments, further passage of fluid into the extraventricular regions. Such an elaboration of fluid, with the ventricles filled with the experimental solution, would result in an increase in the normal ventricular tension. If this be the real explanation, the passage of the fluid into the extraventricular spaces would result in part from the increase in pressure on one (the ventricular) side of the membrane. The process, then, would be one of filtration through the membrane from the point of higher to that of lower pressure. This explanation best seems to cover the results obtained by the replacement method, and is supported by the histological examination of the developing chorioid plexuses and by many other features which are dealt with in other sections of the paper. This view is also strongly supported by the results of injections under mild syringe-pressure .




On the basis that the passage of fluid from the fourth ventricle into the periaxial tissues is in large measure a process of membrane filtration, the phenomenon of the fluid transit of the replaced solutions may be taken as a real index of the circulation and distribution of the cerebro-spinal fluid. It may be assumed, therefore, that the resulting distribution of the prussian-blue granules represents the course and extent of the fluid channels of the embryonic cerebro-spinal fluid.


270
991 604
94 1189 584a
79 230 261


The discussion of the fluid passage outward from the cerebral ventricles into the subarachnoid spaces has thus far been concerned with the processes involved for the transit of the true solutions of the salts. There is, however, an undoubted passage outward, as has already been indicated in a foregoing section, of the protein content of the normal cerebro-spinal fluid. This occurs in specimens in which a trul}- definitive membrane, intact throughout, can be seen inclosing the chorioidal roof. The explanations which suffice for the passage outward of the true solutions will not serve for this phenomenon.


40X30X20


The cells of the body probably are equipped to handle colloidal solutions in several ways, but two methods seem possible as explanatory of the problem at hand.


70X50X50


In the first place, it is conceivable that the cells in the differentiated areae membranaceae could phagocyte the colloidal albuminous particles of the ventricular fluid and excrete them into the subarachnoid spaces on the other side of the membrane; but it does not seem probable that this explanation is correct. Much more likely is it that the colloidal masses may follow the same laws of fluid-passage as the true solutions. But in such a passage through a cellular membrane the rate of passage will be much slower with the colloid.


50X50X50 50 X42 X40 50X50X50 75X60X50 120X70X70


These two theories regarding the passage of the albumen of the ventricular cerebro-spinal fluid into the subarachnoid spaces are not based on any findings presented in this article, but are ventured as being in keeping with current physiological explanations of such phenomena. On the basis of the second hypothesis, the failure of granular material to pass through the cellular membrane of the chorioidal roof must be explained as being due to the inability of the cells to handle the foreign material except in sizes which could be absorbed. The fact that the original unit was not phagocyted or passed through the membrane probably depended on the size of the molecule and the specific character of the lining-cells.


105X65X65


THE PASSAGE OF SILVER NITRATE AND INDIA INK THROUGH THE MEMBRANOUS AREAS IN THE ROOF OF THE FOURTH VENTRICLE.


Hyaline magnia.


Thus far in the discussion of the passage of the experimental fluids through the ventricular roof, true solutions of potassium ferrocyanide and iron-ammonium citrate only have been considered. This solution, as has been pointed out in this and in a previous article^^^)^ jg non-to.xic and is not taken up by the cells. With the dilute solutions (0.25 to 0.5 per cent) of silver nitrate, a far different problem is presented. Replacement experiments with this salt are rendered impossible bj- its intraspinous toxicity and by its precipitating action upon protein; but beautiful preparations may be made by this method by the simple injection with a syringe into the central canal of the spinal cord.




With mild syringe-pressure the result of such an injection with silver nitrate is in all cases a simple ventricular spread, with no extension into the ])eriaxial tissues. This general rule holds in all stages in which the central canal can be definitely entered without causing a spread into the j^erispinal tissues. This failure of the spread to extend into the periaxial tissues under mild pressure is undoubtedly due to the coagulating effect of the silver, which renders further passage of the fluid impo.ssible. The reduced silver collects about the superior membranous area in the roof of the fourth ventricle, outlining it distinctly. This phenomenon is illustrated in figure 115 (a transverse section of a pig emljryo of 19 mm.). At this stage the rejjlacement of cerebro-spinal fluid by a ferrocyanide solution results in a quite extensive spread {cf. fig. 5).




With increa.sed pressures of injection the silver may be pushed into the periaxial tissue through the roof structures of the fourth ventricle. The transit of the injection-mjuss occurs in the area meml^ranacea sujjcrior in practically all cases (c/. fig. 12). The inferior membranous ar(>a, in the earlier stages, is almost invariably impermeable to the silver (unless the injection-jiressure i.s extreme). When the superior area is examined after such an injection under high pressure the silver is found deposited throughout the cells of the area, extending only a short distance into the adjacent tissue. This feature of the injection is pictured in figure 113. In these injections the high pressure undoubtedlj' suffices to force the silver through the coagulated area membranacea. Its coagulating effect on the ependjina is almost equally marked, but the point of least resistance is apparently in the membranous area, allowing the fluid to pass through it.




Replacements of the cerebro-spinal fluid with diluted solutions of mdia ink within the medullary-canal system of small pig embryos never result in any extension of the granules into the periaxial tissues, for under the normal tension in the ventricles of the pig the arese membranaceaj are impermeable to the passage of granular material. After such a replacement the carbon masses maj' be found everywhere throughout the ventricles, but not in the periaxial tissues. However, india ink may be forced into the periaxial tissues by the use of high pressures of injection, as shown in figure 10. In this s])ecimen of a pig embryo (21 mm. m length) the periaxial spread occurred solely from the superior membranous area. This is analogous to the results obtained with silver nitrate, shown in figure 12. Without doubt in the earlier stages the superior area is much more permeable than the inferior. Histological examination of these specimens after an injection of india ink under high pressure reveals that the carbon granules gain the extraventricular space onh' through the area membranacea superior; some cells in this area are crowded with the granules, but for the most part extensive intercellular stomata have been made. The whole process must be viewed as a result of the excessive pressure of injection.


I will now review several specimens illustrating these three varieties of pathological magma. The specimens considered are arranged in table 2. The list could easily be increased to several hundred, but as the specimens with catalogue numbers less than 403 have been iiublished in detail with illustrations in my monograi)h cm monsters, I will allude only to some of them. The pathological specimens with numbers over 402 are being prepared for publication, so that a few selected specimens with higher numbers are illustrated. Pathological sjiecimens from tubal pregnancy with numbers u]) to 1,000 will be found described in detail in my monograph on tubal pregnancy.


In the more advanced stages of the pig embryo (30 mm. and upwards) the pressure necessary to occasion an extraventricular spread of the india ink after intraspinous injection decreases somewhat, so that \\ith mild syringe-pressure a local periaxial sjn-ead from the fourth ventricle may be obtained from an injection into the central canal of the spinal cord. This is in accordance with the observation of MalU'^), who found that the injection flowed "through the medial opening of the fourth ventricle." The opening in these cases is in the area membranacea inferior, and in many instances subsequent examination showed rupture of the membrane with escape of the ink, even though the injection-pressure was moderate.


The first specimen which 1 shall consider (No. 278) consists of an entire ovum, measuring 6 by 4 mm. It was sent me by Dr. Stanton, of Albany, New York. The siiecimen might be viewed as normal, but it contains no embryo, and as it was obtained from a di.seased uterus, it is prol)ably pathological, the magnia having un<lergone minor changes.


Taken as a whole, then, the findings are against the passage of solutions of silver nitrate or suspensions of india ink from the ventricles into the periaxial tissues, except when injected under pressures far above the normal intraventricular tension.


This ovum was found accidentally in curct lings fnmi a woman .supposed to have chronic endometritis following pregnancy. There is nothing in the history from ul,;,l, ti„. a^e of the specimen could be estimated. Part of the specimen had been


RELATION OF THE EPENDYMAL DIFFERENTIATION TO THE PASSAGE OF FLUID.




HUMAN MAGMA RfiTICULf: IX NORMAL AND I'ATIIOI/KIICAL DKVELOPMENT. 19
Under this heading it is jjroposed to discuss the relationship, if any, existing between the stages of differentiation of the ependjona of the roof of the fourth ventricle and the prssage of fluid through the two membranous areas. The discussion must necessarily be of a temporal character, with an attempt to consider possible factors in the process.
cut into sections before it was received at the laboratory, with the statement that no embryo had been found, it having faUen out. I found that the half sent contained a coelom, 3 by 2.5 mm., filled with magma, in which there was a cavity about 1.5 by 1 mm. 1^'ections showed that the cavity was natural and not sharply defined, with nothing to indicate that it had contained an embryo. On the contrary, it was found that the magma reticule was composed of a loose network of mesoderm cells, which bound one side of the chorion with the other. These cells are directly continuous with those of the mesoderm and resemble them in every particular. At one point there is a small grouj) of epithelial cells, which may represent what was originally the embryo. Otherwise, the chorion and its villi are normal in appearance, being encapsulated in decidua which has in it some uterine glands. All in all, this specimen reminds one very much of the Peters ovum. There are some leucocytes in the decidua, but no accumulation of them indicating inflammation of the uterus. Several figures, illustrating this specimen, may be found in my monograph on monsters.




Specimen Xo. 531 is in many respects similar to the one just described ( Xo. 278 ) . It came from a woman who had been pregnant 6 times, her periods having been 17 days overdue before this abortion. The ovum is spherical, 19 mm. in diameter, and is covered only by a mass of villi which appear normal. The ccclom within contains many magma fibrils, the meshes of which arc more or less filled with dense granules, as is i^hown in plate 1, figure 8. Within this mass there is a detached vesicle, 1.5 mm. in diameter, which no doubt represents the umbilical vesicle.
The most important question in this connection is whether the ependymal differentiation is necessary for the passage of fluid through it. In tlie \ng embryo of 13 mm. the area membranacea superior has reached a stage of marked differentiation (fig. 31), but at this same stage (fig. 2) there is no evidence of any pas.sage of fluid through the roof of the fourth ventricle into the periaxial tissue, only an outlining of the oval membranous area. Here, then, the histological differentiation has definitely preceded the assumption of function on the part of the area membranacea superior. The passage of fluid through the lower area occurs at a relatively earlier stage than it does through the superior opening. The first evidence of differentiation of the inferior roof of the fourth ventricle was observed in pig embryos of 15 mm. in length. At 18 mm., even though the process of differentiation was far from complete, some of the replaced fluid was able to pass through the lower area (figs. 4, 16, and 18).




A specimen intermediate between the two just described is X'o. 250, of which several illustrations are published in my paper on monsters. The specimen came embedded in a mass of decidua and was obtained by scraping the uterus. When opened it was found filled with magma reticule just beneath the chorion, in which could be seen a small embryo, and farther away towards the center of the coelom was the umbilical vesicle. The whole ovum was cut hito sections. The chorion and the villi are apparently normal in shape and structure, being also rich in bloodvessels, which are filled with embryo blood. The villi are bathed in mother's blood and covered with an active trophoblast. The decidua is somewhat infiltrated with leucocytes, but there are no abscesses. The front end of the amnion is absent, and its free edge and the embryo are embedded in reticular magma, indicating that the amnion was destroyed before the abortion took place. The general shape of the embryo and its degree of development are practically normal. The heart is well formed and, including the blood-vessels, is filled with blood. The alimentary canal, brain, spinal cord, otic and eye vesicles, myotomes, and branchial arches are much like those of embryo Xo. 12, to be described presently. The septum transversum is well marked and the thyroid gland is just beginning. The tissues of the embryo, however, and the cavity of the front end of the brain are filled with numerous small round cells with fragmented nuclei. All stages of fragmentation are seen, just as may be observed in the leucocytes in small abscesses. IMost of the red blood-cells are within the blood-vessels, but those within the tissues appear perfectly normal. On account of the dimmished number of mesoderm cells, which, in fact, diminish in proportion as the fragmented cells increase, the con
A consideration of these observations leads to the assumption that some histological differentiation of the ependj^ma is necessary for the extraventricular passage of the replaced fluid. In the case of the superior area the differentiation occurs at a considerable developmental interval before fluid passes through it; in regard to the inferior area the assumption of function occurs at a somewhat earlier period in its differentiation. This slight difference between the two areas may possiblj' be explained on the basis that as soon as the stage of 14 mm. is attained (by the pig embryo) a greater amount of cerebro-spinal fluid is produced than can be cared for by the more slowly enlarging ventricular cavities. As soon as this disproportion occurs the excess of fluid is poured into the periaxial tissues through the already differentiated area membranacea superior; therefore, when the inferior area first shows evidence of formation there is still this excess of fluid in the ventricles. The fluid apparently avails itself almost at once of the new opening and its functional existence becomes immediate. It is apparent, moreover, that the capacity of the membranous areas for the passage of fluid is considerably in excess of the demands made upon them, and furthermore, that the provision for the passage of increasing amounts of fluid is completed before the demand arises.




In the passage of fluid from the ventricles into the mesenchyme, there is one other factor which has not as yet been considered. This concerns the iKitentiality of the adjacent mesenchyme to afford channels for the fluid poured into it. Were resistance offered to the flow of solutions through the mesenchymal tissue spaces, fluid could escape from the ventricles in only very small amounts, if at all ; as soon, however, as easily traversed fluid channels became established, the cerebro-spinal fluid could readily escape through the two membranous areas. The question as to what part the embr\'onic cerebro-spinal fluid i)lays in the further development of the meningeal si)aees also arises in this connection. It is at i)resent impossible to assign to any one of these factors a specific role in the passage of fluid from the fourth ventricle into the jieriaxial spaces, but it is important to consider them as possible determining agents. The evidence all indicates that the rate of production of the embryonic cerebro-spinal fluid is the most important factor, by far, in the extraventricular escape of the fluid.


20 iiiMAN MACMA nf:Ticri.f: in nohmai. and i'athoi.oc.icai, development.


==VII. General Histological Differentiation Of The Cerebro-Spinal Spaces==


elusion must bo drawn tliat the fragmented cells arise from the mesoderm cells. Tlie ei)idormis covers the whole embryo. The primary change in this specimen is no doubt in the mesoderm, for all the rest of the embryo appears normal. That the c(|uililM-ium was overthrown is indicated by the necrotic amnion and the great amount of reticular magma in the exoccrlom. \\'hat is especially interesting in this specimen is the partial destruction of the amnion, which brings the embryo directly in contact with the pathological magma of the coelom.


The general problems concerned in the formation of the meninges and of the spaces inclosed within them deal with the gradual adaptation of a primitive undifferentiated mesenchyme to the anatomical and physiological requirements of the adult. Originally the meninges were held to be derived from the same epidermal infolding which gave origin to the central nervous system; then, with increasing knowledge of the structure, the dura alone was said to be a product of the middle germ-laj'er; and finally, by the researches of His^^s) and of K6lliker,'3i) the mesenchymal origin of the three meninges was established. The general process of the differentiation and the stages in this transformation have not been reported in great detail; here, too, the investigations must have an outlook for physiological anatomy as well as for pure morphology.


Embryo Xo. 12, which has been just referred to, may also be discussed in this connection. It was questionable for a long time whether or not the eml)ryo was normal, as the villi and contents of the ccelom and embryo are beautifully preserved and show no pathological change. However, more careful consideration of the specimen shows that there are a few fibrinous masses between the villi, with every indication of uterine inflammation and infection. The extent of the reticular magma is more pronounced than usual, and it w'as necessary to dis.sect it away before the embryo could Ije isolated sufficiently so that it could be well seen. The head is no doubt atrophic, and I am fully convinced that this part of the embryo must have undergone pathological changes a short time before the abortion.


It may be well to comment briefly on the relationships of the three meninges found in adult mammals. The dura is well estabUshed as the fibrous-tissue envelope of the leptomeninges and the central nervous system. But there is a tendency to regard the arachnoid and pia mater as constituting one structure — the leptomeninges or "pia-arachnoid," in the terminology of Middlemass and Robertson'^"). This difference of opinion in regard to the two inner meninges is due to their structural and intimate relationships. The arachnoid may well be a.ssumed to be a single membrane, worthy of being regarded as a single structure if one considers only its outer continuous membrane as the essential structure. But the inner surface of this membrane sends processes inward to fuse with the pia mater, which is so closely applied to the ner\-ous tissue. These processes divide the subarachnoid space (mcluded between arachnoid and pia) into the well-known meshes in which the cerebro-spinal fluid circulates. From the standpoint of these channels (the subarachnoid spaces) the arachnoid constitutes the parietal and the pia the visceral layer. Thus the intimate structural unitj' of the two membranes seems, in the opinion of many investigators, to warrant their designation as a single membrane. This view, however, has been strongly opposed by Poirier and Charpy^*^', who considered the distinction of three meninges very essential. Hence, in considering the transformation of tissues in the embryo, regard must be had for the dura as a well-differentiated structure, and for the leptomeninges as units, but certainly to be regarded from the standpoint of the subarachnoid spaces. In tliis connection Sterzi's'^^') observations on the comparative anatomy of the meninges are of interest. It will be recalled that the dura in lower forms becomes well estabhshed before the leptomeninges emerge from a primitive mesenchyme.


Specimen Xo. 318 is much Uke X"o. 250. The ovum, measuring 20 by 18 b.\- 1 1 mm., is covered with villi which appear to be perfectly normal. Upon opening, it was found to be filled with stringy magma, on one side of which was embedded an embryo 2.5 mm. in length. The head is sharply outlined, but the embryo seems to continue directly with the uml)ilical vesicle, leaving an atrophic tail. Sections .show' that the amnion over the head has dissolved, leaving a picture very much like that shown in Xo. 250. ^^'e ha\e here a small embryo with a very large ca?lom, the ovum being moderately filled with reticular magma and a small embryo


THE PERIAXIAL MESENCHYME.


only partlj' covered with the amnion. Xo. 543 is another embryo of the same type. The magma is a little den.ser than in Xo. 318. The chorionic villi are developed, but markedly pathological, as the photograph show^s. The embryo within is 3 nmi. long, lying (juite free within the mass of magma. It is covered by a ragged anmion; that is, the amnion is jjartly destroyed.


Surrounding the central nervous system in young embryos is a rather thick cushion of undifferentiated mesenchyme, similar m all respects to the undifferentiated tissue in other parts of the embiyo. But verj' soon in the course of development the nuclei in this mesenchyme increase along the clear marginal zone of the spinal cord ami ha.silar structures, forming the initial indication of the pia mater. This phenomenon is indicated somewhat in figure 40, a photomicrograph taken from a human embryo (Xo. 836) of 4 mm., the earliest stage here illustrated.


An interesting specimen in this connection is X"o. 402, which is partly described in my paper on monsters, since the issue of which the embryo and chorion have been cut into serial sections. The villi of the ovum are not well developed, and they are distributed irregularly over the surface. The coelom is filled with reticular magma. The embryo is club-shaped, the head being much too far advanced for the body. The umbilical vesicle is normal in .size; the heart is well outlined, and the extremities are just api)earing. Sections .show that the amnion is greatly distended. Sections of the chorion were stained with cochineal and ^'an (iieson, and show beautifully the fibrillated structure of the chorionic membrane. These fibers take im red stain, as do thoseof the reticular magma. The two are continuous, as .shown by the illustration on plate 3, figure 3. In fact, this continuity is much more pronounced in i)athological than in normal specimetis.


The next essential change in the great differentiation of the meninges concerns a blastemal condensation of this same mesenchymal tissue to form ultimately the bony covering of the central nervous sj'stem and a portion of the dura; but between these two zones of differentiation the mesenchyme remains for a time almost unaltered. A portion of this tissue will go to form the arachnoid membrane and the trabeculaD which mark off the subarachnoid spaces. This process in the formation of the arachnoid will be discussed here; the formation of the pia mater and dura will be detailed in succeeding divisions of the paper. The differentiation will be discussed as a general process, in regard to both human and pig embryos, for in no respect has any essential difference between the two been observed.


Sixrimen Xo. 533 (plate 2, figure 3) .shows a more atlvanced stage ol an extensive development of reticular magma. The villi of the ovum appear to be norinal and tl,,. iLiioular magma is very dense. Hehveeii tlie meshes there are a number of


The general character of the periaxial mesenchyme may be commented upon here. The tissue is of a verj' loose and typical structure, forming a syncytial network of rather small mesh, but fragile. The nuclei of the cells are oval, with a definite chromatin content; the cytoj^lasm is largely devoted to the maintenance of long processes which connect with adjacent cells. Adhering to the cj^toplasmic processes are very tiny albuminous coagula, of such small amount as to be hardly noticeable; also in the meshes of the mesenclwme very small quantities of this albumen may be identified. These albuminous coagula undoubtedly represent the protein of the tissue fluids in the undifferentiated stages.




HUMAN MAGMA KKTICULE IX NORMAI, AMJ I'AIHOLOOICAI, DEVELOPMENT. 21
THE FORMATION OF THE ARACHNOIDEA.
opaque nodules about 0.5 mm. in diameter. \\ illi much difficult^' the embryo




was teased out, but it was practically impossible to clear it entirely of the magma fibrils. The embryo is long and slender, looking more like that of a dog than a human specimen, the head being unusually small and thin for a human embryo of 0.5 mm. long. The fibers are irregularly^ stuck together by small granules, and there is a gap in the center which represents the place in which the embryo
A general consideration of the problems here involved will surely shed light on some of the various factors concerned. It must be noted that in its development this membrane proceeds from an undifferentiated but small-meshed mesenchyme into the adult structure which contains the relatively large cerebro-spinal channels. Then, too, the enlargement of the tissue meshes in certain places — as the future cistcrnae — must be enormous. Besides this necessary dilatation of the spaces in the periaxial m(>senchymc, the outer portion of the tissue must separate from the future dura and form the outer surface of the arachnoid membrane. Here the process must be one of tissue condensation and proliferation. A similar agencj' is involved in the growth of the mesothelial cells which cover the outer surface of the arachnoid and also the inner suliarachnoid spaces.




was located. The illustration shows this condition beautifully. The specimen was sent me bj' Dr. Fewsmith, of Trenton, New Jersey, who obtained it from a woman whose menstrual period had been a month overdue.
The g(>n<'ral process, then, in the formation of the arachnoid membrane concerns a thiiuiing and readjustment of the primitive mesenchyme in certain areas, while in others the process is reversed, the membrane reaching the adult form through proliferative and condi-nsing i)henonieiia. Such alterative processes must naturally result from the apj)lication of certain mechanical or vital agents in the growth of th(? embryo. Is the mere growth of the central nervous system sufficient to furnish these alterative agents, or must we likewise trace the corres])onding development of the bony coverings of the brain and spinal cord? X{>ith(>r factor seems relatively of great importance when comi)ared to the possible influence of the presence and circulation of cerchro-spinal fluid on this periaxial tissue. This seems to be the most important factor, an internally-modifying influence to which the periaxial mesenchyme is subjected in the formation of an arachnoid and its subarachnoid spaces. It will therefore be from this standpoint that the development of the spaces will be discussed; for, as has already been pointed out, the periaxial mesenchyme becomes a functionally active tissue for the circulation of the cerebrospinal fluid at a stage when difi"erentiation has not begim. On this basis, the lack of differentiation shown in the |)eriaxial mesenchyme in the stages before the ventricular cerebro-spinal fluid is poured into the mesenchyme in the neighborhood of the roof of the fourth ventricle is not surprising. The character of the periaxial mesenchyme in the early stages is reproduced in numerous photomicrographs (figs. 25, 49, 51 , and 53). The mesenchj'me is here characterized by a rather dense meshwork of cytoplasmic processes, interspersed b^' a considerable number of oval nuclei. The content of the interstices in albumen, as judged by the persisting coagula, is very small. This picture of the periaxial mesenchyme persists untU cerebro-spinal fluid is poured from the ventricle through the area membranacea superior.




An extremely interesting specimen is Xo. 545, well illustrated in figure 1, plate 3. The magma is not extensive, but it is pronounced. The embryo
As will be seen in figure 3, the first indication of an extraventricular spread of the replaced fluid in the ventricles occurred in a pig embryo  of 14 mm. At this stage the membranous area in the superior portion of the roof of the fourth ventricle has already' become well differentiated. The fluid from the ventricles, however, does not reach any considerable spread until after a length of 18 mm. is attained; the periaxial spread during this period of growth is wholh^ confined to the peribulbar tissues. It is quite important in this connection that the first obvious differentiation of the mesenchyme for the formation of the arachnoid .should appear during this period and should involve the peribulbar tissues.




is atrophic, and the chorion is only partly covered with vilU. The specimen was sent me by Dr. Rand, of New Haven, Connecticut. It was obtained from a woman who is the mother of one healthy child. The last menstrual period began on September 2. Bleeding began on October 22 and ended with the abortion on October 25. The ovum was found embedded in the clots of blood attached to the cervix of the uterus.
The first change to be noted in the transformation of primitive mesenchjone into the future arachnoid is an obvious thinning of the structure with a decrease in the number of nuclei per unit-volume. This is made out in a photomicrograph (fig. 57) of a section from a human embryo 14 mm.* long, when contrasted with a similar mesenchymal area posterior to the ventricular roof (fig. 53). In the pig embryo this thinning of the mesenchyme is as obvious at this early stage.




An extreme case of degeneration of the magma is sho^Mi in No. 660, also well illustrated in figures 4 and 5, plate 3. There is a tendency towards membrane formation, tough strands of fibrils, spaces, and clumps of granules. The chorionic wall is hemorrhagic and degenerated; within there is a collapsed amnion containing a cheesy granular mass.
The process of dilatation of the mesenchymal spaces at this stage hardly seems to concern a direct disruption of the syncj'tial strands, but resembles more the spreading of the cell-bodies by the introduction of more fluid into the tissue spaces. This process would certainly result in an ap])earance similar in everj' way to that represented by figures 35 and 57. It probably also concerns other factors, as, possibly, the growth of the whole embryo without a corresponding degree of mesenchymal proliferation.




I shall use two more specimens to illustrate the nature of granular mass in more advanced stages. The first is No. 605 and the second is No. 584a. No. 605 is a white transparent specimen, covered with a uniform laj'er of \'\\\i which branch two or three times. The entire specimen measures 45 by 40 by 25 mm. ; a small patch of decidua adheres to the outside. The interior is partly filled with coarse strands of reticular magma, having numerous granules attached. On one side of the specimen the umbilical cord is seen, surrounded b}- a ragged amnion. The tip of the cord has a piece of intestine and stomach hanging from it. The larger masses of tissue which are intermingled with the reticular magma must be the renmants of the embryo, parts of which appear to be normal, and judging from the form and size of the arms and legs the embryo is about 10.5 mm. long. The second specimen is unusually interesting because it contains a normal embryo with hernia of the Uver. The exoccelom is unusually large and is filled with a more extensive layer of reticular magma than should be found in an ovum containing a normal embrj^o of this size.
In a human embryo of 17 mm. (Xo. 576) evidences are apparent of such a thinning of the mesenchyme about the medulla. Thus, in figures 58 and 59, from this specimen, the cellular decrease can be made out both in the region of the roof of the fourth ventricle and around the basilar surface of the medulla. It wdll be noted that the differentiation (i. e., the thinning) about the roof has proceeded more rapidly than along the anterior bulbar surface. This is perhaps to be expected in view of the initial pouring-out of the cerebro-spinal fluid into the mesenchyme just posterior to the roof.


*This embryo measured 14 mm. on the slide.


The remaining three si)ecimens are given because they well illustrate various degrees of reticular magma within the ovum.


In this mesenchymal differentiation a slightly increased amount of albuminous coagulum may be noticed. The truth of this is made obvious by an examination of figure 61, a photomicrograph from a human embryo of 17 mm. The almost entire freedom of the mesenchyme from albuminous detritus is most noticeable at earlier stages.


No. 560 (plate 1, figure 6) shows very pronounced reticular magma intermingled with much granidar. Two stages of somewhat later development are given in Nos. 636 and 991. In the former (plate 1, figure 10) the magma is more pronoimced than in normal development, and in the latter (plate 1, figure 7) it is in an extreme amount.


As was pointed out in the description of the results of replacing the cerebrospinal fluid, a marked change in the rate of development of the cerebro-spinal spaces in the pig-embryo ensues just after attaining the length of 18 mm. Within the growth of 2 mm. the injection spreads completely down the spinal cord and about the basilar structures of the cerebral cavity. This rapid extension finds its analogous process in the equally rapid changes which may be traced in the periaxial mesenclnTTie. Thus, in figure 72, a photomicrograph from a sagittal section of a pig embryo of 18 mm., the whole nervous tissue appears surrounded by a very thin, Ughtly staining tissue; this is the periaxial mesenchyme, which is undergoing its rapid metamorphosis. It will be noticed in this figure that the posterior structures (rhombencephalon) are surrounded by a much less dense mesenchyme than are the anterior (mesencephalon). This relative differentiation between the bulbar tissue and that around the mid-brain is only of temporal character; the mesenchyme about the medulla, as has already been pointed out, begins to differentiate first, the differentiation of the mesenchyme about the other nervous structures following somewhat later.


Finally, a unique specimen (No. 1189) throws some light upon the formation of the reticular magma. The ovum came to us within the uterus, having been removed by an operation. At first it seemed to be normal, but on opening it the


Figure 73 is a photomicrograph of higher power, taken from the squared area in figure 72. It shows to what a surprising degree the mesenchymal differentiation has proceeded during a few millimeters' growth. Two striking features of the process are brought out in this reproduction. In the first place, many of the mesenchymal trabecular have apparently been broken down, sacrificed to a few larger remaining strands. The cells connected with the destroyed trabeculie appear to recede until one of the heavier surviving strands is met with, when they adhere and apparently aid in the future development of a permanent arachnoid trabecula. The second feature of importance in figure 73 concerns the large amount of allnimen seen in the periaxial space. There is here a much greater amount of albumen than is ever found in the periaxial mesenchyme before the differentia ting process which results in the future subarachnoid space has become definite. The occurrence of this large amount of albuminous coagulum is apparently related directly to the outflow of the cnibrj'onic cerebro-spinal fluid, for the embryonic fluid is very rich in protein material, a.s can be readily seen by the partial filling of the embryonic cerebral ventricles with the clotted albumen.




22 IIIMAN .\1A(;MA HKTICHI.K 1\ Ndli.MAI. AND l-ATIK i|.( )(;i(AI. DKNKI.orMKNT.
This process of the l)reaking-down of the mesenchymal spaces to form fewer and hirger spaces goes on very rapidly in pig embryos as the\' exceed the length of 18 mm. Thus, figure 75 (from a pig embryo of 2.3 mm.) shows a marked decrease in the mesenchymal elements al)out the medulla; the strands are becoming fewer in number, and the albumen-filled spaces are increasing rapidlj' in size, but decreasing in number. About the mcsencej)halon, however, the process has only just begun (also shown by fig. 74). In this photomicrograjih (fig. 75) the mesenchj^mal elements have broken down somewhat; the spaces are Ijecoming enlarged, and a fine albuminous coagulum fills the interstices between the mesenchymal processes. The whole picture conveys an excellent idea of the forces which convert the many-spaced mesenchyme into the much fewer cerebro-spinal channels.




ombryo was found encircled by a large mass of transparent, tough, stringy reticular magma, which was removed only with great difficulty. It behaved much like the vitreous humor of the eye. On account of its great (juantity we at once suspected that the sju'cimen was pathological, and after the embryo was removed it proved to i)e so. Ahhough (juite advanced in development, its head was found to be smaller than normal, the tissues of the face were dissociated, and the borders of the eye were not sharp but ragged. No doubt the specimen had continued to develoi) normally until shortly before the operation, and the magma increased in quantity and became tough and fibrous. It is an interesting specimen, showing changes in the magma late in di'velopment. Sections of the implanted ovum have not .yet been made. The specimen is from a negress, 45 years of age, who had had 9 previous jjregnaiicies. Her last menstrual jjeriod was 67 days before the operation. Pregnancy was suspected before the removal of the uterus, but a hysterectomy was performed because her periods had become verj' severe, lasting 8 days and causing faint ness and weakness.
This general plan of the formation of the larger subarachnoid canals reaches its maximum in the formation of the various cisternae for cerebro-spinal fluid. The process is probably best illustrated in the case of the cisterna magna, which persists in the posterior cerebello-bulbar angle. Figures 74 and 75, taken from an embryo pig 23 mm. long, give an idea of the initial formation of the cisterna cerebellomedullaris. The mesenchymal strands, as shown in figure 75, are already broken down in part, and are profusely covered with albuminous coagula. The process has not proceeded to any extent in this specimen of 23 mm., but in the course of the next 10 millimeters' growth extensive changes occur, as arc shown in figures 76 and 77, photomicrographs from an embryo of 32 mm. In the space outside the inferior membranous area the mesenchymal trabecula? have almost disappeared; the space — or cistern, as it should now properly be called — is almost completely filled with the clotted albumen. The mesenchjTne is seen running through this embryonic cistern as a few isolated strands, but most of the tissue appears now as a fairly definite membrane on the outer side of the space. This membrane will go to form the inner surface of the dura and the continuous outer layer of the arachnoidea, as it furnishes a visceral layer for the subdural space.




The two types of degeneration which the reticular magma undergoes have been considered above. The magma becomes granular and denser as it lessens and becomes li(|uid. The liquid again either coagulates or remains fluid when the specimen is fixed in formalin. The two fluid types maj' be related to the destruction of the amnion, but as yet I have been unable to reach a conclusion regarding this point.
More laterally in this same specimen the formation of the cistern has progressed to an even greater extent. In figures 78 and 79 the total freedom of the lower portion of the cistern from trabecular strands is seen; above, the mesenchjTne still sweeps down as a supporting structure for the chorioid plexus. A definite differential hne of mesenchymal condensation indicates the future outer border of the arachnoid as it incloses the cisterna cerebello-medullaris. This general process of mesench^Tnal breaking-down, altering the original small spaces into the larger arachnoid channels, holds as the embryo develops into larger forms.




The beginning of the formation of granular magma is shown in specimens No. otJO and 991 (plate 1, figures 6 and 7) as well as in Xos. 533 (plate 2, figure 3) and 6G0 ( plate 3, figure 5) . It appears to extend into the cavity of the amnion, and often forms great crusts, which surrovmd the embr3'o, as shown in several siiecimens l)ictured in my monograjjh on monsters (e. g., Xos. 79, 94, 104, 230, and 261). An extreme sj)ecimen of granular magma w'ithin the exococlom is shown in specimen Xo. 651</ (plate 1, figure 9).
In addition to this formation of the subarachnoid spaces in the adult through the enlargement of the embrA-onic mesenchjTnal spaces, the perimedullary mesench\Tne undergoes in these same localities condensations which result ultimately in the formation of the arachnoid membrane and the trabeculse diAnding up the cavum subarachnoidealo. Mention has already been made of the adhesion of the cellbodies of the disrupted mesenchymal elements to the persistmg strands — the initial step apparenth- in the ultimate differentiation of the mesothehal cells which line these spaces. Gradually ^vith the increasing growth of the embryo these cells seemingly become arranged in definite columns covering the persisting arachnoidal trabeculae. At the same time a differentiation of these primitive mesenchj-mal elements occurs, the cells ultimately being transformed into the very low cuboidal mesothelium of the subarachnoid spaces. This differentiation begins first in the basilar portions of the cranium and spreads upward, in a way similar to the course of development of the cranium and of the enlargement of the pericerebral spaces.




It is extremely difficult to determine with certainty the structure of the granular magma, but in studying pathological ova (especially those obtained from tul)al pregnancy) I have fre(|uentiy ol)served that there are large masses of granular magma which take on hematijxylin stain. The.^^e granules are mixed with a slimy mass which al.so takes on hemat<jxylin stain. My attention was called to the.se granules becau.se they have a characteristic circular stratification and contain within their centers small granules which also stain intensely. 1 am by no means certain whether all granular magma stains in this way with hematoxylin, and what T have just stated may apply only to a portion of the granular magma.
While such a general process as outUned accounts for the formation of the arachnoidal trabeculae and the subarachnoid spaces, it has but little bearing on the development of the outer intact membrane of the arachnoidea. This portion of the arachnoidea (which might be termed the arachnoid membrane as distinguished from the arachnoid trabecule) first appears as a distinct line of mesenchymal condensation separating the mesenchyme into the primitive arachnoid and dura mater, as in figures 76 and 77, dmc. This rather thin zone of cellular density in reality represents not only the outer surface of the arachnoidea, but also the inner surface of the dura mater. At first these develop in close fusion with a later separation of the two membranes. With this cleavage of the two surfaces, the arachnoid membrane rapidly differentiates, forming an intact layer over the subarachnoid spaces. The cells covering the surface membrane seem to change gradually into the low cuboidal type, similar to those covering the arachnoidal trabeculae. The details of these processes may be most easily studied in the region of the cerebral hemispheres; in this situation the transformation of the tissues occurs at a later period than in the basilar regions, for the differentiation of this mesenchyme follows the general plan of development of the cartilaginous and bony cranium.




Specimen Xo. 531 (siiowii in figure 8, i)late 1) has its ccrlom filled with a li(iuid ma.ss in which there is a granular dejwsit that surrounds the embryo anlage. Such specimens are numerous and, without opening them, they may frequently be recognized by the transparency of the chorionic wall, which is covered with but few atroi)hic villi. A more advanced embryo, showing the same condition, is shown in specimen Xo. 512. In it the embryo is atrojihic and macerated, without the l)iesence of an amnion. The chorion is thin and is fully covered with delicate degenerati'd villi. (Jther specimens which come within this group are Xos. 21, 7S. 122, and 244f/. The.se are all illustrated in my monograph on monsters.
The greatest problem in connection with the development of an external arachnoid membrane naturally concerns the separation of this leptomeningeal tissue from the pachymeninx. In the solution of this particular problem gross dissections have been found of benefit. For this purpose, pig embryos of larger size were used, and attempts were made to ascertain at what stage of development a true anatomical separation of the two membranes occurred. It was found that in embryo pigs of about 40 mm. the dura over the calvarium could be well separated from the arachnoid, but areas of unseparated tissue still persisted at this stage. This was also found to be true in pig embryos of 50 mm. ; on the inner surface of the dura at this stage a mesothelial cell pattern could be demonstrated, although areas of attachment to the arachnoid existed. However, the differentiation of the periaxial mesenchjTne into the adult arachnoid does not occur coincidently with the possibility of a forceful separation of the dura from the surface of the brain; ])ut before this separation of the pachymeninx can l)e made the mesechyme which will go to form the arachnoid must undergo some differentiation. This process invohcs a condensation or accumulation of mesenchymal elements directly in the secondary dural thickening; the cells, with oval nuclei, soon form a continuous membrane of two or three cells in thickness. Apparently soon after the cellular accumulation has been accom])lished, a separation of the dnni from the arachnoid may be made. In certain areas, varying greatly in size, there is still an intimate connection between dura and arachnoid. These connections are particularly prominent over the developing cerebral hemispheres, and it is with this differentiation in the formation of the arachnoid spaces that we will now deal.




In a human fetus of 76 mm. (No. 1134) the arachnoid was found to constitute, in the region about the great sagittal sinus, a cellular layer which adhered quite closely to the dura, even though a Une of difTerentiation between the two meninges could be made out. This adhesion could undoubtedly be separated, even bj' gross dissection, although the tendency to adhesion was stronger than the attachment of the pia to the cortex. From its cell-character and general histology' the arachnoid at this stage must be considered as a formed membrane, but in a primitive state.




A somewhat similar but more advanced stage in the formation of the arachnoid membrane is seen in a human fetus of 100 mm. (No. 928-E) and in a fetal pig of 114 mm. In both the arachnoid membrane is verj' cellular, adhering to the dura only along the superior longitudinal sinus and in certain isolated areas. The cells comprising the arachnoidea possess oval, rather large nuclei which stain palelj- with hematoxyhn. No typical arachnoidal trabeculae could be made out in specimens in this cortical region.


HUMAN MAGMA HKTICLLK IX NORMAL AND I'ATHOLOCIICAI. DFAELOPMENT. 23
t^ometimes the entire specimen is filled with a gelatinous mass, which becomes firmer when fixed in formalin and separates into a more sohd mass, and into a liquid when preserved in formalin. This mass appears to lie within the amnion in most specimens, as in cases where it fills the whole ovum the amnion is missing. Specimen Xo. G04 (plate 1 . figures 1 and 2) is (juite typical, as is al.so Xo. 13o. In Ijoth the embryos are atr()i)hic and necrotic, and the jelly-like fluid fell out with ea.se w'hen the ovum was cut open. The chorion is atrophic in both of them and is covered only with a few atrophic villi. Specimen Xo. 604 came to me without a history, and measures 70 by 50 bj' 50 mm. It is fully covered with fibrinous clots, between which there are few large villi, as the picture shows. The chorionic wall is 3 to 4 mm. in thickness, and its interior is entirely filled with jelly-like magma of uniform consistency. On one side of the specimen, lying free witliin the hyaline magma, is a straight embryo, 17 mm. in length, with atro])hic head, arms, and legs. The same description applies equally well to Xo. 135. Specimens like these are quite numerous in our collection of human ova, but usually the jelly is lost when the specimen is ojjened. Figures illustrating embryos of this sort maj' be .seen in mj^ paper on monsters, under the descriiition of embryos X'^os. 79, 94, 230, 261, and 270.


The cellular character of the arachnoid persists in the larger embryos and fetuses as a layer, several cells in thickness, constituting the outer arachnoid membrane. In a fetal pig 190 mm. in length the membrane was practically differentiated, its outer wall being covered by me.sotheUal cells with large nuclei lying about a small fibrous-tissue base. The arachnoid trabeculae were developed only in the larger sulci, where they appeared as typical cellular cords about a core of fibrous tissue. At this stage, too, the vessels traversing the arachnoid spaces were found covered with similar cells. These may now be justly termed the mesotheUal cells.


Xo. 1117 (plate 1, figure 5) contains an embryo well i)acked in the jelly-like magma. The cavity of the ovum is small and its wall is ^■ery hemorrhagic. The specimen came from a woman, age 26 j-ears, who was married at 15. She had two births at term and one previous abortion. She believed she became pregnant about 3 months before the operation, although she had not missed her regular i:)eriods.


Quite similar stages of arachnoidal differentiation occur in human fetuses of 200 (No. 870) and of 240 mm. (No. 1131). The arachnoid has everj-where practically become adult in character, except for a further decrea.se in the number of the peripheral layers of mesothelial cells. The fibrous tissue underlying this covering membrane possesses, as in the adult, almost a minimum of support.


Another specimen belonging to this category is X^o. 813. It consists of a fleshj' mole, well filled with tough, jelly-like magma. All the vilh are destroyed and its surface shows ulceration. Further study of this magma is necessarj' before it can be related to the granular magma which forms with the reticular magma in the exoccelom. I am inclined to believe that the hj-aline substance which is so often found within the amnion of pathological specimens arises from the amniotic liquid, w^hich has become richer in albumen, and therefore congeals into a jelly-likf ma.?s when preser^■ed in formalin.


In certain areas, however, the differentiation of the mesenchA-me into the adult arachnoidea does not keep pace with the general process. In the present study this phenomenon of unequal develoj^ment was especially well shown in fetal pigs of 150 mm. and upwards. It concerns the development of arachnoid trabeculae in the cerebral sulci. As is well known, the arachnoid membrane bridges the cerebral fissures, wliile the pia follows the cerebral contour. In the fetal pigs of the stages specified above, certain furrows showed a typical adult relationship with the covering arachnoid membrane and lining pia, the intervening space being traversed bj' definite arachnoid trabeculae. Other of the sulci were filled with an almost emb^^•onic tj-pe of mesench}Tne — a loose meshwork of cytoplasmic processes containing rather small oval nuclei. The explanation of this embryonic type of tissue seems to be that it occurs in the newly developing sulci and that some time must elapse in this formation before the tissue fully differentiates into the adult arachnoid membrane. Strangely enough, a similar collection of an embryonic type of tissue is sometimes met with, in these stages, between the two hemispheres.


CONCLUSION.


The general process, then, of the formation of the arachnoidea involves both a breaking-down (or thinning-out) of the mesenchymal spaces and a condensation of the cells. The first of these processes results in the transformation of the interstices of the periaxial mesenchyme into the larger subarachnoid spaces, divided off by arachnoid trabeculae; the second finds its final accomplishment in the development of the outer arachnoid membrane which, covered with mesothelial cells, forms the inner surface of the subdural space. The transformation begins in the basilar regions of the cranium and spreads upward over the hemispheres.


The fibrils forming reticular magma are always in direct continuity with those of the mesenchyme of the chorionic wall. This can easily be demonstrated by means of Van Gieson stain, and reticular magma must therefore be viewed as embryonic connective tissue extending into the cavity of the ovum. The stronger strands are accompanied more or less by mesenchyme nuclei, show^ing that the magma itself must be viewed as independent connective tissue identical with the mesenchyme of the chorion. As the amnion extends these strands are pushed aside, their final remnants being seen in that portion of the exoccelom which encircles the umbihcal cord.


THE CIRCULATION OF FLUID THROUGH THE SUBARACHNOID SPACES.


In pathological specimens the reticular magma increases in quantity in the earlier stages of development, continumg for a number of months of pregnancy. Frequenth' the meshes between the reticular fibrils are filled with pecuhar stratified granules which take on an extensive hematoxjdin stain. Often the amnion is


In view of the processes of differentiation involved in the formation of the arachnoidea and the subarachnoid spaces, the circulation of fluid through this pecuhar membrane must be considered. It seems important to ascertain, if jjossible, the relationships between the beginning of the passage of the cerebro-spinal fluid and the onset of the histological changes.




24 IIIMAN MACMA KKTICII-K IN NdU.MAl. AM) l'Ari|()|.()( ;i(AI, DKN i;i,< ll'MKNT.
The conceptions of the development of the circulation of the cerebro-spinal fluid which are presented in this conmiunication are dependent, in large measure, upon the results of the replacement of the fluid, in living embryos, by the ferrocyanide solution. Additional evidence was obtained from the identification of albuminous coagula in the periaxial tissues. The correlation of these findings with the development of the chorioid plexuses and with the results of injections under low I^ressures, from a syringe and so forth, gave evidence of their correctness.




destroyed e:irl>- in (k'veK)i)nient, in wliicli case the magniii may dissolve, but sometimes it inrreases {greatly in (luantity, forming a gelatinous mass. Freciuently pathological ova are encountered in which the development of the embryo is retarded, and the amnit)n is often found filled with a flaky deposit that, as time goes on. increases greatly in quantity and finally forms large crusts which invest the embryo. In other cases there is marked hydramnios, and in certain instances, where the amnion is destroyed, the magma dissolves, leaving only the embryo floating in the fluid encircled by the chorionic wall. Specimens are also found in which the ca\ity of the amnion is greatly enlarged and is filled with a jelly-like substance, which in later stages may form crusts encircling the embryo. The true relation between the ])athol()gical changes of the contents of the exocoelom and of the cavitv of the amnion remains to be determined.
The differentiation of the mesenchyme into arachnoid membrane may be said to keep pace with the establishment of the periaxial channels for the cerebro-spinal fluid. In the main, the passage of this fluid into the undifferentiated mesenchyme about the nervous system precedes the process of histological change. This phenomenon is shown in figure 14, from a pig embryo of 18 mm. The replaced fluid is seen passing out into the mesenchyni(^ through the two membranous areas in the roof of the fourth ventricle. The mesenchyme at this stage has already differentiated somewhat, but hardly in proportion to the length of time during which the fluid has been passing into the space.




There are several features of interest in the course of the fluid through the periaxial spaces. In sections of embryos in which the cerebro-spinal fluid has been replaced by a foreign solution the granules of the precipitated salts may be identified in the periaxial mesenchyme in situations corresponding exactly to the extent of th(! spread shown in the cleared specimens (figs. 1 to 9). The exact location of the prussian-blue granules is of importance in this connection, as the exact form and distribution of the periaxial spaces and their relation to the adult subarachnoid spaces may thus be determined.




Kxamiiiation of serial sections from an embryo in which the embryonic ventricular fluid has been replaced by the ferrocyanide will reveal, if the embryo exceeded 14 mm. in length, granules of prussian-blue in the peribulbar mesenchj'me (fig. 14). The granule.s are not found in any cell-bodies in this tissue; they are made out, in large measure, adhering to the mesenchymal cell-processes or lying free in the mesenchymal interstices. The granules do not penetrate the pia mater or the dura mater, a finding which will be discussed more fully in the sections dealing with these membranes. Everywhere the transit of fluid into the nervous tissue seems to be prohibited by the pia; in some areas, however, the outer condensation of mesenchyme to form the dura-periosteum has not j'et occurred. This is shown particularly well in the region of the roof of the fourth ventricle (fig. 18), where the epidermis offers the only barrier to the passage of fluid from the pericerebral spaces.


IIIM AX MAGMA RfiTICULfi 1\ NOHMAF. AM) I'ATHOI.OUICAI. I^KVELOPMENT.


In the earlier stages in which the phenomenon of fluid passage about the central nervous system may be observed, the outer layer of the arachnoid is not at all differentiated. Here the barrier to the fluid is the blastemal condensation of mesenchyme (fig. 16). In the later stages, when the outer layer of the arachnoid is beginning to appear as a mesenchymal thickening, the fluid (as indicated by the precipitated prussian-blue) is confined strictly within the immature arachnoid membrane.




The course, then, of the fluid which has replaced the cerebro-spinal fluid in the embryo follows that of the aduh cerebro-spinal fluid (as shown by the resultant blue granules). It is everywhere contained within spaces which topographicaUj' and embryologically correspond to the subarachnoid spaces in the adult. The spread of the replaced solution from the embryonic ventricle into the peribulbar tissue is analogous in every way to the passage of cerebro-spinal fluid from the fourth ventricle of the adult into subarachnoid spaces.




2.5
==VIII. A Consideration Of The Embryonic Pia Mater==




Our present conceptions of the embryolog}' of the pia mater are largely due to the work of His^^s) and of Kolliker^^D. who first firmly established the idea that this inner leptomeninx was mesodermal in origin. While generally accepted (Farrar'i^*). this view has not been widely referred to in the literature; but the absence from all embryologies of any information concerning the development of the meninges is quite striking and it does not seem strange, therefore, that our information regarding the pia mater has not advanced in keeping wnth our knowledge of the embryology of other structures of the body. In the present section of this conmiunication it is purposed to present merely a general consideration of the process by which the pia mater is formed and to point out some of its functional characteristics, especially in regard to the fluid channels.


BIBLIOGRAPHY.


The term pia mater is accepted throughout this article as designating solely the cellular membrane which adheres closely to the outer surface of the nervous sj-stem, but it is in direct connection with the arachnoidal trabecular which traverse the subarachnoid space Whether the two membranes should be considered together as the pia-arachnoid or as the leptomeninx is a question in regard to which there is some disagreement ; it will suffice to consider the pia as a separate membrane.






THE GENERAL HISTOLOGY OF THE PIA MATER.


Hhyce and Teachkr: Contributions to the study of the early development and inibeddinc of the hiimau ovum. Glasgow, 1908.


The findings in this mvestigation are wholly in accord with the conclusions of His(25), of Kolliker('i), and of Farrar^^^), that the pia mater is derived from the middle germ-layer. In the earliest stages the mesenchymal elements may be made out adhering to the outer i)ortion of the primitive nervous system. In the course of growth these cells are grouped about the mantle zone of the spinal cord in a rather dense laj'er, two or possibly three cells in thickness, with the tyjjical oval nuclei of the mesench5Tnal elements. Certain stages of this process may be made out in the figures in this paper. Thus, in a human embryo of 4 mm. (No. 836 of the Carnegie collection) the mesenchymal elements form a definite layer around the neural axis (fig. 41). The nuclei are oval in shape, possessing a moderate amount of chromatin, and are found in a layer two cells in thickness. This membrane, with its fairly scant cytoplasm, is sharply differentiated by its existence between two layers, in one of which nuclei are wanting, and in the other somewhat widely separated — the mantle zone of the spinal cord and the periaxial mesenchyme.


Eternod, a. C F. : La gastrulc dans la .seric animalc et plus speeialemeut ehcz I'hommc et les mammif6rcs. Tirage a part du Bull. Soc. Vaud. So. Nat., 1906, xi.ii, 150, Lausanne, 1906.


This typical arrangement of the mesenchj'mal elements about the cerebro-spinal axis holds in almost unchanged form throughout the whole embryonic growth. Thus, about the nervous tissue in figures 48 and 52 (from human embryos of 7 and 9 mm., respectively) the same condensation of the mesenchymal elements to form the pia mater are made out. This ajipearance is so familiar that further description in the later stages seems needless, but certain characters of this embryonic arrangement seem to require comment.


. Dea premiers stades de I'ocuf liumain et de son im
planation dans I'uterus. M6moirc present^ au premier Congrfcs ffdfratif international d'anatomic (Geneve, 6-10 aout 190.5), Nancy, 1906.


The general appearance of the pial layer is greatly altered by the early formation of the capillary blood plexus about the nervous sj'stem. This plexus tends to render the pial tissue more cellular, on first microscopic examination, as the endothehal channels branch greatly outside of the nervous tissue in this mesenchymal pia. The general character of the pial layer, however, as a membrane with prominent nuclei and scanty protoplasm, is not altered at all by the vascular plexuses.


. L'ceuf humain. Implantation et Ke.?tation tropho
derme et placenta. Mf-moirc publi6 ii I'occasion du Jubilfi de rUniversitt, l,i.59 1909, Geneve, 1909.


The ultimate fate of these undifferentiated mesenchjinal elements forming this initial \nti\ condensation is a gradual transformation of the cells into ver}' low cuboidal mesothelial elements constituting the adult pia. The transformation concerns not only the differentiation of the cells but also a rearrangement so that the original layer of two or more cells in thickness becomes finally of but a single cell in thickness. The jjrocess, in a way similar to the development of the subarachnoid spaces, begins in the basilar portions and spreads upward; the process, hence, may often be studied in a single suitable sjiecimen.


. In6galit6s de croissance du chorion ovulaire humain
et localisations consecutivcs en chorion laive et chorion frondosum. C R. de la Reunion de r.\s.-;ociation des Anatomistes (Nancy, 5-7 avril 1909). Lille, 1909.


More imjjortant, for our consideration, is the i)eculiar relationship of the pia mater to the roof of the foiutli ventricle, and in particular to the two area^ membranaceje. In this situation, in place of the slight mesenchymal condensation which characterizes the jjia, and which ^Minot''*"^ pictures in his figure 114, the mesenchyme seems altered. The condensation to form the pia, which takes place in other situations about the true nervous tissue, has not here occurred. This absence of the typical pial arnuigenit-nt may be noted even in very small embryos — those in which the roof of the fourth ventricle is composed of the many-layered, epitheUal-like cells. This is well shown in a photomicrograjih (fi}^. 53) from an injected human embryo of 9 mm. (Xo. 721) of the Carnegie collection. Likewi.sc, in this region in a [)ig embryo of 8 mm. (fig. 25), the same absence of a real pial condensation may be made out. But this peculiarity of the pia is most striking at the period of maximal differentiation of the superior membranous area in the rhombic roof. In figures 37 and 43, photomicrographs from pig embryos of this stage, the mesenchymal condensation, augmented by some vascular endothelium, is shown in adhesion to the ependyma on both sides of the membranous area; but directly behind the differentiated cells of the area membranacea evidence of a condensation of mesenchyme is whollj' lacking, even though both specimens show vascular channels in close appro.ximation. Similarly, in a human embryo of 14 mm. (No. 144, Carnegie collection) a total lack of the true pial thickening is to be observed (fig. 57).


Fkassi, L.; Ueber ein jungcs menschlicheaEi in situ. Archiv fiir mikroskopische Anatomic und Entwicklungsgcschichte, Bd. 70, 1907.


Quite similar is the failure of a pial thickening about the inferior membranous area. This can be made out in figures 83 and 87, from human embryos in which the process of differentiation of the area is proceeding. In later stages of the formation of the area membranacea inferior, the marked absence of a true pial condensation in the mesenchyme in this region is noted in figure 75 (a specimen from a fetal pig of 23 mm.) But this apparent failure to form the typical mesenchymal condensation of the pia mater in certain areas in the roof of the fourth ventricle must not be construed as indicating an absence of pia mater. Such does not seem to be the case here, for in the later stages of the formation of the cisterna cerebellomedullaris the area membranacea inferior is found entirely unsupported, except for a layer of mesenchymal cells. This is shown in figures 77 and 79, both taken from fetal pigs of 32 mm. This mesenchymal laj'er must be considered as pia mater apparently modified for a specific purpose.


GiACoMiNi, C: Probleme aus Entwickelungsanomalien d. Menschl. Embryo. Mcrkel and Bonnet "Ergeljnisse." iv, 1894.


The general process, then, of formation of the pia mater concerns a condensation of mesenchymal elements to form an embryonic membrane about the central nervous system. From its earliest beginning very slight modification is needed to reduce it finally to the histological character of the adult membrane. The general process holds, except in the regions of the area? membranaceic in the roof of the fourth ventricle; here, apparently, a modification of the pia for a specific purpose, involving an absence of the primary pial condensation, takes place.


CJrosser, O. : Eihiiute und der Placenta. Wien und Leipzig, 1909.


THE RELATION OF THE PIA MATER TO THE FLUID CHANNELS.


. The development of the egg membranes and the
placenta; menstruation. Keiljel and Mall, Human EmbryoloE}', i, 1911.


The cerebro-spinal fluid in its normal pathways comes everywhere into contact with the pia mater, which serves as the inner retainer for the subarachnoid space; therefore the functional relation of this membrane to the fluid which bathes it becomes of interest. To some degree the results of the experiments recorded in the earlier portions of this paper throw light upon the relation of the pia mater to the circulating fluid. The most important question in this connection is naturally that dealing with the possible penetration of the normal fluid through this embryonic membrane. In this regard the findings in replacement experiments with ferrocyanide solution serve to elucidate the problem. These observations give no evidence of any penetration of the pia mater by the fluid. This is well brought out in figures 14 and 18. In every respect (as demonstrated by numerous experiments of this type in pig embryos of varying lengths) the pia mater is wholly impenetrable to true solutions of foreign salts when injected so that the normal tension is not altered. The whole subarachnoid sjiace may, in such an experiment, be filled with the prussianblue, but none of these granules are found within the cells of the pia mater or in any layer between these cells and the nervous system. Evidence that the fluid has bathed the outer pial cells is afforded by the adhesion of granules of prussian-blue to the outer cytoplasmic borders of the cells.


Herzog, M.: a contribution to our knowledge of *he earliestknown stages of placcntation and embryonic development in man. American Journal of Anatomy, ix, 1909.


Likewise the cells comprising the embryonic pia have been found to be impenetrable to true solutions (ferrocj'anide) when injected under varying pressures from a syringe. In these cases, rupture of the roof of the fourth ventricle or of the infundibulum may be produced by great pressure, without causing any of the fluid to penetrate the intact layer of the pia mater. The same result is obtained when india ink is substituted for the true solution.


His, W. Anatomie menschlicher Embryonen. Leipzig, 1880-1885.


The pia mater, then, even in its embryonic form, serves as an efficient fluidbarrier. This is demonstrated, in regard to the adult pia mater, in the report^^^) of the observations made on adult cats, dogs, and monkej's. But the barrier which the pia offers to the entrance of fluid from without exists also for fluid coming in the reverse direction. This is shown by the well-known phenomenon of the so-called subpial extravasation, which occurs in blood vascular injections when the injections are continued for too long a time at too high a pressure. The perforating vessels in such cases rupture as they enter the nervous system, and the injection mass spreads extensivelj' beneath the pia, stripping it away from the nervous tissue. Of interest in this discussion is the fact that the injection mass in these extravasations does not rupture the pia, which seemingly is an equallj' efficient fluid barrier to pressure exerted on it from within. Similar subpial spreads of the injection fluids have been observed in the course of this work. These extravasations resulted from the rupture of the whole nervous tissue from within, particularly in the region of the infundibulum, when the inj(>ction was made into the ventricular system under excessive pressure. In this respect, too, the i)ia seems to be wholly efficient as a retainer for true solutions or for granular suspensions. It is realized that the embryonic pia mater will not resist the passage of fluids through it under the highest pressures afforded by the syringe, but the membrane serves as an efficient barrier for all pressures such as are employed in careful anatomic injections.


HocHSTETTEB, !•'. : Bildcr der ausseren Korpcrform einigcr menschlicher Embryonen avis den beidcn crsten Monaten der Entwicklung. Mimchen, 1907.


With this conception of the impenetrabihty of the pia mater to fluids under ordinary pressures, it does not seem strange that there is a variation in the process of formation of the pia mater in the region of the roof of the fourth ventricle. It has been shown in the foregoing paragraj^hs that the phenomenon of mesenchymal condensation which results in the formation of pia elsewhere does not occur in the region of the two area? membranacefe. In view of the passage of cerebro-spinal fluid through these two membranous areas, the pia mater must necessarily be altered in these places. For were it not adapted to the jjurpose of affording fluid passage the cerebro-spinal fluid would, in its course from the ventricle to the subarachnoid space, form a subpial extravasation. It would seem that this modification of the pia is designed to meet the particular need and function of this region.




==The Adhesion of the Pia Mater to the Cerebral Tissue==




Ingalls, N. W.: Beschrcibung cines menschlichen Embrj-os
It is a well-known fact in embryology that tlie pia mater and the peria.\ial mesenchyme in poorly dehydrated specimens split away from nervous tissue, but in adult preparations, if the meninges and brain are dehydrated in a block, the separation of the tissues occurs between the dura and the arachnoid, or (in more exceptional instances) the dura and arachnoid come away, leaving the pial layer closely applied to the cortical tissue. It is quite difficult in any adult mammal to separate the pia from the brain tissue. Realization of these peculiarities in the degree of adhesion of the pia led to an attempt to ascertain what structures were involved in the attachment of this mesodermal layer to the epidermal nervous system. The results of this attempt add nothing to the ultimate solution of the problem, but are perhaps of sufficient interest to justify brief presentation.
von 4:9 mm. Archiv. fQr mikroskopische .\natomie
und EntwicklungsKcschichte, Bd. 70, 1907. JOH.NSTONE, R. W.: Contribution to the study of the early
human ovum. Journal of Obstetrics and Gync
cologj' of the British Empire, 1914. Juno, P.: Ei-Einbettung Iieim menschlichen Weibo. Berlin,
1908. Keibel, F. : Die aussere Korpcrform von Affenembrj-oucn.




.Selenka, Entwickelungsgeschichte, xiv, Wiesbaden,
Two theories in explanation of this adhesion of the pia immediately suggested themselves. One of these concernod a possible growth into the pia of neuroglial elements, causing an intimate association between the pia and the cerebral cortex. Our findings in reference to the neuroglial outgrowth in fetal pigs gave no reUable basis for the assumption. The second theory dealt with a diminution in the elasticity of the walls of the perforating blood-vessels which supply the nervous system. The early embryonic vessels, with walls comjjosed solelj- of endothelium, when subjected to the distortions of poor dehydration, might possibly offer less resistance to the separation, so that the pia would come awaj' from the nervous tissue. In the later stages, however, the thicker-walled perforating vessels w^ould naturally oppose such a cleavage, so that the pia would remain adhering to the cortical tissue. This theory is also purely an hj^Dothesis, although it does not seem unlikely, especially if one takes into account a possible connection of the pia with the perivascular system. In examining blocks of the meninges and brain tissue taken together it was found that the pia mater separated cleanly from the nervous tissue in fetal pigs 15 cm. in length. Beyond this stage the arachnoid might remain in adhesion to the dura, but in such cases there was always found a layer of cells on the outer side of the cortical tissue, constituting a true pia mater.
1900. . The formation of the germ layers and the gastnila
tion problem. Kcibcl and Mall, Human Embr>-
ologj-, I, chapter v, 1911. Keidel und Elsa: Normentafel zur Entwicklungsges
chichtc des Mcnschen. Normentafel zur Entwick
lungsgcschichte der Wirbcltiere. Achtcs Heft.




Jena, 1908. Lewis, F. T.: The development of the intestinal tract and
==IX. The Development Of The Cranial Dura Mater==
respiratory organs. Keibel and Mall, Human
embryo




logj', ii, 1912. Mali., F. P.; Origin of human monster.^. Journal of Morphology, XIX. Published also as a monograph by
The dura mater, like the two other meninges, is derived from the mesenchj-me about the central nervous system. The researches of Sterzi'"' on the comparative anatomj" of the meninges furnish additional evidence for this conception in the higher mammals. The origin of the pachymeninx from the middle germ-layer is now well established. But there is lacking in the literature a comprehensive account of the formation of this fibrous envelope. The gross generaUties of the process are given in pari, liut there is an almost total absence of the more intimate details of the transformation. One of the most essential points in the process concerns the relationship of the dura to the bony coverings of the cerebro-spinal axis. Does the adult dura serve as the periosteum of the bony skull? In the standard text-books of anatomy the adult human dura is described as being composed of two layers. In the skull these layers split, to comprise the walls of the great venous smuses. The outer layer of the dura serves as the periosteum for the bony skull, but below the foramen magnum the two layers separate to inclose the epidural space. The outer dural layer in this spinal region adheres to the inner surface of the bony vertebral column, where it functions as the periosteum; the inner layer here becomes the spinal dura.
theWistarlnstituteof Anatomy, Philadelphia, 1908. . Development from the connective tLssue of the  
syncytium. .Vmerican Journal of Anatomy, i, 1901. . On the fate of the human cmbrjo in tubal pregnancy.




Publication No. 221, Carnegie Institution of Washington, 1915. Peters, H.: Einbcttung des menschlichen Eies. Leipzig und
In this account of the adult dura mater there is indicated a very suggestive periosteal relationship which implies an embryological basis for the disposition of the two laj-ers of the membrane. It must be granted, however, that this division of the cranial pachymeninx into two layers is quite arbitrary; there is nothing in the general histology of the fibrous covering to suggest such a halving except its division about the sinuses and its spinal relationships.
Wien, 1S99. Retzius, G.: Ueber das Magma rfcticulfe des menschlichen
Eies. Biologische Untersuchungen, I, Stockholm,
1890. Stbahl and Beneke: Ein junger menschlicher Embryo.




Wiesbaden, 1910. Velpeau, a. L. M.: Embryologie ou Ovologic Humaine.
THE GENERAL PROCESS OF THE FORMATION OF CRANIAL DURA.




Paris, 1833. Waterston. a young embryo
The first evidence of the development of the pachymeninx is found in the basilar region of the skull, where the mesenchyme thickens, to form eventually the bony covering of the brain. This thickening of the mesenchymal elements results not only in the formation of the chondro-cranium, but also in the final formation of the bony skull and possibly its internal periosteum and dura. In the process of differentiation the condensation of mesenchyme in the early stages gives no index of the varied character of the resultant tissues, so that, in the first place, the study of the process was necessarily related to the more adult stages. In this paper, however, the whole history of the dura will be detailed chronologically, beginning with the earliest stages.




with 27 somites. Journal
Bardeen(2) has given data on the first appearance of the mesenchymal condensations which go to form the blastemal phenomena in both the cranial and spinal regions. The blastemal vertebra? become fairly well differentiated in human embryos during the first month of intra-uterine growth. At the end of the first month, in the occijiital region, three fairly well-marked occii)ital myotomes may be made out; these afterwards disappear. "During the early part of the second month the membranous anlage of the skull becomes extensively developed. The roof of the cranial cavity is formed by a dense membranous layer, which fu'st becomes marked at the side of the head in embryos from 9 to 11 millimeters in length" (Bardeen).
of Anatomy and Physiologj', xlix, 1914.




These evidences of a primary mesenchymal condensation about the central nervous system are concerned in the problem of the differentiation of the dura only in so far as they indicate the onset of the process which will give rise to the bone and possibly the periosteum — a part of the dura about the cerebro-sjjinal axis. Gaupp^'^) has already pointed out that this cranial l)lastemal condensation gives rise to these adjacent Init wholly different structures. These cranial mesenchymal condensations persist in simple form until after the cerebro-spinal fluid begins to fill its extraventricular bed; then, within a short time, the tissue becomes transformed by the development within it of cartilage, so that in the human embrj^o the caudal half of the chondro-cranium forms a ring of cartilage about the posterior portion of the brain. On the inner side of this ring of cartilage the mesenchyme later shows a marked condensation in the midst of the rarefied perimedullary tissues. In this layer the nuclei soon become fewer in number and the cytoplasmic structures fibrillar, the whole resulting ulthnately in the formation of the fibrous adult dura. The mesenchymal condensations in the regions of the skull, where membranous bone formation holds, go directly into a membrane of fibrous tissue, in the outer portions of which bone is laid do\\Ti. The details of these processes will now be taken up.




In figures 30 and 32, photomicrographs from pig embryos of 13 and 14 mm., respectively, the well-established vertebral differentiations and the now poorly differentiated base of the skull are shown. From this stage upward the mesenchymal condensation in the head region proceeds rapidly. Thus at a stage of 17 mm. in the human embryo (fig. 60) the ventral portion of the vertebral canal has become cartilaginous, while the base of the skull has also undergone the chondrogenous transformation in its mere posterior portions. But of especial interest in our problem is the line of mesenchjTnal condensation, which may now be traced whollj^ around the brain-stem and hemispheres (fig. 60). The nature of this condensation is well shown in figure 61, an enlargement of the squared area of figure 60. The mesenchymal nuclei have become closely packed and rather sharply differentiated from the looser mesenchj^me which in part goes to form the arachnoidea. Figure 59 similarlj' shows this condensation proceeding upward to the vault.


EXPLANATION OF PLATES.


Examined in another plane, the process of mesenchymal condensation seems to proceed much more rapidly in the posterior than in the anterior region. This is brought out in a transverse section of a human embryo of 18 mm. (fig. 62). Here the condensation is much more extreme about the medulla and roof of the fourth ventricle than in the more anterior parts of the mesencephalon. The same general appearance, typical of this stage, may be made out in figures 56 and 57 from a human embryo of 14 mm.* (No. 144, Carnegie collection). In the slightly larger stages the process of mesenchj^mal condensation about the nervous system becomes rapidly more marked. This increase in the number of cells comprising the denser membrane is shown in figures 64 and 65, photomicrographs of embryo No. 460 (21 mm.).


Plate 1.


The degree of condensation of the mesenchyme in the various stages of the human embryo is followed quite closely in the pig embryo. The comparative degree of differentiation coincides within a millimeter or two. Thus, in a section from a pig embryo of 19 mm. (fig. 38), the degree of condensation about the roof of the fourth ventricle is practically similar to that in human embryos of the same length.


Fl<58. I, 2. Pliotograplis of the two lialves of o\-um Xo. 604. Natural size. Tlio cavity of (lie ovum is filled with a
jelly-like substance in which a pathological embrj-o is embedded. Fig. 3. Section through a normal ovum Xo. 836, encapsulated in the decidua. X3}. Drawn by Mr. Didusch.


The phenomena just commented upon represent the stages concerned in the formation merely of a cranial blastema and are related to the formation of the dura only so far as it is out of this mesenchymal condensation that the periosteal portion of the pachymeninx may be derived. The degree of condensation referred to in the figure.s has been solely of the blastemal type, but in some of the specimens this simple condensation is seen only in the more cephalic portions of the cranium. Thus, in the figures (64 and 60) taken from embryo 460, the mesenchymal condensation is still of the simjile undifferentiated type, whereas in this same embryo the more caudal sections show a chondro-cranium which is well develojied. The i)rocess of formation of the cranial dura, then, is one which begins in the basilar j^ortions of the cranium and proceeds from these points into the region of the calvarium. In general, all of the phases of this transformation into dura may be found in one specimen of sufficient and suitable size, the basilar differentiation re]:)resenting the advanced stages, while the steps in the differentiation are found in the areas nearer the vertex.


The embryo lies within the coelom, and bands of -magma fibrils radiate from the amnion to the chorionic
wall. The head of the embryo shines through the more transparent portion of the amnion. Fig. 4. Photograph of a block of the ovum, Xo. S36, in situ after the embryo had been removed. X25. The
.«up|)orting strands of magma are strikingly shown. Fig. 5. Pathological embrj-o Xo. 1117, embetlded in hyaline magma. X4. From a tubal pregnancy following
gonorrhea (?J. Fig. 6. Pathological o\Tim Xo. 560, containing a great quantity of reticular magma. X2i. The embrj-o is
normal in form. From a case of retroversion of the uterus. Fig. 7. Pathological ovum Xo. 991, with the cavity completely filled with reticular magma. Xatural size. The
embryo is normal inform. From a negro woman. Sect ions of the embryo indicate that it Ls macerated. Fig. S. Pathological ovum N'o. 531, containing a granular deposit around a nodular embryo. X 1 §. Fig. 9. Pathological ovum with a nodular cmbrj-o (651?). X2. The exocteloni is gorged with granular magma. Fig. 10. Specimen Xo. 636. X"2J. The embryo and chorion are normal in form, but the reticular magma is
marketlly increased in quantity.




Pl.\te 2.
* Measured on slide after sectioning.




Fig. 1. Pathological embryo Xo. 512, lying free within the ovum. X6. The villi are tliin and scattered and the
It is quite difficult to decide exactly what importance the primary condensation of mesenchyme maintains in the formation of the dura, because, coincident with the chondrification of the blastema, there occurs another condensation which forms the line of division between the inner surface of the dura mater and the outer arachnoid membrane. The first evidence of this secondary perimedullary condensation is found in pig embryos of about 17 mm. In these specimens, in the narrow space formed by the mesencephalic flexure, mesenchymal cells collect together in the form of a fairly definite membrane. After its primary beginning in this area, the narrow line of its thickening may be traced to the basis cranii in embryos a little larger. In slightly older stages this secondary fine of condensation is found to be fairly extensive throughout the area between the middle and posterior cranial chambers.
embryo is atrophic. There is no formed magma. Fig. 2. .\n ovum, Xo. 576, obtained from tubal pregnancy, showing a delicate layer of magma fibrils around the  
attachment of the umbilical cord to the chorion. X3. Fig. 3. Ovum Xo. 533, showing ven.- extensive magma. x6.




Pl-.\TE 3.
At a stage of 20 to 21 mm. the whole basilar portion of the cranium shows evidence of this secondary line of condensation lying between the pia mater and the cartilaginous skull. The condensation occurs in the outer portions of the loose tissue which, as shown in a foregoing section (No. vii) becomes the subarachnoid trabeculae. The line of condensation is not broad on section; it comprises a cell-layer from three to six cells in thickness. Between this cellular border and the cartilaginous skull the tissue rapidly differentiates (a process seemingly sj^nchronous with the develojiment of this membrane). This tissue, which maintains dural relationships, is far more cellular and compact than the original perimedullary mesenchyme. Even without the rather dense line of division in the mesenchymal ti.ssue, the dural structure can be easily outlined by its characteristic apjiearanco.




Fig. 1. Ovum Xo. 545. X7. There is a delicate network of fibrils below the amnion and the chorion.
The original dural condensation between the two wings of nervous tissue which unite in the me8encei)halic flexure can be traced in slightly later stages around into the tentorium cerebelli. This structure develops as a wholly similar mesenchymal thickening in the midst of the jK-rimeduliary mesenchyme. The tentorium consists in these embryos of 20 to 2") nun. of two thin lateral plates which widen at their cranial attaclunents into prismatic areas. These areas, which finally lodge, in the two layers f)f dura, the sinus transversus, arc characterized by the same dense tyjie of mesenchyme. The jx-ripheral edges of the prismatic jwrtion of the tentorium sjjreads caudalwards as a definite line of condensation. In the earlier stages this line becomes indefinite as it extends from its teiitori:il attachment, but finnllv a similar line of condensation about the whole posterior chamber may be made out. This lies within the area of the cartilaginous skull and bounds the subarachnoid spaces.




Fig. 2. Kmbryo Xo. .588. XS. Delicate strands are shown radiating from the umbilical cord and yolk-sac. This figure is given to show the appearance of magma in vesicle development. From a woman who has IukI numerous me<;hanical abortions performed upon herself. Uterus badly inflamed.
This same process of formation of dura holds for the formation of the basilar dura in the more anterior portions of the cranium. The appearance of the .secondary zone, narrow and rather dense, may be made out inclosing the more cellular mesenchyme w^hich extends to the cartilaginous skull. The same process also endures for the formation of the dura of the calvarium, but here the addition of tissue from the undifferentiated mesenchyme is undoubtedly very small in amount. This will be discussed in a later paragraph. The various stages in the formation of this secondary condensation which goes to form the major portion of the dura may be fairly well studied in any one embryo of suitable stage, because the process, as pointed out above, begins in the basilar portion of the cranium and extends upward. Likewise, the condensations directly beneath the region of the dorsal membrane are delayed as compared to those of the lateral regions.




Fig. 3. Section through the chorion and magma of Xo. 402. X280. The specimen was stained with Van Gicson stain and shows that tlie fibrils of the magma are continuous witli those of the mesenchyme of the chorionic wall. It came from a case with subinvolution and symptoms of endometritis.
Some of the phenomena shown in the formation of the dura mater are illustrated in figures 46, 76, 77, 78, 79, and 94. Throughout these figures the letters dmc refer to the Une of the secondary mesenchymal condensation, which borders internallj^ the dura and which goes to form the outer membrane of the arachnoidea.




Fig. 4. Outline of the ovum of Xo. 660. Xatural size. The diagram indicates the part of the specimen shown enlarged in figure 5.
In figure 46, a photomicrograph of a pig embryo of 32 mm., the dura mater (dmc) is shown as a somewhat condensed tissue separated a sUght distance from the chondro-cranium. On the basilar surface, the inner line of dural tissue is quite remote from the inner surface of the basioccipital. Tracing this line of condensation forward, it is soon seen to merge more closely with the perichondrium* of the basioccipital. More anteriorly it again leaves the occij^ital plate and after a brief interval it fuses with the temporal perichondrium. Continuing slightlj^ more anteriorly the dural process toward the mesencephalic angle maj' be made out; this appears as a doubled membrane at its basal attachment. In its further prolongations the dural surface is at times a distinct structure; at other times it is completely fused with the perichondrium.




Fig. 5. Xo. 060, showing ver>' extensive changes in the magma. X6. The upper tip of the amnion is shown. The magma is fibrillar and granular, and at places the fibrils seem to form membranes. The chorionic wall is very hemorrhagic.
Posteriorly, in this figure 46, the line of dural condensation (incorporated also with the outer arachnoidea) may be traced upward around the cisterna cerebeUomedulla




ris. This hne of condensation is seen to lose its definitive character as it curves inward toward the chorioid plexus of the fourth ventricle — a phenomenon shown particularly well in figures 77 and 79, taken from the same pig embryo of 32 mm. The dura in this termination maj' be said to be in its formative stage; but dorsally, over the mesencephalon, the inner surface of the dura again becomes a definite membrane, as shown in figures 76 and 78. In the latter figure it is shown inclosing a wide mesh of dural vessels, between the arachnoidal surface and the membranous skull. Anteriorly, again, it seems to lose its definite hne of condensation.




•The term "perichondi.L n" is used tliroiighout this paper to designate only the ver>' dense cellular line delimiting the edge of the cartilngo. This dense zone is composed of the nuclei of the cartilage, crowded together, and represents probably some phenomenon of the growth or resorption of the cartilage. In a much broader sense, the whole dural tissue, lying between the line of secondarj- condensation and the cartilaginous border, could be termed "perichondritmi," as it probably represents the sole internal membrane which could be stripped from the cartilage.




1i ^•%-^^_ _ ^


Quite similar pictures are obtained regarding the dura mater in the human embryo. The relationships of the dura to the cisterna cerebello-medulla




ris are shown in figure 94, a photomicrograph of a human fetus of 35 mm. (No. 199 in the collection of the Carnegie Institution). In this rejiroduction the line of secondary mesenchymal condensation (representing the outer membrane of the arachnoidea and the inner surface of the dura) becomes widely separated from the occipitale superius in its superior portion.




Fi;?. 1 C604)
In a fetal pig of 8 cm. the same general arrangements of the dura mater could be made out. The inner surface of the dura was in places still fused with, the outer arachnoid membrane, but in other places the areas of attachment were lacking, so that a true separation of arachnoid from dura had taken place. Along the ])eripheral points of the tentorium the dura and arachnoid were still closelj' a{)plied to each other. The dura itself was of the same cellular, rather loose tissue, with a dense inner surface. In i)laces, as described in the younger stages, the dural tissue was incorporated with the definitive perichondrium over certain cartilages or even over parts of the same structure. In other places a definitive perichondrium may be wholly lacking; in these areas the indefinite cartilaginous border gradually merges into the dura. In still other situations an intermediate arrangement of dura and perichondrium exists, where the cartilage is bounded bj' a somewhat condensed but not fully developed perichondrium which is continuous with the dura. Everywhere in the membranous sutures between the cranial cartilages or bones, the dura bridges the gap as a loose, cellular tissue. Over the calvarium the dura appears solely as a dense, rather fibrous membrane which is incorporated with and serves as the inner periosteum. This dura over the hemispheres is continuous with the fibrous sutures of the cranial vault.




The findings in a fetal pig of 98 mm. were not dissimilar to those just recorded. The dura was everywhere quite well developed, a rather loose cellular tissue except over the hemispheres, where it showed a more fibrous character. In the region of the occipito-atlantoid ligament the dura was fused with the ligamentous tissue, while above (over the occipitale superius) the dura became a distinct, thick cellular layer. The structure of the tentorium was wholly similar to the occijiital dura. In the basis cranii there are areas in which the dura is wholly fused with the periosteum or jierichondrium; in other areas it bridges the sutures or exists as a definite membrane on the inner surface of a definite perichondrium.


Fi-. 2 (604)


The dura mater in a fetal pig of 15 cm. did not vary greatly from those larger stages already tlescribed. The tissue, however, had become somewhat more fibroiis. The prismatic attachment of the tentorium was no longer as large proportionately, but the dura lining the occipitale superius remained a thick bulbous swelling on the dorsal surface. Hut most striking of all the features in the specimen was the very dense fusion of the dura of the calvarium with the fibrous sutures of the cranium. No line of demarcation betwecm dura and fibrous suture could be made out; the two fibrous layers are anatomically one structure.




Fig. 3 (836)


The falx cerebri forms in the pig and human embrj-o by a process similar to that of the inner portion of the dura mater. In the sulcus between the two cerebral hemispheres the mesenchyme remains undifferentiated until quite late; then there appears in the posterior portion a narrow zone of condensation which soon presents two lateral surfaces separated by a layer of rather loose cellular tissue, similar in all regards to the dural tissue already described. This zone of condensation spreads forward to comprise the whole falx. The double surfaces of this membrane finally separate into two parts, forming the outer surface of the arachnoidea and the inner surface of the falx. At the cranial attachment of the fabc, the loose tissue forms a prismatic base, containing the suj)erior sagittal sinus and spreading laterally over the denser dura of the calvarium. The whole appearance of this region, which will again be referred to, is that the falx has been added onto the dura of the vertex. Its time of initial appearance is later than that of the rest of the cranial dura and there is apparently no additional acceleration of development. Hence the dural tissue in the fabc cerebri presents, in appropriate stages, a more immature type of differentiation than does the adjoining dura.




The process of the formation of the dura is not wholly a simple one due to the relation of the adult dura to, or its function as, the inner periosteum of the skull. In the figures already referred to, the almost complete fusion in some areas of the inner Une of dural condensation with the perichondrium has been commented upon. In other situations definite separations of the inner dural surface from the perichondrium occurred; in still other regions no perichondrium could be made out as a definite membrane. These differences in relationships of the dural tissue to the line of the perichondrium can not at present be wholly explained, but some indication of the meaning of the process can be given.




ri.cr. 4 (836)
Out of the original cranial blastema, as described by Gaupp'^i^\ there develops the cartilaginous and bony skull, the periosteum, and the dura. But the observ^ations recorded above indicate that by far the major portion of the dura is formed by a secondarj^ mesenchymal condensation, which was indicated bj' a thin zone of more condensed cells on its inner border. This inner zone ultimately separated to form the inner surface of the dura and the outer membrane of the arachnoidea. The tissue included between this inner line of condensation and the cranial wall gradually differentiated into a more condensed but still a loose cellular tissue and finally became a fibrous dura.




In all cases the dural tissue extends from the inner line of condensation to the cranial blastema, to the perichondrium, or to the cartilage of the skull. The presence of a definitive perichondrium can not at present be explained, but apparently the perichondrium is manufactured by the cells of the original cranial blastema and not by the dural tissue which lies in approximation to it. When a definite perichondrium is found, it seems quite uninfluenced by the dura; at other times a fusion of an indefinite perichondrium with the dura seems to occur. The fusion of the perichondrium with the dural tissue derived from the secondary mesenchjinal condensation may occur, so that the small outer portion of the dura may be derived from this laj^er. The findings, however, in this investigation, are against any addition of perichondrium to the dural tissue; histologically, a definitive perichondrium is a membrane entirely apart from the dural condensation.




Over the cerebral hemispheres the dura of the cranial vault offers more difficulties of study than does that of the basilar regions. With the formation of a blastcmal condensation over the whole vertex — an extension of the dorsal membrane to form the membranous skull — there occurs very quickly a condensation to form the dura. This condensation may be first detected as a continuation anteriorl)'of the leaflet of the tentorium cerebelli, which stretches forward from the prismatic zone of the tentorial attachment. This zone of condensation is wholly similar to the narrow line of the mesenchymal thickening which was found in the more basilar regions of the skull. This zone of condensation occurs just wathin the cranial blastema and may be traced upward over the mesencephalon and laterally around the rapidly enlarging hemispheres. As the distance from the cerebellar attachment increases, the zone tends to approach the blastema, except in those regions in which the precursors of the dural veins occur. In such a situation this inner dural zone swings inward to encompass the vessels. Between this inner line of the dura (representing also the outer surface of the arachnoid) the same rather loose cellular tissue exists.




lis- 5 (in?)
From the fabc cerebri a zone of dural condensation in the mesenchyme spreads laterally also; this gradually may be traced anteriorly and laterally until fusion with the similar lines of condensation from the basis cranii and the prismatic zone of the tentorium are reached. The condensation connected with the falx cerebri, however, is not an extensive process, the greater part of the hemispheres being covered by the development from the basis cranii and from the tentorium. It must be understood, however, that there is no active migration of this line of condensation, for the whole process is a development in situ. The appearance of an active extension is derived solely from the study of various stages and the increased area of condensation appears as an increment which has developed at the terminal points of the previous condensation.




The amount of dural tissue delimited in the mesenchyme by the secondary zone of condensation is not great in the region of the vertex. It is a thin layer which fuses to the inner surface of the cranial blastema. At the stage of this fusion the blastema has become somewhat fibrous and it constitutes the membranous skull. In this fibrous tissue (the union of the blastema and the dura) bone is deposited, but only in the outer layers. The phenomenon is easily studied in any suitable stage, for the sutures between the flat cranial bones remain incorporated with the inner memliranf!— the dura which includes the periosteum. Hence, over the cranial vault, the dura and periosteum become incorporated as a single membrane; this serves as the membranous .skull, into the outer layer of which bone is deposited.




In the basis cranii, as soon as ossification of the cartilaginous skull takes place, tin; dura becomes inc()r|)orated as the periosteum in a manner similar to that which takes i)lace in the cranial vault. While no definite relationship of dura to the peri-chondrium could bo made out in the earlier stages, the later function of the dura as the inner cranial periosteum is Cjuite obvious. Thus the adult relationships of the dura are obtained. But it is quite difficult to decide to what extent the dura (or internal cranial periosteum) is derived from the primary cranial blastema. It seems probable that this blastemal condensation, in its final resolution into bone, may contribute, in the form of a periosteal element, somewhat to the formation of the dura. Such an addition is verj' difficult of verification; certainly the greater part of the dura is derived by the secondary condensation from the perimedullary mesenchyme.


Fig. 6 (560)


Before giving details of the fibrosis of the dura, it may perhaps be interesting to point out a peculiarity of the primarj' cranial blastema, which does not seem to be connected directly with the formation of the dura. This concerns the tendency of the membranous skull to form more than one layer in its original zone of condensation. In certain areas, as in figure 64, from a human embrj^o of 21 mm., the dorsal membrane is .shown split into two layers. Somewhat similar to this is the occurrence of two zones in the cranial blastema of a pig embryo of 23 mm. (figs. 22 and 101). Inthis latter figure a less cellular outer layer and a more cellular inner layer are seen. Neither of these have particular significance in the formation of the meninges, although the inner layer in early stages actively functions as a fluid retainer.




Pig. 7 (991)
The question of the development of fibrous tissue in the dura mater in the course of its development requires consideration here. This phase of the problem concerning the formation of the pachymeninx has been followed, in this stud}', in the dura of the vertex about the sinus sagittalis superior. The tissue was removed in blocks, including the meninges and cortex cerebri, and was then sectioned in the coronal plane. For the most part the deposition of fibrous tissue was studied in sections stained with hematoxyhn and eosin; the findings were controlled by treating other sections from the same blocks with IMallory's connective-tissue stain. In this way the general histogenesis of the dural tissue could be well investigated.




Sections from such a block from a human fetus of 76 mm. (Xo. 1134, Carnegie collection) showed the dura to be composed of fibrous tissue everywhere except in the region of the great sagittal sinus. About this sinus an immature, almost embryonic, tj'pe of loose mj-xomatous tissue was observed. The fibrous tissue comprising the dura elsewhere is of a quite cellular, somewhat immature type of white connective tissue, with a considerable number of true fibrils. A wholly similar picture is found in a section, stained by Mallory's method, of a block from a fetal pig of SO mm. (fig. 104). Unfortunately the cellular character of the fibrous dura is not brought out, but the photomicrograph shows well the avoidance of the lateral walls of the sinus by the process of fibrosis. The more embryonic type of tissue in the region between the hemispheres is also well presented.




The dura mater of a human fetus of 100 mm. (No. 928-E, Carnegie collection) possesses fewer nuclei in a given area than does the dura from the specimen of 76 mm. (No. 1134). The tissue is fibrous, except in the immediate region of the sinus sagittaUs superior; but interspersed among the connective-tissue fibrils are many stellate or spindle-like nuclei, greatly exceeding in number the nuclei found in the dense dura of the adult. Bone is being laid down in the outer portion of this dura where it merges \\dth the membranous skull. The lateral walls of the great sinus are still free from fibrillar depositions. A somewhat analogous picture is afforded b}' a photomicrograph of a specimen stained after IMallory's method, from a fetal pig of the same length (fig. 105). In this specimen the outer portion of the dura, incorporated into a part of the membranous skull, is quite dense with the fibrous tissue; about the superior sinus, however, the decrease in the amount of fibrous tissue is very striking. The falx is beginning to exhibit a fair degree of fibrillar structure; it forms a definite division between the two hemispheres.




In the larger fetuses, above 100 mm. in length, the process of formation in the dura of denser and denser connective tissue proceeds rather slowly. It is realized, however, that this fibrous transformation in fetuses of 10 cm. is veiy extensive, the region about the sinuses alone remaining comparatively free from the development of the fibrils. The chief difference between the dura of this stage and the dura of the adult is a greater number of cell-nuclei in the fetal membrane. It is well, then, to consider the cellular character of the fibrous membrane and the region about the sinuses in the larger stages.




Fig. S (531)  
In a human fetus of 125 mm. (No. 900-H) the dura is quite fibrous, but still contains an increased number of the stellate and spindle forms of nuclei; likewise, about the superior sinus the tissue is an immature m30iomatous structure, fairly free from connective-tissue fibrils. This increased number of nuclei in the dural tissue holds also for human fetuses of 165 mm. (as in No. 745), but seems slightly decreased as compared with the smaller specimens. The lateral wall of the great sagittal sinus in this stage possesses distinct bands of white fibrils, but the tissue is much looser and more cellular than the fibrous dura over the hemispheres. These phenomena may be made out in similar stages of the fetal pig, as shown in figure 106, a photomicrograph from a specimen of 17 cm. In this specimen, treated by Mallory's stain, the superior longitudinal sinus is shown surrounded b}' a clear zone in which the deeply staining fibrils are comparatively few in number. On each lateral wall of the venous channels distinct fibrous bands may be made out, lying in the looser, more immature tissue. The lower portion of the falx has assumed quite an adult character.




Gradually the conversion of the tissue about the cerebral sinuses into the adult structure progresses. Thus, in both human and pig fetuses of 20 cm. length, the dura mater has acquired practically all of its adult features. Everywhere over the cereljral cortex the dura is characterized ])y dense layers of interlacing strands of white fibrous tissue, but the number of nuclei in the.se bundles may still be slightly greater than in the adult structure. In the more posterior regions, at this stage of 20 cm., the lateral walls of the sinus sagittalis superior are found to be completely occupied by the white fibrous tissue; in the anterior portion of the sinus much thinner tissue, resembling myxomatous structure, appears, as shown in figure 107. But in this specimen the invasion of the area about the great venous channel by fibrils has begun; isolated bundles may be made out everywhere in the lateral walls of the sinus. This freedom from connective-tissue formation does not persist, however, and the area is gradually invaded by the continued growth of the fibrils. The avoidance of the region about the sinuses by the connective-tissue resolution will be further commented on in the following subdivision of this paper.


Fig. 9 (65:-G)


The dura, then, develops probably first in connection with the mesenchjinal condensation which ultimately forms the bony skull and a portion of the dura (the cranial periosteum). It first becomes apparent, as a structural unit, as a more cellular layer differentiated, by a secondary condensation, out of the peria.xial mesenchyme. As the chondrogenous stage is approached it becomes differentiated as a distinct layer, maintaining varying relationships with the inner perichondrium of certain of the cranial bones. At a stage of 40 mm. m the fetal pig, the dura of the vertex may be dissected out as a distinct, somewhat fibrous laj'er. The process of fibrous-tissue transformation, however, is slow; the dura until late in fetal life shows an increased number of nuclei, as does any young connective tissue. The invasion of the region about the superior longitudinal sinus by connective-tissue fibrils is much more tardj' than is the transformation over the hemispheres.




Fig. 10 (636)
THE SUBDURAL SPACE AND THE MESOTHELIAL LINING OF THE DURA.




The subdural space (cavum subdurale) has been the subject of controversy in regard to its role in the pathway of the cerebro-spinal fluid. Before the work of Key and Retzius^^s) ^\^q yj^^^y ^.f^g ]^q.](\ ^)jat the cerebro-spinal fluid occupied the subarachnoid space in the spinal cord, but that in the cranium the subdural space afforded an analogous pathway. This conception was largely due to the fact that, in dissection on fresh material, the dura and arachnoid in the spinal region are found to be in approximation; in the cranium the greater adhesion, bj' trabeculse, of the arachnoidea to the pia renders the freeing of the dura from the leptomeninges the simplest line of cleavage. This view was entirely disproved by the beautiful injections of Key and Retzius, who demonstrated the anatomical and physiological continuity of the subarachnoid spaces.




With the introduction of this latter view by Key and Retzius the conception of the subdural space naturally changed. These Swedish investigators demonstrated a typical mesothehal cell-lining on the inner surface of the dura, as shown by the method of silver reductions. Without an intimate connection with the true cerebro-spinal fluid, the subdural space has come to be looked upon as somewhat analogous to the serous cavities of the body. Quincke^-*^', after a subdural injection of cinnabar granules, ascertained that communications existed between the subdural and subarachnoid spaces, but only in the direction from subdural to subarachnoid. Leonard HilK^^), from the results of physiological experiments, assumed that fluid passed from the subdural to the subarachnoid space, and in the reverse direction, with great ease. The more recent investigations, however, lend evidence to the view that in the normal animal with undisturbed intracranial pressure relations the two spaces are physiologically as well as anatomically separate. The current impression that the subdural space is in manj' respects a serous cavity will probably finally have greatest support; intimate connections with the lymphatic system are, however, entirely lacking in the dura.


Didusch fee.


The development of the subdural space must necessarily follow the develoj)ment of the dura. It has been mentioned that in fetal pigs of 50 mm. the dura can be freed from the arachnoid by gross dissection, but that at this stage many areas of adhesion between the two membranes exist. Such an observation has considerable bearing on the subdural space. For in the development of this space two processes must proceed far enough to permit the separation of the dura and arachnoidea by the capillary layer of fluid. The first of these processes, in order of probable importance, concerns the condensation of mesenchj'mal cells to form the outer membrane of the arachnoidea. This thickening and resolution into a true membrane takes place in close approximation to the inner surface of the dura. The second factor concerns the covering of this inner surface of the dura with mesothelial cells.




The lining of the subdural space by mesothelial cells can be readily demonstrated on the inner surface of the dura by silver reductions, but the outer membrane of the arachnoid does not permit of a similar technique. This technical failure in regard to the outer arachnoid surface is probably to be accounted for by the dissimilarity in cell-structure in the two situations. Similar difficulties have been encountered by other observers.




Fig. 3 (ysS)  
In order, then, to ascertain, if possible, at what stage a really adult subdural space could be demonstrated, the inner surface of the dura from fetal pigs of various lengths was subjected to treatment with silver nitrate. After the reduction had taken place to a sufficient degree, the whole dura was washed with distilled water, stained with hematoxylin, and cleared in glycerin. The i)ictures afforded by this method were quite satisfactory, and the technical procedure was so simple and reliable that considerable faith could be placed in the absence of the intercellular reduction lines.




The smallest fetal pig in which a typical mesothelial cell-j)attern could lie demonstrated on the inner surface of the cranial dura was one of 50 mm. In this specimen the inner surface of the dura was not uniformly covered with the mesothelial cells; certain ragged areas seemed to represent the points of adhesion of the arachnoid to the dura. Figure 108 is a reproduction of a drawing made from one of the areas in this specimen where a mesothelial cell pattern could bo seen. The drawing shows many of the characteristics of mesothelial ci'll-i)ut terns of other parts of the body. The irregularities in the cell-borders, the frefiuent accumulations of the reduced silver in the cellular angles, and thegeneral cellular pattern are quite typical ; but the variation in the size of the cells, as shown in figure 108, is also somewhat different from the usual finding in the adult, where there is considerable constancy in the size of the cells. About half the cells in this fetal pig of 50 mm. are diminutive in size; the smallest are hardly a third the size of the largest. Transitions between the smallest and largest cells are al.so shown in this figure. It i.s difficult to a.scertain whether these smaller cells represent young elements which have not yet reached their maximal growth; no evidence of cellular division, as evidenced by mitotic figures, has been observed, although in this connection it must be granted that the cleared si^ecimens are hardly the most favorable. Und()ul)tedly this explanation of the smaller cells would seem to be the true one, but there is little proof for the view, except their absence from the adult dura and their disappearance in larger specimens.




This disappearance of the smaller mesothelial cells is not rapid, but is seemingly delayed over into the larger fetuses; thus, in figure 109, a similar preparation from a fetal pig of 75 mm., corresponding smaller cells arc outhned. On account of the absence from the field of the drawing of the larger elements, these cells do not appear relatively as great in number as in the preceding figure. Likewise, in figure 110 every gradation in cell-size is shown, in a specimen made in the same manner from a dura of a fetal pig of 90 mm.




Fig. 2 (55S)  
Very slowly in the course of growth of the fetus the cells Uning the inner surface of the dura reach their standard size and compose the mesothehal surface, with very little variation in size. The process, however, is apparently very tard}', even though the fetus at 16 cm. shows an inner surface to the dura which is largely composed of standard cells (fig. HI); but even in this figure, from a relatively large fetus, the standard size of the cells has not been attained, for a few cells of small size appear in the drawing. In other respects the whole pattern, in general appearance, resembles closely the adult.




It seems most fair to assume that the occurrence of a true mesothelial cellpattern on the inner surface of the dura represents the initial estabUshment of a subdural space. On this basis the subdural space may be said to occur in fetal pigs 50 mm. in length; in the present investigation it has been found impossible to demonstrate the existence of the mesothehal cell-pattern in fetuses smaller than 50 mm. The separation of the dura, possible bj'^ gross dissection in pig fetuses of 40 mm., suggests that the space may be found at a sUghtly earUer stage than that in which the mesothehal cells have been demonstrated.


Fig. 1 045)


Anatomically the subdural space in pig fetuses resembles m everj' particular the adult space in cats and dogs; this was described in a paper^^S) pubhshed in 1914. In the large pig fetuses injections of solutions of potassium ferrocyanide and ironammonium citrate were made into the spinal subarachnoid space. After precipitating the foreign salts as prussian-blue, the injection is found to be wholly within the subarachnoid spaces, both in the spinal and cranial regions; the subdural space is absolutely free from any evidence of connection with the subarachnoid space. These findings wholly accord with the opinion concerning the adult subdural space which has been repeatedly expressed.




THE COMPETENCY OF THE EARLY DURA AS A CELLULAR MEMBRANE.




During the stage when the condensation of mesenchyme to form the cranial blastema is prono'inced the spread of the cerebro-spinal fluid becomes more and more extensive. In these stages, when the pig embryo measures from 16 to 25 m m. approximately, the outer membrane of the arachnoid is not 3'et formed, the arachnoid spaces extending from pial to blastemal condensation. WTien in these embryos the cerebro-spinal fluid is replaced by the ferrocyanide solution and the embryo kept alive for some time, the course of the injection may be traced to varying extents throughout the periaxial tissue. To this spread of the injection fluid (a true solution, during the progress of the experiment), however, the blastemal condensation of mesenchjme opposes an absolute barrier. This pecuUarity of the early condensation may be readily seen in figures 16 and 18. At this stage in development the blastemal thickening may be said to play the role of the outer membrane of the arachnoidea or of the inner surface of the dura.




This feature of the blastema as an impenetrable membrane — an absolute barrier to the passage of fluid — is found also to endure during injections of the ferrocyanide solution under pressures sufficient to rupture other parts of the central nervous system. Similarly, the early blastemal condensation resists the inflow of the other injections used (india ink and silver nitrate) under similar pressure conditions. In later stages the injection solutions, from ventricular or subarachnoid mtroduction, do not reach the dura. This is due to the development of an outer membrane of the arachnoidea and the formation of the subdural space. The arachnoid membrane when formed does not permit fluid to pass outward into the subdural space; but the competency of the early blastemal condensation in the mesenchyme affords a very good example of the perfect function of a tissue as a fluid barrier.


^ ■■'■ft-'' '".-• Fig. :? (W2)


An interesting phase of the competency of the secondary mesenchymal condensation (forming dura and outer membrane of arachnoid) may be seen in the region of the cisterna cerebello-medullaris. Here, as shown in figure 77, the zone of secondary condensation, while complete below, does not remain definitive above as the mesenchyme sweeps inward to the chorioid plexuses. At such a stage of 32 mm. in the pig, a replacement experiment would show no penetration of this secondary dural condensation by the foreign solution, where the condensation made a definitive membrane; above, however, in the region of the plexuses, a limited penetration by the introduced fluid could be made out.




X. THE RETURN OF CEREBRO SPINAL FLUID TO THE VENOUS SYSTEM.


The question of the exact mode of return of the cerebro-spinal fluid to the general circulation has interested many investigators. It has occasioned a large amount of work, with the presentation of several hypotheses. Key and Iletzius<2»), from the results of injections of colored gelatin into the spinal subarachnoid space, held that the cerebro-spinal fluid returned through Paccliionian granulations into the great dural sinuses. Other workers, following Key and Retzius, were dissatisfied with this theory, because of the apparent lack of these granulations in infants and in the lower animals. Cathclin(6), with but little evidence, hypothecated an absorption of the fluid by way of the perineural sheaths into the lymphatic system, although the physiological findings of Ziegler(57j^ Reiner and Schnitzler(<«), Leonard 11111*2^), and others made it necessary to consider a direct absorption into the blood system. Cushing(9) premised the drainage of fluid into the great sinuses through a valve-Hke mechanism. Dandy and Blackfan'"' .-suggested its absorption l)y the capillaries of the pia-arachnoid — an untenable hypothesis in view of the work of Kadyi'^Sj, Shroeder van der Kolk(5i), Ekker(i*>, Adamkiewicz^*), and others. Still another conception of the process has been advanced by Mott('**), namely, that the absorption of cerebro-spinal fluid isone of the functionsof the cerebral capillaries. Ina previous investigation'^), making use of a method similar to the one here employed in the rej)lacement experiments, evidence was presented indicating the drainage of cerebro-spinal directly into the great dural sinuses through arachnoid villi. These structures represent an invasion of arachnoid tissue through the lateral wall of the sinuses.


Didusch fee.


In view of the findings in adult laboratory animals, interest naturally turned, during the course of this work, to the process of drainage of the embryonic cerebrospmal fluid. The evidence afforded by the replacement experiments with the ferrocyanide solution indicated that in pig embryos of over 20 mm. cerebro-spinal fluid circulated throughout most of the periaxial tissue, and that in embryos of about 26 mm. the periaxial distribution was complete, the relations of the fluid at this stage becoming adult. With this evidence before us, the question of the drainage of the fluid became important. as the absorption process similar to the normal adult procedure, or was it entirely lacking, the production of the fluid being balanced by the growth of the nervous system and its meningeal spaces?
The question of the absorption of cerebro-spinal fluid was approached in the embryo in a similar manner to that employed in the adult animal. The problems incurred by the use of abnormal intracranial pressureswere eliminated by the method of replacing, without disturbing the normal tension, the embryonic cerebro-spinal fluid with the ferrocyanide solution. The embryo was then kept aUve and was finally fixed in a preservative which would precipitate the replaced fluid as prussianblue. This procedure was carried out in many embryos of varying lengths and the specimens were subsequently stained in serial sections.




The smallest embryo in which any evidence of absorption of the fluid from the periaxial tissue was obtained was a pig embryo, 23 mm. in length. In this specimen granules of prussian-blue could be traced through the mesenchymal spaces (arachnoidal) to the inner wall of the sinus transversus. The sinus is well differentiated at this stage in the human embryo of 21 mm., as demonstrated by Streeter^**). The wall of the sinus in this pig embryo was quite thin, the mesenchvTne lending the endothelium but Uttle support. The prussian-blue granules could be traced directly through the endothehnl wall of the sinus, and a few were identified lying free in the lumen. The conditions of the observations, permitting a flow of venous blood through the sinus, undoubtedly accounted for the fact that but few of the granules were found Ijing free in the sinus. This passage of the replaced fluid into the lateral sinus is portrayed in figure 21, taken from the pig embrj-o of 27 mm.




Coll:ipfPil amnion. Fig. 5 ((IGO)  
The same process of drainage of cerebro-spinal fluid may be obser^-ed in pig embryos more than 23 mm. in length. In all but one particular it corresponds exactly to the process observed in adult laboratory animals. There is the same lack of absorption on the part of the cerebral veins and embryonic capillary plexuses. In the adult, however, the process is not diffuse, but is confined to the arachnoidal villi, while in the embryo a considerable extent of the inner wall of the sinus lying in the mesenchymal tissue, which is breaking down to form the arachnoidal spaces, serves as a site for the fluid passage. In these earlier stages the sinus transversus functions as the chief sinus of absorption. This is probably to be explained by the primary basilar spread of the replaced cerebro-spinal fluid and also by the fact that the true sinus sagittalis superior is a much later addendum. In the human embryo, according to Streeterf^^), it is found in stages of over 50 mm.




The absorption of cerebro-spinal fluid in the embryo seems to follow the directing agencies which operate in the adult. Increase in the pressure employed in the injection of true solutions results in the drainage of more of the fluid, as determined by subsequent microscopical examination. This suggests that the process is determined by factors other than that of difTusion; it seems most likely that here, too, the process is one of filtration, with a possible distension of the cellular membrane, so that intercellular spaces are opened. The histological picture of the sinus waU, however, undoubtedlj^ gives the impression that the fluid has passed almost solely through the cytoplasm of the endoth(>lial cells and likewise through the layer of supporting mesenchyme. These findings are in accord with observations on the adult.


CONTRIBUTIONS TO EMBRYOLOGY, No. 11.


With dilute suspensions of india ink as the injection mass, the results are quite different in regard to the passage of the material into the sinus. Replacement experiments making use of this suspension of particulate matter yield no evidence, as the carbon granules do not leave the ventricular system. Likewise, simple injections of the suspension into either the central canal of the spinal cord or into the perispinal spaces furnish no information unless the syringe-pressure be high. In this case the carbon granules may be traced into the sinus transversus, which is apparently the point of least resistance. Because of the obscuring of the picture by the carbon it can not be determined histologically whether the granules pass into the sinus in the same manner as does a true solution, or whether the passage is effected by numerous small ruptures of the tissue. The impression gained from our study would incline one toward the latter view.




If the injection of india ink be made under very great pressure from a syringe, the segmental veins may be filled with the carbon. This filling is always subsequent to its flow into the sinus transversus. But in no case was an evidence of a flow into lymphatic channels observed.




THE STRUCTURE OF CHROMOPHILE CELLS OF THE NERVOUS
The process of drainage of the cerebro-s])inal fluid into the venous system of fetuses will not l)e (U^tailed here. This undoubtcdl}' concerns a study of the formation of arachnoidal villi and of the differentiation of the lateral walls of the superior sagittal sinus, the best site for this study. The material at hand is not suited for this investigation, so that postponement is necessary.
SYSTEM.






==XI. The Chorioid Plexuses and the Elaboration of Cerebro-Spinal Fluid==




By E. V. CowDRT.
With the realization that at a definite period in embryonic Ufe, cerebro-spinal fluid passes from the cerebral ventricles into the periaxial spaces, it seemed desirable to ascertain what relationship existed between the developing chorioid jilexuses and the elaboration of the fluid; for with the extension of the fluid into the periaxial tissue it becomes obvious that the balance between the development of the intraventricular fluid and the volume of the ventricles is destroyed and that more fluid is being elaborated than can be contained within the medullary-canal sjstem. This relationship between the ventricular volume and the production of cerebro-spinal fluid has been described at some length in a preceding section of this communication.




Anatomical Laboratory, Johns Hopkins University.
The determination, then, of the exact role plaj-ed by the chorioid plexuses in the further extension of the fluid into the periaxial tissue appeared to be of importance, for it could be conceived that the embryonic ependymal cells might be capable of elaborating the excess of fluid. With this purpose in mind the chorioid plexuses were investigated from morphological and cytological standpoints, in the hope that some index might be afforded as to the assumption of function on the part of the developing chorioid plexuses. These methods of study were apphed solely to the chorioid plexuses of pig embryos, for it is from them alone that evidence of the period of extraventricular extension of the cerebro-spinal fluid has been obtained.




THE DEVELOPMENT OF THE CHORIOID PLEXUSES.




The development of the chorioid plexuses is so well understood that only a ven,brief outline will be given here. The general scheme of origin of these glandular structures concerns a gradual histological differentiation in certain localities of the ventricular ependyma. The ependyma of the roof of the fourth ventricle thickens along the transverse invagination (phca chorioidea) and then gradually becomes tufted in villous projections into the ventricle, following the ingrowth of a capillar^' plexus and supporting mesenchyme. This general process of differentiation occurs at first along the lateral portions of the phca; the central portion of the ependj-ma remains unaffected by the villi even when the tufts have become quite well differentiated (fig. 23).


With one jilate.


Quite similar to this process of development of the plexus chorioideus of the fourth ventricle is the differentiation of the other plexuses. The plexus of the third ventricle develops as an infolding of the tela chorioidea of the roof. In every case the process holds of ependymal invagination and subsequent vascularization and suspension by mesenchymal ingrowth.




The histological differentiation of the ependjTnal cells into the glandular tj-pe of plexus, as first determined by Luschka''^^ and Faivre^^^)^ jg hardlj- satisfactory as an index of the production of fluid, as the secretory phenomena of the adult cells have not as yet bten completely established. The researches of Pettit and Girard(^\ dealing with the correlation of histological changes in the chorioidal cells and their functional state, first furnished reliable evidence that these cells give rise to cerebrospinal fluid. Since the publication of their investigations in 1900 many workers — Meek(37), Findlay("), Pellizzi(''2^, Mott^^D, Hworostuchin(26), Engel(>2), and othershave been concerned with this problem and have established on fairly definite bases the relationship of the plexuses to the production of the fluid. The histological appearances of the secretory cells, however, does not rest on incontrovertible ground, as has been stated in a previous paper^^^).




THE STRlTTniE OF ClIKOMorillLK CELLS OF THE NERVOUS SYSTEM.*
The process of diff'erentiation of the ependymal cells which form the glandular elements of the chorioid plexuses occurs with the invagmation and tufting of these structures. The various stages of transformation from the low type of cubical epithelium constituting the ependymal layer are shown in various figures in this paper. The nuclei of these cells assume basilar positions and the outer zones of the C3-toplasm become granular with their greater height. The process is rather a slow one, as might be expected from the fact that the whole villus is gradually enlarging and becoming more and more tufted.




The histological differentiation of the plexuses need hardly concern us here, except as an index of the assumption of function. The final completion of this change into the adult morphology occurs at a much later stage of development than our evidence indicates for the establishment of a cerebro-spinal circulation. It becomes obvious, then, that the final liistological changes are not necessary for the process of elaboration of the fluid. This assumption seems warranted also.b}'' the fact that the embryonic fluid contains much more albuminous material than does the adult fluid.


By E. V. CowDRv.


The time of appearance of the chorioid plexuses in relation to the extraventricular spread of the fluid would surely seem to offer undoubted evidence in regard to the first elaboration of the fluid by the plexuses. It has been shown that in pig embryos over 14 mm. in length the replaced solution in the cerebro-spinal system spreads from the roof of the fourth ventricle into the periaxial tissues. This extraventricular extension occurs practically simultaneously with the first indications, in the pig embryo, of the formation of the chorioid plexuses of the fourth ventricle. Thus, in a pig embryo of 14 mm., the primitive thickening and tufting of the ependyma of the roof of the fourth ventricle may be observed (fig. 32). In earlier stages no definite evidence of this developmental process is found.




From the first indication of a developing chorioid plexus in a pig embryo of 14 mm., the growth of the tufts progresses rapidly, so that at 18 nmi. the process is well advanced. In embryos of 20 mm. and over the tufts of the plexuses in the fourth ventricle are quite marked, as shown in figures 22, 44, 40, and 92.




INTRODUCTION.
The chorioid plexuses of the third and lateral ventricles api)ear at a somewhat later stage than do those in the more caudal v(>ntricle. Thus the first indication of their ai)i)earance in pig eml)ryos is found in si)ecimens measuring 19 mm. in length. This coincides well with the further extension of the replaced fluid in specimens of 19 mm. and over. The definite differentiation of these plexuses, however, does not actually take place until the embryo reaches a length of 23 mm.— a fact suggestive of some relationshif) to the complete periaxial spread found in embryos of this measurement.




It has long been known that certain pecuUar nerve-colls, well characterized by their structural aj^pearance, occur in the normal human brain, and indeed in the brains of all the vertebrates which have been examined. In fi.xed preparations they are slightly shrunken, they stain deeply with both acid and basic dj-es, and their nuclei are obscure and hard to define. Flesch (1887, p. 196) called them "chromophile" cells. Nissl (1896, p. 1154) thought at first that they were artefacts of some sort, but Cajal (1909, p. 210) and others brought forward strong evidence against this view. Cajal (1909, p. 211) concluded tliat they were resting cells. On the other hand, in the light of Dolley's (1910, p. 333) work, they would seem to be in the initial stages of fatigue, as evidenced by the increase in the amount of Nissl substance in them and by their obscure, deeply-staining nuclei. Our knowledge of their structure is incomplete so far as the mitochondria and the canalicular apparatus are concerned. Busacca Archimede (1913, p. 332), alone, has observed that the mitochondria in certain cells in the brain of Testudo grwca stain particularly intensely with iron hematoxylin, and in some cases seem to lose their definite outlines and to form homogeneous masses. Rina Monti (1915, p. 39) has made a comprehensive studj^ of the canalicular apparatus ("apparati di Golgi") in nerve-cells, but she does not mention cells in the chromophilic condition. I shall consequently venture to present in this paper my observations on these two structures in the chromophile cells in the brain of the white mouse.
Considered, then, as a whole, there seems to be a very definite relationship between the developing chorioid plexuses and the periaxial spread of the embryonic cerebro-spinal fluid; for immediately after the first appearance of chorioidal tufting in the roof of the fourth ventricle (at 14 mm.) the replaced injection sjjread appears in the periaxial tissue (fig. 3). This extraventricular spread does not become marked until a length of 19 mm. is attained (fig. 5) — a factor in accord with the elaboration of the villi in the chorioid plexus of the fourth ventricle. The periaxial spread remains localized in the rhombencephalic region until the 20 mm. stage is attained, when it rapidly becomes pericerebral and perispinal (figs. 6 and 7). This coincides with the first indications of the chorioid plexuses in the more cephalic ventricles. But the further spread is here delayed (as in the stages between 14 and 19 mm.) until a length of at least 24 mm. is reached — which is perhaps of importance in the further development of the cerebral plexuses and the greater elaboration of the cerebro-spinal fluid. Thus it seems possible to conclude that coincident with the first appearance of the chorioid plexuses a more rapid production of cerebro-spinal fluid occurs, necessitating the passage of the fluid into the periaxial tissues.




MATERIAL AND METHODS.
THE GLYCOGEN CONTENT OF THE CHORIOID PLEXUSES.




White mice were employed because they are the smallest mammals which can be conveniently used in the laboratory for experimental purposes. The small size of their nervous system permits the study of the distribution and the arrangement of chromophile cells in serial sections. All the mice were of known age and care was taken that they were perfecth- normal.
In the hope that some cytological method might afford direct and incontrovertible evidence of the time of the assumption of function by the chorioid plexuses, stains demonstrating the intracellular presence of glycogen were applied to these structures. The quantitj' of the starch in the chorioid plexuses of rat and mouse embryos, as shown bj' Goldmann, suggested that this substance might be associated with the early elaboration of the cerebro-spinal fluid. Furthermore, the presence in the adult fluid of a definite reducing bodj', demonstrated by XawTatschi to be dextrose, added some weight to the hope that a definite conclusion might thus be afforded.




A modification of the methods of Altmann (1890, p. 27), Galeotti (1895, p. 466), Regaud (1910, p. 296), Bensley (1911, p. 309), and Shirokogoroflf (1913, p. 523) was devised for the study of mitochondria. The method has many advantages. In the first place, the use of a mixture of formalin and ])otassium bichromate as a fixative (Regaud) gives a much more uniform preservation of mitochondria than the osmic acid containing fixatives in general use. The application of the fixative by
Several important studies concerning the presence of glycogen in the cells of the embryonic and fetal chorioid plexuses have been made. Creighton W found that the glycogen of the chorioid plexus was verj^ abundant about the middle of embryonic life, while von Loeper concluded that the great content in the cells of the fetal plexus was characteristic. Goldmann^^o) found large quantities of gh'cogen in the plexus in rats and mice, not only in embryonic life but also in animals from two to three weeks old. In the adult plexuses the cells contained no trace of glycogen.
'The work was aided by the Department of Embrj-ology of the Carnegie Institution of Washington, and part of it was done at the Marine Biological Laboratory. Woods Hole. Massachusetts, where, through the kindness of the Director, Dr. Lillie. a room was placed at my disposal.




29
The observations here included were made after fixing the chorioid plexuses of various pig embryos in absolute alcohol and staining the sections (cut either from celloidin or paraffin blocks) by Best's carmine method. This technique is similar to that employed by Goldmann. The staining reaction is such that a very striking differentiation of the glycogen occurs, but the shrinkage of the embryonic tissue in the fixation in absolute alcohol is a disadvantage. In these obsers'^ations the plexuses from the fourth and lateral ventricles were used.




As shown in the table on page 94, glycogen could be identified in the cells of the chorioid plexuses in pig embryos varying in length from 28 to 155 mm.


30 'I'llK STIUCTIKK OV CllHOMiilMIIl.K CELLS or TUF-: NF.UVors SYSTKNL
iiijrctiuii tlinuitili the hlood-vi'ssi-ls (r^hirokogorotTi cliniiiiiitcs tiiany vciy ohjectionahk' artrfacts dui' (u faulty poiu'tration. The use of pcnuaiiganatc and oxalic acid (Benslfv) facilitates the staining of the mitochondria with the anilin fuchsin (Altmann), and the countcrstaining with methyl green (Galeotti) ix'rniits of the demonstration of the Xissl substance in the same cell with the mitochondria. The fact tliat the method gives good results in the case of other tissues where the mixtures of Altniann. Flomming. and others are useless on account of their ])o(»r i)()W('rs of penetration, justifies the following detailed statement:
Fixation:
Chloroform tlic aiiiiiial. Inject wanned U.8o ix-r cent Na( '1 sDlutiim into the aorta tliroufjli the ventricle. If tlie lirain alone is to be stuciieil clamp the descending aorta. If the entire nervous system is to be fixed, clamp the cceliac, the renals, the superior antl inferioi- mesenteries, the iliacs, and the brachials. Continue the injection until the salt solution is returned uncolored through the jugulars. During this time lay t)are the arch of the aorta and the carotids from connective tissue, so that they may expand easily ami carry more fluid to the brain. (Jravity pressure of not more than 6 feet may be employed. Cut the inferior vena cava and the jugulars so that the salt .solution may run through e:i.-;ily.


Below the first measurement no glycogen was demonstrated bj' the method employed; above the higher limit in only one instance (series No. 41) was glyocgen found. This finding of a limited period in the embryonic hfe of a pig during which glycogen occurs in the cells of the chorioid plexuses does not coincide with Goldmann's observations on the rat and mouse. Furthermore, it was found here that in stages up to 100 mm. the glycogen was practically generally distributed throughout all the cells of the chorioid jilexus, occurring with gn^it regularity in every villus and cell. This general distribution was not found in the plexuses of embiyos over 110 mm. in length; in these more advanced stages the cells containing starch occurred in clumps, giving a localized distribution. In the stages under 100 mm. the glycogen was present in very large amount, as estimated histologically. As the stages advanced the quantity of glycogen decreased rapidly. This great amount of starch was present in the same stages in wliich the general distribution of the cells occurred.


Follow the .salt solution with the formalin and bichromate mixture: 3 per cent i)otassium bichromate. 4 parts; neutral formalin. 1 i)art. The pota.ssium l)ichromate acts best when freshly preparetl. Neutral formalin is made from the commercial variety l)y the addition of magnesium carljonate, a deposit of which should always remain at the i)Ottom of the formalin ijottle. It is important that tlie pressure should be at the maximum when the mixture is first injected, so that the blood-ve.s.sels may be fixed in a state of dilation. If the pressure is low when the fixative comes in contact with the ve.ssel-walls they will l)e fixed in a condition of collap.se. It will then be difficult, or even impo.ssil)le, to obtain a complete injection. The injection should be continued for about an liour.


Occurrence of glycogen in the chorioid plexuses of embryo pigs.


The brain is then dissected out antl iniinersed in the fluid. In the ca.sc of the mouse's brain it is sufficient to divide it longitudinally into halves. The fixative must be changed every day for 4 or ') days, otherwi.se it undergoes a ciiange eviilenced by a darkening in color. This change is accelerated by light and by heat, so that the tissue should be kept in the dark and in a cool place. Fixation may also be effected by simple immersion of the tissue in the fixative, instead of by injection, but this procedure is not recommended.
C.P.




After this prolonged fixation the tissue is mordanted in a fresh 3 per cent solution of potassium bichromate, in which it remains for 8 or 9 days, changing every .second day.
series,  
No.


C.R.


Wasli in running water for 24 hours. The object of this careful washing is to remove most of the formalin ami bichromate, for otherwise the tissue will be extremely lirittle and hard to cut.


naeasure,
mm.


hrhijdratiitu find einbeddimj:
.')0 per cent alcohol 12 hours; 70 jx-r cent and 9") jier cent alcohol 24 hours each; absolute alcohol G to 12 hours: half al>solute an<l xylol (> hours; xylol .'^ hours: paraffin ()()° ( '. 3 hours: cut in 4/i serial sections.




Staining:
Glycogen.
(1) PaKS the .sections, niounleil on slides, duuii 1 hrowgli toluol, absolute, it."), 7(1. .iimI .">() per cent alcohol to distilled water.




(2) 1 jK'r cent u(|ueous .solution of pola.'^siuiii pcrnianganalc 30 seconds; but ihi^ tinie must lie determined experiment, -dly.


Globular forms of glycogen.


(3) .') jMT cent a(|ueous .solution of oxalic acid also about 30 .secomU.




(4) Hin.se in .s<'veral changes of distilled water about a minute. Imnnipletc washing prevents the staining with fuchsin.
Plaques of glycogen.




(o) Slain in .Mtmatm's anilin fuchsin, which is to lie niaile up as follows: Make a saturatecl Milution of anilin oil in distilled water by shaking the two together (anilin oil goes into solution in water in about 1 per cent). Filti'r aiul add 20 grams of acid fuchsin to 100 c.c. of the filtrate. The stain should 1)C ready to use in about 24 hours. It goes bad in about a month. To stain, dry the slide with a towel, except the small area to which the sections are attached. Cover the sections on the slide wit h a small amount of the slain .-ind heat over a spirit lamp until fumes, smelling strongly of anilin oil.


Amount of glycogen.






Distribution
of glycogen
throughout
plexus.


rilE STIUCTUHE OF CFiROMOPHILP: CELI>^ OF THE NERVOUS SYSTEM. 31
conic off. .Vlluw to cool. Let the stain icmain on the section.-* for about minutes. Return the stain to the bottle.




(()) Dry ofT most of the stain with a towel and rinse in distilled water, so that the only stain remaining is in the sections. If a large amount of the free stain remains it will form a troublesome precipitate with the methyl green; on the other hand, if too much stain Ls removed the coloration of the mitochondria will Ix; impaired.
Intracellular distribution of glycogen.




(7) Again dry the slide with a towel, except for the area covered by sections. Allow a little 1 per cent methyl green, added with a pipette, to flow over the sections, holding the slide over a piece of white pajicr so that the colors may be seen. Apply the methyl green for about o seconds at first and then modify the time to suit the needs of the tissue.


16
3
13
12
14
6
9
8
4
1
42
17
15
20
10
18
27
39
25
32
40
41
24
19
23
11
21
22
26
5


(8) Drain off excess of stain antl plunge the .slide into 9.5 per cent alcohol for a second or two. then rinse in absolute for the same time, clear in toluol, and mount in balsam.
Several difficulties may be met with: (1) The methyl green may remove all the fuchsin. even when it is only aii]>lied for a short time. This is due to incomplete mordanting of the mitochondria by the chrome salts in the fixative. It may often be avoided, either by omitting the treatment with permanganate and oxalic acid, or by treating the sections with a 2 per cent solution of pota.ssium bichromate for a few minutes immediately before staining (as advised by Bensley). The action of the permanganate and oxalic is to remove the excess of bichromate. (2) Or the fuchsin may stain so intensely that the methyl green removes it verj' slowly or not at all. This, on the other hand, is due to too much mordanting. It may be corrected by prolonging the action of the permanganate and oxalic. (3) Sometimes, after obtaining a good differentiation, the methyl green is wa.shed out before the slide is placeil in toluol. This may be avoided by omitting the 95 per cent alcohol, by passing from the methyl green to the ab.solute direct. (4) Unfortunately the stain Ls not very permanent. Under favorable conditions it will last for 3 or 4 j-ears. The fading in color is hastened by light and by heat, and it proceeds verj' rapidly in a damp atmosphere.


18
23
28
33
36
39
40
55
66
80
90
100
105
118
132
155
155
155
158
160
163
170
173
185
195
209
213
223
244
260


Cajal's (1912, p. 211) uranium-nitrate method was emploj'ed for the canahcuhir apparatus in its original form, except for the substitution of nu^thyl sreen in the place of carmakmi as a counterstain.




Control preparations were fixed in a variety of Huids and were stained in many ways, as will appear later.




The figures have been made from specimens prepared by the above-mentioned fuchsin-methyl green method, by which the mitochondria are stained red, the Xissl substance green, while the canalicular apparatus remains uncolored; and also from specimens made by the uranium-nitrate method, which blackens the canalicular apjxiratus and colors the Ni.ssl substance green.


Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present . .


OBSERVATIONS. Chromophile cells, as the name implies, possess an unusual affinity for stains, which may be either acid or basic. Their structure is variable. A glance at the figures is sufficient to show this. The variations may represent stages in a process, which, when pushed to an extreme, results in a cell in an advanced stage of chromophilia, but of this we have no conclusive i)roof. Neither can we as.scrt that the process proceeds in this direction, for the changes observed may ecjually well be interpreted as taking place in the reverse order. We do not yet know whether the series is homogeneous; that to say, whether we are not arbitrarily grouping several processes of different nature under the same heading. For instance, a mitochondrial increase (figures 1 and 2) may not precede a diffuse staining of the whole cell with mitochondrial d^es (figure 6), which may be brought about in an entirely different waj'. Nevertheless, the cells are all chromophile in the sense already defined, that they stain deeply.


Present
Present
Present
Present
Present
Absent




Present
Present
Present
Absent
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present




:V2 THK STHlCTrHK OF CHROMOPHIM-: C'ELI.S OF THK NKinOUS SYSTEM.
Present
Present
Present
Present
Present
Present
Absent
Present
Present
Absent
Present
Absent
Absent
Absent
Absent
Absent




Some chromophilo cells differ only from other cells hy a slifrlit increase in the amount and in the intensity of the staining of the mitochondria ( figure 1 ). There is apparently no correspontling change in the Xissl substance and the morphology of the mitochondria is unaltered.
Great
 
Great
Great
Great
Great
Great
Great
Great
Great
Great
Small
Small
SmaU
SmaU
Small
SmaU


Other cells show a remarkable increase in the number of mitochondiia. l"or examjile, a cell (figure 2) frequently contains three or four times as many mitochondria as its neighbor: this increase in mitochondria is associated with a slight but perceptible increase in the amount of the diffuse Nissl substance in the cytoplasm and with a darker staining of the acidophilic and basophilic nucleoli and the grovnul-.^ubstance of the micleus. Cells in this condition show no evidence of shrinkage. They may be recognized in Cajal preparations (figure 7) by the changes in the nucleus and the Nissl substance. The Cajal preparations show that the canalicular apparatus is unaltered.


General
Localized
General
General
General
General
General
General
General
Genearl
General
Localized
General
LocaUzed
Localized
Localized


There may be a great increase in the Nissl substance, which is present as a diffuse dei)Osit (figure 3). At the same time some of the mitochondria often lose their discrete outlines and .seem to merge into the surrounding cytoplasm. Mitochondria may not be very numerous in cells of this kind. The nucleus stains intensely and a few clear canals may be seen in its vicinity. The cell has ai^parently shrinkage spaces on either side of it. Preparations, made by fixing in alcohol and staining with tohiidiii l)lue, contain cells in which the Nissl substance is in this condition and Cajal ])reparations show that the canals are unaltered.


Basilar. Basilar. Basilar. Basilar. Basilar. Basilar. General. General. General. General. Basilar. General. General. General. General. General.


Figure 4 illustrates a cell in a rather more advanced stage of chromopliilia. In this cell there is an unusually large amount of Nissl substance and there are further evidences of the disappearance of formed mitochondria, esi^ecially in the cell process. The outlines of the nucleus can barely be made out. The canalicular apparatus shows no modifications either by this method or bj' the Cajal technique.




A very interesting condition is shown in figure 5. Here, with this degree of differentiation, only a few tyjiical mitochondria i)ersist near the origin of the cell ])rocess. The Nissl substance is overshadowed by a cloud of material staining the .same way as the formed mitochondria do in adjacent cells. Figure S illustrates a similar cell in a Cajal prei)arati(m. The Nissl substance in it is increased and there is no modification in the blackened canalicular apparatus. Cells in this condition are often shrunken. Il is difficult to determine whether the shrinkage is the expression of an actual diminution in the size of the cells during life, or whether it is simply the result of a difference in the reaction of chromo])hile cc^lls to the fixation and subse(|uent treatment. The presence of what apjx'ar to be shrinkage spaces around the cells seems to indicate that it is in reality due to the techniiiue em])loyed, because if. on the other hand, it was due to a decrease in the size of the cell during life, one would expect the space to hv filled up by a shifting of neighboring structures. It may be emi)ha.sized that the fact that other cells, in actual contact with chromophile cells, .show no signs whatever of shrinkage must be regarded as one of the distinctive |)roperties of cells in the chromophilic condition. There is. of course, still another interpn'tation. namely, thai the sjjaces in (luestion are unusually large perineuronal si)aces, the enlargement being in some way connected wild the <lilTereiice in the phvsiological c(»ndition of chromophile ci'lls as contrasted with other cells.






Absent




THE STUUCTURE OF CIIKOMOPHILE CKLL.S OF THF. NFRVOFS SYSTEM.. 33
The mitochondria maj- apparently- disappear more or less completely in certain cells, and their place be taken by a mass of amorphous material with the same staining properties (figure 6). The nucleus may or may not be visible. Cajal preparations of cells in the same condition (figure 9) show that the canals are unaltered. The nucleus is obscured by the cloud of Xissl substance. The appearance of these cells, in advanced stages of chromo])hilia, would jjerhaps lead one to suppose that they arc degenerating and that their nuclei have disapjjeared. That this is not the case may be seen if one of the mitochondrial preparations is stained with hematoxylin and eosin. The hematoxylin and eosin does not color either the amorphous deposit or the Nissl substance, which, in the mitochondrial and in the Cajal preparations, hides the nuclei. The nuclei have in reality distinct and definite outlines and apjiear to be quite unaltered, except that they contain rather more than the usual amount of chromatin. In fact, the change in the mitochondria and the increase in the amount of the Xissl substance would never have been suspected if hematoxylin and eosm had alone been used.




The distribution of chromophile cells is important. They often occur singly. They may be surrounded on all sides by cells which show no tendency toward an assumption of the clironiophilic condition. They may, on the other hand, occur in clumps. The clumps vary greatly in size. They contain cells in all stages of chromophilia in addition to a variable number of unaltered cells, which are always present, scattered among them.
Present
Absent




The neuropil in which the chromophile cells are embedded differs m no way from the neuropil elsewhere. It seems, by all the mitochondrial methods, to be studded with mitochondria. But it must not be thought that the mitochondria occur in anything like equal num])ers in the neuropil of different regions, because there is a remarkable variation in this respect. The mitochondria appear to be intercellular, but unhapi)ily a source of error is introduced by the fact that the unmedullated, and to a les.ser extent the medullated, processes stain in much the same way as the mitochondria, so that in .some cases it is impossible to distinguish between them. Undoubtedly a large number of the mitochondria in the neuroi^il are contained in nerve-cell processes cut in section, but there is no a priori reason wh}' they should not occur free from the cells as an intercellular deposit. This important question can only be .solved by a detailed study of staming reactions, possibly by the elaboration of new methods, or b}' taking advantage of the differential solubilities of mitochondria. It has a direct bearing upon the role of intercellular material in the metabolism of the central nervous sj'stem.
Present




Cells in the chromophilic condition are comparatively rare in the olfactory bulb as compared with the cerebral cortex. In fact, they are more abundant in the cerebral cortex than in any other part of the brain. Clumi)s of them are more common here than in other regions. The clumps vary in size, in shape, and in position in the brains of animals from the same litter, apparently treated in exactly the same way. Chromophile cells are also numerous in the hippocampus. They are, on the contrary, comparatively rare in the corpus striatum and in the thalamus, in both of which they are more frequentlj'^ met with singly than in groups. In the midbrain they are found in about the same number. It is interesting to note that they are
Absent




Small


34 THE STRUCTURE OF CIIUOMOPHILE CELLS OF THE NERVOUS SYSTEM
(juite numerous in the eerebellar cortex. The Purknije cells are ])articularly lial)le to show this condition. They are infrequent in the medulla and they scarceh' ever occur in the spinal cord, in the spinal ganglia, or in the sensory ganglia of the cranial nerves, as, for example, the Gasserian ganglion. In other words, this remarkable condition of the nerve-cell is more prevalent in the higher centers than in the lower ones. This is particularly true of chromophile cells in advanced stages of chromophilia.


Localized


The question at once arises as to whether these changes in the appearance of the cells are indicative of real alterations in the cells themselves or whether they are merely the result of the treatment to which they have been subjected.


General.


Unfortunatclj' it w^as found impossible to confirm these observations by the stud}' of unstained, living cells by reason of the difficulties met with in attempts to isolate the cells without injuring them. Attempts to stain the mitochondria in living cells by injecting a solution of janus green into the brain through the bloodvessels did not yield satisfactory results because the janus green was almost immediately reduced, first to the leucobase, and then to the red diethylsafranin, by the reducing action of the brain-substance and the absence of an adequate supply of atmospheric oxj'gen, so that observations could not be made. Pure ox3^gen was bubbled through the janus-green solution while it was being injected, in the hope that the reduction of the janus green might thus be retarded, but without success. Attempts to tease out indi\idual cells in the nervous system and to stain them bj' simple immersion in the janu.><-green solution resulted, of course, in a coloration of the mitochondria, but it was on the whole unsatisfactory on account of the unavoidable injury to the cells. Consequently I have had to rely solely upon the study of fixed material.




The results obtained with the fuchsin-methyl gr(>en method and with the Cajal techni(|ue have been confirmed by the detailed examination of material stained liy the Benda method, the Altmann method, and with iron hematoxylin. Chromophile cells are, I think, not artefacts due to alcohol fixation, as Barker (1899, p. 124) suppo.ses, becau.se I have observed them in tissues fixed in a great variety of fluids not containing alcohol. Moreover, Flesch (1887, p. 197) found years ago that they could be identified in the fresh, unstained condition as well as in tissues stained vitally with methylene blue.
Absent




The fact that the chromophile cells are very abundant in the superficial layers of the cortex would at first .s(>em to indicate, as some investigators believe, that they arc artefacts due to mechanical maniinilation. The clusters of chromojjhile cells are .sometimes cone-shaped, with the base on the surface of the cortex and the apex of the cone extending inwards, which looks as if they might have been produced by pressure from without which radiated inwards. But isolated clumjis of chromojihile cells occur in deeper i)arts of the brain, which can not be explained in this way. Moreover, a number of other facts seem to be incompatible with this view. In the first place, since all the brains were fixed, l)efore removal, by the injection of the fixative through the blood-vessels, it follows that there could be no mechanical injury until after fixation. The invariable occurnnice of unaltered cells, side by side with the chromophile cells, is hard to explain on tiie basis of mechanical injury, because wh:itever pressure had been brought to bear upon the t i.ssue must necessarily


Absent




THE STRUCTURE OF CHROMOPHILE CELLS OF THE NERVOUS SYSTEM. 35
have acted upon both; but one shows the condition and the other does not (as is shown in all the figures). Furthermore, if mechanical injury is the cause of the condition, it is difficult to understand why chromophile cells are so rare in the spinal cord and in the ganglia of the cranial nerves, which are l)f)und down by membranes and which in removal are consequenth^ subjected to greater mechanical injury than the cortex of the brain.




In order to settle the question the results of intentional mechanical injury brought about by bruising the cerebrum and the spinal ganglia with a blunt instrument were studied. It was found that the lesion produced was characterized by the flattening or comin-e.ssion of many cells in the same direction, at right angles to the direction in which the pressure had been exerted. All the cells in the area were uniformly affected. Normal cells were not ."scattered among them. The injured cells stained intensely, but they did not simulate the chromophile cells. The neuropil between them showed marked changes and could readily be distinguished from the neuropil elsewhere in the same section.
Absent




Chromophile cells are not the result of differences in the time or in the degree of fixation. The whole brain is uniformly fixed by the methods of technique employed. The distribution of chromophile cells is not related to the arrangement of the bloodvessels, which are the avenues of approach of the fixative. Neither do the mitochondria vary in number with the vascularitj' of the region.


Absent


The condition is not due to irregular mordanting with the potassium bichromate, because complete extraction of the bichromate by prolonged treatment with permanganate and oxalic acid does not essentially modifj- the appearance of the chromophile cells when the sections are stained.




Another possibility is that the intense staining of the chromophile cells results from incomjjlete differentiation. Even if this were the case the differences in the rate of decolorization must be the visible expression of real differences in the cells themselves. I have found, however, that the same differences obtain in undifferentiated specimens stained lightlj' with fuchsin, crystal \'iolet. and iron hematoxylin. I have made a number of experiments to determine whether more complete differentiation would bring to light formed mitochondria in cells in which they appear to have been replaced by the amorphous deposit which stains in the same way.


Absent


Specimens were stained in the usual fashion with fuchsin and methyl green and were mounted in balsam. Drawings were then made of chromophile cells which had been stained intenselj^ with the fuchsin and in which no formed mitochondria could be seen. The cover-glass was then dissolved off and the slide was passed down through toluol and graded alcohols to water. It was then restained with fuchsin. differentiated more strongly with the methyl green, mounted in balsam, and examined. The same condition was apparent, except that the homogeneous deposit had a distinctly greenish color. The same process was repeated as many as five times with the same cell, increasing each time the extent of differentiation, until the cell stained intenseh' with methyl green and very httle trace of the fuchsin was ■left; still no formed mitochondria were observed; this was repeated with other cells with the result that in some of them formed mitochondria were brought to hght, while in othei-s thev were not.




Line 1,393: Line 1,596:




36 THK STRUCTIRE OF CIIHOMOPHILK CELLS OF THE NERVOUS SYSTEM.
Cioldmann'20) pictures the glycogen as occurring throughout the cells of the chorioid i)lexus in the form of globules of larger or smaller size. Some of these globules may be .seen even in the s\n'rounding cerebro-sjiinaj fluiil. This general intracellular disposition was observed in this series in specimens measuring 6G mm. and over (fig. 95). Below this measurement the glycogen occurred practically entirely in the basilar portion of the cell, central to the nucleus. Furthermore, in the stages between 30 and 00 mm. the glj'cogen globules were present in but small numbers and the glycogen was found in crescentic plaques (fig. 96). This formation of definite plaques is ai)parently to be ascribed to the fusion of the globules when the amount of glycogen becomes extreme. As far as is known tliis plaque formation with glycogen has not previously been noted; in one of Goldmann's figures the fusion of some of the globules has apparently taken place.




Similar experiments were performed with individual cells stained a homogeneous black color with iron hematoxylin. The results obtained are easier to interpret because the differentiator, iron alum, does not itself color the tissue like the methyl green. This advantage is counterlialanced by the fact that both the mitochondria and the Xissl substance stain in the same way and it is often difficult to distinguish between them. In many cases, jjarticularly in slightly undifferentiated specimens, the extraction of the stain from chromophile cells by further differentiation brought to light a variable number of formed mitochondria. Moreover, it is worthy of note that the chromophile cells in the cerebral cortex are the last to become decolorized and that the differentiation occurs with unequal rapidity in different parts of the cell, thus indicating that the homogeneous deposit is not present in the same concentration in all parts of the cell.
The table on page 94 records the findings in these observations.




The end-result of this experimentation is that chromophile cells, particularly those in advanced stages of the condition, contain a diffuse deposit, which stains in a tyi^ical way with all mitochondrial dyes, and which is probably formed by the solution of some of the mitochondria in the cell.
The occurrence of glycogen in the cells of the chorioid plexus only during a certain portion of embryonic life is, as shown l)y the foregoing table, a fairly definite phenomenon, but there is surelj' no indication that this temporary presence of the animal starch bears any relation to the assumption of function on the part of the chorioid plexuses. The evidence afforded by the extraventricular flow of the replaced fluid, with the apparent relationship of the developing chorioid plexuses to the periaxial extension of the fluid, argues strongly against such an assumption.




The condition is not due to technique and it is not associated with a visil)le pathological change on the part of the animal.
XII. PERIVASCULAR SPACES IN THE EMBRYO.




All the mice employed were api)areiitly normal. They ate well and showed no signs of sickness. They were killed with chloroform, and it may at once be said that the changes are not due to acute chloroform poisoning, because animals killed in other ways, by decapitation, for example, showed the same condition. The mice were not excited, or disturbed or exercised in any unusual way before they were killed. A careful autop.sy of each mou.se was made to make sure that it was quite normal. Some were found to contain a parasite, present in the cysticercus stage in the liver; these were invariably discarded. The chromophile cells were found in mice of both sexes in almost all seasons of the year. They were found in mice Aarying in age from 25 days to adults, so that they can not be regarded as an expres,sion of .senility. It was thought that they might occur in consequence of abnormal conditions due to captivity. In order to settle this point a wild field-mouse was captured alive and in good condition and its brain was prepared in the usual way. It, also, showed chrom()i)hile cells.
In 1865 His' using a puncture injection, found that each nerve-cell existed in a so-called space. These pericellular spaces connected, as demonstrated by the flow of the injection mass, with an extensive perivascular network, more complex in its gray matter than in the white. In all of His's cases continuation of the injection led to a peripheral spread toward the pia, both in the spinal medulla and in the brain.




An apparentlj'^ analogous partial solution of mitochondria was observed in liver-cells poisoned with i^hosphorus by INIayer, Rathery, and Schaeflfer (1914, |). G09). Accordingly, W. J. M. Scott tried the effect of experimental phosphorus ])oisoning on the nervous system of white mice. The chromophile cells were apparently entirely unaffected and a solution of mitochondria was not brought about. Dr. Bensley made the interesting suggestion to me that this partial solution of mitochondria in chromoi)hile cells might be due to a swing of the reaction in them toward the acid side, with the liberation of free fatty acids. I therefore made some preliminarj^ experiments on acidosis in mice produced by the subcutaneous injection of dilute hydrochloric acid, all of which yielded negative results as far as the chromf)phile cells were concerned. I have, further, found that slight exercise does not alter the appearance of the chromophile cells in the brains of white mice to any noticeable extent.
]Mott('*^\ working on the brains of animals in which an experimental cerebral anemia had been produced by ligation of the head arteries, found the perivascular spaces enormously dilated and the perineuronal spaces Likewise verj' evident. Direct connections between the perivascular and perineuronal spaces are pictured in Mott's communication.




The deduction which ^Nlott made from his findings, regarding the possible absorption of cerebro-spinal fluid by the cerebral capillary bed from this perivascular and perineuronal system, was discussed by the present author in a paper two years ago(55). It was there shown that, with the use of true solutions as the injection (potassium ferrocyanide and iron-ammonium citrate), the whole perivascular sj'stem could be filled. This injection of the spaces, however, occurred only when the pressure conditions within the cranial cavity were such that the subarachnoid pressure exceeded the vascular tension. This reversion of the pressure relations was accomplished by maintaining at normal the subarachnoid pressure with the injection fluid, and occasioning a simultaneous and complete vascular anemia. Under the routine conditions of injection (with undisturbed pressure relations) no injection of the perivascular system from the subarachnoid space resulted. It was found impossible to mject the perivascular system, using granular suspensions as the injection mass, without employing pressures far above the normal.




From these results here recorded briefly, the belief was expressed in tliis former paper that each nerve-cell was surrounded by a capillary space which drained along the perivascular channels into the subarachnoid spaces. Probably this sj'^stem represents a mechanism for accessory tissue drainage comparable ph3^siologically to the lymphatic channels of the other parts of the body.


THE .STRUCTURI': OV CIIUOMOI'HILK CELLS OF IIIK NEHVOUS SYSTEM. 37
It seems highly probable, therefore, that chromophile cells occur normally in the bram of the white mouse and that we have to reckon with a partial solution of mitochondria just as we have for many years recognized a chromatolj'sis, or solution of the Nissl substance.


In view of these findings in the adult mammal it seemed desirable to ascertain at what period of intra-uterine life such function was acquired. It also seemed not unhkely that information of interest might be acquired from the embryonic intramedullary circulation which would amplify our knowledge of this sj^stem in the adult. It was thought that there might be a correlation between the production of the perivascular fluid and the enlargement of the subarachnoid channels, similar to the evident connection between the chorioidal invagination and the extraventricular spread of the fluid.


DISCUSSION.


Experiments to demonstrate possible perivascular and perineiu-onal spaces were first attempted on rather large fetuses (pig), as follows: The spinal meninges were exposed in a fetus in which the heart was still beating vigorously. Into the spinal subarachnoid space was introduced a needle connected with a small reservoir, containing the injection solution (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm. ; water, 100 c.c). The reservoir was then adjusted so that a pressure of 160 mm. of water was maintained in the subarachnoid sjiace. The arteries and veins in the neck of the fetus were then severed, and the subarachnoid pressure maintained at its former level. At the end of 20 minutes the head was placed in a fixative containing 1 per cent hj^drochloric acid.


This work on chromophile cells has, I believe, an important bearing upon (1) the question of differential nerve-cell activity; (2) the phenomena of chondriolysis and hyperchromatism; (3) the functional independence of the mitochondria and the canalicular apparatus; and (4) our conception of the structure of hving nerve-cells.


This procedure, as outlined above, in the adult laboratory mammal, usually resulted in a complete injection of the perivascular system. In the embryo, however, the procedure was uniformly unsuccessful. The injection solution, as shown subsequently by the precipitated prussian-blue, rarely ascended over a centimeter above the point of injection. This indicated that the existent cerebro-spinal fluid was not replaced by the injection solution, and that the failure to demonstrate the perivascular system was to be explained on this basis, if the system were functional at this stage. Attempts were then made to replace the subarachnoid fluid with the injection solution before the cerebral anemia occurred. These attempts likewise m^t with failure, because of the impossibility of keeping the heart beating for any length of time in the larger pig fetuses. Other attem]its were also made to demonstrate these channels, in larger pig embryos, by means of a procedure which in the adult gave at times good injections of these intracortical canals. This method differed from the method first employed only in the maintenance of a high pressure (100 mm. Hg) in the spinal subarachnoid si)aces. It likewise met with failure, due ai)|)arently to the same causes which occasioned its failure in the adult: the high sul)arachnoid pressure o})erated chiefly to compress the cerebral and spinal tissues, rendering the injection of the i)erivascular spaces impossible.


(1) The distribution of chromophile cells in the different parts of the brain is interesting. The fact that they occur most abundantly in the cerebral cortex and in the cerebellum, and that they are rarely found m the lower centers hke the spinal cord, would seem to indicate that the central neurones differ in some way from the more peripheral ones. The difference may be one of lability, for DoUey (1914, p. 56) has found that more highly speciahzed cells are more prone than less specialized ones to respond with structural changes to physiological experimentation. Moreover, the occurrence of these cells in groups, which \'ary m size and in position in different brains, is in accordance with our conception of the alternation of rest and activity in the higher centers and may well have some bearing upon the vexed problem of cortical locahzation, for as yet neither the mitochondria nor the canahcular apparatus have been considered in this connection.


The same procedures were attempted in smaller pig embryos (15 to GO mm.). The method usually successful in demonstrating the spaces (subarachnoid pressure slightly above normal, with subsequent cerel)ral anemia) failed, ajjparently because the cranial cavity at these stages is in no sense a rigid closed box. as in the adult.


(2) We must recognize a "chondriolysis, " or a partial solution of mitochondria, in nerve-cells as well as a "chromatoh'sis." The word "chondriolysis" was first employed by Romeis (1912, p. 139) to describe the disintegration of certain mitochondria which escaped from the cells into the uterine fluid of Ascarifi. It is, to my mind, more appropriate than the term "chromatolysis," which is frequently apphed to the so-called solution of Nissl bodies, for the simple reason that I am of the opinion (1914, p. 20) that the Nissl substance is usually in solution in the hving nerve-cell, whereas the mitochondria are assuredly present as definite formed bodies (except of course in the chromophilic condition).


Any method of service in the adult — which must have in consideration the physical character of the skull as a closed box — was here necessarih' doomed to failure.


Chemical changes are undoubtedly involved in the phenomena of conduction (Tashiro and Adams, 1914, p. 329) and, in view of the distinct differences in the chemical constitution of the mitochondria and of the Xissl substance, the one being of a lipoid albumin nature (Faurc-Fremiet, Mayer and Schaeffer 1910, p. 95) and the other bemg apparently a complex nucleoprotein containing iron (Scott, 1905, p. 507), it seems probable that the studj^ of mitochondria and the changes which they undergo may bring to light variations in the activity' of the nerve-cell which could never be detected by the stud}' of the Xissl substance alone. Quite apart from the role of the nucleus in the elaboration of the Xi.s.sl substance and the purely cytoplasmic nature of mitochondria, there is further evidence of a functional diversity between the two structures. I have found that m the nerve-cells of the mouse the mitochondria s cry directly ^dth the volume of the cytoplasm and that the Xissl substance varies inversely wdth the nucleus cytoplasmic ratio; also that the mitochondria are of more general occurrence in nerve-cells than the Xissl substance.


Together with these technical failures to demonstrate a perivascular system, it must be borne in mind that these are merely failures to demonstrate the existence of the perivascular system in the pig embryo. The system wall probably be demonstrated as soon as a suitable technique is devised. The spaces are very likely present soon after the capillary plexus invades the nervous system, but the observation in many histological preparations of the spaces around the cerebral vessels must not be considered as offering proof of their existence, because of the likehhood of shrinkage influencing the picture. It is interesting, how^ever, to note that elasticity of the cerebral tissues seems greatest along the course of the blood-vessels, for here the phenomenon of shrinkage is most frequently observed. The existence of the perivascular and perineuronal spaces, probably of only capillary thickness, must remain — in the embryo as in the adult — a subject of physiological demonstration; histological e\adence, except with proper physiological regard, is of no value.




The early development and function of such a system as the perivascular and perineuronal canals afford seems most likely from the standpoint of pure speculation. It is not improbable that fluid is poured from this system into the embryonic subarachnoid space at a period soon after the capillary plexus invades the cerebrum. There is no evidence, however, from the observations recorded in foregoing paragraphs, that adequate subarachnoid channels are afforded until the pig embryo reaches a length of about 25 mm. The hypothesis of Essick^^^) regarding the damming of the perivascular fluid as the cause of the two cava corporis striati is of extreme interest in this connection. It remains, however, for future work to afford real evidence in regard to the embryonic perivascular system.




38 THK STIUfTrHK (tl" CUUi i.Mi >l'llll.i; CKI.LS Ol' TIIIC MOHVUU.S .SYSTEM.
XIII. THE PERINEURAL SPACES IN THE PIG EMBRYO.




They are present in the granule-cells of the cerebellum, as is also evident from the earlier work of Altmann (1890. i)late xiii, figure 1) and Nageotte (1909, i). 826), and in the granule-cells of the olfactory bulb of mice and rats, which are well known to he devoid of Xissl substance. ^Moreover, in certain cell-groups, under normal conditions, there is often a variation in the mitochondria, as between different cells, without any corresponding change in the Xissl substance, ^litochondria occur abundantly throughout the length of the axone, where no Nissl substance has ever been seen. They also occur in certain dendritic processes which do not contain any Xissl substance. Evidence of this sort may be multiphed.
The question of the existence of potential or actively functional spaces around the peripheral nerves is of great interest, partly because of the possible relation of these spaces to the developing lymphatic system, and also on account of the anatomical evidence of the possible existence of such spaces.




Just how the mitochondria are concerned with the activity of the nervous system is unknown. I have presented evidence elsewhere (1914, p. 18) that they jilay a part in the basic i)rocesses of metabolism which are common to all cells, but this is unfortunately a very broad statement and we natvu'aliy desire to learn something rather more specific about them. Coghill's (1915, p. 350) belief that the mitochondria are concerned in the constructive (anabohc) side of metabolism is of interest in this connection, particularlj- since it falls so well in line with the wellknown "eclectosome" theory of Regaud (1911, p. 699), which, in turn, is an exteni^ion of the "side chain" theory of Ehrlich. M. R. and W. H. Lewis (1915, p. 393) make the interesting suggestion that the mitochondria take part in cellular respiration, wliich is also a fundamental process common to all cells.
It is realized that before much dependence can be placed on any theory regarding these potential spaces around the cerebro-spinal nerves, the possibility of their being purely artifacts must be dealt with. The methods of demonstration, in the adult, in the hands of the earliest workers were such as to favor the production of artifacts. As far as can be ascertained, Cotugno('), dealing with the ner\-us ischiadicus, was the first to conceive of these possible spaces. His method of demonstration consisted in filling the spinal subarachnoid space with mercurj' (in a cadaver placed in the erect posture) . Globules of the mercurj^ were subsequently found about the sciatic nerve in what then became the perineural spaces.




We may confidently expect that this new avenue of approach to the study of the activity of the nervous system will yield results of importance, not only because our histological methods of technique are now sufficientl}' accurate to permit of the actual enumeration of the mitochondria, a thing which can not be accompli.shed in the case of the Xissl substance, but also because AValdemar and JNIathikle Koch (1913, p. 427) have recently succeeded in devising chemical methods for the (jualitative and quantitative estimation of substances, ^■ery closely related, perhaps identical with mitochondria, in the nervous system. These substances are phospholipins. Hoppe Peyler long ago pointed out that lecithin (a typical phospholipin) and cholesterol are to be found almost everywhere that life jihenomena exist. In fact, a great wave of revi\-ed interest is manifested in recent chemical and pathological literature in these com])lex comjiounds of fatty acid, phosjihorus, and nitrogen. Mathews (1915, p. 88) very aptly remarks that tlic i)li(is])h<)lii)ins are the most important substances in living matter :
Modern anatomical interest in these spaces was aroused by the remarkable injections of Key and Retzius^^s). These investigators, by means of gelatin injections into the spinal subarachnoid space, were able to demonstrate perineural spaces around the cranial nerves, especially around the optic pair. Their results, however, are open to criticism, because of the excessive pressures employed ("not over 60 millimeters of mercury") and because the injections were made in fresh cadavers kept warm for periods of 10 or more hours.
" For they arc found in all cells, and it is undoubtedly their fiinctimi to ])n)(iuee, with eliole.sterol, the peculiar .semifluid, seniisohd state of protopla.sni. Tlie hitler iiolds niucli water in it, but it does not dissolve. Indeed it may be said tiiat the pliospholipins witii cholesterol make the essential substratum of living matter. This physical substratum of ])hosi)holii)in dilTers in ditTerent cells and prohal)iy in tlie same type of cell in different animals, but everywhere, from the lowest plants to the hinlily differentiated brain cells of mammals and of man himself, it jiosstisses certain fundamental chemical and physical properties. In all cases tiie ])hospholipin sutistratum is soluble in alcohol containini; some water," etc.




In view of these considerations it is interesting to iiuiuire whc>ther the distribution of mitoclu»nrlria in cells corresponds with that of the i)hospholi])ins. It is certainlv true that mitochondria are more widclv distributed than anv other kind
Some of the difficulties concerned in the problems of the perineural spaces were cleared up in a studj'^^^) of the cerebro-spinal circulation published in 1914. In this work injections of true solutions (similar to those used in the present study) were introduced into the spinal subarachnoid space in living cats and dogs, under pressures but sUghtly exceeding the normal intraspinal tension. These injections were continued for several hours, and the course of the injection fluid was then estabhshed bj'' precipitating the solution in situ. By means of this procedure, which it was beUeved approached the physiological, the perineural spaces around the cranial nerves could be demonstrated. In these adult laboratory mammals the cerebral nerves without exception showed prussian-blue granules in a perineural relation, extending outward along the nerves bej^ond the termination of the dural cuff. This extension of the injection mass outward was more striking around the first two cranial nerves than about any of the others. Thus, the olfactorj'^ nerves uniformlj'^ showed perineural deposits beyond the cribriform plate, extending downwards into the nasal epitheUura, while the optic nerves were surrounded by the granules in the inf ravaginal sheath, which spreads out over the posterior surface of the eyeball. The caudal cranial nerves were likewise characterized by extensive perineural injections.




These findings were interpreted as e\'idencing a true perineural space, probably of only capillary thickness, which could be injected by filling the cerebro-spinal spaces with a demonstrable true solution. As far as could be made out under the microscope, they had no appreciable existence except when fiUed with the precipitated true solution. These spaces were not filled in the early moments of the injections under low pressures, and could be demonstrated only when the injection had been continued for several hours.


THE STHUCTUKK OF CIlKOMUl'HILK CELLS OV THE NER^-QUS SVSTE.NL 39
of cell grauulution now known to us. They occur in almost all cells. Yet certain cells, like the full}- differentiated non-nucleated red blood-cell, unquestionably contain a large amount of phospholii)in, though no formed mitochondria can be seen. The mitochondrial substance is probal)ly ])resent in solution, just as it appears to be in chromophile cells, for it would olniously l)c absurd to state that it must always occur in that state of condensation which makes it visible with the aid of certain powers of the microscope. The recent investigations of Levene (1915, p. 41) on cephalin are of interest. A new field of investigation is evidently opened up. It may thus be possible to pursue this line of work with chemical as well as with histological and physiological methods, a combination which has been but rarely effected.


The perineural spaces are quite different from the spaces surrounding the spinal ganglia and the gangUa of the cranial nerves. These ganglia he in the true subarachnoid space, wath the dura investing the arachnoid membrane. Distal to the ganglion the dura ends upon each nerve. In the injection under low pressure with the ferrocyanide the cranial and spinal ganglia were all surrounded bj^ the precipitated salts; the cranial nerves .showed extensive perineural injections, whereas the spinal nerves rarely showed a true perineural injection, and then only of hmited extent.


A\'ork along these lines seems the more desirable since, as will be seen, it may throw new light upon certain problems in the pathological anatomy of the nervous system as well. Wells (1907, p. 460), in his discussion of mental fatigue, writes:
"Since the lecithin forms so important a part of the nervous system, it is tempting to imagine that in fatigue excessive quantities of its toxic decomposition product, cholin, and the still more toxic derivative of cholin, neurin, are formed in considerable amounts and cause part, at least, of the
intoxication."
Now we have seen that, in the opinion of certain investigators, mitochondria are largelj^ composed of lecithin. It is possible, therefore, if Wells's reasoning is correct, that the symptoms of mental fatigue are the result of their decomposition. Moreover, Halliburton (1907, p. 74) and others are convinced that organic diseases of the nervous system may be distinguished from functional neuroses on account of the formation of cholin in the one and not in the other. This opens up the possibility of a differentiation between these two great groups of diseases on the basis of cell structure, as to whether or not there is a change in the mitochondria.


The existence of perineural spaces in the embryo, however, has been under dispute. The larger nerv^es in sectioned embryos almost invariably show spaces about them, either a complete separation of the surrounding mesenchyme or a partial dilatation of the mesenchymal interstices. Sabint'*^), in 1902, noted that in pcrispinal injections with inflia ink the spinal nerves could be outlined by the carbon granules, but in no case did such an injection run into true lymphatic channels. No evidence was afforded by her work of any lymphatic channels arising from these apparent perineural channels.


(3) The persistence of the canalicular apparatus in chromophile cells is of interest in general cj'tology. In chromophile cells, in which there are marked structural changes, the canahcular apparatus remams without SLuy great modification. This is rather surprising, since investigators have gradually come to regard the canahcular apparatus as the most labile cell organ; but it is m conformity with Key's as yet unpublished observations on degenerative changes in spinal ganglion cells. Key finds that the canalicular apparatus persists without much modification for from 12 to 24 hours after death in spinal-ganglion cells left in the animal.




I have shown (1912, p. 494) that a canahcular apparatus, in the form of a system of clear, uncolored canals, occurs in the same cell with tj^iical mitochondria and that consequentlj'^ the canalicular apparatus and the mitochondria are structurally distinct. This conclusion is strongly supported by my observation that they may likewise be seen together in chromophile cells, the difference being that while the mitochondria are greatly changed, the canalicular apparatus remains with httle or no modification, so that thej' arc functionally as well as structurally different. ]My positive impregnations of the canahcular apparatus hy the uranium-nitrate method of Cajal confirm this observation.
In the course of this investigation of the cerebro-spinal spaces interest naturally turned to the perineural spaces. In the typical experiments (a replacement of the embryonic cerebro-spinal fluid with a demonstrable true solution in the living embryo), there was evidence of a spread of the replaced solution around the cranial nerves. Because of the procedure u.sed (merely a filling of the ventricles and central canal of the spinal cord) no evidence of a perineural spread occurred until the foreign solution passed into the periaxial tissues. Here the spread chiefly involved the caudal cranial nerves curving around the lateral surface of the medulla in fanshaped processes (figs. 5, 6, 8, and 9). The spread, however, was not extensive. In figure 8 a similar slight spread along the spinal nerves is to be made out. Closer study of these cleared specimens, and examination of the same and of similarly injected embryos after serial sectioning, convinces one that the apparent perineural spread in these cases extends around the sensory ganglia and not further toward the periphery. In no case, cither in the caudal portion of the cranial or in the spinal region, has the replaced injection fluid passed the blastemal condensation of mesenchyme. This finding is well shown by the distribution of the injection fluid in figures 9, 16, and 18.




Now, Cajal (1908, p. 123) is so certain of the identity of the clear canals (described originally by Holmgren) and the "Apparato reticolare interno" of Golgi
The optic nerves, however, jjossessing gangha in the retina, usually show, in the typical replacements in the living embryo, a partial or complete surrounding of the nerves by the prccij)itated pru.ssian-bluc. An incomplete example of this — more typical, according to these observations, than a total circumvention — is given in figures 19 and 20. The higher-power reproduction of this field is very interesting. It shows in the central portion the fiber bundles comprising the optic nerve, surrounded by mesenchyme and the developing ocular muscles. In the region between the ners'e and the muscles is an undiflferentiated mesenchyme which is characterized by a crescent of the precipitated granules of prussian-blue. The non-penetration of the surrounding tissue by the ferrocyanide is verj- well brought out in this drawing. The prussian-blue has reached its position about the nerve by extension from the pericerebral spaces; actually it has still the same distribution as noted in figure 8 above. The adult dura will completelj- surround the optic nerve in its whole extent; the subarachnoid space will likewise extend unbroken to the posterior surface of the eyeball. Hence it must be assumed that in this case the perineural space does not extend beyond the peripheral ganglion. With regard to the olfactory nerves, no evidence of a perineural spread was obtained in specimens of pig embryos up to 45 mm. in length.




It seems obvious, then, that in the embryo pig true solutions, when substituted for the cerebro-spinal fluid, do not extend peripherally along the nerves any further than does the dura in the adult. The replaced fluid (if, as appears most likely, it indicates the true circulation of the cerebro-spinal fluid) extends only through the future subarachnoid space. Such a conclusion is best supported bj' the observations. The only discreparcv between the findings in the pig embryo and those in the adult with the same method lies in the fact that in the adult the cranial nerves showed a much more extensive permeural injection. This seeming discrepancy may be accounted for in two ways. In the first place, the experimental replacement in the embryo pigs lasted at most one hour (due to the fact that the embryo's heart frequently ceased beating at the end of this time), while in the adult cat or dog they were continued for several hours; and it was only in the long-continued experiments in the adult that the extensive perineural injections were obtained. On this basis it seems more than likely that the communications between peripheral perineural spaces and the subarachnoid space are very small and that diffusion must account for the slow filhng of the peripheral system. The second explanation seems undoubtedly to concern the time of development of these perineural spaces in the embryo. It may be that the spaces are morphologically non-existent until late in fetal life; in that case, of course, it is not strange that they have not been filled wdth the injection fluid.


40 Tin: .><Tui( Ti IU-: or ciiuomoi'Hilk cells of thk nkhvous syste.nl
that he refci^i to them us "coiukiits de Golgi-Huhugieii." But Kina Monti (,1915, p. 40) has made the statement that the large internal reticular ai)paratus corresponds to the chondriome (/. c, to mitochondria) in the nerve-cells of mammals; to (juote her own words: "II grande apparato reticolare interno dal Golgi ncUe cellule nervose di mammiferi corrisijontle adunque al condrioma, comme il grande api)arato descritto dal Pensa nelle cellule cartilaginea." If Cajal is correct in his identification, it would ai^jiear that the canalicular apparatus and the mitochondria are identical. I have already discussed (1912, p. 490) the older statements of Popoff (1906, p. 258), Smirnow (1906, p. 153), Van Durme (1907, p. 84), Meves (1908, p. 846), and Hoven (1910, p. 479), who are incUned to believe this to be the case.


From the observations recorded above it is quite apparent that in the typical experiment in which the normal cerebro-spinal tension is not increased no evidence of the perineural space, as injected by Miss Sabin, has been adduced. However, the possibility of mjecting these spinal spaces as was done by Miss Sabin is easily demonstrated. The injections may be made with ease, either with . granular suspensions or with true solutions. Success invariably attends such an injection into the perispinal tissues. The injection solutions easily run out around each nerve, more readily, apparently, in the younger embryo than in the older. It is not clear whether this difference is due to the fact that in younger embryos the resistance is greater to the perispinal flow and less peripherally, or merely to the fact that a greater amount of fluid must be introduced in order to attain the same result. Careful repetition of these observations has led to the conclusion that such a demonstration of the spinal perineural spaces results from excessive pressures of injection. WTienever the pressure exerted by the injection is but slightly above the normal, or does not exceed the normal (as in replacements), the perineural spaces are not injected around the spinal nerves. Miss Sabm's conclusions from her results, that no connection exists between the spaces and the lymphatic system, seem to be wholly substantiated by these observations.


It is hard to see how these two views can be reconciled. I am incUned to think that the well-known lack of specificity of the methods of silver imj^regnation which Pensa (1913, j). 5(30) and Rina ]\lonti (1915, p. 45) have employed are the cause of the confusion. I do not beUeve that the Golgi method can be trusted invariably to demonstrate a certain structure within the cell, like the canalicular apparatus; and, for this reason, I can not accept unreservedly Cajal's identification of the canalicular apparatus with the Golgi apparatus. I agree with Duesberg that a more l)recise definition of the "Apparato reticolare interno" is highly desirable, but I do not agree with him in his attempt (1914, p. 37) to define it in terms of its relation to the centrosome, because our knowledge of the centrosome itself is so deplorabh' inadequate, ^^'e retiuire. above all else, more accurate methods before the matter can be cleared up.


The apparent perineural spaces around the embryonic nerv^es must be looked upon as artifacts. In tissue carefully fixed, dehydrated, and embedded, there is no real evidence of these spaces. Theh- size apparently varies with the care observed in the histological technique.


(4) This discussion of the structure of chromophile cells may be profitablj' concluded by a statement of our present knowledge of the cytoplasmic structure of living nerve-cells of vertebrates not in the chromophilic condition, ^litochondria unquestionably occur and may be seen as such in living nerve-cells even without any vital stain. The Nissl substance is usually {)resent in solution, not in the form of discrete masses as seen in fixed prei)arations. I believe that there is also an amorl)hous argentophilic material which (when treated by appropriate but very capricious methods) as.sumes the form of fibrils. The canalicular apparatus, like the neurofiV)rils,is an unknown ([uantity in living nerve-cells, although it maybe demonstrated in fixed tissues with considerable regularity. These structures, or more correctly speaking substances, are distinct and should not be confu.sed with one another. Although the mitochondria alone have a definite morphology and can u.sually be seen in living nerve-cells, under ordinary conditions, with the i^rescnt means at our disposal, it would be arbitrary in the extreme to say that the others can never be seen. Pigment, fat, li])oid, etc., may of course be seen in variable amount in living nerve-cells. It is the more fundamental constituents with which we are concerned.




The recent work in l)io-cli(niistry, sunnnarized by F. (iowland Hopkins (1913, p. 663) in his presidential address before the Physiological Section of llu> British As.sociation, has, I believe, an imi)ortant bearing here. The cell is regardeil as a dynamic system of co-existing phases in more or less stable equililirium, the condit ion of which is altered, from moment to moment, by jirocesses of oxidation, reduction, desaturation, conden.sation, etc., which naturally result in physical changes in the cell, with the building-up and breaking-down of molecular aggregates which may or
==XIV. General Summary==




In the foregoing sections of this communication some of the problems concerned with the embryology of the cerebro-spinal spaces have been discussed and observ^ations have been presented in the hope that a better conception of the processes might obtain. It is purposed to present here briefly the results of these obser^^ations and to attempt to correlate the findings so far as is possible; and in this, as in the detailed reports in the preceding pages, the relationship of the jihysiological processes concerned will be referred to the morphological changes in the developing embryo.


THE STKLCTU]{K 01 CIlK(JM(>rillI.l-; CKl.I.S OK TIIK NiaiVOrS SYSTEM. 41
may not bo visible with tlie ]nicrosco]Jc or the uhni-iiiicroscopc. The Xissl substance, argentophilous niaterial, etc., doubtless undergo changes of this sort from liquid to fluid and semi-solid phases. It seems right and proper, therefore, to steer an intermediate course, as I have done, between those, on the one hand, who assert thattheNissl substance and the neuro-fil)rils occur in li\irig cells in approximately the same form as they appear in fixed and stained i)reparations, and those, on the other hand, who claim that they are artefacts pure and sim])le and that they can never be seen in the living cell. Our problem is more one of material than it is of form. In this connection the solution of mitochondria in chromophile cells is a phenomenon of considerable significance. In addition to variations in the chemical constitution of mitochondria, there is also evidence to the effect that there may be variations in the condensation or density of the mitochondrial sub.stance (vide Duesbcrg, 1915, p. 40). This is a factor which has been too often ignored. We are inclined to look for mitochondria in all cells which arc functionally active and in which metabolic changes are taking place. The fluidity of the mitochondrial substance varies and I am prepared to beUeve that further investigatioii wUl bring to light cells which are active functionally, but in which no trace of formed mitochondria may be seen.


As a means of studying the physiological extent of the embryonic cerebro-spinal spaces, a method of replacing the medullary fluid with a foreign solution was devised. The procedure consisted in substituting, in the hving embryo, a solution of potassium ferrocyanide and iron-ammonium citrate for the cerebro-spinal fluid. The embryos were then kept alive, for periods of about an hour, by placing them with the attached placenta; in an incubator at 38°. At the end of this time, which varied in the many experiments, the whole embryo was fixed in a medium containing hydrochloric acid, thereby precipitating an insoluble prussian-blue. Specimens prepared in this manner were studied after sectioning or after clearing by the Spalteholz method.


CONCLUSIONS.


Pig embryos, subjected to such experimental replacements, exhibited only an intraventricular retention of the foreign solution until after a stage of 14 mm. was attained. In the earliest specimens, embryos of about 9 mm., there was no characteristic distribution of the foreign solution, except that it remained within the medullary-canal system. In stages of about 13 mm. the replaced fluid also was retained within the cerebral ventricles, but in these specimens a dense accumulation of the precipitated prussian-blue may be made out in a distinct oval in the superior portion of the rhombic roof. This granular aggregation occurs against a histological differentiated area in the roof of the fourth ventricle — an area which represents apparently the more epithelial-Uke elements of the earher roof-plate. This area must be considered solely as a differentiation of the epidermal lining of the medullary-canal system.


(1) Chromophile cells occur under normal conditions in the brains of white mice.


In hving pig embryos of 14 mm. and over, the result of the routine replacement of the ventricular cerebro-spinal fluid was a sUght extraventricular spread into the tissues posterior to the rhombic roof. The passage of this foreign solution outward occurred through the same area of ependjTnal differentiation, outlined bj' the collection of granules against its inner surface in the previous stage. The extraventricular spread remains definitely locahzed to a ver}' small conical area which does not rapidly increase in size.


(2) They are distributed luiccjually in the different parts of the nervous system. They are most abundant in the cerebral cortex. The}' are progressively less abundant in the cerebellum, corpus striatum, thalamus, midbrain, and medulla. They are of very rare occurrence in the spinal cord, spinal ganglia, and sensory ganglia of the cranial nerves.


The factors which cause this initial flow into the pericerebral spaces are of interest. It follows that in the growth of the embryo the production of the intraventricular and intraspinal cerebro-spinal fluid must necessarily keep pace with the increasing size of the cerebral ventricles. It is also necessarj' for the occurrence of an extraventricular spread of the fluid that the production of the fluid within the ventricles must exceed the amount required to keep the medullary-canal sj'stem filled. From our knowledge of the elaboration of the adult cerebro-spinal fluid, it is impossible to conceive of the production of a true cerebro-spinal fluid in the perimcdullary mesenchyme. Such a view would be a reversion to the old hypothesis of Haller, who regarded the leptomeninges as the elaborators of the fluid. Likewise, the passage of the replaced foreign solution into the extraventricular spaces would render such a hjiDothesis untenable.


(3) This restriction of the chromophile cells to the higher centers is in full accord with the well-known lability of the central, more highly specialized cells as contrasted with the more primitive, peripheral neurones.


Hence, it becomes incumbent to regard such an extraventricular spread of the experimental solution as an mdication that the production of the cerebro-spinal fluid within the cerebral ventricles exceeds the capacity of the ventricles to care for the fluid. This argues strongly that the process of elaboration of the fluid in these pig embryos of 14 mm. is no longer sluggish, but that an active production, sufficient to cause a sUght extraventricular flow during the observation, is now taking place. This acceleration of the flow is not great, but it represents a marked change in the relationship of the process of fluid elaboration to the increasing volume of the ventricles.


(4) Chromophile cells, as seen in fixed and stained preparations, vary greatly in structure. There is usually more or less shrmkage of the cell-body. The nucleus vasiy also be shrunken. The acidophUic and basophilic nucleoli are particularly prominent and the ground-substance of the nucleus stains intensely with both acid and basic dyes. There is an increase in the amount of Xissl substance. The Nissl bodies become confluent and form a homogeneous mass. The cell is hyperchromatic. The canalicular apparatus is unaltered. The mitochondria either increase in number and stain more intensely', or else some of them lose their discrete outlines and form a diffuse deposit which stains intensely by the mitochondrial methods of technique. This change in the mitochondria occurs in the cell processes in the neighborhood of the cell, as well as in the cell-body. Although the nucleus may be completely obscured by this cloud of mitochondrial substance, it still remains and stains in the usual way with hematoxylm and eosm.


It seemed desirable to endeavor to correlate this extraventricular spread of the experimental fluid with the morphology of some intraventricular structure at this critical stage of 14 mm. m the pig embryo. The first evidences of villous tufting in the chorioid plexus of the fourth ventricles were found to occur at this stage in the pig. Other studies of this plexus, particularly those which concerned the occurrence of glycogen in these glandular cells, were found to offer no additional evidence of value in regard to the onset of function in these structures. The correspondence between the initial tufting of the ependyma to form the rhombic chorioid plexuses and the initial extraventricular spread must be regarded as of the utmost importance. It would appear most Ukely that as soon as the chorioid tufts occurred an increased production of cerebro-spinal fluid took place, necessitating an extraventricular expulsion of the excess of fluid. Such a view receives the utmost support from these recorded observations; it is in keeping with the best conceptions of the processes of production of cerebro-spinal fluid in adult mammals.


(5) The labilit}' of the mitochondria and the constancy of the canaUcular apparatus in chromophile cells confirms my earher contention by showing that the two structures are physiologically as well as anatomically distinct.


With the initial pericerebral extension of the experimental fluid occurrmg in pig embryos of about 14 nmti., the further extension of this spread did not occur until after a length of 18 mm. was attained. At this stage the replaced foreign solution passed from the fourth ventricle through two areas in the roof-plate. The chorioid plexuses now have divided the roof into two portions; from each, fluid escaped. The superior area of fluid passage is the same which was concerned in the mitial outpouring of the ventricular fluid. The inferior area, like the superior, is an area of ependymal differentiation, of which the first evidence may be made out in pig and human embryos of 15 mm. This differentiation consists in the transformation of the densely staining ependymal elements into cells with larger nuclei, poor in chromatin, and with more abundant cytoplasm.




After the functional employment of the two membranous areas is established at about 18 mm. in the pig, the further pericerebral spri.'ad of the replaced solution occurs very rapidly. The peribulbar tissues are filled with the fluid and from this region extensions occur downward into perispinal spaces and upward into the more basilar pericerebral spaces. Thus, the spinal spaces must be considered as develop>ing physiologically from above, and not from below upward, as Reford found. The complete filling of these perispinal spaces is found in pig embryos of 21 mm. At this stage the pericerebral spaces are filled, with the exception of those around the superior portion of the midbrain and about the cerebral hemispheres.




EXPLANATION OF FIGURES.
The final filling of all the periaxial spaces occurred in pig embryos of about 26 mm. This phenomenon may be taken to indicate the estabUshment of the true cerebro-spinal relationships of the adult, for in this case there is an intraventricular production of the fluid and an extraventricular spread. Likewise, the fluid returns to the venous system in embryos of over 23 mm., and this escape of the fluid from its periaxial bed is, as in the adult, directly into the venous sinuses of the dura mater.




All the figures have been drawn with Zeiss aiKxhromatic objective 1.5 mm. compensating ocular C and camera lucida giving a magnification of l.-'iOO diameters. The figures have not been reduced in reproduction. In all of them unaltered cells are represented side by side w-ith chromophile cells just as they occur in the preparations.
The rapidity of the further extension of the replaced solution after the stage of 18 mm. is passed is apparently due to a second marked acceleration in the rate of production of the ventricular cerebro-spinal fluid. As in the first instance, this increased elaboration seems connected intimately with the formation of the chorioid plexuses of the third and lateral ventricles. As soon as these tufts develop, the cerebro-spinal fluid is produced in amounts which far exceed the quantities for which the more slowlj' enlarging ventricles can provide.




Figures 1 to represent cells in the cerebral cortex of a male white mouse. 26 days old and weighing o grams. The brain was cut into serial sections 4 /u in thickness and stained with fuchsin and methyl green. .\11 the figures were drawn from cells in the same section to insure uniformity in the action of the stain and of the differentiator. The mitochondria are stained red, the Xissl substance green, and the canalicular apparatus persi.«ts, in some of the cells, in the form of clear, uncolored spaces.
The histories of the two areae membranaceae of the fourth ventricle are dissimilar. Both are areas apparently differentiated from the normal lining ependyma for a specific functional purpose — the passage of fluid from the ventricles into the future subarachnoid spaces. The superior membranous area reaches its maximum functional importance in the stages of 18 to 20 mm. in the pig and also in the human embryo and from these stages on it slowly regresses. The final obUteration of the area, if it do not persist as an occasional small remnant, is due to the increasing growth of the cerebellum and the enlargement of the chorioid plexuses of the fourth ventricle. On the other hand, the inferior membranous area continues to increase both in size and functional importance after its initial differentiation from the ependyma; it finally occupies the greater portion of the velum chorioidea inferior. These observations can not solve the interesting question of a perforation of the inferior velum to form the foramen of Magendie.




Figures 7 to 9 represent cells from the cerebral cortex of a male white mou.se, 29 days old and weighing 10 grams. Portions of the brain were prepared by the uranium-nitrate method of Cajal and were cut into serial sections 4 /i thick. These figures were also drawn from a single section to insure uniformity in the .action of the counterstain, methyl green. The canalicular apparatus is in the form of a blackened network and the Nissl substance is colored green.
Of the factors which influence the passage of fluid outward into the periaxial spaces, it must be reahzed that probably there is difference in this regard between the true solutions of the salts and the colloidal suspensions. For the true solutions (as in the experimental replacements) diffusion probably plays some role; but that this is not the sole factor is shown by the failure of the fluid to pass through the membrane in the stages under 14 mm. The findings of the granules of prussian-blue within the cytoplasm of the cells of this membrane mdicates that the fluid passage is similar in every way to that through a true membrane. There is also a possible site of fluid passage between the cells of this membrane. But, surely, the most important factor in this process is one of filtration of the fluid from the point of higher pressure to one of lower. This is mdicated by all of the findings : that the mcreased production of the fluid or the increased mtra ventricular pressure (whether due to normal or experimental agencies) causes a marked extraventricular spread seems firmly established. For the colloidal suspensions (particularly the protein of the normal ventricular fluid) a slower process of diffusion and filtration seems the probable agencj^ for passing the ventricular colloids into the subarachnoid spaces.




Fig. 1 . Two cells, having a distinct increase in amount and intensity of the staining of the mitochondria. This change may mark the first stages in the assumption of the chiomophilic condition.
That the results obtained by the method of replacement were not solely due to diffusion, but represent a fiUing of the physiological extent of the cerebro-spinal spaces, has been shown in many ways, but probably the chief argument against such a view is that whollj^ similar extensions of the foreign solution may be obtained by injections under mild pressures from a syrmge; with increasing pressures these injections show the same type of spread, but always in a smaller embryo than the replacement method demonstrates as the standard for a given stage of the extension. The results recorded in the foregoing pages indicate also that suspensions (India ink) and true solutions (when powerful precipitants) are valuable only for affording comparisons in problems concerning the normal processes of absorption.




2. X much greater increase in amoimt of mitochondria and a slight increase in intensity of the staining of the Kissl substance and the nucleus.
Of interest in any discussion of the results of injections into the perispinal spaces or into the spinal central canal are the findings in regard to the perineural spaces. It is possible to inject such spaces around each of the segmental nerves, but only when the pressures of injections are extreme. In no case, however, were such injections found to enter the lymphatic system — a finding in accord with the observations of Reford and Sabm. The physiological importance of these spaces in the adult is probably great, but the same methods of demonstration (with carefully controlled pressures) which suffice in the adult are unavailing in the embryo.




.3. The Xissl substance is more abundant. It is diffuse and stains more brightly. The outlines of the mitochondria are indistinct. The nucleus stains darkly. -A few clear canals are visible near it. There is what appears to be a shrinkage space on either side of the cell.
The origin of the three meninges from the perimedullary mesenchjTue is well established. His, Kolliker, Sterzi, Farrar, and others have placed this conception on a very firm basis. Most of the investigators have been concerned with the differentiation of the spmal meninges, while the observations here reported have been concerned solely with the cranial portion of these membranes. In general, the same phenomena in the transformation of the primitive periaxial mesenchyme as recorded by these earlier workers may be found in the cranium. The division of the primitive mesenchyme by a secondary condensation, a view advanced chiefly by Salvi, seems well supported. The findings in the cranium are in accord with this concej^tion; the outer portion of this primitive meninx becomes the dura mater, the inner forms both the pia and arachnoid. The processes in the formation of the arachnoid are, however, quite diversified and concern both the formation of the subarachnoid spaces and the outer membrane of the arachnoid.




4. Still greater changes. The mitochondria appear to be going into solution; outlines of nucleus barely distinguishable.
Out of the rather loose-meshed periaxial mesenchyme, the subarachnoid spaces develop. The process concerns the transformation of the small " tissue spaces " of this mesenchyme into the larger subarachnoid channels, which are interrupted by the well-known arachnoid trabecule. Well-marked stages in this metamorphosis, which begins in the basis cranii, can be made out. The first appearance of a differentiation is seen in the gradual increase in the size of the mesenchj'mal mesh. This is closely as.sociated with an increased amount of an albuminous coagulum which in a measure fills the larger interstices. Following this initial dilatation of the spaces occurs a breaking-down of some of the syncytial strands; these ruptured mesenchymal processes then retract and adhere to the persisting trabeculue. The process continues with the formation of larger channels in this mesodermal tissue, with also the formation of the permanent arachnoidal trabeculae. Throughout these larger spaces, in the smaller fetuses, the coagula of protein material are everyT\-here found, the remains apparently of the albuminous portion of the circumambient fluid.




o. The mitochondria have almost all gone into solution. The Nissl substance is almost entirely obscured by the cloud of mitochondrial material which stains with the most energetic of the two dyes, acid fuclisin. The nucleus is invisible.
In the formation of the various cisternse, particularly the great cistema cerebellomedullaris, the process of the dilatation and confluence of the original mesenchymal spaces reaches its maximum. Here the breaking-down of the original sjTicytial strands proceeds to such an extent that very few of the strands remain to persist through life.




(). .V complete " chondriolysis" or solution of the mitochondria. The canalicular apparatus is present in the vicinity of the nucleus.
Such a process of the enlargement of mesenchj^ual spaces to form the larger subarachnoid spaces, as described in some measure by His for the spinal meninges, is apparently intimately connected with the circulation through these spaces of the embryonic cerebro-spinal fluid. The fluid flows everj- where through the spaces, as evidenced by the replacement experiments and by the increased content in albumen, before the process of enlargement of the mesenchymal spaces begins. It seems most likely that this circulation of the fluid acts as the causative agent in initiating and probably also in completing the enlargement of the "tissue spaces." The great content of albumen in the embryonic cerebro-spinal fluid has greatly facihtated the investigation, as the presence of the coagula from this protein has permitted the absolute exclusion of artifacts in the process of the tissue-dilatation.




7. The increase in amount of the Nissl substance indicates a slight degree of chromophilia. The canalicular appa
This mechanism of enlargement of the tissue spaces finds its analogue in the formation of the anterior chamber of the eye and in the perilymphatic spaces of the ear (Streeter). In both these situations, as in the meningeal spaces, there are special body-fluids, more or less characteristic in their physical and chemical characters, obviouslj' subserving specialized functions. In both the eye and cranium, the absorption of the fluids is by way of special organs, directly into venous sinuses; in both, the origin of the specialized fluid is from epidermal organs; this fluid is at fijst poured into epidermal spaces and then subsequently into mesodermal spaces (subarachnoid space and anterior chamber of the eye). Thus, m these situations, the characteristic fluids have certain definite channels through rather larger spaces, connected finally with the venous system, and only indirectly with the Ij-mphatic system.
ratus is blackened and shows no changes.




8. Greater increase in tlie Ni.-vsl substance. It is diffuse, with marked hyperchromatisin. The nucleus stains
In no sense must the cerebro-spinal circulation be taken as a portion of the lymphatic system. Increasing knowledge of the cerebro-spinal fluid, of its physiology and chemistry, and of its pathway, have separated it permanently from any connection with the lymph of the lymphatic system, variable though that be. No longer may the meningeal spaces be compared to serous cavities, except possibly in the case of the subdural space, and this space is really a space apart from the true cerebro-spinal or subarachnoid spaces. Quite s i milarly, in place of the many varjdng channels in the dura and to a lesser extent in the leptomeninges, which older writers considered lyonphatic in nature, our increasing knowledge has caused the introduction of specialized arachnoidal cell-chains running throughout the pachymeninx. Unquestionably, the cerebro-spinal fluid i)ossesses its own peculiar and characteristic pathway, analogous in no way to the lymphatic vessels of other tissues.
diffusely willi methyl green. Its outlines are obscure. The canalicular apparatus, in black, is unaltered and the cell as a whole is shrunken.




9. Cell so intensely stained with the methyl green that the nucleus can not be seen. Canalicular apparatus
The outer continuous membrane of the arachnoidea forms as a mesenchymal condensation, at first in common with the inner surface of the dura mater, but soon separated from it by the subdural space. The very low cubical mesothelium which covers the arachnoid membrane on both surfaces and also invests the arachnoid trabeculae differentiates apparently from the original mesenchymal elements in the periaxial tissues.
slightly coiidciisid, othcnvise unchanged. There is a considerable shrinkage of the cell.




IJ
One of the most interesting features of this study has been the relation of the various mesenchymal condensations to the foreign true solution which was introduced into the medullary-canal system. This fluid circulated throughout the periaxial spaces which enlarge to form the subarachnoid channels, but it never penetrated the primary blastema which served as a primitive dura, nor did it ever invade the pial cells which so closelj^ adhere to the nervous tissue; hkewise, as soon as the secondary mesenchymal condensation dividing the dura from the arachnoid spaces appeared, this condensation was impervious to the true solution. No evidence of any penetration, as might be expected as due to diffusion, could be made out.




This summary has been included in order that some correlation between the topics discussed separately in the foregoing sections might be made. No attempt has been made here to present the findings in abstract form; these have been summarized at the end of each division of this communication.




==XV. Conclusions==




Based on the observations recorded in the foregoing sections, the following conclusions seem warranted:
# During the earlj' part of the growth of the pig embryo there is no extraventricular spread of the cerebro-spinal fluid. The first extension of the ventricular fluid into the periaxial tissues occurs in pig embryos of 14 mm.; the adult relationship of the ventricular and meningeal cerebro-spinal fluid is established in pig embryos of about 26 mm.
# The ventricular cerebro-spinal fluid escapes into the periaxial tissues through two areas of ependjmal differentiation in the roof of the fourth ventricle. Both of these areas differentiate at a shghtly earlier period than that at which they function actively. The area membranacea superior undergoes a gradual regression and obliteration due to the changing form of the rhombic roof; the area membranacea inferior gradually occupies the major portion of the velum chorioidea inferior.
# The embryonic cerebro-spinal fluid, as evidenced by the replacement with a true solution, spreads from the ventricles into the mesenchymal tissue about the central nervous system. It docs not penetrate the cranial or vertebral blastemal condensations, nor does it invade the pial cellular layer.
# The subarachnoid spaces arise by a process of breaking-down of the perimedullary mesenchj'mal sj'ncytium and a dilatation of the existent mesench3Tnal spaces. This phenomenon of the enlargement of the mesenchymal spaces is associated with the presence in the spaces of an increased amount of albumen. The process occurs at a period shghtly later than that at which the initial flow of the cerebro-spinal fluid into the spaces is recorded.
# The dura mater, arachnoid, and pia mater develop out of the perimedullary mesenchj'me. The arachnoid trabeculae are left by the breaking-down of the original mesenchymal strands, while the outer arachnoid membrane is formed, together with the inner surface of the dura, by a separate mesenchymal condensation. The dura develops between this secondary Une of condensation and the embryonic skull.
# There is indicated a very close relationship between the tufting of the chorioid plexuses of the fourth ventricle and the first extraventricular spread of the cerebro-spinal fluid.
# By means of the method of replacement it is possible to demonstrate perineural spaces as far out along the nerve trunks as the peripheral gangUa. The extensive injections of the perineural spaces along the segmental nerves are not obtained by the method of replacement.




i
The work, of which this paper forms the report, was done in the Anatomical Laboratory- of the Johns Hopkins Medical School. It was largely due to aid received from the Department of embryology of the Carnegie Institution of Washington that the completion and scope of this paper were possible. The wTiter gladly acknowledges his indebtedness to the Carnegie Institution. January, 1916.




==Bibliography==


Ad AMKiEwicz, Albert. Die Blutgef &S8e des menschlichen Rilckenmarkes. I Theil: Die Gefassc der RQckenmarks-Substanz. II Theil: Die Gefassc der Rlickcnmarksobcrflache. Sitjungshcrichte der k. Akad. d. Wiss. in Wien. Math.-naturw. CI., 1H81, Lxxxrv, pt. 3. 469; 1882, Lxxxv, pt. 3, 101.




{{Ref-Bardeen1910}}


Bardeen, C. R. The development of the skeleton and of the connective tissues. In Kcibel and Mall, Manual Fluman Embryology, Philadelphia and Ix>ndon, 1910, I, 292.




BuAKE, J. A. The roof and lateral recesses of the fourth ventricle considered morphologically and embryologically. Jour. Comp. Neurol., Granville, Ohio, 1900, X, 79.




Canniec, a. Contribution & I'dtude de la voOve du quatrifeme ventriculc chez Ics mammifires; le trou de Magendie. Jour, de Mfed. de Bordeaux, 1897, xxvii, 547.




Cappelletti, L. L'^coulemcnt de liquide cerebrospinal par la fistule c^phalo-rachidienne en conditions normales ct sous I'influence de quelques m6dicaments. Arch. Ital. de Biol., Turin, 1900, xxxv, pt. 2. 463.




//•
.L'efflusso del liquido cerebro-spinale dalla fistola cefalo-rachidiana in condizioni normali e sotto I'influenza di alcuni farmaei. .Vtti dell' / ccademia d. Scienze Med. e Nat. di Fcrrara, 1900, ^xxiv, 85.




Cathelin, F. La circulation du liquide ctphalorachidien. Paris, J. B. Bailli^re, 1912.


-~3


CoTL'GNO [CoTUNNius], DoMENico. De ischiade nervosa commentarius. Viennae, R. GriifTcr, 1770.




Creighton, C. Microscopic researches on the formative properties of glycogen. London, A. & C. Black, 1896-99.




CusniNO, H. Physiologische und anatomische Beobachtungen iiber den Einfluss von Hirnkompres-sion auf den intracraniellen Kreislauf und ilber einigc hiermit verwandtc Erscheinungcn. Mittcilungen a. d. Grenzgeb. d. Med. und Chir., Jena, 1902, ix, 773.


:•)■ 4


. Some experimental and clinical observations concerning states of increased intracranial tension. Am. Jour. Med. Sciences, Philadelphia and New York, 1902, cxxi\-, 375.




. Studies on cerebro-spinal fluid: No. 1. Introductory. Jour. Med. Research, Boston, 1914, xxxi (n. s. XXVI), 1.




Dandy, W. E., and K. D. Buckfan. An experimental and clinical study of internal hydrocephalus. Jour. Am. Mod. Assoc, Chicago, 1913, Lxi, 2216.




. Internal hydrocephalus: An experimental, clinical and pathological study. Aiii. Joiir. Dis. Children, Chicago, 1914, viii, 406.




■ ^ •
Ekker, H. De cerebri et meduUte spinalis systemate vasorura capillari. Trajecti ad Rhenum, 1853.




Kngel, E. A. Uelx^r die Secretionscrscheinungen in den Zellen der plexus chorioidei des Mcnschen. Arch. f. Zellforschung, Leipzig, 1909, ii. 191.


#


Essick, C. R. Transitory cavities in the corpus striatum of the human embryo. Contributions to Embryology, 1915, II, No. 6, 95. (Carnegie Institution of Washington, Pub. No. 222.)




«^\
14. Evans, H. McL. The macrophages of mammals, Am. Jour. Physiol., Baltimore, 1915, xxxvii, 1.


15. Faivre, J.-J.-A.-E. Des granulationes miningiennes. Thfee de Paris, 1853. (No. 142, vol. 540.)


16. Farrar, C. B. The embryonic pia. .\m. Jour. Insanity, Baltimore, 190& 7 LXiii, 295.


r>
17. FiNDLAT, J. W. The choroid plexuses of the lateral ventricles of the brain — their histolog}-, normal and pathological. Brain, London and New York, 1899, xxu, 161.


18. Frazier, C. H. The cerebro-spinal fluid in health and disease. Jour. Am. Med. Assoc, Chicago, 1915, LXIV, 1119.


19. Gaupp, E. Die Entwickelung des Kopfskelettes. In Hertwig's Lehrbuch der Entwickelungsgeschichte d. Menschcn u. d. W'irbcltierc. Jena, 1906, K od., 692.


^
20. Goldmann, E. E. V'italfarbung am Zentralnervensystcm. Berlin, G. Reimcr, 1913.


21. Haller, a. Elementa physiologix corp^jris humanis. Lausanne, 1757-78.


22. Hess, Carl. Das Foramen Magendii und die Oeffnungen an den Reces.iu.s laterales des \*ierten Ventrikels. Morphol. Jahrbuch, Leipzig, 1885, x, 578.


y
23. Hecser, C. H. The development of the cerebral ventricles in the pig. Am. Jour. Anat., Philadelphia, 1913-14, XV, 215.




24. Hill, L. Physiology- and pathology of the cerebral circulation. London, J. & A. Churchill, 1896.


^*^


25. His, W. Ueber ein perivasculiires canalsystem in den nervoseu Ccntralorganen. Leipzig, Engelmann, 1865. . Die Hfiute und Hdhlen des Korpers; ein academisches Programm. Basel, Schweighauser, 1865.


. Die Entwickelung des mcnschlichcn Rautenhirns, von Ende des ersten bis zum Beginn des dritten Monats. Leipzig, Hirzel, 1890.


\  
. Die Hiiute und Hohlen des Korpers. Arch. f. Anatomie u. Physiol., Anat. .\bth., 1903, 368.


. Die Entwickelung des menschlichen Gesims. Leipzig, Hirzel, 1904.




p>.
26. HwoROSTUCHiN, W. Zur Frage uber den Bau des Plexus chorioideus. Arch. f. mikr. Anat., Bonn, 1911, Lxxvii, 232.




27. Jacobson, C. Unpublished.




28. Kadvi, Heinrich. Ueber die Blutgefisse des menschlichen RQckenmarkes. Lemberg, Gubrj-nowicz 4 Schmidt, 1889.


\t0^


29. Key, E. A. H., and G. Retzius. Anatomie des Nervensystems und des Bindegewebcs. Stockholm, Samson & Wallin, 1875-76.




}
30. Knower, H. McE. a new and sensitive method of injecting the vessels of small embrjos, etc., under the microscope. Anat. Record, Philadelphia, 1908, ii, 207.




31. KoLLlKER, A. Entwiekelungsgeschichte des Mcnschen und der hoheren Thiere. 2ed. Leipzig, Engelmann, 1879.


■^
. Gnmdriss der Entwiekelungsgeschichte des Mcnschen. I^eipzig, Engelmann, 18S0.




32. KoLLMANN, J. Lehrbuch der Entwiekelungsgeschichte des Menschen. Jena, Fischer, 1898.


^c>-^


33. Lewandowskt, M. Zur Lehre von der CerebrospinalflQssigkcit. Zeitachr. f. klin. Med., Berlin, 1900, XL, 480.




^ ■^y
34. vov LrscHKA, H. Die .\dergeflechte des menschlichen Gehimes. Berlin, G. Reimer, 1855.




35. Magen-die, Francois. Recherches sur le liquide csfephalorachidien. Paris, 1825.


THE STRUCTURE OF CHROMOPIIILE CELI>,S OF THE NERVOUS SYSTEM.


. Recherches philosophiqucs et cUniques sur le liquide c^phalo-rachidien ou c6r6bro-spinal. Paris, 1842.


{{Ref-Mall1904}}


36. Mall, F. P. On the development of the blood-vessels of the brain in the human embryo. Am. Jour. Anat., Baltimore, 1904, rv, 1.




43
37. Meek, Walter J. A study of the choroid plexus. Jour. Comp. Neurol, and Psychol., 1907, xvii. No. 3, 286.




38. Mestrez.\t, W. Le liquide c^phalo-rachidien. Paris, Maloine, 1912. Thise de Montpellier, 1911, No. 17. .


BIBLIOGRAPHY.


39. MiDDLEM.tss, James, and W. F. Robertson. Pathologyof the nervous system in relation to mental diseases. Part III. Morbid conditions of the dura mater. Edinburgh Med. Jour. 1895, xl, pt. 2, 704.




MiNOT, C. S. Human Embryology. New York, W.


Wood A Co.. 1892. MoTT. F. W. The Oliver-Sharpey lectures on the cerebro-spinal fluid. Liincct. London. 1910, ii, 1 ; 79. Peluzzi. B. Experinientelle histologische Untersuchungcn ilbcr die Plexus chorioidei (Adcrgeflechte).


.\ltman.v, It. 1S90. Die Elemeiitarorganismcn und ihrc
beziehungen zu den Zcllcn. 145 pp. Leipzig,
Veit & Co. HvRKF.R, LEWEL1.Y8 F. 1899. Thc nervous system and ita
constituent neurones. 1122 pp. New York,
.\ppleton & Co. Bk.vslet, U. II. 1911. Studies on the pancreas of the guinca
pig. Amer. Jour. Anat., vol. 12, pp. 297-388. Bls.\cc.\, .\rchimede. 191.3. L'apparato mitochondriale
nelle cellule nervosc adultc. Arch. f. Zellf., Bd.


Foli.i neuro-biologica, Haarlem, 1911. v, No. 4, 305. Pf.ttit. a., and J. Giiuhd. Sur la fonetion .s4or6toire ct la morphologie des plexus chorioldes des ventrilulcs latfraux du Syst^nie ncrveux central. Arch.


11, pp. .327-339. Caj.^l, S. R. 1908. Les conduits dc Golgi-Holmgren du
protoplasma nerveux et le rfiseau pericellulaire
de la membrane. Trab. Lab. Invest, biol., Vniv.


d'anat. micr., Paris, 1902, v. 214. PuLT, F., O. Kehm, and H Schott.mOller. Leitfadcn zur Untcrsuchung der ZerebrospinalflOssigkeit.


Madrid, t. 6, pp. 123-136. — 1909. Histologie du systime ner\-eux de I'homme et
de-s vertebras, t. 1, 986 pp. Paris, .\. Maloine.


Jena, Fischer, 1913. PoiBiER, P., and A. Charpy. Traits d'antomie huniaine.


1912. F6rmula de fijaeion para la demostraci6n
fScil del aparato reticular de Golgi y apuntcs sobre la disposicion dc dicho aparato en la retina, en los nervios y algunos estados patol6gicos. Trab. Lab. Invest, biol., Univ. Madrid, t. 10, pp. 209-220.


Pari,'*, BattaiUe, 1892-1902. Qci-vcKE, H. Zur Physiologie der Cerebrospinalflussigkeit. Arch, f . Anat., Physiol, u. wissensch. Med. (Du Bois-Reymond), Leipzig, 1872, 153. Refobd, L. L. Unpublished. Work referred to by Sabin (49) and Gushing (9). Reiner, M., and J. Schnitzler. Ueber die Abflu.sswege des Liquor cerebrospinalis. Centralblatt f. Physiol., Leipzig and Wien, 1894, viii, 6H4. . Zur Lehre vom Hirndruck. Wiener klin. Wochenschr.. 1895, viii, 371.


CoGniLL, George G. 1915. Intracellular digestion and assimilation in amphibian embrj-os. Science, vol. 42. pp. 347-3.iO.
Sabin, F. R. Development of lymphatic systcm. In Keibel and Mall's Manual Human Embryology,  London 4 Philadelphia, 1912, ii, 709. Salvi, G. Histogenise et structure des meninges.




CowDRY, E. V. 1912. Thc relations of mitochondria and other cytoplasmic constituents in spinal ganglion cells of the pigeon. Intern. Monatssch. f. .\nat. u. Phys., Bd. 29, pp. 473-504.
Thise de Pari.", 1898. . Quoted in Poiricr et Charpy. Traitt d'Anatomie humaine. Paris, 1901, tome iii, 113. ScHROEDER VAN DER KoLK, J. L. C. Bau und Functionen der Medulla spinalis und oblongata. Aus dem Holliindischen iihertragen von Dr. Fr. W.




1914. The comparative distribution of mitochondria
Theile. Braunschweig, Vieweg & Sohn, 1859. Spin.i, a. Expcrimenteller Beitrag zur Kenntniss der Hyperamic des Gehirns. Wiener med. Blatter, 1898, XXI, 17; 247. . Experimentelle Untcrsuchungen iiber die Bildung des Liquor cerebrospinalis. Arch. f. d. gesamte Physiol., Bonn, 1899, Lxxvi, 204.
in spinal ganglion cells of vertebrates. .\mer. Jour. Anat., vol. 17, pp. 1-29
DoLLEY, D. H. 1910. Thc pathological cytology of surgical shock. Jour. Med. Ke^., vol. 22, pp. 331-378.




1911. Studies in the recuperation of ner\-e-cells
52. Spina. A. Uebcr den Einfluss des hohen Blutdrucks auf die Neubildung des CerebrospinalflQssigkeit. Arch, f. d. gcsamtc Physiol., Bonn, 1900, i.xxx, 370.
after functional activity from youth to senility. Jour. Med. Res., vol. 24, pp. 309-344.




1914. Fatigue of excitation and fatigue of depression.
. Untersuchungen uber die Resorption des Liquor bei normulcm und crhohtem intracraniellem Drucke. .Vrch. f. d. gesamte Physiol., Bonn, 1900-1901, Lxxxiii, 120; 415.




A comparison of the reactive effects of function and of the by-products of function on the ner^■ecell. Intern. Monatssch. f. .\nat. u. Phys.,^d. 31, pp. 35-62. DuESBERG, J. 1914. Trophospongien und Golgischer Binnenapparat. Verh. d. Anat. ges. in Innsbruck, pp. 11-80.
53. .'^TERZi, G. Recherches sur I'anatomie comparfee et sur rontogent-se des m6ninges. Arch. Ital. de Biol., 1902, XXXVII, 257.




1915. Recherches cytologiquejs sur la ffcondation
. Ricerche inforno all' anatomia comparata ed air ontogenesi dcllc mcningi, e considerazioni suUa filogenesi. Atti del R. Instituto Veneto di scienze, letterc ed .\rti, 1900-1901, LX, parte ii, 1101.
des Ascidiens et sur leur d^veloppenicnt. Contributions to Embrjologj', Carnegie Institution
of W,i.shington, No. 8, pp. 35-70. F.\ur6-Fremiet, Mayer, and Schaeffer. 1910. Sur la
microchemie des corps gras. .\rch. d'Anat. Micr.,  
t. 12, pp. 19-102. Flesch, M. 1887. Ueber die Verschiedenheiten im chem
ischen Verhalten der Ner\-enzellen. Mitth. d.




Natiu-f. Ges. in Bern Nr. 1169-1194, pp. 192-199. G.^leotti, Gino. 1895. Ueber die Granulationen in den
54. Streeter, G. L. Development of the nervous system. In Keibel and Mall, Manual Human Embryology, London & Philadelphia, 1912, ii, 1.
Zellen. Intern. Monatssch. f. Anat. u. Phys.,  
Bd. 12, pp. 440-557. HALLiBrRTOx, W. D. 1907. Die Biochemie der peripheren
Nerven. Ergeb. d. Phys., Bd. 4, pp. 1-88. Hauschild, M. W. 1914. Zellstniktur und Sekretion in
den Orbitaldrilsen der Nager. Ein Beitrag zur
Lehre von den geformten Protoplasmagcbilden.


. The development of the venous sinuses of the dura mater in the human embryo. Am. Jour, of Anat., Philadelphia, 1915, xviii, 145.


Anat. Hefte. Bd. 50. pp. 533-629. Holmgren-, Emil. 1909. Studien iibcr die stofflichen Veran
deruugen der quergestreiften Muskclfasem. Skan.


55. Weed. L. H. Studies on cerebro-spinal fluid. No. II: The theories of drainage of cerebro-spinal fluid with an analysis of the methods of investigation. Jour.Med. Research, Boston, 1914, xxxi (n. s. xxvi), 21. . Studies on cerebro-spinal fluid. No. Ill: The pathways of escape from the subarachnoid spaces, with particular reference to the arachnoid Wlli. Jour. Med. Research, Boston, 1914, xxxi (n. a. XX VI), 51. . Studies on cerebro-spinal fluid. No. IV: The dual source of cerebro-spinal fluid. Jour. Med. Research. Boston, 1914. xxxi (n. s. xxvi). 93.


.Vrch. f. Phys., Bd. 21, pp. 287-314. Hopkins, F. GowutxD. 1913. The dynamic side of biochem
istrj-. Proc. Brit. .\sso. Adv. Sci., Birmingham
Meeting, pp. 652-668.


56. Wilder, B. G. Notes on the Foramina of Mageudie in man and the eat. Jour. Nerv. and Ment. Dis., New York, 1886, xiii, 206. . The metapore (foramen of Magendie) in man and an orang. Med. News, Philadelphia, 1893, LXiii 439. . Meninges. Ref. Handbuch Med. Sciences, New York, 1893, ix (suppl.), 606.




57. ZiEOLER, P. Ueber die Mechanik des normalen und pathologischen Hirndruckes. Arch. f. klin. Chirurgie, Berlin, 1896, Liii, 75.


==Explanation of Plates==


HcvEN, Henri. 1910. Sur I'histogehfese du systime nerveux
peripheriquc chez le poulet et sur Ic rdle des chon
driosomes dans la neurofibriUation. Arch, de Biol.,
t. 25, pp. 426-492. Koch, W. and M. L. 1913. Contributions to the chemical
diflerentiation of the central ncr\'ous sj'stem.


KEY FOR FIGURE-LEGENDS.


III. The chemical differentiation of the brain of
the albino rat during growth. Jour. Biol. Chem.,
vol. 15. pp. 42.3^48. Levene, P. A. 1915. Cephalin. II. Brain cephalin. Jour.


ami, area membranacea inferior. dmc, dura mater cerebri (inner surface, in pme, pia mater cerebri.


Biol. Chem.. vol. 24. pp. 41-53. Lewis, M. R. and W. H. 1915. Mitochondria (and other
cytoplasmic structures) in tissue cultures. .\mer.


amt, area membranacea .superior. approximation with arachnoid). j»p6, precipitated prussian-blue.


Jour. .Vnat., vol. 17, pp. .339-401. M.tTHEWs. A. P. 1915. Physiological Chemistr>-. New
York. William Wood A Co., 1040 pp. Mayer, Rathery, and Schaeffer. 1914. Les granulations
ou mitochondries de la cellule h^patique. Deuxitme
partic. Jour. Phys. et de Path, gfn., t. 16, pp.


cbl, cranial blastema. epe, epithelial-like cells lining ventricle. pun, reduced silver nitrate.


607-622. MEVE.S. Fr. 1908. Die Chondriosomen als Trager erblichcr
.\nlagen. Cytologisiche Studien am Huhnerem
brj-o. .\rch. mikr. .Anat., Bd. 12. pp. 816-867. MoN'Ti, RiNA. 1915. I condriosomi e gli apparati di Golgi
nelle cellule ner\'o5e. Arch. Ital. di Anat. e di
Embrj.. vol. 14, pp. 1—15. Nageotte, J. 1909. Mitochondries du tissu ner\-cux. C.


cent, cisterna cerebello-mcduUari.i. epe, ependyma. s<u, subarachnoid spaces.


rend, soc Biol., t. 66. pp. 825-828. NissL, Fr. 1896. Die Beziehungen der Ner\enzellensub
stanzen zu den thatigen, ruhcnden imd ermiideten
Zellzustanden. Allg. Zeitsch. f. Psj'chiatrie, Bd.


chp, plexus chorioiduus. 4"^, vcntriculus quarlus. tir, sinus trans\ersu3.


52, pp. 1147-1154. Pessa, \. 1913. La struttura della cellula Cartilaginea.


===Plate I.===
[[File:Weed1917 plate01.jpg|800px]]


Arch. f. Zellf., Bd. 11. pp. 557-582. PopoFF, M. 1906. Zur Frage der Homologisierung des Bin
'''Fig. 1.''' Drawing of a pig embryo of 9 mm, into the spinal central canal of which an injection of 0.5 per cent solution of potassium ferrocyanide and iron-ammonium citrate was made under very mild syringe-pressure. The embryo was fixed in Camoy's fluid to which 1 per cent hydrochloric acid had been added. The specimen was carefully dehydrated and cleared by the Spalteholz method. The resultant precipitate of prussian-blue is found w holly within the central canal of the spinal cord and within the cerebral ventricles. Enlargement, 11 diameters.
nenetzes der GangUenzellen mil den Chromidien
(Mitochondria etc.), der Geschlechtzellen. Anat.




.\nz., Bd. 29. pp. 249-258. Reo.\uu, Cl. 1910. Etude sur la structure des tubes s£mi
'''Fig. 2.''' Drawing of a pig embryo of 13 mm, in which the cerebro-spinal fluid was replacc<l by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. The embryo was kept alive for 90 minutes after this replacement and was then fixed in 10 per cent formol containing 1 per cent hydrochloric acid, .\fter dehydration the specimen was cleared by the Spalteholz method. The occiurence of a definite oval, outlined by the denser mass of the granules, in the roof of the fourth ventricle, is characteristic of this stage. Enlargement, 9 diameters.
niferes et sur la spermatogenese chez les mam
mif^res. Arch. d'.Ainat. Micr.. t. 11, pp. 291-131. 1911. Les mitochondries organites du protoplasma
considerfes comme des agents de la fonction eelec
tique et pharmacopexique des cellules. Rev. de
Med., t. 31, pp. 081-699. RoMEis, B. 1912. Beobachtungen iiber Degenerationser
scheinungen von Chondriosomen. Arch. f. mikr.


'''Fig. 3.''' Drawing of a pig embryo of 14.5 mm, in which the cerebro-spinal fluid was likewise replaced by the ferrocyanide solution, After the replacement, the embryo was kept alive for 60 minutes; it was fixed in Camoy's fluid (with 1 per cent hydrochloric acid added) and after dehydration it was cleared by the SpaltehoU method. The earliest indications of a peria.idal spread of the replaced fluid from the roof of the fourth ventricle is here shown. Enlargement, 8 diameters.


Anat., Bd. SO, pp. 129-170. Scott, F. H. 1905. On the metabolism and action of nen-e
===Plate II.===
cells. Brain, vol. 28, pp. 506-526. ScHiROKOGOROFF, J. J. 1913. Die Mitochondrien in den
[[File:Weed1917 plate02.jpg|800px]]
erwachsenen Ner\cnzcllen des Zcntralner\ens>'S
tems. Anat. .\nz., Bd. 43, pp. 522-524. Smirnow, a. V. 1906. Ueljer die Mitochondrien und den
Golgischen Bildungen analoge Strukturen in
einigen Zellen von Hyacinthus orientalis. Anat.


'''Fig. 4.''' Drawing of a pig embryo of 18 mm., in which a typical replacement of the spinal fluid had been made. The animal was kept alive for 45 minute." and was then fixed, dehydrated, and cleared in the usual manner. The extra ventricular spread of the replaced fluid from two are;is in the roof of the fourth ventricle is well illustrated. Enlargement, 9 diameters.


Hefte, Bd. 32, pp. 143-153. T.iSHiRO, Shiro, and H. S. Ad.vms. 1914. Comparison of the
carbon-dioxide output of the nerre fibers and
ganglia in Limulus. Jour. Biol. Chem , vol. 18,
pp. 329-334. Van Di'RME, M. 1907. Les mitochondries et la methode de
Sjovall dans I'ovogenfee des oiseaux. Ann. de
MM. de Gand, vol. 87, pp. 76-86. Wells H. Gideox. 1907. Chemical Pathologj-. 549 pp.


'''Fig. 5.''' Drawing of a pig embryo of 19 mm., in which likewise a typical replacement of the cerebro-spinal fluid by the ferrocyanide solution had been made. After this procedure, the embryo was kept aUve for 55 minutes and was then carried through the routine technique for the Spalteholz method. The ftirther pericerebral spread of the replaced fluid is recorded. Enlargement, 8 diameters.


Philadelphia, W. B. Saunders Co.


===Plate III.===
[[File:Weed1917 plate03.jpg|800px]]


'''Fig. 6.''' A frank lateral drawing of a pig embryo of 21 mm. The typical replacement of the embryonic cerebro-spinal fluid by the ferrocyanide solution was effected in this embryo and it was then kept alive for 45 minutes. At the end of this time the embryo was fi.xed in an acid fluid, dehydrated, ana cleared. The almost complete periaxial spread of the replaced fluid is indicated by the precipitated granules. Enlargement, 7.6 diameters.




'''Fig. 7.''' A dorsal view of the embryo illustrated in fig. 6. The perispinal spread of the replaced fluid is well shown. Enlargement, 7.8 diameters.


CONTRIBUTIONS TO EMBRYOLOGY, No. 12.


===Plate IV.===
[[File:Weed1917 plate04.jpg|800px]]


Fig. 8. Drawing of a pig embryo of 26 mm. in which the typical replacement of the cerebro-spinal fluid has been made. After the introduction of the ferrocyanide solution the embryo was kept alive for one hour; at the end of this time it was fixed in an acid solution, subsequently dehydrated, and cleared in oil of wintergreen. The specimen shows a complete periaxial spread of the replaced fluid, as evidenced by the precipitated granules, in addition to a total filling of the intramedullary system. Enlargement, 6.5 diameters.




Fig. 9. Drawing of a pig embryo of 16 m.m., in which the central canal of the spinal cord was injected with the ferrocyanide so.u'ion under moderate syringe-pressure. After fixation in an acid mediiun the embn,-owaa dehydrated and cleared by the Spalteholz method. The extraventricular snread in the peribulbar region Ls easily made out. Enlargement, 9 diameters.


ON THE DEVELOPMENT OF THE LYMPHATICS OF THE LUNGS IN THE EMBRYO PIG.


===Plate V.===
[[File:Weed1917 plate05.jpg|800px]]


''Fig. 10.'' Drawing of a pig embryo of 21 mm., in which an injection of diluted india ink was made into the central canal of the spinal cord. The preasure employed was the highest obtainable from the syringe, yet below the tension causing rupture. The specimen, after injection, was fixed, dehydrated, and cleared. The slight extent of the periaxial spread of the carbon granules can be easily seen. Enlargement, 7 diameters.




'''Fig. 11.''' Drawing of a pig embryo of 16 mm., in which an injection (under moderate syringe-pressure) of 0.5 per cent solution of silver nitrate was made into the central canal of the spinal cord. The silver was reduced in the sunlight, the embryo then fixed. After dehydration, the embryo was cleared in benzol and oil of wintergreen. Enlargement, 7. .5 diameters.


By R. S. Cunni.vgham.


'''Fig. 12.''' Drawing of a pig embryo of 13 mm.; into the central canal of the spinal cord a dilute solution of nitrate of silver was injected under strong syringe-pressure. Reduction of the silver was accomplished by exposure to sunlight; the embryo was then fixed, dehydrated, and cleared. Enlargement, diameters.




===Plate VI.===


[[File:Weed1917 plate06.jpg|800px]]


With five plates.
''Fig. 13.'' Photomicrograph of transverse section of a pig embryo of 15 mm. Specimen obtained from an embryo in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and ironammonium citrate. After this replacement the embryo was kept aUve for 65 minutes. The resultant priL'Ssian-blue precipitate is not included in this photomicrograph. Enlargement, 13 diameters.




'''Fig. 14.''' Drawing of blocked area in fig. 13, under higher magnification and including the resultant precipitate of prussian-blue. The typical ependymal cells (epc) lining the fourth ventricle are shown on either side; between them occurs the area membranacea superior {ams). The transit of the replacement fluid through the membranous area and the spread through the adjacent mesenchyme are illustrated. Enlargement, 245 diameters.




'''Fig. 15''' Photomicrograph of transverse section from embryo pig illustrated in fig. 13. Section taken from more caudal plane than that given in the former figure. The prussian-blue spread is not illustrated. Enlargement, 10 diameters.


45


'''Fig. 16.''' Drawing, under higher magnification, of the rectangular area in fig. 15. The passage of the replaced solution, as shown by the resultant precipitate of prussian-blue, through tlie area membranacea inferior (ami) is here illustrated. The extension of the replaced fluid through the adjacent mesenchyme and the nonpenetration of the solution into the condensed mesenchyme are shown. Enlargement, 140 diameters.




CONTENTS.
'''Flg. 17.''' Photomicrograph of sagittal section of a pig embryo of 18 mm. Specimen obtained from an embrj-o in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. After this replacement the animal was kept alive for 45 minutes. Fixed for 5 minutes in 10 per cent formol containing 1 per cent hydrochloric acid; then over night in modified Boiiin's solution (saturated aqueous solution of picric acid 75, formaldi-hyde 10, glacial acetic arid 10). Dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraffin. Serial sections, stained by hematoxylin and eosin. The resultant precipitate of prussian-blue has not been reproduced in the photomicrograph. Enlargement, 8 diameters.




PAGE.
'''Fig. 18.''' Drawing of blocked area in fig. 17 under higher magnification. The granules of prussian-blue are here represented by the blue stenciling. The transit of the fluid, as shown by the granules, into the periaxial mesenchyme through the two membranous areas {arm and ami) in the roof of the fourth ventricle are well shown. Enlargement, 35 diameters.




Methods 50-52
===Plate VII.===
Vessels arising from the left (hict ". 52-64
[[File:Weed1917 plate01.jpg|800px]]
Lyniphaties of the hioiichi 03
Lymphatics of the veins 63
Lymphatics of the pleura 63
Summan- 64-66
Bibliograph}' 66
Explanation of plates 67-68


'''Fig. 19.''' Photomicrograph from a sagittal section of a fetal pig of 27 mm. The cerebro-spinal fluid in this specimen was replaced by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate; the fetus was kept alive for 40 minutes; fixed in 10 per cent formol containing 1 per cent hydrochloric acid for 15 minutes; then over night in modified Benin's solution; dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraffin. The prussian-blue granules are not represented in this photomicrograph. Enlargement, 8 diameters.




ON THE DKVHLOI'MKVr (>l- TIIK LVMI'llATICS OK TIIK I.I Mis l\
'''Fig. 20.''' Drawing of squared area in fig. 19. The center of the field is occupied by the optic nerrve; around it the developing extrinsic optic muscles are shown. The precipitate of prussian-blue occurs in the perineural mesenchyme. Enlargement, 190 diameters.
KMBKVO I'Ki.




'''Fig. 21.''' Photomicrograph of rectangular area in fig. 19. The passage of the ferrocyanide solution into the sinus transvcrsus {sir) is represented by the precipitated blue granules. Enlargement, 133 diameters.




'''Fig. 22.''' Photomicrograph of a transverse section of a pig enibrj-o of 23 mm. The cerebro-spinal fluid was replaced in this embr>'o with a 1 per cent solution of potassium fenocyanide and iron-ammonium citrate. The embryo was kept alive for .50 minutes and was then fixed over night in 10 per cent formol containing 1 [ler cent hydrochloric acid. The granules of pnissian-blue are not shown in this reproduction. Enlargement, 13 diameters.


By R. S. Cunningham.


'''Fig. 23.''' Drawing of squared area in fig. 22. The area membranacea superior (rima) is shown, sunounded on either side by tufts of the chorioid plexus (rhp) .ind the typical ventricular ependyma. The transit of the solution is shown, as represented by the resultant granules, through the areji, with the subsequent spreail into the periaxial mesenchyme. Enlargement, 125 diameters.




===Plate VIII.===
[[File:Weed1917 plate08.jpg|800px]]


'''Fig. 24.''' Photomicrograph of a transverse section of a pig embryo of 8 mm. Fixed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent grades of alcohol, embedded in xylol-parafiin. Enlargement, 30 diameters.


From an analysis of the literature on the development of the lymphatic sj'stem, it is clear that there is a general agreement among recent workers that the mammalian lymph-sacs precede the lymph-vessels in the time of their appearance, and hence constitute what may be called a primary lymphatic system. This system consists, in mammals, of 8 sacs: 3 paired, the jugular, the subclavian, and the posterior iliac lym]ih-sacs; and 2 unpaired, the retroperitoneal sac and the cysterna chyli.


'''Fig. 25.''' Photomicrograph, retouched, of the blocked area in fig. 24. The character of the cells (epc) composing the roof of the fourth ventricle Uve) is showTi in this reproduction. Enlargement, 16.5 diameters.


The further development of the lymi^hatic system — that is, the formation of the thoracic ducts and the peripheral vessels — has been discussed at length by numerous workers during the paot decade. These workers have been grouped into two general schools: the one holding that the lymphatics grow by a centrifugal sprouting of pre-existing endothelium, the other belicA-ing that these vessels are formed by a coalescence of numerous isolated spaces developing in the mesenchj-me.


'''Fig. 26.''' Photomicrograph of a sagittal section from a pig embryo of 11 mm. Fi.xed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent gracles of alcohol, embedded in xylol-parafhn. Enlargement, 11 diameters.


According to the centrifugal theory, briefly stated, the sacs arise from the veins and are johied together bj' vessels that sprout out from their endothelial walls. Thus the thoracic duct arises from both the retroperitoneal sac and the left jugular sac, and the two elements unite somewhere between the two points of origm. Supporters of the centrifugal theory claim that the secondary lymphatic system (the capillary bed) arises by the sprouting of the endothelial walls of the sacs and of the right and left thoracic ducts. These sprouts invade the organs and, becoming progressively more complex, assume the adult form of the lymphatic system. The supporters of the raultiple-anlagen theories (whether they beUeve in coalescing tissuespaces, multiple venous origins, or degenerating veno-lympha tics) agree in claiming that lymphatics do not grow by the centrifugal sprouting of the pre-existing endothelial walls.


'''Fig. 27.''' Photomicrograph of the blocked area in fig. 26. The area membranacea superior (ams) in the roof of the fourth ventricle is shown sharply delimited from the two processes of typical ependyma {epe). Enlargement, 67 diameters.


It is not my intention to review here all the various theories that have been advanced, but only to call attention to the two general ^-iews, in order to correlate my findings with them. A very thorough discussion of these two views, as well as a comprehensive review of the literature, may be found in the Ergebnisse der Anatomie und Entwickelungsgeschichte, 1913. (Dr. F. R. Sabin. Der Ursprung und die Entwickelung des Lymphgef assy stems.)
Though primarily concerned with the problems of origin and the method of growth of the lymphatic vessels, the supporters of both theories have aided in establishing the morphology of the primary system and have laid the foundation for the further study of the development of the system as a whole. If the centrifugal


'''Fig. 28.''' Photomicrograph of a more lateral section of the pig embryo of 11 mm. given in fig. 26. Enlargement, 11 diameters.




4S DKVEI.OrMENT OK TlIK LYMI'IIATICS or TMH UNOS IN PHK F.MBHYO Pin.
'''Fig. 29.''' Photomicrograph, under higher magnification, of the blocked area in fig. 28. The lateral border of the area membranacea superior {ams) of the roof of the fourth ventricle is given. Enlargement, 50 diameters.




theory is correct, it is clear that it should be possible to follow the growth of lyml^hatics from the sacs into anj- organ or group of organs. It should also be possible to demonstrate in progressively older stages constantly increasing lymphatic zones and decreasing non-l}-mphatic zones. The development of the lymphatics of the skin, of the intestine, and of the lung has now l^een studied in this manner.
'''Fig. 30.''' Photomicrograph of a sagittal section from a pig embryo of 13 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 8 diameters.




In 1904, Dr. F. R. ^abin demonstrated that the skin received its lymphatic supply from the two jugular sacs and the two iliac sacs. From each of these sacs a group of radiating vessels invade the skin and form there a close-meshed plexus. These four jilexuses gradually increase in size and finally unite, so that the entire skin is supi)lied with lym])hatics. The differentiation which takes place varies with the location and deix>nds upon the adaptation which the vessels must make to the other structures. Continuing the work of Baetjer (1908) on the retroperitoneal sac, Heuer (1909) studied the development of the intestinal lymphatics by the injection of this sac. He observed and described progressive changes in the intestinal supply, finding more complex injections possible in each older stage. He interIjreted these results to mean that the lym]ihatics had not extended beyond the point which his injections reached and that the n^gion beyond this point constituted a non-lymphatic zone.
'''Fig. 31.''' Photomicrograph, under higher magnification, of the squared area in fig. 30. The reproduction comprises a sagittal section of the area membranacea superior (ams) of the roof of the fourth ventricle. Enlargement, 67 diameters.




There is, therefore, a i)riniary and a secondary lymphatic system. The former consists of a series of sacs formed from the veins and connected by the right and left thoracic ducts. The secondary system consists of the peripheral vessels, which are held by some to be outgrowths from the sacs and by others to be formed in situ. With regard to the development of these perijiheral vessels, only those of the skin and the intestine had been studied. There was need, therefore, for the study of the other abdominal and the thoracic lymphatics. This work was begun to establish a. clearer concei)tion of the development of the secondary system.
'''Fig. 32.''' Photomicrograph of a sagittal section of a pig embryo of 14 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 11 diameters.




In presenting this study, I do not claim to have found any new e^'idence as to the mode of growth of lymjihatics. This work supports the centrifugal theory in the same manner as does that of Heuer (1909) ; and it is certain that the theor.y is sufficiently well established to .serve as a basis for this work. It is the object of the i)resent ])aper to follow the gross morphological changes in 1h(> develoi)ment f)f the Ij'mijhatic vessels of the lung from the primary' stage to the adult form. It is desired to indicate the general lines of growth and the various stages which the system passes through in the course of its development. No attempt has been made to study the finer structure of the vessels or the mode of growth.
'''Fig. 33.''' Photomicrograph of the blocked area in fig. 32 under higher magnification. The area membranacea superior (ami) in the roof of the fourth ventricle is reproduced. Enlargement, 75 diameters.




It is im]iortant to note that complete injections are very difficult to make, and that it is also difficult to be certain whether the injection in a i)articular specimen is comjjlete or not. Therefore it is not claimed that any of the injections are complete; and the limits of the lymphatic and the non-lymi)hatic zones ai any stage are defined in a general uiaiincr, depending on the comparison of a lunnber of specimens.
===Plate IX.===
[[File:Weed1917 plate09.jpg|800px]]


'''Fig. 34.''' Photomicrograph of a transverse section of a pig embryo  of 18 mm. Fixed in Camoy's fluid (6:3:1), deh3'drated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffin. Enlargement, 13 diameters.


The lymphatic supply of the lungs develops from three sources: the tlioracic duct, the right thoracic duct, and the ce])halad portion of the relro])(Mit()iieal sac. In 19Ki, Sabin remarked: "The right lymphatic duct curves ventralward and grows to the heart and lungs." This is llic only statement whicli I \va\v Ix'cn iibic to find


'''Fig. 35.''' Photomicrograph, under higher magnification, of the blocked area in fig. 34. The area membranacea superior (atns) is here given, flanked on either side by typical ependyma (ept). Enlargement, 170 diameters.




devf:l()pment of the lymphatics of the lungs in the embryo pig. 49
'''Fig. 36.''' Photomicrograph of a transverse section of a pig embrj-o of 18 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 13 diameters.
in the literature regarding the development of the cardiac and i)ulmonary lymphatics from the right duct, or the morjihological fate of the right duct in mammals. In the same report attention ^vas called to the fact that the lung- vessels can be injected from the retroperitoneal sac, but this was not studied further at that time. The right duct grows primarily to the heart, just as the left grows to the aorta, this asjTTimetry depending upon that of the cardio-vascular system, according to the general rule that the i)rincipal lymphatic trunks follow the large blood-vessels, and grow with the greatest rapidity where the blood-sujjply is most abundant.




In the beginning I wish to lay emphasis upon the fact that the lung lymphatics develop partly from the retroperitoneal sac. and to call attention to the fact that these vessels persist in the adult as part of the permanent drainage of the lung, and hence may be of importance in pathology. On accoimt of the complexity of the development of the lung lymphatics, it has seemed best to present this work, not by describing and figuring a series of jirogressively more complex specimens, but bj' describing the development as a consecutive growth and illustrating with those preparations that may seem best to clarify the text. However, as a matter of reference, the following table of periods has been arranged, in order to offer a brief outline of the complexity at varying stages. These stages are selected with regard to the more important principles of growth and are as follows:
'''Fig. 37.''' Photomicrograph of rectangular area outlined in fig. 36. The extent of the area membranacea superior (ams), with its adherent coagulum of albuminous material, is well differentiated from the adjacent typical ventricular ependyma (epe). Enlargement, 100 di.imeters.
(1) The (lowngrowth of the two ducts, completion of the primitive system, and the first vessels to the trachea and lungs. Embryos 2.3 to 3.5 cm.




(2) The migration of the heart ; the coalescence of the cardiac drainage with that of the lungs, by the formation of the tracheal plexus and the plexus on the arch of the aorta; the growth of the vessels in the lung from the earliest sprouts along the bronchi to the primitive pleural plexus, and the earljmarking-off of the connective-tissue septa; and the growth up from the retroperitoneal sac through the liganientum latum and the anastomosis in the primitive septa into which the vessels grow. Embrj-os 3. .5 to 4.5 cm.
'''Fig. 38.''' Photomicrograph of a transverse section of a pig embryo of 19 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 13 diameters.




(3) The completion of the primary lymphatic sj'stem; that is, when the entire organ is suppUed, and the further development is in an increasing complexitj' of the plexuses already present, incident to the increase in the .size of the organ and its assumption of mature activities. During this period the formation of the valves and nodes begins. Embrj-os 4.5 to 7 cm.
'''Fig. 39.''' Photomicrograph, under higher power, of the rectangular area in fig. 38. A small break in the integrity of the lining ependyma of the roof of the fourth ventricle, representing the irregular boundary of the area membranacea superior (ams), is given. Enlargement, 290 diameters.




(4) The remainder of the development is considered a period, as it is, in reality, an adaptation of the system already present to the increasing needs of the organ. This includes the differentiation of the drainage-lines and the final development of the nodes.
'''Fig. 40.''' Photomicrograph of a transverse section of a human embryo of 4 mm. (No. {{CE836}} of collection of Carnegie Institution of Washington). Enlargement, 33 diameters.




In describing the development of the lymphatics of the lung, the growth of the left duct dowTi to the aorta, of the right duct to the heart, and the formation of the primitive tracheal plexus and the early vessels to the lungs from both ducts vriU be considered first; the further development of the tracheal plexus, together with the changes incident to the descent of the heart, will follow; then the origin of the vessels from the retroperitoneal sac and their growth up through the liganientum pulmonale into the lungs will be considered. After the anastomoses of the two sets of lymphatics, the lung will be considered as a whole, inasmuch as the further development is symmetrical for the entire organ, with the exception of the final lines of drainage and the development of the nodes.
''Fig. 41.'' Photomicrograph, retouched, of the blocked area in fig. 40. The epithelial-like cells (epe) composing the roof of the fourth ventricle (4^) are here shown separated from the denser nervous tissue. Enlargement , 100 diameters.


===Plate X.===
[[File:Weed1917 plate10.jpg|800px]]


I wish to express here my indebtedness to Professor F. R. Sal)in for her constant advice and assistance throughout this work. Also I wish to thank ^Ir. James F. Didusch and Miss Flora SchaefTer for the illustrations.
'''Fig. 42.''' Photomicrograph of transverse section of pig embryo of 19 mm. Fixed over night in modified Bouin's solution, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in x>-lol-paraffin. Enlargement, 13 diameters.




'''Fig. 43.''' Photomicrograph of squared area in figure 42, under higher magnification. The area membranacea superior (ams) with the attached coagulum of albumen is reproduced. Enlargement, 115 diameters.




'''Fig. 44.''' Photomicrograph of sagittal section of pig embryo of 23 mm. Fuced in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 5 diameters.


50 DEVliLOPMlCNT OV TIIL: LYMPHATICS OK THE LUNGS IX THE EMBRYO PIG.


'''Fig. 45.''' Photomicrograph, under higher magnification, of squared area in fig. 44. The area membranacea superior (atns) is here shown, delimited by the cells of the chorioid plexus (chp) on one side and by the further ependymal prolongation (epc) of the cerebellar lip. Enlargement, 88 diameters.


METHODS.


'''Fig. 46.''' Photomicrograph of sagittal section of pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Certain portions of the dura mater (dmc) are indicated. Enlargement, 5 diameters.


The injection method ha.*; lieen iirincipally used, but it has been supplemented and supported by eAidence from l)oth single and serial sections. The collection of pig embryo of the Anatomical Laboratory has been at my disposal, and I have also studied a number of especially prepared series. Many of the series have been of embryos in which the blood-vessels have been injected, and this has materially aided in their interpretation; in fact, in all the especially prepared series the blood-vessels were injected. All these embryos were fixed in Carnoy's fixing fluid, consisting of G parts of absolute alcohol, 3 parts of chloroform, and 1 part of glacial acetic acid.


'''Fig. 47.''' Photomicrograph of blocked area in fig. 46, under higher magnification. The small remaining area membranacea superior (ams) is quite surrounded by encroaching ependyma in the chorioidal folds. Enlargement, 88 diametere.


The method of fixation is as follows: Place the embryo immediately in the fluid and allow it to remain there 6 to 8 hours; then transfer directly to 70 per cent alcohol; dehydrate by ascending grades of alcohol with 2 per cent difference until 95 jjer cent is reached; then change to absolute. This gives excellent fixation w'ith very little shrinkage. The stains used were Ehrlich's hematoxylin and a mixture of eosin, aurantia, and orange (1.


'''Fig. 48.''' Photomicrograph of transverse section of human embryo of 7 nun. (No. {{CE617}} of the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.


The injection masses used were india ink, a saturated solution of prussian blue, a 5 per cent aqueous solution of silver nitrate, and an aqueous suspension of lampblack. The india ink and prussian blue give about the same results, except that the si)ecimens injected with prussian blue are more easily studied after clearing, as the ink renders them more opaque. The india ink, however, flows more easily and hence the injections are more nearly complete. The silver-nitrate injections are easiest to analyze and give beautiful preparations, but its caustic action prevents the finer vessels from filling, so that only the larger trunks are injected; however, it furnishes an extremely valuable method of following the principal drainage-Unes at different stages. The lampblack is the mass which gives the most nearly complete injections, but unfortvmately it precipitates in fine flakes and gives a feathery appearance to the specimen, thus rendering it difficult to use for illustrating.
'''Fig. 49.''' Photomicrograph of squared area in fig. 48, under higher magnification. The epithelial-like cells (epc) composing the roof of the fourth ventricle at this stage are well shown. Enlargement, 100 diameters.




It will be necessary to review the methods used in injecting the various stages, as they differ considerably and are of especial importance in interpreting the results. The earliest injections were made by filling the jugular sacs from the superficial plexuses and then gently moving the embryo. I have succeeded in injecting the early vessels to the trachea and the lungs in only a few pigs less than 3 cm. long, because the injection mass usually follows the path of least resistance, wliich is into the jugular vein.
'''Fig. 50.''' Photomicrograph of transverse section of human embryo of 7 mm. (No. 617 in the Caniegie Institution of Washington). Enlargement, 10 diameters. Fig.




In injecting embryos between 3 and (> cm. in length, three general nietluKls have been employed:
'''Fig.  51.''' Photomicrograph of blocked area in fig. 50. The marked invagination of the roof of the fourth ventricle (4if ) with the lining of epithelial-like cells (epc) is given. Enlargement, 33 diameters.  
(1) The best and by far the easiest method of obtaining good preparations of the left part of the tracheal i)lexus is to inject through the retrojieritoneal sac in the manner described by Ileuer (1909); but this seldom gives good prei)arations of any of the ve.s.sels of the lung except those of the lower lobe. However, this method has been of particular importance in following the lymjihatics uj) from the r('tro])eritoneal sac to the posterior poles of the lower lobes.


'''Fig. 52.''' Photomicrograph of transverse section of human embryo of 9 mm. (No. {{CE721}} in the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.


(2) One may inject the tracheal plexus, especially the left i)ai(, l)y plunging (Inneedle deep behind the aorta and injecting cerebralwards; the right plexus is sometimes filled also, and often the vessels of the left lob(> of the liuig.


'''Fig. 53.''' Photomicrograph of squared area outlined in fig. 52. The pale, large cells {epc) comprising the roof of the fourth ventricle characterize the reproduction. Enlargement, 50 diameters.




'''Fig. 54.''' Photomicrograph of sagittal section of human embryo of 11 mm. (No. {{CE544}} in the collection of the Caniegie Institution of Washington). Enlargement, 6 diameters.




DKVEI.OI'MLNT Ol' TilK I.VMI'IIATICS OF THK M NOS IN THE E.MHRVr) PIG 51
'''Fig. 55.''' Photomicrograph of blocked area in fig. 54. The apparent break in the continuity of the roof of the fourth ventricle with exudation of the ventricular albumen into the mesenchyme is brought out. Enlargement, .50 diameters.
(3) Finally, the vessels of the lung are best injected by a puncture just ventral to the trachea (the tracheal plexus) and behind the arch of the aorta. Here the tracheal plexus is always extravasated, but the lung-vessels fill up nicely.




The embryos older than these mentioned, that is, longer than 7 cm. (or after the valves are formed), are much more difficult to inject, and this difficulty increases with further development. The method employed has been to inject directly into the connective-ti.ssue sei)ta of the lung and to continue the injection .slowly until there is some extravasation at the point of puncture, when a part of the lung surrounding the area of extravasation is well injected. This method has been very satisfactory in all specimens that were obtained very soon after the removal of the uterus; most of the injections were made while the heart was still beating.
===Plate XI.===
[[File:Weed1917 plate11.jpg|800px]]


'''Fig. 56.''' Photomicrograph of sagittal section of human embryo of 14 mm. measured on the slide (No. {{CE144}} of the collection of the Carnegie Institution of Washington). Enlargement, 8 diameters.


In order to study the relations l)etween the l)lood-vessels, bronchi, and lymphatics, multiple injections had to be made, ^'arious combinations were employed. In some, the lymphatics were injected together with veins and arteries; m others with either veins or arteries alone. Again, the lymphatics and the bronchi were injected; and in still others the lymphatics were combined with either veins or arteries. In these multiple injections prussian blue, India ink, and carmine were used, the lymphatics being injected with either the blue or the ink.


'''Fig. 57.''' Photomicrograph, under higher magnification, of blocked area in fig. 50. The greater part of the ventricular wall shown is composed of the area membranacea superior (ams), bounded below by typical ventricular ependyma {epe). Enlargement, 67 diameters.


The specimens in which three systems were injected were difficult to clear, unless only the large bronchi and blood-vessels were filled.
'''Fig. 58.''' Photomicrograph of sagittal section of human embryo of 17 ram. (No. {{CE576}} of the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.  


'''Fig. 59.''' Photomicrograph of rectangular area in fig. 58, showing the area membranacea superior {ams) of the roof of the fourth ventricle. Enlargement, 50 diametere.


In order to trace the vessels more accurately, many of the injected lungs were embedded in paraffin and cut in tliick serial sections (100 to 500 m); these were mounted in balsam but not stained. Other lungs were cut at 10 to 20 ai and stained similarly to the series already referred to.


'''Fig. 60.''' Photomicrograph of sagittal section of human embryo of 17 mm. (No. {{CE576}} of the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.


All measurements of embryos refer to crown-rump diameter and were taken before fixation, as is customary in this laboratory. The illustrations are labeled "C. R. — "; this refers to the crown-rump measurement.
'''Fig. 61.''' Photomicrograph of tlie blocked area in fig. 60 under higher magnification. The aggregation of epithehal-like cells (epc) on the lateral border of the area membranacea superior is here portrayed. Enlargement, 67 diameters.




In 1900, Flint published his study on the development of the lungs in the pig, and his work has been taken as a basis of the general structure of the lungs, esijecially with reference to the development of the bronchi and blood-vessels. He reviewed all the important literature on the embryology of the mammalian lung, studied the lymphatics in sections, and brief!}' summarized their structure and distribution at various stages, but he did not attempt to inject them. I have been able to confirm most of his observations. However, he labored under the difficulty of having neither reconstructions nor injections. He gives a short summary of each stage, and of these summaries I quote the more important parts :
'''Fig. 62.''' Photomicrograph of transverse section of human embryo of 18 mm. (No. 409 of the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.
Stage 3 cm.: At the root of the lung a few dilated lymphatics may be noted near the bronchi and pulmonary vessels; however, they have not grown bej'ond this point into the substance of the
lung wings.




Stage 5 cm.: From the root of the lung the lymphatics have gone some distance into its substance. The}' have thin walls composed of young fibrils Hncd with endothelium with occasional valves. They are confined, however, to the immediate neighborhood of the main bronchi and their chief subdivisions.
'''Fig. 63.''' Photomicrograph, under liigher power, of squared field in fig. 62. The peculiar inversion of the roof of the fourth ventricle (/fVe) indicated in fig. 62, has resulted in a marked dislocation of the area membranacea superior (nm«), shown in this figure. Enlargement, 75 diameters.




Stage 7 cm.: The most interesting change, however, hes in (he further growth of the lymphatics, which in the earlier stages are found in the root of the lung in the neighborhood of the pulmonary ve-s-sels and the large bronchi. .As they grow in. they accompany these structures for a distance;
===Plate XII.===
[[File:Weed1917 plate12.jpg|800px]]


'''Fig. 64.''' Photomicrograph, retouched, of a transverse section of a human embryo of 21 mm. (No. {{CE460}} of the collection of the Carnegie Institution of Washington). The field taken consists of a portion of the fourth ventricle with the lining of typical ependyma (cjk) on either side. The area membranacea superior {ams) is shown between the two lips of ependyma. Enlargement, 33 diameters.




52 ni:vKi.0PMKNT of tiik i.vmi'iiatics of the hn"c:s in the kmbijvo ric.
'''Fig. 65.''' Photomicrograph, retouched, of a similar section to that given in fig. 64, but taken from a more anterior plane from the same embryo. The field shown is analogous in everj' way to that in the preceding figure.




thou apiJioaching the end branches tln-y leave them and run in a plexifonu manner midway between the broneliial tubes until tliey reach the j^leura. This gives the kiiiK now an indefinitely lobulatcd appearance in which the periphery of the simple lobule is indicated by the lynii)hatic vessels and the center by the bronchi. Tiie lymphatics are lined with flattened endothelium; their walls are formed by the young connective-tissue fibrils, and here and there valves are beautifully shown which, in general, point away from the pleura.
'''Fig. 66.''' Photomicrograph of a transverse section of an embryo chick of 121 hours' incubation. Fixed in Bouin's solution. Enlargement, 15 diameters.




Stage IS an.: The lymphatics, forming a plexus around the bronchial veins and arteries at the root of the lung, accompany them towards the periphery, giving off branches to the interlobular spaces en route. * * * On reaching the periphery of the lung they leave these structures and pass out, as in the preceding stages, to the pleura. They have a plexiform arrangement and may be traced at times into the substance of the lobules. This course may be observed in the deeper lobules of the lung as well as in those on the surface under the pleura.
'''Fig. 67.''' Itetouched photomicrograph, under higher magnification, of the blocked area in fig. 66. The area memhranacoH superior (ams) is here given, delimited sharply from the lips of ependyma {cpc) which line the rtmf of the fourth ventricle. Enlargement, 133 diameters.




Stage 19 cm.: In general the relations of the lymphatic sy.stein have not changed.
'''Fig. 68.''' Photomicrograph of a more caudal section from the same embryo as portrayed in fig. 66. Enlargement, 15 diameters.




Stage 23 an.: At 23 cm. the first evidence of the submucous lymjihatic system is seen in the stem bronchi. It may, however, be found earlier, but the ves.sels are difficult to follow. It would seem thus that we have in the jiig's lung, liesidcs the lymphatic plexu.ses accompanying the bronchi, arteries, and veins, an interlobular system which IMiller has been unable to find in the human lung. Injections pointing to such a relationship he has interpreted as artefacts. If Miller's conclusions prove correct, then the Ij-mphatics of the human lung must develop, so far as the interlobular system is concerned, in some other way.
'''Fig. 69.''' Retouched photomicrograph, under higher magnification, of the blocked area in fig. 68. The area membranacea superior {am») is shown at the point of its greatest transverse diameter. Enlargement, 88 diameters.




I quote at length from Flint because he alone, of the many workers on lung lymphatics, has api)roached the subject from the embryological side. As I have said. Flint was seriou.sly handicaj-)ped by having only sections from which to draw his conclusions. He was especially struck by the prominence of the vessels lying in the interlobular septa, and attempted to explain their apparent change of course (j. €., from the bronchi to the septa) hy the theory that the density of the tissue was greater around the bronchi and vessels and that the IjTiiphatics chose the path of least resistance. He did not call attention to the relation of the veins to this point in the development of the lymphatics, which will be discussed later, but emphasized the fact, .so amjily .^liown by injections, that these interlobular vessels grow much more rapidly than the vessels around the bronchi and arteries.
'''Fig. 70.''' Photomicrograph of a migittul section of a pig embryo of 15 mm. Fixed in modlfied Bouin's solution, dehydratiil by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement , S diameters.




It will be necessarj'^ hereafter to discuss the work of IMiller on the adult lymphatic system, in connection with the later stages; therefore it will suffice to refer here to the statement which Flint discussed in the quotation given above. Miller has called attention to the fact that the terminal vein lies in the jieriphery of the lol)ule and that the lym])hatics accompanj'ing the vehi communicate with those of the pleura. He cites Councilman's (1900) description of the interlobular vessels, but does not claim to have found the same vessels. I think that these different views will be reconcilable when w^e have followed the development of the lymphatics through the various stages that lead to the adult form. The literature on the lymi)hatics of the adult mammalian lung is very large, and for a comprehensive review of it the reader is referred to the pai)ers of :\Iiller (1893, 1S9(), 1900, 1902, 1911). It seems needless to di.scuss it more at length here.
'''Fig. 71.''' Phutomicrograph, under higher magnification, of blocked area in fig. 70. The earliest evidence of the area membranacea inferior (ami) in the roof of the fourth ventricle is here shown. Enlargement, 12.'i diametere.




THE VESSELS ARISING FROM TI IE LEFT DUCT. As has been said, the lymphatics of the lungs arise partly from tlic iwo thoracic ducts l)y sprouts. These vessels grow to the mesenchymal wall of the trachea and form there a jjlexus which sends vessels down into the lungs. Other vessels grow directly into the lungs.
'''Fig. 72.''' Photomicrograph of sagittal section of a pig embryo of 18 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 11 diameters.




'''Fig. 73.''' Photomicrograph, under higher power, of the rectangular area outlined in fig. 72. The enlarging area membranucea inferior (ami) is shown in the midst of the typical lining ependyma of the roof. Enlargement, 100 diameters.




'''Fig. 74.''' Photomicrograph of sagittal section of a pig embryo of 23 mm. Fixed in modified Bouin's solution, dehydrated by 2 :\iid 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 6 diameters.


de\i:loi'ment of the eymi'Hatics or the lungs in the embryo pig. 53
The thoracic duct, as has been shown by Sabin (1913), Baetjer (1908), and Kampmeier (1912), is complete — that is, it connects the jugular sac with the retroperitoneal sac — in a pig; embryo 2.5 cm. long. Very soon after this the first evidence of the pulmonarj' supply may be found. I have obtained partial injections at 2.8 cm., and have found some small vessels in serial sections at 2.6 cm.; so it is evident that these sj^routs are either formed from the thoracic duct as it grows down or very soon after the primary system is completed.


'''Fig. 75.''' Photomicrograph of blocked area in fig. 74. The area membranacea inferior {ami) is, at this stage, quite extensive, as shown in the reproduction; the early stages in the development of the cistema cerebellomedullaria may also be seen. Enlargement, 75 diameters.


About midwaj^ between the jugular anastomosis and the arch of the aorta the thoracic duct leaves its position lateral to the trachea and bends dorsalward to lie near the dorso-lateral border of the esophagus. In this position it comes down behind the arch of the aorta. This transition is shown by Sabin (1913, figures 12 and 13). Just at the point where the duct begins to bend dorsally the earliest sprout to the lung is formed. At this point a single large vessel buds off from the thoracic duct and passes down over the arch of the aorta to reach the hilum of the lung. This vessel unites with the vessels that grow up from the thoracic duct just caudal to the arch of the aorta and forms the lower part of the tracheal plexus. This vessel usually persists in the adult as one of the drainage trunks from the hilac nodes to the thoracic duct. It is shown m figiu'e 5, plate 1, and figure 2, plate 4, marked with an asterisk. From the region of the thoracic duct, where this vessel buds off to a pomt about the level of the aortic arch, a number of other ^'essels are formed xery soon afterw'ards. These vessels arise very close together and grow across to the lateral wall of the trachea, where they anastomose and form the primitive left tracheal plexus; thej' lie in the undifferentiated mesenchymal tissue that surrounds the tracheal lumen. These lymphatics have formed a plexus by the time the embryo has reached a length of 3 cm. From this plexus vessels grow across the trachea to anastomose with other vessels from the similar plexus on the opposite side; other IjTnphatics grow up the trachea and form a coarse-meshed plexus around it. This is the anlage of the adult supply of that structure. But the most important of the branches of this plexus, as far as the present work is concerned, are those from the lower part. These pass down the trachea and, being joined by other vessels that leave the duct near the arch, ixiss up over the bifurcation and into the lung. The left tracheal plexus is shown in figure 5, plate 1, and figiu-cs 1 and 3, plate 2. Here must be noted the fact that the plexus of the left side supplies the greater jwrtion of the ventral surface of the trachea and forms the largest part of the great sheet of lymphatics around the primary bronchi. Later these vessels anastomose freely with those from the right side. It is important to call especial attention to the difference in the richness of the supply of the dorsal and the ventral surfaces of the trachea. There are vessels that grow to each from the left plexus, but a much greater number pass to the ventral surface than to the dorsal. Thus the plexus formed from the two lateral groups is much more closely meshed on the ventral surface, and from it is derived the greater part of the lung supply. Over the bifurcation there is a very complex group of vessels, and these form tubes around the ])rincipal bronchi as they grow on into the lung.


'''Fig. 76.''' Photomicrograph of sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 7 diameters.


Below the level of the hilum several vessels, three or foiu- in number, grow up from the thoracic duct and its plexus surrounding the aorta, to join with the large


'''Fig. 77.''' Photomicrograph, under higher magnification, of blocked area in fig. 76. The unsupported character of the area menibranarca inferior and the formation of the cistema cerebello-medullaris is here reproduced. Enlargement, 67 diameters.




.")4 m;\ i;i,()i'Mi-:.\T ok ihk i.vmi'iiaiics oi' tiik i.l'ncs in thi: iimhkmi tig.
===Plate XIV.===


[[File:Weed1917 plate14.jpg|800px]]


vessel which has Ikhmi tlescribod as the first tu the king and wliicli comes over the arch to reach the hihnii. These vessels from the duct below the hilum form a plexus with the vessel from above, as has been described. It is well known that the thoracic duct is double below the level of the arch of the aorta and that the two divisions are connected by numerous anastomostic vessels (figure 1 , i)late 2 ) . This system is the anlage of the vessels that surround the aorta in the adult. This relation has l)een figured by Heuer (1909). One of the lymi)hatics that i:»ass uj) from below to join the first vessel from the thoracic duct above leaves the duct near the diai)hragm and is consequently very conspicuous in injections of this region. Heuer has figured this lymphatic as one that goes to the heart, a conclusion entirely justifiable from the general appearance of the injected specimen. Figure 1, plate 2, is from a di.ssected embryo 4 cm. long, in which the lymphatics were injected from the retroi)eritoneal sac. The thoracic duct and \n\rt of the left tracheal plexus are injected, and the extension of the plexus down on the bronchus is also shown. JJelow the arch may be seen some of the vessels that grow up to meet the branch from above. These vessels have been cut off, with the arch, to expose the tracheal jilexus. The double duct is also shown, the more ventral olemont being the one figured by Heuer.
'''Fig. 78.''' Photomicrograph of a sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xj'lol-parafiin. Enlargement, 7 diametera.




The i)ulmonar\' vessels reach the hilum when the embryo is al^out 2.S cm. long, and can be seen in .sections at 3 cm. (see figure 1, plate 1). The lung-tissue is at this time very slightly differentiated mesenchyme, containing the early bronchi and blood-vessels. For a further description of the structure of the lung at this stage see Flint (1906). These early Ij^mphatics are grouped in an irregular manner in the hilum of the lung and may be found at 2.9 and 3 cm. in sections. But I have not been able to inject them earlier than 3.3 and 3.5 cm. Figure 1, plate 1, is of a .section from an embryo 3 cm. long, in which the blood-vessels were injected while the embryo was still li\ing. The lymphatics are sliow'n as a few dilated spaces (blue) in the hilum. These vessels are beginning their invasion of the lung-tissue while the tracheal i)lexus is forming. It is necessary, however, to complete the description of this plexus before considering the portion of this study which relates to the lung proper. The development of the vessels within the lung-substance will be considered after the formation of the right lymjjhatic j^lexus has been described. It is important, however, to note here that all the vessels to thi' left lung come from the closely united grouj) of vessels on the trachea and around the aortic arch, as has been described. This will be studied in relation to the first vessels to the lung on the right side, which will next be considered.
'''Fig. 79.''' Photomicrograph of the blocked area in fig. 78, under higher magnification. The intact area membranacea inferior (ami), unsupported by any mass of tissue, is shown separating the ventricular cavity from the developing cisterna ccrebello-medullaris. Enlargement, 67 diameters.




On the right .side the development is, in general, similar to that on the left, but difTers in a few particulars, chiefly relating to and in conseciuence of the a.synunetry (»f the va.scular system. The right duct is primarily to the h(>art. or i)erhaps to the vena cava, since it follows that ve.s,sc"l to r(>ach the cardiac base. Hut while the heart supi)ly is at first only from th(> right s'nU\ the vessels to the lung and the trachea develop ut about the same time. The right duct grows caudalwaid i)arallel to the thoracic duct to the iwint where the vena cava arches venlralward to reach the heart. There it divides, and one branch follows the posterior wall of the vena cava to reach the cardiac ba.se, while the (.Hut p;is.scs into the hiluiii of tlu' lung. The
'''Fig. 80.''' Photomicrograph of a sagittal section of a human embryo of 16 mm. (No. {{CE406}} in the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.




'''Fig. 81.''' Photomicrograph of the area outlined in fig. 80, but under liigher magnification. An early stage in the differentiation of the area membranacea inferior (ami) is given. Enlargement, 50 diameters.


DKVEUJP.MKNT Ol THK LV.Ml'llATICS UF TlIK IA"N(iS IN Till-: EMUKV*.) IMG. OO
cardiac division, after reaching the base of the heart, along the posterior wall of the vena cava, passes around the bulbous arteriosus to reach the anterior surface of the heart, where it divides to form the primitive pericardial plexus. By introducing a canula dorsal to the vena cava and injecting towards the heart, I was able to fill this plexus in a p'\g 3 cm. long. At this stage it extends about one-fourth of the distance from the base to the apex of the heart. Figure 13 in Volume \ of the Johns Hopkins Hospital Reports, Monograph Series (Sabin on "The Origin and Development of the Lymphatic System"), shows the right duct near the heart in an embryo pig 2.5 cm. long. In that paper attention was called to the fact that the duct grows towards the heart and that it probably represents the cardiac supply.


'''Fig. 82.''' Photomicrograph of a sagittal section of a human embryo of 17 mm. (No. {{CE576}} in the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.


The second of the two terminal branches of the right duct passes down parallel to the dorsal wall of the trachea in about the same general jwsition as that occujjied by the duct above the point of division. Thus it might seem proper to consider the lung division as the more fundamental of the two, as it appears to be the continuation of the undivided duct. However, the heart branch is probably the more fundamental and the earlier of the two, since it is a general principle in the growth of lymphatic trunks for the principal vessels to follow the larger blood vascular channels. Hence we consider the left duct as primarily aortic and the right as primarih' cardiac in distribution.


'''Fig. 83.''' Photomicrograph, under higher power, of the area blocked in fig. 82. The chorioid plexuses of the fourth ventricle Ue in the central portion of the field; above is the thick cell-layer on the lateral side of the area membranacea suf)erior (ams), while below the upper limit of the area membranacea inferior (ami) appears. Enlargement, 67 diameters.


This vessel enters the hilum of the lung and breaks up into a few- branches that are grouped around the bronchi and blood-vessels as on the left. The nature of the grouping and the further development are similar on the two sides, and hence both will be considered together. There is, however, an interesting difference between the two u])per lobes, which is dependent upon the relation of the aortic arch to the hilum on the left. On the right the lung is distinctly higher (i. e., nearer the neck) than on the left, because on the latter side the aortic arch lies in the groove made at the juncture of the upper lobe with the trachea. Thus the vein to the upper lobe on the left passes close to the bronchus under the aortic arch, while on the right it is well above the bronchus. This allows more freedom in the honphatic growth on the right, so that the vessels to the upper lobe come down directly into it instead of growing back from a single grouj), as they do on the left. It must be understood that the stage referred to is between 2.5 and 3 cm., when the heart is still higher than the bifurcation. Later the heart passes still farther down into the thoracic cavity, and these differences disappear as the cardiac and aortic relations to the lung begin to assume their adult form. There is, however, one verj^ important effect of this asynometry; the lymphatics of the right duct pass directh' into the lung, while tho.se of the left must course up over the arch of the aorta and the bifurcation of the trachea to reach the lung-tissue. This has been mentioned briefly before. It is clear that the principal supply of the bronchi, and therefore, ultimately, of the lungs, comes from the left duct. This is in large measm-e the result of the asymmetric relations of the heart and aorta.


'''Fig. 84.''' Photomicrograph of a transverse section of a human embryo of 18 mm. (No. {{CE409}} in the collection of the Carnegie Institution of Washington). Enlargement, 5 diameters.


The development of the first vessels to the trachea and lungs on the right side will next be described in detail. From the heart Umb of the right duct a few vessels arise and grow down over the vein to the upper lobe on the right side; after crossing the vein thev enter the lung near the hilum and divide into several branches, some


'''Fig. 85.''' Photomicrograph of the blocked area in fig. 84. The cellular character, and especially the clumping of cells, of the area membranacea inferior (ami) is shown. Enlargement, 25 diameters.




■)() ni:vi;u»i'.Mi;NT tM- tmk i.v.\irji.\Tic.s ui- tiik lincs in tiik lmuhvu ric.
'''Fig. 86.''' Photomicrograph of a sagittal section of a human embryo of 19 mm. (No. {{CE431}} in the collection of the Camegie Institution of Washington). Enlargement, 5 diameters.




iif which anastomose with those mentioned above as growing down into the liiluni of the lung from the pulmonary limb of the right duct. Other vessels turn outwards along the bronchi and veins and grow into tlio lung-tissue of the upper lol)e. This process will be described later.
'''Fig. 87.''' Photomicrograph of the blocked area outUned in fig. 86. The area membranacea inferior (ami) appears separating the fourth ventricle from the developing cistema cerebello-medullaris. Enlargement, 25 diameters.




Along the right duct, cephalad to the division into the two Ijranches, other vessels are given off; some grow down to anastomose with ascending liranches lying along the tracheal wall and coming from the vessels described above, while others grow to the tracheal wall at varying positions along the section lying between the jugular anastomosis and the bifurcation, corresponding somewhat to the vessels on the other side, with which their branches anastomose, forming the tracheal supply. The earliest injection of the lymphatics of the right side were at 2.8 and 2.9 cm.
===Plate XV.===
[[File:Weed1917 plate15.jpg|800px]]


'''Fig. 88.''' Photomicrograph from a sagittal section of a human embryo of 17 mm. (No. {{CE576}} of the collection of the Carnegie Institution of Washington), representmg an enlargement of the second blocked area in fig. 58. The area membranacea inferior (ami) appears sharply delimited from the adjoining tj^pical ependjina. Enlargement, 67 diameters.


P'igure 2, plate 3, shows an embryo of 3 cm., where the injection was made into the right sac, which illustrates the relative position of the vessels to the upper right lobe and the limb that follows the vena cava to the heart. This drawling is diagrammatic and does not show the different vessels to the lobes on the right side, though some of them were injected. The left duct is shown without any branches.


'''Fig. 89.''' Photomicrograph of a sagittal section of a human embryo of 23 mm. (No. {{CE453}} of the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.


In figure 1, plate 2, the right tracheal plexus is represented. Though it is verjincomi)let('. it shows the general form of the plexus and its relation to the similar plexus on the other side. The right tracheal plexus, in its simplest form, consists of a few vessels which are beginning to anastomose along the lateral wall. These anastomoses become more and more complex and numerous until, along the right side of the trachea, a jjlexus somewhat similar to that of the other side is formed. They differ, however, in that on the right there is no aortic arch to complicate the form. Therefore tlie plexus is a simple sheet-like group of vessels which lie along the lateral wall of the trachea, but do not extend up over the ventral surface of the bifurcation, except by a few anastomosing vessels. It anastomoses frcelj' with the larger plexus from the other side on the ventral surface of the trachea, and later the combined plexui^es lose their individuality and appear continuous. In the meantime the two tracheal ])lexuses have Ix'gun to anastomose. This will next be described.


'''Fig. 90.''' Photomicrograph of the blocked area in fig. 89. The area membranacea superior (ams) appears in the stage of closure, while the area membranacea inferior (ami) is becoming well differentiated from the typical ependyma lining the other portions of the fourth ventricle. Enlargement, 26 diameters.


Between 3.3 and 4.5 cm. the two tracheal plexuses anastomose by means of numerous vessels which grow around the trachea, both dorsally and ventrally. AI)ove the level of the aortic arch these connecting vessels are far less numerous than below, where the two are merged into a sheet-Ukc plexus that surrounds the trachea and i)as.ses down into the lungs as tubes of vessels surrounding the bronchi. Above the bifurcation the dorsal surface of the trachea has fewer vessels than the ventral, while the two original lateral plexuses are much more clo.sely meshed, representing the anlagen of the two lateral grou|)s of lymph nodes of the adult.


'''Fig. 91.''' Photomicrograph of a sagittal section of a human embryo of 26 mm. (No. {{CE1008}} of the collection of the Carnegie Institution of Washington). Enlargement, 4.5 diameters.


From the clo.se-meshed plexus on the left side of the trachea just at the bifurcation a group of lymjihatics pass up over the left stem bronchus and sweej) across to the right bronchus, forming the upper grouj) of vessels lying on the bronchial wall. These grow down on the side and anastomo.se with the vessels coming down from the j)lexus on the right side. Thus it will be se(>n that the left supply is a more important part of the gr-neral origin than the right, supplying, as it does, all of the left lung and part of the right.


'''Fig. 92.''' Photomicrograph, under higher magnification, of the blocked area in fig. 91 . The area membranacea superior has been almost completely closed by the dense ependyma of the superior half of the roof of the fourth ventricle, while the inferior area (ami) has become a membrane lacking wholly the character of ependyma. Enlargement, 23 diameters.




'''Fig. 93.''' Photomicrograph of a sagittal section of a human embryo of 35 mm. (No. {{CE199}} of the collection of the Carnegie Institution of Washington). Enlargement, 3 diameters.




DEVELOPMKNT OF THK LVMIMIATICS OF TUF: LUNGS IX THE EMBUVO PIG. 0/
'''Fig. 94.''' Photomicrograph, under higher powers, of the blocked areas in fig. 93. The formation of the cistema cerebello-medullaris is shown in relation to the ventricular roof. Enlargement, 23 diameters.
It is of importance to note here that the heart is migrating downwards (i. e., caudalwards) during this period, and, by the time the embryo




has reached 4.5 cm. in length it has come to lie almost directly over the hilum of the lung. Hence the vessels that formerly ran in a long course from their point of origin in the heart limb of the right thoracic duct to reach the ui)per lobe and the hilum of the lung have become a part of the common tracheal i)lexus, and the formerly distinct duct to the heart has also been absorbed by the plexus over the bifurcation.
'''Fig. 95.''' Drawing of cells of the chorioid plexus from the lateral ventricles of a fetal pig of 132 mm. The specimen waa fixed in absolute alcohol, and stained by Best's carmine stain for glycogen. The glycogen occurs in the form of globules within the epithelial cells. Enlargement, 950 diameters.




The cardiac vessels then (at 4.5 cm.) drain directly into the plexus over the hilum of the lung (figures 1 and 3, plate 2). This relation remains in the adult in the drainage of the cardiac vessels into the mediastinal nodes and the union of the efferent trunks of these nodes with those from the hilum of the lungs.
''Fig. 96.'' Drawings of the cells of the chorioid plexus from the lateral ventricles of a fetal pig of 36 mm. The specimen was fixed in absolute alcohol and stained by Best's carmine method. The glycogen appears in the epitheUal eclls in the form of basilar plaques. Enlargement, 950 diameters.




Here must be mentioned, though not bearing ixirticularly on the lymphatics of the lungs, the connection between the right and the left ducts. In specimens of about 3.5 to 4 cm. in length, I have regularlj^ found a vessel arising from the dorsal part of the right tracheal plexus and joining the thoracic duct behind the aorta. As has been said, it seems best to consider the vessel to the heart as the continuation of the right thoracic duct; hence this vessel must be considered, as was the one to the lung, as a i)art of the collateral supply.
===Plate XVI.===
[[File:Weed1917 plate16.jpg|800px]]


'''Fig. 97.''' Photomicrograph of a transverse section of a pig embryo of 10 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent f;radcs of alcohol, and embedded in xylol-paraffin. Enlargement, 10 diameters.


The lung, as has been stated also derives lymphatics from another source — the cej^halad portion of the retroperitoneal sac. These vessels are growing into the lung during the period when those already described arc differentiating, but it seems best to i:)osti>one the discu.s.sion of this portion of the pulmonic supply until we have studied the early changes that take i)lace in the lung itself, following the invasion by the vessels already described. The desirabihty of this is evident when it is remembered that the vessels from below must follow a similar course in the lung, with the exception that this course is reversed, due to the fact that these vessels invade the lung through the pleura instead of the hilum, and must reach the other supph' through the interlobular septa, to be described later.


'''Fig. 98.''' Photomicrograph, under higher magnification, of the blocked area in fig. 97. The double condensations of mesenchyme to form pia mater (pmc) and cerebral blastema (cbl) appear separated by a region of mesenchyme which is breaking down. This central area of mesenchyme, with the marked albumen-content, is to become the arachnoid spaces. Enlargement, 133 diameters.


At 3 cm. there are two primary bronchi and two veins on either side, one of each to each upper lobe and one to each lower lobe. From these the secondary branches are beginning to form. From 3 cm. to o cm., these secondary branches are developing rapidly and are very large in comparison to the size of the lung. The arteries are very much smaller, and the veins are somewhat larger than the arteries, but much smaller than the bronchi. It is of great importance to note the relations of these structures to each other during this period. Flint has studied their development very thoroughly, but he does not call attention to the fact, so im])ortant with reference to the lymphatics, that the developing vein is separated as widely as possible from the bronchus with which it is morphologically associated. The artery, on the other hand, follows the bronchus very closely and is distributed with it to the center of the developing lobule. The two primary branches of the pulmonarj' vein lie close to the corresponding bronchi. This is, indeed, as far separate as is possible, since there is almc?*^ no lung-tissue at this period, while the secondary vessels which may be considered the terminal branches lie about equidistant from the two adjacent bronchi. The arteries follow the bronchi more closely. This fact is of the greatest importance in the development of the lymphatics and also m the relation of the veins to the periphery of the lobule in the adult, as has been shown bj' Miller (1900).


'''Fig. 99.''' Photomicrograph of a transverse section of a pig embryo of 20 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffm. Enlargement, 10 diameters.




'''Fig. 100.''' Photomicrograph, under higlior powers, of the blocked areas in fig. 99. The relations of the pial condpn.sation ipmc) of mesenchyme to tlie nervous system, as well as the infiltration of tlie arachnoid mesenchyme (sas) with albumen, is reproduced. Enlargement, 133 diameters.




58 DliVLLOPMIiNT OF THE LYMPHATICS OF THK LIXCIS IN TIIK KMUKYO I'K!.
'''Fig. 101.''' Photomicrograph, under higher magnification, of the blocked area in fig. 22. The reproduction is included here to show the double condensjition {cht) of mesenchyme which goes to form ultimately bone and possibly a portion of the dura. Enlargement, 132 diameters.




As till' lung iiKToascs in size and (Ik' \t'in.s and Ijionclii wliicii \vc have termed secondary give off other branches, these in turn become the terminal ones and assume the relations that have been described. The others are, bj' the increasing amount of lung-tissue, forced closer together. Thus it is seen that it is only the terminal veins that occupy the position described; that is, jiass along the perii)hery of the k)l)ule. In the i^ig there is considerable connective tissue forming definite lobules in the adult lung; and these .septa, bounding as they do the area sui)plied by terminal bronchi, divide the lung into a large number of irregular cones or pyramids, which have the bronchus and artery in the center and the veins passing along the periphery until close to the apex, where thej^ enter veins of the next larger size. For further discussion of this arrangement see Miller's article (1900).
'''Fig. 102.''' Photomicrograph of a transverse section of a pig embryo of 18 mm. The embryo was one in which the cerebro-spinal fluid was replaced by the ferrocyanide solution. Subsequently the embryo was fixed in 10 per cent formol containing 1 per cent hydrochloric acid for a few minutes to precipitate the prussianblue. It was then transferred to modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. The granules of prussian-blue are not represented in this figure. Enlargement, 10 diameters.




As we have seen, a few dilated lymphatics are found in the hilum of the lung at 2.9 and 3 cm. These are the first branches from the vessels that are forming the plexus on the trachei and bronchi aln^ady described. The bronchi, as has been said, are surrounded by lymphatics which follow them into the lung-tissue; and, as secondary bronchi are formed, lymphatics from these plexuses branch off to follow them.
'''Fig. 103.''' Photomicrograph of the squared area in fig. 102. The relation of the thinning mesenchyme in the arachnoid areas to the caudal cranial nerves is shown. The granules of prussian-blue, scattered through the area of thin mesenchyme (sas), are not reproduced. Enlargement, 40 diameters.




The primary veins lie very close to the corresponding l)ronchi at tliis stage, and are accompanied by a few lymphatic trunks which arise from the same general plexus that covers the bifurcation. These vessels anastomose very richly with those of the bronchi, and, close to the point where the trachea divides, they merge together. We have seen that the secondary veins lie midway between the adjacent bronchi, and represent the outer border of the primitive lobule of the developing lung. Along these veins the lymi)hatics grow towards the pleura; thej^ are derived both from the plexus that follows the i)rimary vein and from the vessels that surround the primary bronchi. The lymi)hatics from the bronchial supply join those from the vein, and the combined grou)) passes along the vein, spreading out on either side to form a sheet, until the vessels reach the pleura. Flint observed these sheets of lymphatics, but thought that there must be some difference in the density of the tissues to account for their leaving the bronchi to run midway between. He did not recognize the relation between the veins and the lymphatics. It will be clear, when it is remembered that the smaller branches of one \ein spread out fan-like to meet tho.se of the other vein, that the sheets of lymphatics lying between the bronchi are directed by the veins as well as the separate lymph-ve.s.sels directly associated with them.
'''Fig. 104.''' Photomicrograph of a coronal section of a tissue block which includes the meninges and cerebral cortex in the region of the sinus sagittalis superior. The block was obtained from a fetal pig of 80 mm., fixed in Zenker's fluid, and stained, after embedding in celloidin, by Mallory's technique for coimective tissue. Enlargement, 27 diameters.




In this manner the true primitive lobules are formed by the interpolation of a sheet of rapidly growing lymphatics between the bronchial tubes. It is along the distal margin of the.se jjlexu.ses that the pleural marking begins. When these vessels reach the pleura there is a marking-out of the characteristic coar.sely-meshed i)lexus, each interspace corresixmding to the .sheet beneath (figure 3, plate 1). It must be remembered that these vessels, growing as they do very rapidly, reach the pleura very early, and hence th(! i)leural plexus is developing while the above-mentioned interlobular plexuses are forming. We have so far described only the formation of the largi' i)arallel plexu.ses shown in figure 1, i)late 4, figure 2, plate 5, and figure 1, plate 3. I*)iit ihe formation of \-eiiis in oilier ])lanes directs the growth of the lym
===Plate XVII.===
[[File:Weed1917 plate17.jpg|800px]]


'''Fig. 105.''' Photomicrograph of a coronal section of a tissue block including cerebral cortex and meninges in the region of the sinus sagittalis superior. The block was obtained from a fetal pig of 10 cm., fixed in Zenker's fluid, and stained by Mallory's technique for connective tissue. Enlargement, 13 diameters.


'''Fig. 106.''' Photomicrograph of a coronal section, similar to that in figs. 104 and 105, except in that it was obtained from a fetal pig of 17 cm. The same technical procedures employed in the other specimens were used in this. Enlargement, 27 diameters.


DKVKLOP.MENT OI' TIIK LYMPHATICS OF THK I.INCS IN TUK K.MBRVO I'lr; 59
'''Fig. 107.''' Photomicrograph of a similar section to those of the foregoing figures. The specimen was obtained from a fetal pig of 20 cm. and was treated in the manner outlined above. Enlargement, 20 diameters.  
phatics, so that with each bronchus there are several veins and several sheets of lymphatics developing. Thus the series of cone-shaped or pyramid-shaped lobules are surrounded by plexuses of lymphatics. Along these plexuses the differentiation of the connective-tissue layers takes place, for, when the lymphatics invade these areas, there is only an undifferentiated tissue, which is characteristic of the lung. Flint suggested that the lymj)hatics followed the bronchi for a certain distance and then turned away midway between them, because of some relative difference in the densitj' of the tissues. It is quite impossible to observe the relation to the veins in uninfected sections, and conseciuently this point was not discussed in relation to the problem of the question of tissue density. Notwithstanding this phase of the development which Flint was unable to follow, there still remains considerable probability in his suggestion. The fundamental reason for the direction of growth is as yet entirely a mystery, but there .seems to be little doubt that the principal lines of lymphatic development are along the larger blood-channels; and, in general, the veins are chosen, though the left duct may be considered as following the aorta.




The much slower-growing lymph-vessels on the bronchi follow each branch out towards the periphery. The i)rimary bronchus is surrounded by a very close-me.shed jjlexus, which consists of a large number of vessels; in cross-section one can count from 50 to 75. However, this numl)er is very greatly reduced on the secondary bronchi, each of which has four or five trunks following it. These are closely bound together by anastomosing collaterals.
'''Fig. 108.''' Drawing of the cell pattern from the inner surface of the dura mater of a fetal pig of .5 cm. The specimen was prepared by the reduction of a dilute solution of silver nitrate in sunlight. The preparation was subsequently stained by hematoxylin. Enlargement, 190 diameters.  




With reference to the .secondary bronchi, almost the same series of events occur as given above for the primary ones. These secondary bronchi are likewise marked off by interlobular septa in which the lymphatics develop more rapidly than along the bronchus whose lobule they mark off. The lymi)hatics around the bronchus give off small vessels near each branch of the bronchus, and these pass across to join the plexuses that surround the area of the lobule (figure 1, plate 3). As the new-formed bronchi grow larger they are, in turn, followed by two or three lymphatics, which end, as did those around the secondary' bronchi, by passing over to join the septa or, if close to the pleura, the vessels there. These lymphatics that pass from the bronchial system to join those in the septa follow the branches of the veins which bend in from the septa to reach the capillary bed of the arterial tree. These persist in the adult as the vessels that pass from the bronchus to the vein and thence to the pleura (figure 2, plate 1).
'''Fig. 109.''' Drawing of a preparation, similar to that of fig. 108, but obtained from the inner surface of the dura mater of a fetal pig of 7.5 mm. Enlargement, 28.5 diameters.  




We will consider now the lymphatics that grow up from the retroperitoneal sac into the caudal pole of the lower lobe.
'''Fig. 110.''' Drawing of a preparation, similar to those of figs. 108 and 109, obtained from the inner surface of the dura mater of a fetal pig of 90 mm. Enlargement, 28.5 diameters.  




In 1906 F. T. Lewis described, in rabbit embryos, a lymphatic sac just median to the mesonephritic vein. Baetjer (1908) showed that it arises from the ventral surface of the large vein which connects the two Wolffian bodies (embryos 17 to 23 mm.); Heuer, following Baetjer, found that numerous lymphatic sprouts arise from this sac and invade the intestine through the mesentery. This sac supphes lymph-vessels to the stomach, the liver capsule, the AVolffian bodies, and the reproductive glands.
'''Fig. 111.''' Drawing of a preparation from the inner surface of the dura mater of a fetal pig of 16 cm. The specimen was made in the same manner as outlined in fig. 108. Enlargement, 285 diameters.




The lower pole of the lower lobe of the lung is contmuous with the mesentery in the earlj' stages. As the embryo develops, this comiection becomes a thin band
'''Fig. 112.''' Photomicrograph of a sagittal section of a pig embryo of 17 mm. An injection of an 0.5 per cent solution of  nitrate of silver was made into the central canal of the spinal cord; the silver was reduced in sunlight  and the embrjo fixed in formalin. Enlargement, 13 diameters.


'''Fig. 113.''' Photomicrograph, under higher powers, of the blocked areas in fig. 112. The accumulation of the reduced  silver (p»n) again.st the area membranacea superior is represented in black. Enlargement, 117 diameters.




00 DEVELOPMENT OF THE LVMI'IIA'I'ICS OK THE LLNCS IN THE EMIJKVO I'lG.
'''Fig. 114.''' Photomicrograph of a transverse section of a pig embryo of 19 mm. An injection of 0.5 per cent solution of  silver nitrate was made into the central canal of this embryo and the silver immediately reduced in  sunlight. The embryo waa fixed in formalin, carefully dehydrated, and embedded in xylol-peiraflBn. Enlargement, 10 diameters.  




of tissue that passes down Ix'liiiid the (Uaphragni to end in the tissue around the aorta; it corresponds to the hgamentuni puhnonale in the human. It is through this prolongation of the lower lobe that the lymphatics from the retroperitoneal sac grow uji to reach the lung. These vessels arise from the cephalad portion of the sac and iias.s u]) l>ehind the dorsal w'all of the stomach to enter this long jiosterior or lower pole of the lung (figure 2, ])late 4). There are three or four vessels that grow out from the sac and up into the lung; these are closely associated with those that pass to the diaphragm and, in adult life, join with them just before reaching the nodes into which they drain. They pass upward and divide, on reaching the lung, into two groups, one of which passes up over the diaphragmatic surface and the other over the outer or lateral surface of the lower lobe.
'''Fig. 115.''' Photomicrograph, under higher magnification, of the blocked area in fig. 114. The collection of reduced  silver {psn) against the cells at the inferior end of the area membranacea superior is illustrated. Enlargement, 100 diameters.  




The anlage of the ligamentum pulmonale is connected not only with the lower l)ole of the lung, but also with the median border of the lower lobe. Thus the lymi)hatics grow directly up about one-third of the way to the hilum in this medial extension of the ligament, and from there sweep out in a fork-Uke division which jiroduces the two plexuses on the two borders of the lung (figure 3, plate 5) . I have injected these vessels at 3.4 cm.; but I think that they reach the lung border a little earlier.
'''Fig. 116.''' Photomicrograph of a tranverse section of a pig embryo of 16 mm. The central canal of the spinal cord of this embryo was injected with a 1 per cent ferrocyanide and citrate solution under mild syringe-pressure; the embryo was then fixed in 10 per cent formol containing 1 i>er cent hydrochloric acid. Enlargement,  
10 diameters.  




From the two plexuses described above vessels grow into the lung in exactly the reverse order to that followed by those develojjing from the hilum. They grow in just where thej' wtU meet the veins, and along these form the septal plexuses, exactlj' similar to those described above. These rapidl}'' anastomose with the other lymi)hatics. and, by the time the embryo has reached 4 cm. in length, the entire lung is uniformly supplied.
'''Fig. 117.''' Photoniicrogrnph of the blocked area in fig. 116, under higher magnification. The accumulation of the  
 
precipitated injection fluid against the area membranacea superior is represented in black. A slight
 
extravcntriculai Bprciul of the fluid, which is found in this as in all embryos of this stage, can not be  
It is very pertinent to infjuire why the lymi)hatics that reach the lung from below select the.se points for the; im'asion of the deeper tissue of the lung. However, when it is recalled that the lymphatic vessels which lie in the mesenchymal tissue (the pleural anlaga) are very large in proi:)ortion to the other structures and that the budding ves.sel would be in direct relation to the outgoing veins, it is easily imderstood that exactly the same causes must be acting here as those which direct the growth from above. So here, as above, the position of the veins controls the direction of growth. Of course, the i)lexuses on the two surfaces become more complex as the lung is invaded and follow the same steps as the pleural supplj' in general. As has been .said, there are branches along the pleura, and these anastomose with the other jjleural vessels, so that the supplj' becomes general. The drainage in the early stages — that is, before the formation of the valves— is pr()hal)ly divided; the flow of lym])h might be to the retroperitoneal sac via the vessels that grow up from that structure, or to the thcjracic ducts through the tracheal i)lexu.ses and the vessels accomi)anying the veins and the bronchi.
miule out in the reproduction. Enlargement, 07 diamuters.




We have seen how the lymphatics grow into the lung-tissue and there form two distinct groups, and how one of these rapidly reaches the pleura and there forms the characteristic plexus-pattern marking off the boundaries of the lobules; also how the vessels grow into the i)osterior i)oles of th(> lower lobes and anastomose with the system from al)ove, which follows the veins in the connective-tissue .septa.
{{Footer}}
 
[[Category:Draft]]
 
 
 
 
DEVELOPMENT OF THE I.YMI'HATICS OF THK LlXflS IN THE EMBHYO PIG. 61
Xow, it will be well to review briefly the state of the development of the lung lymphatics at the time that the primary system is complete — that is, in 6 cm. embryos. At 6 cm. the lymphatics around the trachea form a dose-meshed ple.xus near the bifurcation, extending down into the lung around the Ijronchi. Above the bifurcation there are only a few connecting vessels on the ventral and dorsal surfaces of the trachea, but the two plexuses on the lateral surface are very close-meshed. From the left plexus the j^rincipal supply of both lungs is derived, but there are numerous vessels passing down into the right lung from the right plexus, and the two are closely bound together, especially near the bifurcation, where they have fused into one plexus. The vessels surrounding the bronchi follow them towards the periphery, giving off branches to the venous tree at eveiy division of the bronchial tree. Ear-h smaller lironchus derives its lymj^hatic supply from the jilexus that accompanies the i)arent bronchus. These vessels are very difficult to inject.
 
 
Accompanying the primary divisions of the pulmonarj- vein there is another group of vessels that is closely bound, by anastomoses, to the honphatics around the principal bronchi (figure 4, plate 1). Along each of the tributary veins vessels pass to the pleura and spread out in the region that has been described as the septa between the lobules. Each of these dividing sheets anastomose with other sheets and with the pleural vessels. Th*^ vessels derived from the retroperitoneal sac are continouus \\-ith those derived from the two ducts; there can be determined no line of differentiation either within the lung-tissue or on the pleural surface. The posterior pole is connected with the retroperitoneal sac by three or four vessels that pass down in the fold of tissue that precedes the ligamentum inilmonale (figure 3, plate 5). The pleural plexus has begun to form within the gross markings that we have described as corresponding to the connective-tissue septa. These vessels are very superficial and are not connected, at this time, with the deeper vessels.
 
 
The further development is chiefly due to the multiphcation of the lung units and the increase in volume of the intcrbronchial tissue. As new bronchi are formed, new groups of lymphatics bud off from the plexus that accompanied the parent bronchus and follow the new-formed structure towards the periphery. These lymphatics leave the bronchus and pass to the venous group when they reach the region where the air-sacs are developing.
 
 
As the lung-tissue differentiates further and further, the larger veins become more closely associated with the bronchi and only the terminal vessels are peripheral with reference to the lobule. This brings about the relations that are found in the adult, where the principal veins and bronchi are closel}- associated, while the terminal ones have the same relative positions that have been described for the developing structures.
 
 
The arteries in early stages lie very close to the bronchi and arc associated with the plexuses that follow that structure. As these blood-vessels increase in size the bronchial plexus differentiates into two parts, following the arteries and the bronchi. This is accompUshed by the growth of vessels around the arteries, and, as the artery increases in size, tiie two plexuses become entirely distinct, but are still connected bv numerous anastomotic vessels.
 
 
 
 
 
02 DKVKLOPMKNT OF THK LYMPHATICS OV THF: UNCIS IN THE EMBRYO PIG.
 
 
In the meantime, the vessels of the jileurii, which at from o to 6 cm. we have seen hefjinning to form the true jileiiral plexus, continue to proliferate, and thus form a tine-meshed plexus in the ])leura between the bli)ckin<2;-off of the lobules.
 
 
The completion of the primary plexus is shown in figure 3, plate 1. This is the surface of the lung in a pig embryo of 6 cm. with the pleural vessels injected. Each of these uninjected areas re];)resents a primary lobule, and the surrounding lymphatics mark out the connective-tissue plexuses. Figure 1, plate 5, illustrates one of the jirimary lobules, and the close-meshed i)lexus is the true pleural sui)ply. It is still seen to be connected with the deep vess(>ls of the septum.
 
 
Here and there one finds vessels passing from the terminal bronchi to thc^ surface, in the lobule proper, to join with the fine-meshed plexus of the pleura. These pass around the air-cells, but are never found on their walls, and, uniting with the terminal vessels of the end veins, pass to join tho.se in the pleura. These are the vessels described by Hint (1906) as seeming to di]) down into the lobules from the pleura; these, he said, he could follow only a little way into the lobule. This is easily understood from the information gained from injections, for the vessels around the bronchi can not be seen, in uninjected specimens, and consequently those which remain patent in sections seem to end abruptly in the midst of a lobule, whereas they in reality connect with those following the bronchi and terminal veins. The lymjihatics that follow the terminal bronchi leave them just before the atria are reached and cross over to join the lymphatics which follow the veins. The lymphatics which accompany the veins ])ass to the ])leura just where the veins bend to reach the center of the lobule.
 
 
Flint first observed the submucous plexus of the bronchi and trachea in embryos 23 cm. long. It was surprising that injections did not reveal this i)lexus very much earlier. I have not been able to demonstrate any lymphatics in the .submuco.sa before the embryo reached a length of 19 cm. This plexus develops, as do all the secondary i)lexuses, by the outgrowth of vessels from the primary one and their coalescence to form the new group. This process has been carefully studied by Heuer in the formation of the mucosal plexus in the intestine. The submucosal jjlexus is complete just before birth and consists of numerous fine vessels that lie ju.st beneath the bronchial epithelium. From this plexus numerous vessels i)ass down between the cartilaginous rings and join the lymphatic trunks which follow the bronchi, as has been descrilx'd. In those bronchi having no cartilaginous rings there is only the one grouj) of lym])h-vessels to be found, and these have already been described.
 
 
The lymphatics of the adult lung were first described by Olaf Rudbeck in 16.51-1054 ((juoted from Miller, 1900). Since that time numerous workers have .studied these vessels. In 1900 W. S. Miller reviewed the literature very thoroughly, and it will, therefore, be unneces.sarv to repeat that here. Miller studied the lymiihatics in the lungs of adult cats and dogs by injecting them from one of tlu> pleural vessels. He divided the lymphatics into four groups, as follows:
.\. The lynipliatics of the bronchi. C'. The lymphatics of the arteries.
 
 
K. The I\iiii)hati('s of thf \('iiis. D. 'I'lic l\inpliatics of the pleura.
 
 
 
 
 
DEVELOPMENT OF THE LYMPHATICS OF THE LUNGS IN THE EMBRYO PIG. 63
The lymphatics of the bronchi. — Miller describes two sets of lymph-vessels associated with those bronchi which have cartilaginous rings and only one with those which have no rings. In the former the two sets are connected by vessels that pass between the rings and join the trunks situated on the outer side of these structures. These trunks drain the lymi)hatics that accompany the smaller bronchi and empty into the nodes which are situated at the hilum of the lung. While there are several Ij'^mphatics accompanying the larger bronchi, only three are to be found with those nearer the air-sacs. The.se end by leaving the terminal bronchus just before it ends in the atria; one of them passes to the artery, while the other two join the lymphatics of the vein.
 
 
The lymphatics of the veins. — There is a .single group of vessels that extends from the terminal vein to the hilac nodes. Along the larger veins there are several vessels, but the terminal ones are accompanied by only one or two lymphatics. Anastomotic vessels pass from the bronchial lymphatics to join those of the vein at each branching of the bronchial tree. The lymphatics that accompany those veins which go to the pleura join the pleural lymphatics.
 
 
The lymphatics of the arteries. — The lymphatics which accomi)any the arteries are very similar to those of the veins, with th(> exce])tion that none of them pass to the pleura.
 
 
The lymphatics of the pleura. — There is only one plexus in the pleura, and this drains through several large trunks to the nodes at the hilum. There are anastomoses with the lymphatics of the veins, as has been said, but the drainage probabhdoes not pass through these. ]Miller put his canula into a large pleural vessel and injected towards the hihmi. After some time the deep h'mphatics, as well as those of the pleura, were filled. He thought that the injection mass backed up into the deep ves.sels from the nodes at the hiknn, since both the sets of vessels drain into the same nodes.
 
 
]\Iiller does not confirm the findings of Sappey (,1874) and of Councilman (1900) with regard to the interlobular IjTnphatics. Sappy thought that it was wrong to divide the lung lymphatics into superficial and deep groups on account of the rich anastomosis of these vessels. He thought that the lobules were surrounded by lymphatics which formed networks between the adjacent lobules in much the same manner as the blood capillaries do around the air-sacs. Councilman divided the deep lymphatics of the lung into two sets, the bronchial and the interlobular; the latter he interpreted as verj' important in infections.
 
 
While ^Miller does not agree with these observers in regard to the interlobular lymphatics, he does describe anastomoses between the hTuph-ve-ssels of the venous radicles and those of the pleura, and he emphasizes the peripheral location of the veins. It might well be that the vessels which Sappey and Councilman found in the interlobular septa were the hTnphatics of the veins, since they did not have verj' accurate methods for the differentiation of these structures. It becomes more difficult to reconcile Miller's findings with those of FUnt and the results of this study. Both Flint and I have found distmct groups of vessels in the interlobular connective tissue in embryo pigs. These groups of vessels are directed in their growth and location by the position of the veins, but are not limited in their distri
 
 
 
r>4 DEVELOPMENT OF THE I.VMl'lI ATICS OF THE I.TXOS I\ THE EMBRYO PIO.
 
 
bution to (he venous trunks. The fact tliat so careful an observer as Miller does not find these lymphatics in the sejjta suggests the possibility that the assumption of mature activities in some way brings about the atrophy of all of the interlobular lymjihatics except those that accompany the veins. Again, this plexus may be peculiar to the pig. It seems necessary that this question must remain unsettled until studied bj' some method other than simple injection.
 
 
The question of the drainage of the lung lym])hatics is of exc{'i)tioual interest and importance; and while we must deixMul, for the final settlement, upon physiological methods, there is much e\'idence available from morphological observations. In the larger vessels on the bronchi, the veins, and the arteries there are valves which point towards the hilum. This is assumed to be very good evidence that the flow is in that direction. No \alves have been described in the lymph-vessels which accompany the smaller bronchi, veins, and arteries. Hence it can not be stated whether the lymph flow, in the lymph-vessels of the veins, is towards the ])leura of the hilum; and, in like manner, the flow in the bronchial vessels might be either towards the hilum or towards the veins and arteries. With regard to the vessels on the pleura, all of the lymphatics above a certain regional level of the lower lobe drain either towards the mid-line and then course up in the pulmonary hgament to end in the nodes at the hilum, or joass by direct paths to these nodes. Those below this level drain to the nodes lying in the mesenterj- of the lesser curvature of the stomach. Home of these drain as do those above — towards the median and ])ass down in the ligamentum pulmonale — while others pass directly down from the posterior pole. This group of vessels which pass to the preaortic nodes drains about one-third of the lower lobe of the lung. This varies considerably; in some specimens as much as half of the lung has been found to drain in this direction.
 
 
This peculiar drainage of the lower lobes seems especially important from the bearing that it may have on the jiathology of the lungs. It has long been kiu)wn that the diaphragmatic vessels drain to these nodes, but there is no connection iK'tween these vessels and those of the lung jiroper. The lymphatics that pass through the pulmonary ligament apparently drain only the pleura; but, as has been shown, the deep lymphatics anastomose with those of the pleura, and therefore it seems possible for substances to pass from the lung-tissue to the preaortic nodes. What bearing this may have upon the pathology of the lungs or of the abdomen remains to be settled.
 
 
SUMMARY.
 
 
The lymphatics of the lungs are derived from three sources — the right and the left thoracic ducts and the retroperitoneal sac.
 
 
In embryos 2.6 to 3 cm. long, vessels bud ofT from the thoracic duct and grow across to the trachea, forming ther(> a plexus that gradually extends over the ventral surface of the trachea, and especially down over (lie hifurcnlion. I'roni this plcwis ve.s.sels pass into both lungs and into the ])leura.
 
 
The right thoracic duct divides, in embryos about 2.5 cm. long, uito two ves.sels. One pa.s.ses to the heart, while the other breaks up to form a plexus on the right lateral wall of the trachea. Some vessels from this plexus pass down into the hilum
 
 
 
DEVELOPMICNT OK TIIK I.VMI'H.VTICS OF TlIK UNCS l.\ TIM:; KMHKYO I'IG. Go
of the right king;, while others anastomose with the plexus from the left side, which extends up over the trachea. The development of the lymphatics within the lung depends upon the division of the vessels into two groups — those associated with the veins and connective-tissue septa, and those associated with the arteries and the bronchi.
 
 
The former grow verj- rai)idly, and following each of the branches of the jjuhnonary vein, j^ass to the pleura. There are at first onlj' two or three h'mphatics with each vein. In the early stages the terminal veins He about midway between the adjacent bronchi, and in this plane a sheet of lymphatics develops from the vessels surrounding the veins and passes to the pleura, where they mark out the l)oundaries of the distribution of each bronchus. These vessels anastomo.se with those thot grow direct to the pleura from the i)lexus on the trachea.
 
 
The bronchial vessels develoj) more slowly and at first are to be found only around the larger bronchi. As these structures increase in size and number,, the lymphatics surrounding the main bronchi send vessels to the smaller ones and these form a plexus around each of the bronchi, so that the bronchial tree is surrounded by a continual series of branching tubes made up of lymphatic vessels. From every point of division of the l)ronchi. lymphatic vessels pass to the lymphatics of the veins; those around the terminal ])ronchus leave it near its ending in the atria, and pass to join the lymphatics of the veins or sei)ta, or, more rarely, those of the pleura.
 
 
Lymphatics also arise from the retroperitoneal sac and grow up posterior to the diaphragm to enter the lower pole of the lower lobe of the lung. These vessels form a plexus on the median surface of the lower lobe, and send branches both to the pleura of the other surfaces and into the lung along the veins. Plexuses develop here as with those that come from above and the two groups .soon anastomose.
 
 
The further develojiment consists m the multijjlication of the plexuses on the bronchi and blood-vessels, following their continued differentiation. As the lung increases in size, the larger veins become approximated to the bronchi and only the terminal ones are separated from them; these lie in the periphery of the lobule. Connective tissue is formed along the sheets of lymphatic vessels, and these become the septa of the lung, containing a definite set of vessels which develop from the early vessels following the veins. The lymphatics accomj^anying the veins remain connected with those of the bronchi and septa.
 
 
The common plexus surrounding the artery and bronchus is separated into two mdividual plexuses, incident to the increase in size of the artery-; however, these continue to have anastomosing branches.
 
 
The vessels of the pleura mark out the early connective-tissue septa, but later there develops a fine-meshed i)lexus between these larger vessels, which is not connected with the vessels of the lung-tissue. The valves begm to form at about 6 cm. and. in general, point away from the pleiu-a. None, however, ha^•e been found in the smaller vessels which accompany the terminal bronchi.
 
 
In the adult there are lymphatic vessels accompanying the bronchi, the arteries and the veins; these anastomose freeh'. There are also vessels in the connectivetissue septa which drain chiefiy into those around the veins, and, to some extent.
 
 
 
 
 
06
 
 
 
DEVKLOPMEXT OF THE LYMPHATICS OF THE LUNGS I\ THE EMBRYO PIG.
 
 
 
 
 
into those of tlio bronchi and arteries, near the i)oint where the vein and the bronchus separate to take their rehUive positions with rehition to the lobule. There are numerous anastomoses between the deep vessels and those of the i)leura, but i)robably most of the flow is towards the hilum. All the deep vessels, together with the greater number of the pleural vessels, drain into the nodes at the hilum; but the vessels of the lower half of the pleura of the lower lobe drain through several vessels to the preaortic nodes. These vessels pass through the ligament of the lower lobe and behind the diaphragm.
 
 
 
 
 
BIBLIOGRAPHY.
 
 
 
 
 
B AETJER, Walter A. : On the origin of the mesenteric s.ic and
thoracic duct in the embrj-o pig. Amer. Jour. Anat.,
Philadelphia. 190S, viii. Clark, .V. H.: On the fate of the jugular lymph sacs, and the
development of the lymph charmels in the neck of
the pin. Amer. Jour. Anat., Philadelphia, 1912, ix. Clark, E. R.: Observations on living growing lymphatics
in the tail of the frog lar\-a. Anat. Record, Philadelphia, 1909, HI. . An examination of the methods used in the study
of the development of the lymphatic system.
 
 
.■Vnat. Record, Philadelphia, 1911, v. . Further observations on living growing lymphatics:
their relation to the mesenchyme cells. Amer.
 
 
Journ. Anat., Philadelphia, 1912, xiii. Councilman, W. T.; The lobule of the lung and its relations
to the l.vmphatics. Journ. Boston Soc. Med.
 
 
Science, Boston, 1900, iv. Criikshank, W.: The anatomy of the absorbing vessels
of the human body. London, 1790. Delemere, G., p. Poirier, and B. Ci'neo: The lymphatics.
 
 
1904. Flint, J. M.: The development of the lungs. Amer. Jour.
 
 
Anat., Philadelphia, 1900, vi. Hei'Er, G. J.: The development of the l.vmphatics in the
small intestine of the pig. Amer. Jour. Anat.,
Philadelphia, 1909, ix. Kampmeier, O. F.: The development of the thoracic duct
in the pig. Amer. Jour. Anat., Philadelphia,
1912, XII. Klein, E. : Anat. of the lymphatic system. London, 1875, ii. Lewis, F. T.: The development of the lymphatic system in
rabbits. Amer. Jour, of Anat., Philadelphia,
1906, V. McClire, C. F. W. : The development of the thoracic duct
and right l.vmphatic duct-s in the domestic cat.
 
 
Anat. Anr... Jena, 1908, xxxii. Mabcaoni, Pai'l.: Vasorum lynijihaticoruin corporis humniii
hiHloria ct ichnographia. Senis 1787.
 
 
 
 
 
Miller, W. S. : The structure of the lungs. Jour. Moiph., 1S9;{, viii.
 
 
. The lymphatics of the lungs. Anat. .\nz., Jena,
1896, xii.
 
 
. Das Lungcnldppchen, seine, Bliit- und Lymph
gefiisse. Archiv. filr Anat. und Physiologic, Leipzig, 1900.
 
 
. Anatomy of the lungs. Reference Handbook of
the Med. Sciences, 1902. 575-586.
 
 
. Lymphoid tissue of the lung. Anat. Record, Philadelphia, 1911, V.
 
 
Pappenheim, — : Sur les Ijmphatiques des poumons et du diaphragmc. C'ompt. Rend., 1860, xxx.
 
 
Poirier and Ch.\rpey: Treatise of human anatomy.
 
 
Sabi.n, F. R.: On the development of the superficial lymphatics in the skin of the pig. Amer. Jour. Anat., Baltimore, 1904, iir.
 
 
. The lymphatic system in human embryo, with a
consideration of the morphology of the system as a whole. .■Vmer. Jour. Anat., Philadelphia, 1909, ix.
 
 
. Der Ursprung und die Entwickclung des Lymphge
fass.vstems. }>gcbnisse der Anatomic und Entwickelungsgeschichte, Wiesbaden, 1913, xxi.
 
 
. The origin and development of the lymphatic system. The Johns Hopkins Hospital Reports, Monographs, new scries, 1913, v.
 
 
S.VPPEY, P. C. : Anatomic, physiologic, pathologic des vaisscaux lymphatiqucs. Paris, 1874.
 
 
SiKORSKi, J.: Uebcr die L.vmphgefiisse der Lungcn. C'entralbl. f. Medicin. Wisscnsch., 1870.
 
 
Teichman, L. : Uebcr Lungcnlyn'phgcfiissc. Anz. d. Akad. d. Wissen.sch., in Krakau, 1896.
 
 
vox WiTTiCH, W.: LTebcr die Bczichungcn der Lungenalveolcn zum Lymphsystem. Mitth. Anx. d. Konigsbcrger Phys. Laborat., 1878.
 
 
Wywodzoff, — . : Die L.vmphgewege der Lunge. Wiener Medicin. Jahrbucher, 18G6, xi.
 
 
 
 
 
EXPLANATION OF PLATE5. Plate 1.
 
 
Fig. 1 . DiaKram of transverse sect ion of left lung of an embryo
 
 
pig 3 cm. long, in which the blood-vessels were injected through the umbilical arter>- with india ink. The lymphatics appear as dilated spaces (blue). The section is 20 ;i thick and is stained with hematoxj'lin and eosin, aurantia, and orange G. X55. Ao, aorta; 7', trachea.
 
 
Fiii. 2. Diagram of section through lobule of lung of an erabrj-o pig 7 cm. long, in wliich the lymphatics were injected with india ink through the left tracheal plexus. The veins were slightly injected by the rupture of a lymphatic vessel into a vein near the hilum. The section is lOO/i thick and is unstained. X47.5. A, arterj-; B, bronchus, ]', vein; PI, pleura.
 
 
Fig. 'i. Surface of lung of an embrj-o pig cm. long, in which the lymphatics were injected with india ink through the left tracheal plexus. The section was taken from the ventro-lateral surface of the left lower lobe and is about 200m thick and is unstained. X29.4. /' L, primarj' lobule.
 
 
Fig. 4. Longitudinal section of lung of an embryo pig 6 cm. long, in which the lymphatics were injected with india ink through the left tracheal plexus. The veins contain some blood pigment. The section is 400 /i thick and is un.stained. X3.'?. V, vein; B, bronchus.
 
 
Fig. .). Diagram of left tracheal plexus in an embryo pig 6 cm. long, in which the lymphatics were injected through the thoracic duct. Cleared by Spalteholz method. Note that part of the vessel marked with an asterisk (*) has been removed in di.ssecting the body-wall away. This vessel is the one described as the first to the lung. X15. *, first vessel to lung; T, trachea; Th D, thoracic duct; L T P.. left tracheal plexus; Ao, aorta.
 
 
Pl.\tk 2.
 
 
F'iG. 1. Dissection of an embryo pig 4 cm. long, in which the lymphatics were injected with prussian blue through the retroperitoneal sac. The heart, aortic arch, left lung, and the body-wall have been removed. Cleare<l by the Spalteholz method. X19. Th D, thoracic duct; R Th D right thoracic duct; R T P, right tracheal ple-vus; L T P, left tracheal plexus; C L, cardiac lymphatics; Ao, aorta; B, bronchus; Oe, csophagtis.
 
 
Fig. 2. Section of a small area of the lung of an embryo pig 7 cm. long, in which the lymphatics were injected with Prussian blue through the retroperitoneal sac. Drawing to show the relation of the peri-bronchial h-mphatics to the wall of the bronchus. Section is 20fi thick and is stained with hematoxj-lin and eosin, aurantia, and orange G. X93. B, bronchus.
 
 
Fi<;. 3. Dissection of an embryo pig 4 cm. long, in which the Ij-mphatics were injected with prussian blue through the retroperitoneal sac. The left lung, the arch of the aorta, the pulmonary arterj-, and the body-wall have been removed. Cleared by the Spalteholz method. The left tracheal plexus is shown as a soUd blue mass because the meshes are so close that they could not be analyzed in the drawing. X 15. Th D, thoracic duct; R Th D, right thoracic duct; Ao, aorta; L T P, left tracheal plexus; B. bronchus.
 
 
Fig. 4. Longitudinal section of upper lobe of right lung of an embrj-o pig 6 em. long, in which the hTnphatics were injected with prussian blue through the retroperitoneal sac, and the veins were injected with india ink through the pulmonary vein. The section is 400m thick and is unstained. Cleared by the Spalteholz method. X39.
 
 
Pl.\te 3.
 
 
Fig. 1. Small block of an embryo pig lA cm. long, in which the lymphatics were injected with prussian blue by puncture of an interlobular septum. The arteries were injected with india ink through the pulmonarj- artery. Cleared by the Spalteholz methotl and mounted in balsam. The specimen was mounted at a convenient angle to best show the interlobular septum; unfortunately, it was jarred out of position while being drawn and hence the group of lymphatics in the septum is shown bent to one side. X40. PI, pleura: .1, artery; / L S, interlobular septum.
 
 
Fig. 2. Diagram of a dissection of an embrjo pig 3 cm. long, in which both the right and left jugular sacs were injected and, from them, the right and the left thoracic ducts respectively. India ink was useil. The body-wall, heart, and left lung have been removed. Cleared by Spalteholz method. X30. T/i D, thoracic duct ; R Th D, rig' t thoracic duct; V C .S, vena cava superior; Ao. aorta; P A, pulmonarj* arterj'; C L, cardiac branch of the right thoracic duct.
 
 
Fig. 3. Diagram of a section of the right lung of an embryo pig 6 cm. long. This is the same specimen from which figure 4, plate 2, was made; the part of the section shown in that figure is indicated b>- an X. X20.
 
 
 
 
 
68 DEVELOPMENT OF THE LYMPHATICS OK THE LINOS IN THE EMBRYO PIO.
 
 
Plate 4.
 
 
Fig. 1. Longitudinal spction of the left lung of an embryo pig 5 cm. long, in wliich the lymphatics were injected with Prussian blue througli the rctro|)eritoneal sac. The veins have some blood pigment in them. The section is 400/i thick and is unstained. X22. V, vein.
 
 
Fig. 2. Dissection of an embryo pig 4 cm. long, in which the lymphatics were injected with prussian blue from the retroperitoneal sac. The right lung, esophagus, and body-wall have been removed. The stomach was pulled to the left side of the embryo in order to expose the retroperitoneal sac. Cleared by the Spalteholz method. X20. Ao, aorta; L L, left lung; Th D, thoracic duct; D, diaphragm; li P S, retroperitoneal sac; *, first vessel to the lung.
 
 
Pl.\te .").
 
 
Fig. \. Surface of lung of an embryo pig 23 cm. long, in which the lymphatics wore injected with prussian blue, by puncture of an interlobular septum. Cleared by Spaltcholz metliod. The interlobular septum is mdicated by a very large lymphatic trunk. X2S. / L S, interlobular septum.
 
 
Fig. 2. Longitudinal section of the upper portion of the lower lobe of the left lung of an embryo pig 5 cm. long, in which the lymphatics were injected with pru.ssian blue through the retroperitoneal sac, and the veins have retained a little blood pigment. The section is 400 ;u thick and is unstained. X57. V, vein; .4, art en,-; PI, pleura; B, bronchus.
 
 
Fig. 3. Lower portion of left lung of an embryo pig 5 cm. long, in which the lymphatics were injected with india ink through the retroperitoneal sac. Cleared by the Spalteholz method and mounted in balsam. X28.
 
 
 
 
 
CUNNINGHAM
 
 
 
 
 
 
 
 
 
Fis. 5 (C. R. 6s;
 
 
 
Fig. 5 (C. R. 6^-=)
 
 
 
Tlira SJtatfer ftdt
 
 
 
CsmfltU An Omfany
 
 
 
CUNNINGHAM
 
 
 
RThD.,
 
 
 
R.Th.D.
 
 
 
 
 
 
Fig. 4 (C, K. 6™)
 
 
 
/. f. DiJuu/i /«,
 
 
 
Camfhtlt ji'i CffJTi/.J
 
 
 
4^'
 
 
 
CUNNINGHAM
 
 
 
F.Th.D.
 
 
 
 
 
 
Y\z 1 (.C. R. ■. ^::ii
 
 
 
 
 
Fig. 2 (C. R. 3s)
 
 
 
 
 
FJ-. 3 (C. R. 6es)
 
 
 
7 r. Dkuirh fiiit
 
 
 
CAmfltU Art Czmr^nf
 
 
 
CUNNINGHAM
 
 
 
\
 
 
 
I
 
 
 
 
 
 
 
 
 
 
 
 
CUNNINGHAM
 
 
 
 
 
Fig. 2 (C. R. 752!)
 
 
 
Fig. .3 (C. R. 5™)
 
 
 
7. f. ZJ/Ji/jcA ^uV
 
 
 
CamthrU At± r«,
 
 
 
CONTRIBUTIONS TO E.MIiRYOLOGY, No. 13.
 
 
 
 
 
BINUCLEATE CELLS IN TISSUE CULTURES.
 
 
By Ch.\rles C. ^Macklin.
 
 
 
 
 
Four plates, eoiitaining seventy figures.
 
 
 
 
 
CONTENTS.
 
 
 
 
 
Introduction "1
Method:
ill) Preparations of the cultures:
(1) Tissue and media 71
(2) Cultivation 72
(6) Observations:
(1) Living:
Continuous observation 73
Vital standing 74
(2) Fixed 75
The binucleate cell 76
(a) Incidence 7G
(6) Morphology 77
 
 
 
The binucleate cell — Contitiued. page.
 
 
(,c) Origin:
(1) Theoretical 80
(2) Observations:
Living 82
Fixed 83
(3) Mechanism 86
(d) Fate:
(1) Observations:
Living and fixed 89
Xuclear Fragmentation 98
Summary 100
Literature cited 103
I'^xplanat ion of plates 105
 
 
 
BIMTI.KATK CELLS IN TISSFE CI I.TCHI-S.
 
 
 
 
 
By CiiAKLE-s C. ]\Iacklix.
 
 
 
 
 
INTRODUCTION.
 
 
In examining a living tissue culture, or a preparation from the same, one frequently finds a cell which contains two or more nuclei, of about equal size, slighth' separated or in contact. In a communication of Lewis and Lewis (1912 c, fig. 12) a binucleate cell from a tissue culture is shown; more recently these authors (1915, p. 391) have referred to the occurrence of such cells in tissue cultures under the heading "Amitosis and giant cells." That they may be quite numerous in an area of new growth is seen by referring to figure 1, where one quadrinucleate and six binucleate cells ai)pear in a small field.
 
 
The question of their origin and fate in cultures of embryonic tissue, involving as it does the idea of direct nuclear division, gathers interest from the fact that such cells are found in embryonic tissue developing in vivo, and from the further fact that thej' probably represent the first stage in the formation of certain giant cells. The problem of inquiring into their history bj- prolonged observation of the living cell was suggested by ]\I. R. and \Y. H. Lewis.
 
 
METHOD.
 
 
Cultures were grown in ordinary hanging-drop preparations, the technique of W. H. and M. R. Lewis (1911, 1912«. 19126, 1915) being employed. The tissue was obtained from embryo chicks of from three to ten days' incubation. Heart tissue was most frequently used, and gave very satisf actor}- results.
 
 
Locke solution as a culture medium was used. A stock saUne solution was first made up as follows: NaCl, 18 grams, 0.9 per cent; KCl, 0.84 gram, 0.042 per cent; CaCL, 0.5 gram, 0.025 per cent; NaHCOs, 0.4 gram, 0.02 per cent; HoO, 2,000 c.c. Freshly distilled water and absoluteh^ clean bottles are indispensable. The solution will keep apparently good for months.
 
 
Culture media was made up, from time to time, as required, from this stock solution, in 100 c.c. lots, by dissolving from 0.25 to 1 gram of dextrose in 100 c.c. of saline, thus making a solution of 0.25 to 1 per cent of dextrose. The media was then placed in clean plugged test-tubes, 10 c.c. in each, and steriUzed in the Arnold sterilizer for 30 minutes, after which it was stored for use as required. It should be made up fresh every two weeks.
 
 
The best results were obtained by diluting this media, when it was being used, by the addition of 20 to 25 per cent of freshly distilled, sterile water, since some evaporation went on during the jjlanting, and further concentration of the media often occurred in the preparation from evaporation of the hanging drop and conden
71
 
 
 
72 lUNTC'I.K.VTK Ci:i.l.S IN TISSIK CILTLHKS.
 
 
.sation (»f the vapor about the walls of the moist chamber on the depressed slide. By using a slide with a dceji dejiression, containinfj; a littl(> distilled water, this eva])oration was lessened.
 
 
Cultures planted in a lianjiinfi drojiof tliis media, upon a sterile, clean cover-slip, inverted over a depressed slidt', which was sealed with \aseline, grew very well at a temperature of 39° to 40°. However, it was found that if a small (juantity of extract of chick embryo (Carrel, 1913) were added to the media better growths were obtained, /. e., cells migrated out from the original piece earlier, growth was more rapid and vigorous, mitoses were more fre(}uent, and a larger percentage of growths was obtained. Hence this addition to the UKnlia was generally made. Ordinary l>ouillon had a .similar activating effect.
 
 
The embryonic extract was prepared as follows: After tlie ciubryo iiad been removed from the egg, under sterile conditions, and with as little contamination with yolk as possible, it was placed in a Petri dish containing 10 c.c. of sterile Locke's media and washed. The ti.ssue to be planted having been dissected out and removed to another dish of media, the remainder of the embryo was cut up and ])laced in a small, sterile test tube with a little media, and carefully ground u]) with a glass rod. This mixture was next centrifugalized, and the supernatant fluid adiletl to the culture, generally in the proportion of ecjual parts of this fluid and Locke. Too high a proportion of embrj'^onic extract was undesirable on account of its richness in food material, in that it jiroduced a cell overloaded with fat globules, which interfered with observation.
 
 
The advantages of glycosaline over plasma ha\'e been notetl by Lewis and Lewis ( 1912«, J). 10). It is more transparent and practically all of the growth is upon the lower surface of the cover-slip — not scattered throughout the hanging drop, as in the ca.se of the plasma clot. The cells, unimpeded by the fibrin network, migrate freely along the cover-slip, upon which they spread themselves flat and thin, thus facilitating observation. The quantity of fat being much less than in plasmagrown cells, the cytoplasmic constituents, such as the centrosi)here and mitochondria, are much more ea.sily observed and studied. There are, too, the additional very considerable advantages that the media is more convenient to handle, lixed i)re])arations are more easilj* made and are not marred by staincul fibrin and coagulated albumen, and exjierimental o])erations, such as staining with xht\\ dyes, are more satisfactorily carried out.
 
 
Immediately after i)lanting, the cultures were transferred to a warm box, kej^t at a constant temperature of 39 to 40 degrees by means of an electric thermoregulator. ^^'ith the niicrosco])e inside this box it was not necessary to remove the cultures from tiieir warm environment for purpo.ses of observation, and it was owing to this manifest advantage that the earlier method of observation of the living cultures ujion a warm stage outside the incubator was discontinued. In adilition, the warm-stage method of heating, from one .side only, was found to be inferior to that of the warm box, in which the culture was (•nini)let('ly .surrounded by an environment of uniform temperature.
 
 
Illumination was furnished by daylight. Tung.sten glol)e, or Welsbach burner. A ray filter, consisting of ii glnss vcs.scl filled with a solution of (•oi)per sul|)hate or
 
 
 
BINUCLEATI': CKLLS IX TISSUE CULTURES. 73
copptT acetate, placed l)et\veeii the source of illumination and the condenser, when artificial li^ht was used, was found to be an advantafi;e (Kite 1913ii, p. 149).
 
 
In studyinji; the grosser changes, such asthe variation in shape of the nucleus, the 4 mm. Leitz apochromatic objective was used; for the finer details the Leitz i\ oilimmersion objective was found to be satisfactory. Oculars were Leitz Compensating Nos. 4 and 6.
 
 
For observation of the living cells the cultures of the second day were generally the most favorable; the growth was then usually abundant and the cells in a healthy condition, with a fair jjroportion of mitoses.
 
 
It is quite evident that continuous observation of the living cell, provided it can be carried out successfully, is the ideal method of studying the sequence of changes occurring therein. Indeed, for the stud^^ of amitosis it has been regarded as indispensable, as witness the statement of Richards (1911, p. 125): "For amitosis there is but one absolutely certain criterion, the observation of living material and subsecjuent study of material fixed under observation;" he adds, "this is, of course, impossible in most cases."
The method has already been used in the study of living multinucleate cells of tissue cultures, Lambert (1912o) having attempted to settle the ciuestion of the origin of giant cells growing from explanted tissue by its aid, and the character of the results attained through its use was sufficiently encouraging to warrant its application to the i)roblem in hand, though not altogether satisfactory in view of the obvious difficulties. It was hoped, too, that these difficulties would be minimized by the use of glycosaline media, which produced films of tissue sufficiently thin for study in the living condition.
 
 
It was first planned to ascertain the full history of the binucleate cell by selecting a cell with a single nucleus and observing it continuoush' on the stage in the warm box till either the nucleus divided and formed a double-nucleated cell or the cytoplasm became merged with that of another mononucleate cell to form a single cell containing two separate nuclei. Observations upon this binucleate cell were then to be continued until the ultimate fate of the double nucleus was disclosed. Drawings were to be made from time to time with the camera lucida.
 
 
This ideal was found to be impossible of realization, on account of the technical difficulties. Cultures under continuous observation, exposed, as they were, to strong light, often showed e\idences of degeneration; even daj-light seemed to cause this and the use of ray filters did not altogether eliminate it. Degeneration was noticed at times even when the plan was followed of making short observations and immediately turning the light off, the culture remaining continuou.sly on the stage.
 
 
Living cells show a marked tendency to migrate; hence the cell under observation had to be closely watched to prevent its escape from the field of vision. Other cells often wanderd over the cell under inspection, and so interfered with the work. Added to these difficulties is the length of time involved in the process, which necessitates many hours — even days — of continuous observation. Then, too, the minuter cell changes are very difficult to follow, even for short periods, the only
 
 
 
74 BINUCLEATE CELLS IN TISSIE CULTURES.
 
 
optical picture presented being verj' delicate shades of difference in refractivity. The obscure and peculiar ojjtical properties of living matter, as Kite (1913, p. 148) jwints out, give rise to an important source of error.
 
 
The ideal jjrocedure having to be abandoned, the alternative practice was adojited of following shorter i)eriods of change and piecing the records of these together. A start was made with the formation of the double nucleus, and here another difficult}' was encountered; it was manifestly impossible to tell which of the thousands of nuclei in the culture was about to divide, and by selecting nuclei at random, months might be spent without getting one which ultimately divided. It was thus necessary to select a nucleus which gave some indication of being on the way to division, /. e., by elongation, or equatorial constriction; such a cell was observed continuously until it divided or became degenerate. The subsefjuent history was studied by selecting a double-nuch^ated cell and observing it continuously.
 
 
In the observations the shape of the nucleus was particularly noted, and with this was considered the behavior of the centrosphere, mitochondria, fat globules, nucleoli, shape of the cell generally, and whether or not the cell itself ultimately divided following nuclear division. Cells on or near the outer border of the new growth were found most favorable, since they were larger and more free from surrounding cells. They appeared to be usually (luite healthy during the first 48 hours at least.
 
 
As has here been noted, the morphology of the cell is difficult to make out in the living and unstained condition, and it was thought that inspection would be much easier and more accurate if the details could be rendered visible by the use of stains which would not impair the vitality of the cell.
 
 
Churchman and Russell (1914) and Russell (1914) have recorded satisfactory results with gentian violet in staining embryonic and adult tissues of the frog growing in vitro. They stated that endotheUum from adult frog pericardium in frog's ])lasma to which gentian violet had been added grew definitely when the strength of gentian violet was 1 in 2,000, and actively in a dilution of 1 in 20,000. Furthermore, their records .show that they were able to follow cell division in their stained jireparations. Clear karyokinetic figures were not seen in growing adult frog tissue, but in embryonic frog tissue these figures were found in the divitling stained nuclei. They believe that the nucleus is stained intravitally and that growth continues in the jiresence of the dj-e. Toxic action was shown when this stain was u.sed upon Paramecium, even in dilutions of 1 in 1,000,000. They believe that "the use of stains in the plasma in which tissue is grown will i>robably facilitate the study of nuclear growth."
My results with gentian violet in chick tissue growing in Locke do not bear out those of Churchman and Russell, for the staining could not be considered as in an>sense intra vilmn, under the conditions existing in my experiments.
 
 
A solution of Criibler's gentian violet was made up in a strengtli of 1 in 100,000 with slightly diluted s-ilinc. Without removing the culture from the warm box, the cover-slip was lifted olT and a small drop of the stain (warmed to the same tem
 
 
 
BINUCLEATE CELLS IN TLSSUE CULTURES. 75
ix'niturc as the culture and of about the size of the hanging drop) was added to the hitter, the excess fluid being withdrawn. The dilution of the stain was thus approximately 1 in 200,000. The culture was immediately examined under the microscope. The dye rapidly diffusc^d through the cytoplasm into the nucleus, the nucleoplasm taking on a finel,y granular appearance; this latter was, apparently, the result of the coagulative action of the dye. The nucleoli were distinctly marked out, and stained much more darkly than the nucleo])lasm or cj'toplasm. The nuclear membrane, too, was sharply outlined as a dark violet ring. This staining was very valuable in delineating indistinct nuclear boiuidaries, since these, in the living unstained culture, are often obscure. Irregularities in outline, such as indentations, were rendered very plain, and the method was of assistance in studying the relationship to one another of double nuclei.
 
 
The cytoplasm, after this treatment, consisted of coarse violet granules in a very faintly stained matrix, showing at times a slightly fibrous structure. Cell borders were well marked, especially the pseudopodia, which, however, lost their power of movement upon being stained. Intercellular ]:)ridges could be studied. jVIitochondria were not specifically stained, and degenerated in a short time. In a culture so stained evidence of hfe, such as pseudopodial and mitochondrial movement, cell migration, and mitosis, ceased almost at once, and in a few minutes vacuoles formed in the cells and the entire culture became degenerate.
 
 
Though gentian violet staining is of great assistance in obtaining a conception of the morphology of the cells rapidly, under the conditions of the experiments its toxic action precludes the possibility of the stained cell undergoing vital changes. Owing to its coagulative action the appearance of the livmg protoplasm is not accurately reproduced in the stained preparation. In spite, however, of these disadvantages, the use of gentian violet enables one to inspect portions of the cells which, in the Hving condition, are almost, or quite, invisible, and also to examine more accurately and easily some of the visible parts.
 
 
M. R. and W. H. Lewis (1914 and 1915, p. 376) used janus green as a vital stain in tissue cultures growing in Locke solution, and found that, although the mitochondria were specifically stained, the dj-e was toxic in as low dilutions as 1 in 200,000, and caused speedy death of the cells, as well as distortion of the mitochondria.
 
 
Janus green (Hoechst), di-ethyl saflfranm azo di-methylanilin, in Locke's solution, in a strength of 1 in 40,000, was applied to living cultures in the same manner as the gentian violet, and was found to stain the mitochondria specifically in about 5 minutes, but no movements of these bodies could then be noted, and the threads broke up into a row of granules. The cells soon died, as evidenced by their vacuolated and degenerated appearance.
 
 
While janus green staining provided a rapid and convenient method of observing mitochondria, its toxic action rendered it valueless as a means of studying vital changes; moreover, the stained mitochondria soon lost their normal optical characters, thus prohibiting extended observation.
 
 
Some of these living cultures were fixed and stained. Osmic-acid vapor was used for fixation, and Heidcnhain's iron hematoxylin was found to give the best staining.
 
 
 
 
 
76 BiNUCLKATE CKi.i.s IN TissiK cri/rri<i;s.
 
 
'I'lu' c'ultuir, growing in the hanging choj), was renuncd from the vaselined slide and exposed to the fumes from a 2 per cent atjueous sokition of the tetroxid of osmium. This may be done by placing the cover-slip, droj) down, over the mouth of a bottle containing the fixative (the vasehne adhering to the cover-slii) and preventing the escape of the vapor) or, as suggested l)y M. H. Lewis, by floating the cover-slip, drop up, upon the osmic solution. Fixation is complete in 5 or 6 minutes, and the preparation is then dark brown or black. It is now rinsed off with distilled water, and passed rapidly through ethyl alcohol solutions of 35, 50, and 70 per cent. To the latter a few drops of hydrogen peroxid are added, which bleaches the preparation. It is then passed back rapidly through the same alcohols, rinsed in distilled water, and washed in running tap-water for 5 minutes. Too long immersion in alcohol will cause the mitochondria to become dissolved out.
 
 
The cover-slip, culture downward, is now floated upon a solution of 4 per cent iron alum and allowed to remain for 12 to 24 hours; next it is washed in running tapwater for 5 minutes and then immersed in 0.5 per cent aqueous hematoxyhn for 24 to 48 hours, after which it is washed in running taji-water for 1 minute, differentiated in 2 per cent iron alum and again washed in tap-water for 10 minutes, dehydrated through the alcohol series, cleared in xylol, and mounted in balsam. The hematoxyhn solution is prepared as follows: Hematoxylin (10 \wr cent in absolute alcohol), 0.5 c.c; distilled water, 10.0 c.c.
 
 
These fixed preparations were used to make clear the morphology of the hving cells, especially such details as nuclear membranes, nucleoli, mitochondria, and centrosomes. An attemjit was made to pick out the successive stages in the process of direct division of the nucleus for comparison with the observations upon hving material, and thus to build uj) a series exhibiting the various changes. The phases of mitosis were also studied, and drawings were made of interesting cells. For statistical purposes cell counts were made of some of these preparations by placing a glass disk, upon which scjuares had been ruled, in the ocular, and using the mechanical stage.
 
 
THE BINUCLEATE CELL. INCIDENCE.
 
 
Th(» fretjuency of occurrence of binucleate cells varies within wide limits in cultures from different tissues. They were found to be most numerous in membranes growing from the heart, and were not uncommon in cells of the connectivetissue type from this and other tissues, but in the endodermal membranes from stomach and intestine they were exceedingly rare. They ma>- be even altogether absent from the new growth. Lewis (1915, ]\ 150) notes that in growths from the leg of chicks no amitotic forms were noted.
 
 
To get an idea of the relative number of these cells as compared with the total number of cells in the new growth, careful counts were made of 20 fixed cultures from chick heart. Imperfect cells and those situated so close to the original piece as to be indefinitely outlined were omitted. In these 20 i^reparations there was a total of 41,725 cells, of which 375 were binucleate, or an average of 1 binucleate to
 
 
 
niNUCLEATE CELLS IX TISSUE CULTURES. 77
eacli 111 cells; thus the binucleate cells made up 0.9 per cent of the total cells appearing in the new growth.
 
 
Even in different preparations from the same tissue binucleate forms were found with varying fretiuency. Among the 20 cultures of heart mentioned above, one preparation showed 1 double nucleus to each 28 cells, while in another the ratio was 1 to 1,180.
 
 
Age of tissue, too, in these 20 heart preparations, had a bearing upon the incidence of ])inucleate forms, new growths from the younger hearts showing a somewhat greater i)roi)ortion of doul)le nuclei than those from older cardiac tissue. In hearts from chicks of 5 days' incubation there was, on the average, one liinucleate to each 105 cells; in T-day hearts the ratio was 1 to 123, and in 8-day hearts it was 1 to 233.
 
 
Finally, duration of growth seemed to be related to the relative frecjuency of occurrence of these cells. In the same 20 ]:)reparations it was found that cultures of the first 2-i hours showed one double nucleus to each 183 cells; in cultures of the second 24 hours the ratio was 1 to 86 cells. This seems to point to a considerable amount of nuclear splitting in the second 24 hours, of which some at least probably occurred within the new growth. Cultures of older duration were, in the slides counted, not sufficiently numerous and typical to base accurate conclusions upon.
 
 
MORPHOLOGY.
 
 
The average binucleate cell (figs, la, 7, and 9) is somewhat larger than the average mononucleate, the area occupied by the nucleus being approximately twice as great. Each nuclear part is, in size, shape, and general appearance, verj* similar to the nucleus of the mononucleate cell. The nuclear parts are often pressed close together (figs, la, 60) and their adjacent surfaces are consequently flattened, the intranuclear pressure in each being evidently equal. When thus related, the appearance of the double nucleus in the living preparation (and indeed in some of the fixed prei)arations) simulates a single nucleus which looks as though it were separated by an equatorial membrane, ^^uch an appearance has been interpreted as a nuclear plate, or intranuclear membrane, and so described by Child (1904, 19076, and 1911, p. 283), and others; but for reasons which will appear later on, I beUeve that such appearances in tissue cultures are due to the apposition of nuclear surfaces, as above described.
 
 
In an elongated nucleus which has become bent upon itself the folded free edge of nuclear membrane, projecting into the karyoplasm, may simulate a partition which seems to be growing across the nucleus from one side to the other. A nuclear configuration of this character is presented in Child's (191 1) figure 16, page 293, and in other of his figures. It is not to be wondered at that the approximated areas of nuclear wall at the folded edge are somewhat attenuated and appear thin, as Child (1911) has observed (p. 283). Such reduphcations of nuclear membrane are not to be looked upon as intranuclear membranes which cleave the nucleus by growing across its equator. I have .seen no evidence of a type of amitosis of this kind.
 
 
Sometimes an equatorial membrane is simulated bj^ an elongated nucleolus h^mg across the nucleus. Again, as Richards (1911, p. 124) suggests: "A strand of
 
 
 
78 HIMCLEATE CELLS IX TISSUE CULTURES.
 
 
liiiiii strctt'lu'd across a nucleus with chromatin granules upon it often gives the appearance of a membrane dividingthe nucleus amitotically(endogenousdivision?)." He also states that he has found no evidence of the "endogenous" division of Chikl (1907(7, p. 95); nor have I seen anything of this kind in tissue cultures. Optical appearances similar to Child's (1911) figure 6 have been seen in living cells and interpreted as indentations and infoldings of the nuclear membrane. All these conditions can be made clear by the use of a dye like gentian violet upon the living culture, or by proper fixation and staining. In no case has a bona fide intranuclear membranous i^artition been found in any kind of preparation.
 
 
I may also state here that my observations upon fixed and stained cells in tissue cultiu'es have not disclosed cases where one nuclear half was more darkly stained than the other, such as those mentioned by Child (1904, p. 549; 1906, p. 595; 1907f, p. 171 ; and other places) and which he believes to indicate "a certain degree of physiological independence before separation of the parts." In the living condition, too, the nuclear portions present no evident difference in cytoplasm. The contents of the nuclear parts are in every way similar to those of the single nuclei. The nucleoi)lasm apjiears homogeneous during life and when fixed with osmic-acid vapor is finely granular. This method of fixation ])reserves most accurately the details of the living cell (Lewis and Lewis, 1915).
 
 
There is usually at least one nucleolus or karj^osome in each nuclear portion, and more often two (figs. 7 and 9) or even more. The nucleoli of the connectivetissue type of cell are irregular in shape, often elongated, and var}^ greatly in size (fig. 8). In the living cell they are highly refractive. They continuously undergo changes in shape, size, and number during the life of the cell (figs. 24 to 35, and ]ilate iv), as can be seen by watching the living nucleus. It is then apparent that their outline is "ragged," as Lewis and Lewis (1915) describe it. The bodies even ai)i)ear to break up from time to time, and afterward to recombine (figs. 24 to 35). At times the nucleolus comes to lie very close to the nuclear membrane (fig. 29) and it may even api)ear to be attached to it. These bodies take the gentian violet dye very well and stain darkly with hematoxylin. If overdifferentiated with iron alum the nucleolus appears as an agglomeration of small granules of about equal size (fig. 10); it is probably to be regarded as a gel of varying density, the densest |X)rtif)ns being represented by these darkly staining granules.
 
 
During mitosis the nucleolus disappears with the formation of the si)ireme, and the daughter nucleoli reai)iM'ar in the reorganizing daughter nuclei. The nuclear portions may be .separated by an interval (fig. 9), or simply touching one another (fig. 7), or may be ])ressed so clo.se together that their adjacent surfaces are flattened, similar to the condition in the early cleavages of Moniczia, as mentioned by Ilarman (1913, p. 221). They tend to remain close to one another, and do not migrate far apart, as nuclei in a syncytium. A\'hen sej^arated, the nuclear portions show mitochondria Vx'twcen them (fig. 9) and usually the centrosphere is situated either in the interval between the nuclear jjortions, or opposite this interval, as in figures 7 and 59.
 
 
In the living condition the centrosijhere or "central body" of liCwis and Lewis (1915) ajjpears as an area of slightly greater refractivity situated at one .side of the nucleus in mononucleate cells; this side is freciuently concave, with the centro
 
 
 
BIXrCLEATH CELLS IX TISSUE CULTLHES. 79
sphere situated in the concavity (fig. 24c). This concave side then appears indistinctly marked out in the living culture, the close proximity of the centrosphere and mitochondria ol)scuring the nuclear outline. Its relation to the parts of the double nucleus has been noted.
 
 
I have not observed the centrosome (centriole) in the living cell, but when stained with iron hematoxylin this body appears usually as two minute dark granules, lying close together (fig. 7). The centrosphere takes a slightly darker stain than the area surrounding it, and thus appears to be a somewhat more concentrated area of the protoplasm. From this area mitochondria radiate, as seen in figure 8. In the living condition the centrosphere shows an indefinite, irregular, apparently serrated edge, the toothlike processes of which undergo a curious constant, slow, almost imperceptible indrawing and outpushing. The mitochondria seem to be intimately connected with this body, as observed by Lewis and Lewis (1915, p. 349), but they differ from it in their reaction to janus green and to certain methods of staining in the fixed condition, such as iron hematoxylin.
 
 
Mitochondria in tissue cultures have been described at length by Lewis and Lewis (1914, 1915). Their curious movement, mentioned b}' these authors, is plainly evident. The special relation of these bodies to the binucleate cell is their position between the nuclear portions, as in figures 8 and 9, unless, as in figure 7, the parts of the nucleus are too close together to permit of this. The relationship of the mitochondria and adjacent centrosphere to the portions of the double nucleus is similar to that of the Xetzapparat of Deineka (1912. figs. 2 and 12) under similar conditions.
 
 
Fat, though not so abundant as in plasma-grown cultures, nevertheless occurs as fine globules which tend to crowd together at the nuclear poles (fig. 32) and often become arranged in rows between the mitochondria.
 
 
The other details of the binucleate cell are very similar to those found in the mononucleate.
 
 
Occasionallj- cells are found which contain three or more distinct nuclei (fig. 16) and the evidence seems to indicate that the binucleate cell is the first stage in the formation of the giant cell; this stage, however, is seldom passed, for giant cells are comparatively rare. Such multinucleate cells are quite different from the foreignbody giant cells of Lambert (1912 a and b), which have been showni by him to arise by fusion of previously separate wandering cells.
 
 
Binucleate cells, and the intermediate stages leading up to them, have long been known in embryonic tissue. Child (1907c) shows several such from chick embryos in his figure 12. ]Maximow (1908) describes and figures double nucleated cells, similar to those found in tissue cultures, in mesenchj-me of embryo rabbits of IH to Idk days, and he has found amitosis also in the guinea pig in the same region and stage. Patterson (1908) shows illustrations of cells of the same type in developing pigeon's eggs, and such cells have been described by many others. Thus it is certain that, since the paired nucleus occurring in the tissue-culture cell is similar to that found in the cells of embryonic tissue, it can not be considered as an abnormaUtv due to its artificial mode of Ufe.
 
 
 
 
 
so lUXrCLKATE CKI,1.S IX Tli^SrE CTI/niiKS.
 
 
Harrison (1913, p. 67) has shown that the behavior of cells j^rowing in culture media is comi^arable to that of cells growing in the embryonic body, and it is reasonable to assume that the behavior of these liinucleate cells in tissue cultures ai)i)ro.\imates the behavior of similar cells in the developing embryo. Hence the vital i)henomena manifested by such binucleate cells in tissue cultures afford reliable evidence as to the changes which take i)lace in siniilnr cells living vuu1(M' normal conditions in the corresponding embryo.
 
 
ORIGIN.
 
 
If we inciuire as to the origin of these Innucleate cells of the new growth we are confronted with four possibihties, viz:
(a) Migration as a binucleate cell from the explautcd tissue. (6) Fusion of the cytoplasm of two pre\-iously separate cells without fusion of the nuclei.
 
 
(c) Division of the nucleus by mitosis without division of the cytoplasm.
 
 
(d) Division of the nucleus l)y amitosis without division of the cytoplasm.
 
 
This list does not include the llie<iretical origin of luiclei de nnro from the cytol)lasm, or their develoi)ment from extruded chromidial substance (Young, 1913). These hyiwtheses do not appear to have been substantiated, and no evidence in favor of either apjjears in tissue-culttu-e prejjarations.
 
 
First, considering (a), we find that twin nuclei occur in the area of new growth immediately surrounding the original tissue, and such forms are well known in embryonic tissue. Thus it is probable that many of the binucleate cells in the new growth have migrated as such from the explanted tissue. The great increase in proportion of double-nucleated cells in the second 24 hours, however, as has been noted, suggests that not all of the.se cells are of migratory origin, but that some have probably arisen in the new growth itself. This view is borne out by observations upon the living cell, as will be shown, where a single nucleus has been seen to become divided directly into two parts, and also by the linding of nuclei in the act of direct division in the fixed preparations.
 
 
The bimicleate cells which have migrated as such from the original piece have probably originated therein in the .same manner as those arising in the new growth, (iiant cells can hardly be considered to have migrated as .such from the original piece, for in the zone immediately surroimding the latter they are not found.
 
 
Regarding (6), it may be said that no appearances which coukl be interiireteil as transitional forms have been found in fixed and stained preparations or in cultures vitally stained, neither has th(> process been observed in the living cultiuv. I therefore regard it as an imi)rol)able hyi)othesis. This could hardly be considered as an exi)lanalion of the formation of giant cells, for that would ])ostulate the fusion of a nudtitude of previously separate cells, of which there is no evidence in the material examined.
 
 
It may Vx- noted that Lambert iMII'J/m. who brought alx.ut the formation of giant cells by fusion of mononuclear cells in cultures fioiii chick spleen, failed to get such cells in cultures from chick heart. Kurlhermore. Lambert ( 1912'/) recognized three other tyi)es of giant cells in ti.ssue cultures, besides this.
 
 
 
 
 
BIXUCLEATE CELLS IX TISSUE CULTURES. 81
Considering next (c), wo find tliat tliis also is improbable. It is ea.s}' to observe the jjrocess of mitosis in vilro, and to follow the various changes. Many such cases have l)een oljserved, and in none has there been seen a failure of the cytoplasm to divide following separation of the chromf)somes. This jjrocess of cytoplasmic constriction is well shown in figures 68, 09, and 7U, and in the living culture it is very evident and easy to watch. In no case has it been observed, in following these cells dividing by karyokinesis, that a binucleate cell was formed; ahvays the end result was two distinct daughter cells, often widely sejiarated, connected by a thin strand of protoj)lasm (fig. 1, t). If crowding of the cells occurs, .separation of the daughter cells may be interfered with to some extent, but it is doubtful if this interference ever is so serious as to jiievent cyto])lasmic fission altogether and thus result in the formation of a single cell containing two nuclei. At least no evidence has been found from observation of the cells of tissue cultures that tliis is ever the case.
 
 
Upon this point my observations agree with those of Child (1911, p. 283). He says: " In Mornezia nuclei w^hich arise by mitosis are separated by an appreciable distance when they form." Again (p. 292), in describing a "double" nucleus, represented in his figure 11, he says: "The two ])arts of the nucleus . . . are in immediate contact and flattened against each other. It is difficult to understand how they could attain such a position as the result of mitotic cleavage, like that of the earlier stages."
It must be said that my observations upon living cells have principally been made with cells of the connective-tis.sue type. In the case of membranes, however, there is always a well-marked dividing line between the cells, which is made evident l)y staining with iron hematoxylin or the use of silver; also this potential Isolation of the cells is made apparent by the fact, when cells do .separate, that the cleavage is along this line of partition, as is shown from the study of fixed prei)arations (Lewis and Lewis, 1912c, figs. 14, 13, and 12). Xo such partition is ever found between the nuclear parts of binucleate cells.
 
 
In fixed preparations of connective-tissue cells there is no indication of any failure of the cytoplasm to divide in the later stages of mitosis; that is to say, we find no telophases where separation of the cytoplasm is not evident (fig. 17).
 
 
Again, these double nuclei almost always have onlj' a single centrosphere (fig. 7), whereas nuclei arising by karyokinesis have each a centrosi)here. Tliis finding as to the centrosphere agrees cntirelj' with that of Deineka (1912) for the Xetzapparat in the dividing epithehal cells of Descemet's membrane and connective-tissue cells of the cornea. This author is of the opinion that the Netzapparat .surrounds the centrosome, and its changes appear to follow the variations of the latter body. In liinucleate colls of these tissues, in which the nucleus divides by amitosis, the X'etzai)parat remains single, whereas if the nuclear division takes place by mitosis each of the daughter nuclei obtains a separate Xetzapparat. By reference to this di.sjiosition of the Xetzapparat, Deineka is even able to tell the manner of origin of such double nucleus, whether bj' amitosis or mitosis in which cleavage of the cytoplasm has been delayed. I have never observed this cell organ in living tissue-cultures.
 
 
The fact that the centrosjihere in the liinucleate coll is single seems to indicate that the twin nucleus is single so far as its rejiroductive cajiacity is concerned. This inference is borne out bv observations, later to be referred to.
 
 
 
 
 
82 BINUCLEATK CELLS IN TISSUE CULTURES.
 
 
Considt'rinfi finally (d), it seonis probable that these twin nuclei arise through direct ctjual binary fission of the nucleus without division of the cytojjlasm. The evidence upon which this assumption rests is, first, the inadequacy of other explanatory hypotheses; and, second, the observation in living cells of a process which is ajii^arently direct nuclear division, and the occurrence in fixed preparations of tissue cultvuTs of what must be regarded as transitional forms l^etween single and double nuclei.
 
 
It is true, as Harman (1913) remarks (p. 219), that "the fact that two nuclei lie in contact is no evidence that they have arisen by amitotic division," and in the material which she studied, viz, early cleavage stages of Toenia teniceformis and Moniezia, she undoubtedly presents convincing evidence that nuclei which have arisen by mitosis may lie quite close to one another within the same cell. This, however, is a case of delayed cleavage, for she states (p. 215) :
"In cleavage, nuclear division takes place very much in advance of cytoplasmic division. In the early divisions it is the exception and not the rule to find even a constriction in the cytoplasm. This gives rise to a syncytial condition. This syncytium persists until very late cleavage."
This is ciuite a different condition from that obtaining in the cells of tissue cultures. Then, too, many of her nuclei contain spiremes. Her contention in no way counts against the view that the double nuclei of tissue cultures are of amitotic origin.
 
 
Observations on direct nuclear fission will now b(> recorded, first to be described being the process as it was seen to occur in the living cell. As has been pointed out, it is impossible to tell from inspection of the living culture which of the thousands of mononucleate cells will divide directly, and so to follow the process of nuclear amitosis in the living cell it is necessary to select a cell which shows some indication of beginning direct division, i. e., by elongation and constriction. Figures 24 and 25 ajjpcar to be typical of the early stages of direct division of the nucleus.
 
 
Many attempts to trace the changes in such a cell were made, with, however, only partial success, for in almost every case the nucleus lost its constriction and became rounded again, or the cell degenerated. However, one case was found where what ajipeared to be direct division of the nucleus occurred during observation. The various phases are shown in the series of figures 24 to 35, which were drawn at 1.5-minute intervals from a single cell growing in a culture from a 5-day chick heart in Locke solution with extract from chick embryo. The culture was of 57 hours' duration. A cell was first selected which contained an elongated nucleus with a marked notch in one side. In this notch the centrospherc was situated, and conse(juently this side was somewhat indistinctly outlined (24) . Instead of dividing, the cell si raightened out, almost losing the indentation (25) . It contained two nucleoli, one situated in the uppermost jiole, and tlu> other, which was imired, about the efjuator. The nucleus next became rounded (2(; and 27) and, after one hour's observation, its outline was almost circular (2S). In the latter figure there appeared to be only a single piiind nucleolus.
 
 
 
 
 
binucli:atk cells in tissue ctltires. 83
The nucleus now })ccame elongated and a refractive mass appeared in the lowermost pole- apparently another nucleolus: at the same time the central nucleolus Ix'oamc a single mass, and was somewhat longer than before (29). Xext, a shallow notch formed in one side, and the nucleus became shorter and thicker, its nucleoli undergoing minor changes (30 and 31). At the end of two hours the nucleus again elongated and a deep notch appeared, indistinctly marked out on one side (32). This .seemed to become shallower in 33, but the presence of the centrosphere prevented this portion of the nuclear membrane from being well defined.
 
 
The next change was the formation of another notch on the opposite .side, both notches forming what seemed like a zone of constriction about the nucleus. A refractive mass stretched across the equator of the nucleus between the.se notches (34). This is ai:)parently a strand of mitochondria rather than a nucleolus, for, in the next drawing (35), 2f hours after the observation began, this strand is situated between two apparently separate nuclear portions, the nucleus having divided directly. In no fixed and stained cell has a nucleolus been seen to occupj' this position; on the other hand, mitochondria have frequently been seen between these nuclear parts, as in figure 8. There was here no evidence of the formation of either a spireme or an amphiaster. and thus AA'ilson's (1900) criterion for amitosis was fulhlled. It may also be noted that the centrosphere did not divide and the nuclear membrane remained intact.
 
 
The final division apparently took place very rapidly, since the actual separation was completed in the lo-minute interval between 34 and 35. This rapidity of the end process of nuclear cleavage accounts for the infreciuency of such terminal constricting forms as figures 6 and 8, and makes the relatively small number of these later transitional forms adequate to account for the number of binucleate cells which originate therefrom. The cell was allowed to remain on the microscope stage all night, but unfortunately wandered away and was lost, so the subsequent changes could not be followed. The drawings were made from direct observation, but not with the aid of the camera lucida. Mitochondria and centrospheres are partially diagrammatic. This process, though traced with difficulty, and though somewhat obscure, seems to follow the classic descriptions of amitotic division of the nucleus, viz, elongation with eciuatorial constriction, forming a somewhat dumb-bell shaped figure, and final separation of the two nuclear portions.
 
 
A similar elongated nucleus in a comiective-tissue cell was followed for 6| hours, and did not divide, but finally degenerated; in the meantime it underwent various changes in shape and was rounded when last observed. The changes in nucleoli were similar to those in figures 24 to 35.
 
 
Thus it appears that a nucleus in a condition of elongation and constriction may remain undivided for a long time and may even return to the rounded form without dividing at aU. In cases, however, where the constriction has passed a critical point, as apparently was the case in the nucleus represented in figure 34, the iirocess of divis-on proceeds ra])idly.
 
 
The study of fixed preparations, too, throws some light on nuclear amitosis, for in these one frecjuently finds nticlei evidently undergoing direct division. Such
 
 
 
84 BINUCLEATE CELLS IN TISSUE CULTURES.
 
 
forms are to be regarded as transitional stages between the mononueleate and binucleate cell. Figure 2 shows a nucleus which has undergone elongation and equatorial constriction, so that there is an indentation on either side. The nucleolus appears to be dividmg also; this condition of the nucleolus is, however, not constant. Figure 3 shows a cell in which constriction is somewhat farther advanced; here the luicleoli have ajjparently di\ided, two being seen in each nuclear portion. In these cells the method of jireparation does not show cytoi)lasmic details.
 
 
Figures G and S show nuclei in which direct division is almost complete, the nuclear parts being held together onh' by the finest filament. Similar nuclear figures were found bj^ Maximow (1908) in embryonic rabbit tissue, as shown in his figvu-e 1, and the upper two nuclei in his figure 10. In figure 8 the nucleus has divided unequally, and the larger portion contains two nucleoli, while the smaller has but one. In figure 6 only one nuclear portion contains a nucleolus, and this is single. In both cells the unchanged centrosphere is situated in its characteristic position between the two nuclear portions, while the mitochondria radiate out from this body, and a strand of mitochondria passes over the bridge connecting the nuclear parts.
 
 
There is nothing in the appearance of these nuclei to suggest the late telophase of an intranuclear mitosis, such as those shown by Gary (1909) and referred to by Richards (1911, p. 158). The clearness characteristic of the cells of tissue cultures prevents confusion of nuclear amitosis with the late telophase of mitosis, such as has been shown by Richards (1909) to be possible in the cells of Tcenia.
 
 
In figure 9 nuclear separation has been completed, the two portions being quite free from one another. These are of about equal size and appearance, and each contains two nucleoli. Mitochondria and centrosomes occupy their typical positions; the former are short rods, this being a cell from heart membrane.
 
 
In figure 7 the separate nuclear parts have come together and their surfaces are just touching. Mitochondria have been forced out, but the centrosphere is characteristically opposite the area of contact of the nuclear portions. Figure 4 shows a somewhat similar binucleate cell, from a Zenker and Mallory preparation.
 
 
It would appear that the nucleus may sometimes divide by a gradually deepening cleavage from one side, which finally cuts it into two pieces. This may be regarded as an asymmetrical type of constriction. Figure 5 may be taken as representative of the beginning of this jirocess and figure 6 the end. The centrosi)here is found tyi)ically in the notch, as has many times been recorded in amitotically dividing imclei, as by Maximow (1908). In the rare exceptions to this rule the centrosphere may have been originally situated in the notch and subsequently have left it. No evidence of sei)aration of the centriole-pair during nuclear amitosis has been found.
 
 
Richards (1911, p. 150) finds constricted and indented nuclei in his material only in cases of imjK'rfect fixation. AN'hatever may l)e said as to the nuclear distortion brought about by many fixatives, this is not an explanation of such figures as G and 8 seen in tissue cultures, for here osmic-acid vapor was used as a fixative and this does not change the nuclear outline, as may be proved by observing a living nucleus and the .same nucleus after fixation (Lewis and Lewis, 1915). Then, too, only a small proportion of nuclei ajipear thus, whereas if the appearances were to
 
 
 
BIMCLEATE CELLS IN TISSUE CULTURES. 85
be interpreted as due to the fixative they should be abundant. Again, the actual observation of such nuclei in living cells is proof absolute that they are not artifacts.
 
 
The only tyi)e of nuclear fission which I have observed in tissue cultures is that which occurs, apjjarently, bj^ constriction.
 
 
An estimate of the frequencj'^ of occurrence of such transitional amitotic nuclear forms as those shown in figures 2, 3, 6, etc., was made by making careful counts, the aforementioned series of 20 heart cultures being used. Out of a total of 41,725 cells in this series, 50 cells were found to contain constricted nuclei of such a character as would warrant their being considered as amitotic. This is a proportion of one amitotic nucleus to 835 ordinary nuclei, or 0.1198 per cent. In the same series there were 375 binucleate cells, which are regarded as end products of nuclear amitosis. The proportion of transitional forms to end products is thus 50 : 365, 1 : 7.5 or 13.33 per cent. So high a percentage of transitional forms seems to indicate that the nuclei remain a long time in this condition, and the observations upon living cells bear this out. The final stages of direct nuclear fission, as shown in figures 6 and 8, are, as has been noted, rarely found.
 
 
In this connection it is of interest to compare the incidence of amitotic with that of mitotic nuclei. In the same series there w'ere found to be 170 cells undoubtedly in mitosis. The ratio of mitotic cells to total cells is thus 170:41,725, or 1 : 245, or 0.4 per cent. There were probablj^ many more mitoses than this, for some are undoubtedly rubbed off in preparation, since their rounded and thickened form exposes them to friction in washing, etc.; also 62 doubtful mitotic forms were not included.
 
 
It is an eas3' matter to calculate the relative proportion of amitotic to mitotic forms. As has been stated, the ratio of amitotic nuclei to total cells is 1 : 835, while that of mitotic nuclei to total cells is 1 : 245. It is evident that the mitotic forms are 3.4 times as numerous as the amitotic, even when we leave out the doubtful forms and the cells in mitosis which have been rubbed off. Again, when we consider that the amitotic j^rocess is a slow one, as has been shown, and that mitosis is relatively rapid (If to 2h hours according to Lewis and Lewis, 1915, p. 371), it will be realized that the amitotic method of nuclear division is unimportant, so far as nuclear multiplication is concerned, as compared wath mitosis.
 
 
Thus, examination of living and fixed preparations makes reasonable the view that direct division of the nucleus occurs where this structure is elongated, and sometimes bent upon itself, bj' a karyoplasmic streaming, away from the nuclear equator, and a gradually deepening constriction wliich encircles the nucleus more or less symmetrically and cuts it into two parts, the constricted area becoming a narrow tube and finally a thread, which ultimately disappears.
 
 
The behavior of the nuclear membrane during amitosis in the cells of tissue cultures seems to be essentially the same as that of the same structure in the cells of the trematode described by Cary (1909) during intranuclear mitosis. There is, however, no intrannolear spindle in the cells which I have examined.
 
 
The final separation of the constricted nucleus takes only a short time, as has been noted, but a nucleus may remain for a long time apparent^ about to di\'ide without actually doing so.
 
 
 
 
 
86 BINTCI.EATK ('1:1,1,8 IX TISSUE CrLTl'RICS.
 
 
In the process of direct division of tlie nucleus various factors maj' play a ])art. First, we may refer the different changes in form of the nucleus to changes in form of the cell as a whole. It is of frequent occurrence that a cell, by reason of the tension exerted by attached cells, or of its own amoeboid movement, becomes elongated. In consetjuence of this stretching of the cell the nucleus also becomes drawn out, it being simj^ly a sac of fluid, and it is )K)ssible that it may b(>come broken into two parts much in the same manner that an oil globule, floating upon water, becomes broken up if stretched. It may be assumed that there is a streaming of protoplasm away from the equator, with a constriction in this region, which becomes deeper and deeper until the nucleus is divided into two more or less equal liortions, these now^ tending to assume a more globular shape. This view of the cause of nuclear amitosis is somewhat similar to that of Maximow (1908), who belie\es that amitosis in the mesenchyme cells of developing rabbits may be brought about by the stretching of such cells consecjuent upon rapid growth of the adjacent liver.
 
 
The process of direct division of the nucleus as described is strikingly like the division of the cytoplasm of ova which had been replaced in normal sea-water after having been treated with hypertonic sea-water (J. Loeb, 1906, p. 66, figs. 10, 11, 12. and 13). It aj^pears that here the cell first becomes incut from one side; the protojilasm thereupon streams off in oj^posite directions, forming two globules connected by a narrow isthmus. This soon becomes reduced to a mere thread composed of the attentuated cell membrane, which finally disappears, so that there remain two sacs of protoplasm, quite without connection one with another. The jihysical changes involved in this process seem to be very much like those seen in direct division of the nucleus. Lo(>l)'s figures ar(> very similar to those illustrating nuclear amitosis.
 
 
That the size of the nucleus is not a material factor in this process is seen by the variation in size of the twin nuclei, some of which are quite small. Although a twin nucleus is frequently found in a cell which is not elongated, it may be assumed that such a cell has subsequently changed its form, but that it was extended when the separation of the nucleus occurred. This hyj^othesis would not, however, explain the formation of giant cells, multinucleated muscle-cells, etc., and it does not provide an explanation for the evident activity of the centrosphere and mitochondria in direct division.
 
 
A second hypothesis to account for the separation of the nucleus directly postulates the active participation of the centrosjjhere, or mitochondria, or botli, and here we may a.ssume a i)urely mechanical and a i)urely ])hysic()-chemical activity. It has been noted that the centr<)s])here is found connnonly in the invagination of the nucleus; moreover, its edg(> shows evidence of a curious type of movement — a .slow, indefinite retraction and elongation of the marginal processes — which seems to be associated with movements of the mitochondria. It is possible that, through this mechanical influence of the centrosjihere uiwn the adjacent nuclear membrane, the constriction of the latter is favor(>d and the nucleus ultimately divided, and it is easy to conceive how the niitochondria may assist in this nuclear sei)aration
 
 
 
BINUCLEATt: CELLS L\ TISSUE CULTURES. 87
through their own movements (as they have been described b\- Lewis and Lewis, 1915), since they are tj'picallj' found between the nuclear ])arts when these are separated to any extent (fig. 9), and a strand of mitochondria may even be seen lying across the constricted isthmus of the nucleus, when this has not become completely di\idcd (figs. 6 and 8). This position of the centro.sphere and mitochondria undoubtedly seems to have some significance in separation of the nucleus, and is seen even where the nucleus is dividing irregularh', as in figures 48 to 58.
 
 
The relation of the Netzapparat of Deineka (1912, fig. 3) to the nucleus is .similar to that of the centrosphere as just described, viz, it is found in the cleft separating the nuclear portions. This author, however, does not ascribe to it any function in nuclear .sei)aration. He believes that it surrounds the centrosome.
 
 
The position of the centrosphere and mitochondria maj', of course, be without significance, so far as the actual division of the nucleus is concerned, and it is possible that the relationship of these cj'toplasmic bodies to the amitotic nucleus is purely fortuitous, or, at most, occasioned through their adjustment to conditions of intracellular pressure. The occasional absence of the centrosphere from the cleft (once in each 50 cases as determined by counts) and the presence of a cleft opposite the one in which the centrosphere is found are points which count against this second hypothesis. Again, not all nuclei, in which the centrosphere appears in a concavity on one side, divide directly; indeed, this relationship of centrosphere and nucleus has frequently' been noted and illustrated in cells developing, without nuclear amitosis, in their normal environment. It would seem, therefore, that this relationship, of itself, can not bring about nuclear amitosis.
 
 
In no case has there been noted a ring-shaped centrosphere, like that described by Meves (1891), which encircles the constricted zone in the dumb-bell-shaped, amitotically dividing nucleus.
 
 
The centrosphere and mitochondria may be assumed to act in another way in accomphshing direct division of the nucleus, viz, by bringing about a change in the surface tension of the area of the nucleus to which they are opposed, through the elaboration of a chemical substance, and it may be jiossible to explain direct division of the nucleus upon some such hypothesis as that used b}- Robertson (1909, 1911, and 1913) to account for division of the cell in mitosis, viz, that there is produced in the region of cleavage some chemical substance which lowers the surface tension, such as soap, and that there results, in consequence, a streaming of protoplasm away from the equator, leading to separation of the cell. Robertson postulates a cholin-fatty acid soap, the cholin being derived from the splittmg-up of lecithin. Since it has been shown that mitochondria are lecithinoid bodies (Cowdry, 1914, p. 18) it is not beyond the range of possibihty to assume that they maj- act in the formation of a cholin soap. Indeed, the relation of mitochondria to the production of cholin in nerve-cells has recently been discussed by Cowdry (1915). The position of the mitochondria, lying across the zone of nuclear constriction (fig. 8) is eminently favorable for the action of such a soap, should it be formed there.
 
 
A third theory to account for direct division of the nucleus is based upon the assumption that some intranuclear change inaugurates the process. As long ago
 
 
 
88 BINICLICATK CKI.I.S IN TISSUK ClI/riUKS.
 
 
as 1855 and 1858, Romak set forth a theory to account for the division of the cell, which may be stated in the words of Wilson (1900, j). 63) as follows:
'•Cell-division proceeds from the center toward the jieriphery. It begins with the division of the nucleolus, is continued by sinii)le constriction and di\'isi()n of the nucleus, and is completed by division of the cell-body and membrane."
A type of di\ision which bears a close resemblance to this has recently been described by Howard and k""chultz (1911) in the cells of a giant-celled sarcoma from the human oesophagus. To this type of division Hchultz (1915) has proposed the name "promitosis," and these investigators believe it to be intermediate between amitosis and mitosis. This form of cell division seems to have an interesting jiarallel in certain protozoa, and they regard it as a reversion to a primitive biological condition in which the division sjihere is permanently intranuclear — an idea analogous to that of Wieman (1910, p. 175) for a similar form of nuclear division.
 
 
The first step in the division of the nucleus here is taken to be a separation of the karyosome into two or more parts, of equal or unecjual size, followed by a breakingup of the nucleus into portions corresponding in number and size w'ith the fragments of the karyosome, each nuclear part coming to contain a portion of the latter. This function of the karyosome in initiating division of the nucleus is analogous to that of the centrosome in mitosis.
 
 
This form of nuclear division is essentially the same as that described by C onklin (1903) in the foUicular ei)ithelium of the common cricket, and that it is by no means infrequent is gathered from the numerous references to it which this author has found in the literature. Conkhn, however, has never seen actual cell division following nuclear amitosis, and from the fact that the cells in which direct division of the nucleus is found speedily degenerate after the egg is laid he believes that it is, in the material examined, "one of the last functions of these cells and that it is therefore an accompaniment of cellular senescence and decay." Conklin, however, believes that in most cases of amitosis the nucleolus does not divide.
 
 
The evidence from tissue-culture cells does not lend much support to a hypothesis ascribing to the fission of the nucleolus the initiation of nuclear division; true, we have in figure 2 a nucleus which shows lateral constrictions at the equator, and within it, lying with its long axis parallel to that of the nucleus, is an elongated karyosome, which also appears to be undergoing division in the same plane as the nucleus. This somewhat resembles the nuclei described by Howard and Schultz; the kar>'osomes of tissue-culture cells, however, are decidedly simpler in structure than those of the cells of the giant-cell sarcoma. Again, the fact that in the binucleate cell each nuclear i)orti()n is usually su])plie(l with one or more karyosomes seems to ])oint to this body having been divided before or during the division of the nucleus; but against this circumstance, weighing in favor of the view that the division of the karyosome acts to excite direct nuclear division, is the occurrence of such division where the karyosome has evidently not divided (fig. 6), since it is present in only one of the nuclear parts. Such nuclei are not uncommon in tissue cultures. The peculiar condition of the nucleolus in figure 2 may thus be ])ur('ly accidental, since it is not at all constant.
 
 
 
 
 
BINUCLKATE CELLS L\ TISSUE CULTLUES. 89
The (li\i.si()ii of the nucleolus thus seems to have nothing to do with the separation of the nucleus; indeed, after the nucleolus has divided, the nucleus may not divide at all. It may, however, have to do with the size of the nuclear i>ortions; where these latter are erjual they each contain one or two nucleoli, of about equal size, whereas where they are unequal one portion — usually the smaller — may not contain a nucleolus.
 
 
That direct division of the nucleus maj' take place without prehminary fission of the karyosome in tissues developing normally is evident from the statement of Wilson (1900, p. 115): "In many cases, however, no preliminary fis.sion of the nucleolus occurs; and Remak's scheme must, therefore, be regarded as one of the rarest forms of cell division." It is interesting to note that Hchultz finds evidences of such a simple form of direct division in the nuclei of cells of the .same tumor in which he finds "promitosis."
Summing up, then, the process of direct nuclear fission, it is probable that various factors are involved. Elongation of the nucleus is undoubtedly sometimes followed by its cleavage, and, since it is always present in nuclear amitosis, it maybe regarded probably as an essential in this. The activity of centrosjjhere and mitochondria must also be considered as a factor in equal, as well as unequal, nuclear fission, and this activity is appareiitly made effective by nuclear elongation.
 
 
Fission of the nucleolus, while possibly concerned with the relative size of the nuclear parts, is not necessarily associat(>d with the initiation or carrying out of luiclear cleavage.
 
 
Inasmuch as binucleate cells, and constricted nuclei which must be regarded as their precursors, are found in apparently normal embryonic tissue, they can hardly be considered as abnormal or as evidence of a reversion to a more primitive tj'pe of cell division; furthermore, their healthy condition is manifest from their capacity to divide b}' mitosis, as will be shown hereafter. Thus it is reasonable to suppose that the factors operative in nuclear division in tissue cultures are those which function in embryonic cells in vivo.
 
 
8ince these binucleate cells seem to represent the first step on the road to certain giant cells it may be concluded that the latter are the result of a repetition of the same processes which bring about the formation of the former. This view is in accord with that of Lewis and Lew'is (1915), p. 391, w-ho state: "These giant cells appear to be formed by an amitotic division of the nucleus without a coincident division of the cytoplasm."
FATE.
 
 
The nucleus having divided directlj', what becomes of it? Obvioush' the most certain method of settling this question is to select a Uving binucleate cell and watch it constantly as it passes through its various changes. This course has been followed with several cells, and the evidence at hand does not show that the cell as a whole divides otherwise than by the regular process of mitosis; in the earh' stages of this process there is a combination of the two nuclear portions to form a single mitotic figure.
 
 
Plate IV is a series of camera-lucida drawings representing successive stages in the historj' of one of these twin nuclei, in a liA'ing connective-tissue cell, grown from
 
 
 
90 lUMCLEATE CELLS IN TISSUE CULTURES.
 
 
a 7-day chick heart in glycosaliiie with autogenous embryonic extract, the culture lieing 19 hours old when the observation commenced.
 
 
At 11'' oo'" a. m., when the observation began, the nucleus (fig. GO) was seen to be composed of two portions, approximately equal, separated I)}' what appeared to be a single membrane, but what really represents, as has been shown, the apposed areas of nuclear membrane of the two portions. This double partition was seen, by focusing at different levels, to be a plane surface. The first three drawings show roughly the api^earance of such a double nucleus during life. The parts are of about the same size and each at first contains a single nucleolus. These latter undergo obvious changes in size, shape, and number. There is a single centrosphere (c). Fat globules are numerous, and the mitochondria are thread-like and ]3lainly visible, and show their characteristic movement.
 
 
The nucleus remained in much the same condition, undergoing minor changes in outline, for about 2 hours, when, at li" 50"" p. m. (63) the division between the nuclear parts was seen to become less clearly defined at one side and, gradually, refractive material from the nucleus accumulated in this equatorial plane until, at 5'' 05"' p. m. (65), there was a distinct refractive mass in this region, which was evidently chromatin. Soon the entire cell began to contract, to become rounded, and to draw in its processes; the nuclear outUne became indistinct, the position of the nucleus being represented by a clear si)ace surrounded by a ring of fat globules and mitochondria (66). Bj- focusing up and down it is seen that the cell is much thicker than before — in fact, it is almost spherical, the mitochondria and particles of fat forming a hollow globe which incloses the nuclear space. The portions of the twin nucleus have quite evidently fused and (from our knowledge of mitosis) it is plain that the cell is now in the prophase. A spireme, however, could not be made out. The refractive material which had been seen between the nuclear portions has become indistinct. This stage was seen at 6 p. m.
 
 
If we could see the cell represented in 65 in the fixed and stained condition we would doubtless find something like figure 22; here the spireme is forming in a binucleate cell and the nucleoli are becoming smaller and are breaking up. It is evidently composed of two such nuclei as are seen in figure 14, an earlj' prophase in a mononucleate cell. The accumulated chromatin in the plane of contact of the two nuclear portions is clearly evident; this is obviously not the equatorial plate of mitosis. The nuclear membrane has almost disa])peared, but the chromatic material is somewhat more concentrated about the perijjhery.
 
 
Figure 23 evidently rei)resents a somewhat later stage of spireme formation in a double nucleus. Here the skein is well marked and the nuclear membrane has completely disajjpcared. These figures bear a striking resemblance to figure 6 of Kuba.schkin (1905), in which he .shows a sjjireme in a double nucleus.
 
 
The stage rejjresented in 66, if fixed and stained, would probably resemble figure 19, <lrawn from a mononucleate cell in the late i)rophase. Fronj this point on the behavi(jr of the combined double luicleus is identical with that of an ordinary single nucleus.
 
 
As the cell was watched it was seen that a line, refractive in character, formed across its eciuator; this line, represented in 67, was somewhat irregular in outline, its borders being serrated. Tl di<l not remain unchanged, but on the contrary
 
 
 
BINUCLEATE CELLS IX TISSUE CULTURES. 91
showed almost constant minor variations in contour; it seemed to be composed of a row of small refractive bodies (chromosomes) iinderfi;oin^ constant, slow, and very shght movements. From this characteristic formation, situated as it was in a diamond-shaped field, surrounded, as before, by a granuhir ring of refractive globules and mitocliondria, the metaplia.se of mitosi.s was ea.sily recognized. Thi.s stage was drawn at Q^ 50"" p. m. (67) and would appear like figure 15 if fixed and stained. The cell is somewhat smaller and more condensed than that seen in 66, and the appearance plainly indicates that the centrosome has divided and that each part is performing its usual function at a pole of the sjjindle. The actual division of the centro.some was not observed.
 
 
After a short time the plate was seen to split, and the two halves, retaining their parallel relationship to one another, moved to opposite poles of the cell, and there remained, thus marking the anaphase. Figure 16, from a fixed preparation of a mononucleate cell, represents this stage. Almost immediately thereafter the granules and fat globules midway from the poles of the cell were seen to move inward as though a constriction were occurring about the nuclear area at this zone; the result was a dumb-bell-shaped mass within the elongated cell, formed of the nuclear area and surrounding protoplasm. Almost at once the cell membrane itself was seen to be undergoing constriction at this point, as shown in 68, at 7'' 05"" p. m. At the same time the nuclear areas at either end of the cell commenced to become free from granules of fat and other refractive material and the cell outline became larger, showing that the cell was flattening out and that the daughter nuclei were becoming reconstituted in the telophase.
 
 
That the intracellular pressure is considerably increased during this process is shown by the bulging outward of certain portions of the cell membrane, as illustrated in 68, to form bubble-hke protuberances. Frequently the granules and fat globules may be seen to rush out into these evaginations, indicating the formation of cell currents, where pressure has been suddenly released, through giving way and stretching of localized areas of the cell wall. These protuberances soon flatten out, lie close to the cover-slip and expand, becoming armed with hj'aline borders possessed of amoeboid movement (Harrison, 1913, p. 67). The end of the ceU opposite the connection with the daughter cell thus appears fimbriated, as shown bj' Lewis and Lewis (12c, figs. 8 and 10). These refractive borders act as p.seudopodia to anchor the cell to the cover-slip and to drag the daughter cells apart.
 
 
The reformuig nuclei, now more widely separated, and showing wider and clearer areas in the cell protoplasm, are seen in 69 at 7'' 25"' p. m., and at this time the cell was very much constricted, with the nuclei more widely separated. The constricted zone is somewhat more higiily refractive than the surrounding tissue and resembles a short thread. Here also the cell processes are seen to be feeUng their way outward and to be pulUng the two daughter cells apart. The stage corresponding to this in the fixed preparations is shown in figure 17; here the chromatin is a closely clumped, darkly staining mass, and the individual chromosomes are becoming resolved into smaller granules. These subsequently become scattered, and appear in the later definitive, more lightly staining, nucleus as in figure 18. A marked expansion of cytoplasm is here to be noted.
 
 
 
 
 
92 niNlCLEATK CKLLS IN TISSUE CULTURES.
 
 
There have thus been formed two seiiarate and distinct daughter cells, in each of which the nucleus is becoming gradually reconstituted. As the cell was watched the nuclear areas became clear and the membranes distinct; nucleoli also appeared, two in each nucleus. Separation of the cells continued, their hyaline borders becoming very active, stretching away into the outlying media and writhing in a sluggish, eel-Uke manner. Soon the fat globules took up their characteristic arrangement in the cytoplasm, mitochondria appeared, and, in 70, at 8 p. m., 8 hours after the observation commenced, we have to recognize two cells, a])parcntl}' normal, each with its own centrosome.
 
 
The process of mitosis was identical with that followed many times in mononucleate cells, except for the variation in the introductory stage, occasioned by the formation of the spireme from two nuclear parts instead of one. The various stages of mitosis, as it is found in the mononucleate cell, are well shown in the series, figures 14, 15, 16, 17, and 18, selected from a fixed preixiration.
 
 
I have been unable to ascertain whether such si)indle formations arising from the fusion of two nuclear portions are possessed of a double numljer of chromosomes, but the apparent identity of the mitotic process, after nuclear fusion has taken place, w'ith that occurring in mononucleate cells, does not suggest any material variation in the chromatin arrangement. I am in agreement with Maximow (1908), when he says regarding similar si)iremes (p. 95) : '" Aus diesen Spiremen entstehen immer regelmassige normale Mitosen."
These cells were not followed farther. The history for the jjeriod of 8 hours, however, shows conclusively that spiremes from these double nuclei may combine to form a single equatorial jDlate and division may occur by the ordinary mitotic process. That such mitosis occurs in all cases it is impossible to state from this isolated observation, but the presence of double nuclei (with spiremes like those shown in figures 22 and 23) here and there in the fixed jireparations no doubt i)oints to the occurrence of such nuclear fusion as a part of the ])rocess of mitotic division in the binucleate cell.
 
 
Cases have not been found where one portion only of a bipartite nucleus was in a condition of mitosis; hence it seems reasonable to conclude that both parts are always involved in the process. This much is demanded by our conception of the I)otential unity of the double nucleus, so far as its reproductive capacity is concerned.
 
 
In the case of the cells from which figuri>s 22 and 23 were drawn, it may be argued that these represent telophases in which the daughter nuclei failed to separate. Many mononucleate cells have been followed entirely through the mitotic process, and failure of the daughter nuclei to separate has never been noted. Again, in figure 23, drawn from an iron-hematoxylin preparation, there is onlj^ a single centriole-pair, not two, as would be the case in a teloi)hase.
 
 
It might even be suggested that such daughter miclei have reconibined, as observed by Kite and Chambers (1912); here, however, artificial conditions were existent, since the cells were being forcibly separated in the Barber moist chamber by mechanical means. Moreover, entire absence of constriction of the cytojilasm, as would occur in the telophase, points to the condition we are considering as representing the i)rophasc. More than this, the fact that the process has been followeil
 
 
 
BINUCLEATE CELLS IN TISSUE CULTURES. 93
in the living cell, from resting twin nucleus through mitosis to two separate and distinct daughter cells, would seem to be proof absolute that these figures 22 and 23 (which represent a phase of this process) are prophases of combining double nuclei.
 
 
The mere contact of two sjjireme-bearing nuclei (such as ajijiear in figure 23), is of itself no evidence that they will combine, but when we bring to bear upon the interpretation of such a figure the evidence derived from a series such as that shown in plate iv, in which a nuclear formation, like that of figure 23, represents a stage, it seems obvious that these nuclear parts are undergoing fusion to form a single plate of chromosomes. Harman (1913) shows several figures of such nuclei in early cleavages of Tcenia teniaiormis and Moniezia (her fig. c, plate 8), but here the separate nuclei have arisen by mitosi.'^, according to her observations, and cleavage, which is delayed, will eventually separate the blastomeres. The nuclear membranes are here quite intact, and show no evidence of beginning dissolution.
 
 
It may be objected that the condition of spireme is no indication that mitosis is beginning. To this the reply may be made that in the cells of living tissue-cultures a nucleus showing a spireme of this kind, no companion cell in the same condition being present, always represented the prophase of mitosis.
 
 
Since mitosis occurs in binucleate cells in vitro, it might be assumed that it would also occur in such cells in vivo, and indeed this is the case, for ^laximow (1908) has found figures in fixed preparations from the mesenchyme of embrj^o rabbits which strongly resemble those just described. In his figure 7 (p. 93) the spireme is forming in a dumb-bell-shaped nucleus, and in his figure 8 the nuclear fragments in which the spireme is found are quite sei:)arate. Maximow believes that his pictures represent the prophase of normal mitosis occurring in amitotic nuclei; this belief is supported by nw observations upon the living cell shown in plate IV. In his figure 8 he finds the centriole-pair situated between the two coils of the spireme — a position corresponding to that characteristic for it in the amitotic nucleus, viz, in the cleft. In my figure 23, which is shghtly later, the centrosome has shifted its position to the pole. He states that his results resemble the findings of Karpow (1904) for urodele amphibia; this latter author described a process of nuclear amitotic division, with subsequent formation of a spireme from the fragments (which may be two or more in number), with fusion to form one "mutterstern." It may therefore be concluded, from the finding of such double spiremes in embryonic tissue, that this process of mitosis in binucleate cells occurs in normal development. It is thus to be found in differentiating as well as non-differentiating cells.
 
 
I regret that I have seen no other living examples of combination of the parts of a double nucleus during mitosis, but the process is so rare that its observation thus is largely a matter of chance hitting upon a favorable cell. ]\Iitosis occurs rather infrequently in the mononucleate cell, and when it is considered that the proportion of binucleate cells to total cells is very low (1 to 111) the remoteness of the possibiUty of finding a binucleate cell which will divide by mitosis may be realized. It is onlv in those cultures showing abundance of both binucleate cells and mitotic figures that there is any hope of finding such compound mitoses.
 
 
To ascertain the relative frequency with which mitosis occurred among the binucleate cells, as compared with the mononucleate, a study, by careful counting
 
 
 
\H BINTCLEATE CELLS IN TISSUE CULTURES.
 
 
ami clat^sification of cells, was made of the 20 preparations from chick heart mentioned before. In these estimates only the piophases were counted, since it is imjiossible to say of the cells in the later stages of mitosis whether they arose from a monopartite or bijjartite nucleus. Degenerate cells were omitted, and also the area close to the original piece was not counted, since the cells here were usuallj^ too small and closely i)acked to be seen clearly. Nuclei with more than two i^arts of e(|ual size were rare; such were grouped with the binucleate cells in this estimation.
 
 
It was found that there was a total of 41,100 mononucleate cells (excluding the later mitotic and amitotic forms); of these 47 were in the proi)hase of mitosis, or 0.114 per cent of the total. In the same series there was a total of 375 binucleate cells, 2 of which were in the prophase, or 0.53 per cent.
 
 
In spite of the raritj' of occurrence of binucleate cells in jjrophase (there being only 2 in a total of 41,725 cells) it will be seen from this result that mitosis occurred even more fretiuently among the binucleate cells than among the mononucleate — ■ in fact, 4.65 times as frequently. Thus, while it can not be stated definitely that mitosis with recombination of the nucleus always follows amitotic nuclear division, or, indeed, that it frequently does,, it may nevertheless be affirmed with confidence, even allowing for the limited extent of the observation, that the incidence of mitosis in the binucleate cells is at least as high as that among the mononucleate.
 
 
If, in addition to this division by mitosis which these binucleate cells show, they be considered as also i)roliferating by direct division of the cytoi)lasm, it will l)e readily seen that their rate of ])roliferat ion would then be very much greater than that of the mononucleate cells. The imi)robabilit3' of this excessive multiplication strengthens the negative evidence to be put forward later that there is, in these binucleate cells, no division of the cytoi:)lasm following direct division of the nucleus.
 
 
We have seen that a single mitotic figure can be formed from two nuclear portions, previously separate, but contained within the same cell. It has also been found that the sjjireme ma}' form in a bent nucleus of a shape similar to those undergoing direct division. Figure 20 represents an early spireme in such a nucleus. There is apparently but a single centrosphere, situated in the cleft. Figure 21 shows a somewhat more advanced spireme. The nuclear membrane has disappeared and the chromosomes are more definite. One centrosplu^re is situated above, in the cleft, and there is an indistinct trace of a second in the clear area below.
 
 
We can thus construct a series, from cells taken from fixed specimens, illustrating propha.ses in single nuclei, in double nuclei, and in the intermediate forms connecting these. Figures 14 and 19 show spiremes in single nuclei. In the last are two well-marked centrosphcres, indicating that a spindle is about to be formed. Figures 20 and 21 show the process in intermediate forms, and figures 22 and 23 .show it in the double nucleus.
 
 
In figures 20 and 21 it is reasonable to sui)pose that the amitotic process has cea.sed, since the nuclear membrane has almost or quite disapi)eared, and for the .samerea.sonthejjroce.ssof karycjkinesis, which is so obviously taking place in these cells and in those rej)resented in figures 22 and 23, must in all of these cases be considered as starting up imder circumstances where amitosis of the nucleus was under way, or was completed, rather than as having the i)rocess of amitosis superpo.sed upon it.
 
 
 
 
 
BINUCLEATE CELLS IN TISSUE CULTURES. 9o
Altogether the various forms of the nucleus in which spiremes are found in tissue-culture preparations resemble strikingly the findings of Karpow in the leucocytes of urodele amphibia. This similarity is brought out in the following paragraph from Maximow (1908, p. 95):
"Nun ist es aber nach Karpows Untersuchungen zienilich sicher, class hier die Kernainitose zwar zur Kcrnpolyniorphic und sogar zu sicherer Kernteilung fiihrt, dass sie aber doch keine richtige ZeUvonnehnuig nach sich zieht. Wenn die Leukocyten mit amitotisch zerschniirtein Kern sich teilen , so geschieht dies eben aiif dem Wege der Karyokinese, und aus einem zcrschnurten Kern oder sogar aus mehreren einzelneii, ganz getrennten, durch Amitose erzeugten Kernen entsteht dann eine einzige, gewohnlich regelmjissige, mitotische Figur. Man findet Spireme in ring-, hantel-, rosenkranzformigen Kernen, oft auch zwei oder mehrere cinzclne Kerne in einer Zelle, alle gleichzeitig ini Zustande des Spirems, woraus dann immer ein gewohnliclier ^lutterstern resultiert."
]\Iaximow also shows a spireme in a dumb-bell-shaped luicleus found in his own material, and observes:
"Die tief eingeschniirten, oder auch schon ganz zerteilten Kerne konnen in Mitose treten und man bekommt dann hantelformige Spireme (fig. 7) oder zwei kugelige Spireme nebeneinander in ein und derselben Zelle (fig. 8)."
Thus it would seem that the nucleus enters upon the process of mitosis whenever the stimulus initiating this process occurs, whether rounded, bent, undergoing constriction, or divided into two parts, and in all of these, after the single spireme has been formed, the process is apparently identical.
 
 
The question as to whether or not the cytoplasm of the cell divides following direct fission of the nucleus, to form two separate and distinct cells, has been much discussed by various authors, among them IMaximow (1908), who found — besidesthe cases in which the amitotically divided nuclear portions formed a single combined mitotic figure and divided by karyokinesLs— also instances where such portions simply became separated from one another and surrounded by protoplasm, to form ordinary mononucleate cells. In short, Alaximow believes that, though amitosis of the nucleus may be followed by cell division arising through a process of mitosis involving the directly divided nuclear fragments, yet it can lead directly to cell proliferation without intervening mitosis. As such a method of actual cell multiplication, Maximow believes that amitosis functions in certain areas of the normal developing tissue of the rabbit, and he has found it also in the guinea pig. Furthermore, he expresses the view that cells arising by direct division can later divide by mitosis, but his reasons for the latter assumption are not given.
 
 
On the other hand there are those who oppose this view and beheve that nuclear amitosis is never followed by cell amitosis. For instance, Karpow (1904), according to ^Maximow (1908, p. 89) came to the conclusion, based upon his observations upon the leucocytes of urodele amphibia: "dass in den Fallen, wo richtige Amitose wirklich ^•orliegt, man eigenthch doch nur Kernvermehrung annehmen kann, keine Zellvermehrung." This view is in agreement with the findings of Conklin (1903, p. 671) for foUicular epithelial cells of the common cricket.
 
 
No reliable evidence that fission of the cytoplasm follows that of the nucleus has been found in the tissue cultures examined by me. It is quite true that so-called "paired" cells (i e., cells closely resembling one another in form, staining, etc..
 
 
 
 
 
96 BINTCLKATE CELLS IN TISSUE CULTURES.
 
 
lying side by side, but se])aratcd by cleavage of the cytoplasm) may be jjicked out in the fixed preparations, and it might be urged that such were of amitotic origin. This contention can not be proved, however, and it is more jjrobable, in view of the lack of positive evidence of amitotic division of the cytoplasm, that these cells are either of mitotic origin or have migrated together.
 
 
The problem was attacked by the method of continuous observation of binucleate cells (in which the double nucleus has been shown to arise by direct fission), the object being to see if the cytoplasm would diA'ide, and in this way give rise to two separate mononucleate cells. Several such cells containing twin nuclei were followed, but in every case the cell finallj^ degenerated without dividing, after an observation of shorter or longer duration. As an example, the following may be recorded: In a connective-tissue binucleate cell from an 8-day chick heart of 24 hours' growth, the portions of the nucleus were at first pressed closely together, but after 30 minutes they separated slightly-, as in figure 9, and remained apart for 2 hours, when they again became pressed together. The cell was observed for 11^ hours, and the process of separation and reapproximation of the nuclei occurred four times during this period. There was no trace of cytoplasmic division and the only changes noted were those mentioned — some shifting of position of the nuclei and a slight decrease in size of tlie nuclear parts; the latter is believed to be due to prolonged exposure to light. Continuous change in shape of the cell was followed by change in shape of the nucleus.
 
 
This observation shows conclusively tliat the binucleate cell may remain a very long time without direct division of the cytoplasm, and has been confirmed in the case of other binucleate cells. In living cultures the absence of evidence of direct division of the cytojjlasm, combined with similar absence in the case of fixed preparations, leaves us with no ground for the assumption that such direct division ever occurs. Even granting that cytoplasmic division occurs at all, the process aj^pears to be so long delayed that it can not be of much imi)ortance as a method of cell jiroliferation.
 
 
This view is in accord with that of ( 'onklin (1903, y>. G70), for follicular epithelial cells of the common cricket, but does not coincide with that of Child (1907, c, d, and e), who concluded from this (examination of the cells of Moniezia and other animals that amitosis was a rn])id nu'thod of division which occurred where the stimulus to divide was very great and the supply of nutrition was inadeciuate. Patterson (1908) and others hold similar views. From the evidence which tissue cultures afford, however, I am inclined to agree with Harman (1913, p. 219) that the assum])tion that amitosis is a more rapid method of cell ])roliferation than mitosis is hardly justified.
 
 
Tl\e observation just recorded also shows that the interpretation of ''double'' nuclei (such as those seen in my figures 4, 59, and 60 a.s separate nuclear sacs touching one another) is correct, for the sacs have been seen to move ajiart and afterward to return to their original contact with one another, and to rej^eat this process. As has been already mentioned, the a]ii)osed surfaces of such i^aired nuclei give rise to an appearance reseml)ling an intranuclear jjlate; such a plate has. however, not been found bv me in the cells of tissue cultures.
 
 
 
 
 
WXICI.EATE CELLS IX TLSSUE CfLTLRES. 97
The twin nucleus is, then, to be regarded as potentially a single nucleus, in which the nuclear material is separated into two or more sacs. This nuclear material is not to be considered as in anj^ waj^ equallj^ divided between the nuclear portions, which are by no means daughter nuclei. This view is strengthened by the fact that the centrosome, as has been observed, is single in binucleate cells. Before the cell containing such a single twin nucleus can divide, it seems to be essential, judging from the observations, that the nuclear material should recombine and a spireme be formed from the chromatin material in its entirety.
 
 
It may be asked whether nuclear fusion, in these binucleate cells, ever occurs without an accompanying mitosis. I have seen no evidence of such recombination, either in living or fixed preparations, and regard it as improbable, because (among other reasons) the parts increase in size following their division and the single nucleus, which would result from their reunion, would be unusually large.
 
 
Nothing was brought to light, in the material examined, which would in any waj' support the assumption that there are two distinct types of cell division, amitosis and mitosis, for the tj-pe of amitosis which I have described involves onh' the nucleus, and mitosis was the only process which resulted in the formation of two separate cells.
 
 
These observations upon nuclear amitosis do not i)oint to its being an evidence of cell degeneration, for the cells in which it is found are not highly specialized and do not show any more tendency to degenerate than the other cells of the culture. It is generally assumed that mitosis takes place only in normal cells, so that the occurrence of mitosis in amitoticalh' di^^ded nuclei hardly allows them to be considered as degenerate. So, too, the occurrence of amitosis and mitosis in the same preparation (as in the culture from which figure 2 was drawn), where the conditions under which the cells are growing are apparently identical, militates against the view that the environment is not favorable, for the two processes are going on .side bj' side, and mitosis demands suitable conditions. The statements of ^Yieman (1910, p. 174), "amitosis occurs usually under abnormal metabolic conditions which are unfavorable to normal metabolic processes" and "it can occur under circumstances that make mito.sis impossible, " are out of harmony with his finding of both direct and indirect division side by side in the same field, as shown in his figure 13. This coincident occurrence of mitosis and amitosis has been noted by other investigators.
 
 
The conception of amitosis which I have advanced thus differs radicalh' from that of Flemming (1892 and 1893), vom Rath (1891 and 1895), Ziegler (1891), and Ziegler u, vom Rath (1891). They believed that amitosis occurred in cells which were of a transient character and in those which were very highly specialized or on the way to degeneration; and that in cells of amitotic origin the process of mitosis was not believed to take place. In their scheme the condition which I shall speak of as nuclear fragmentation seems to have a place.
 
 
According to this conception, then, amitosis constitutes simply a change in form of the nucleus without increase in its reproductive capacity, and not an actual cell division; and division of such an amitotic cell occurs only by karyokinesis in which there is a recombination of the nuclear material. If this view be correct,
 
 
 
98 BIMCLEATE CELLS IN TISSUE CULTURES.
 
 
and of uni\ersal ai)plication, it may be i)ossible to reconcile amitosis witli the cliromosonie hypothesis, for, since mitosis would he the only method of actual cell i^roliferation, an uneciual distribution of chromatin material to the daughter cells would not be possible, according to our conception of the mitotic process.
 
 
NUCLEAR FRAGMENTATION.
 
 
A note may here be made regarding a curious form of nuclear division which bears some resemblance to the one just described, but which differs from it in many important particulars. It is known as nuclear fragmentation or unequal multiple nuclear fission, and was found to occur where the conditions for growth were not favorable — for instance, in old cultures, in which the food and oxj^gen supply had become depleted and katabolic products had accumulated (figs. 36 to 47) and in those to which a toxic constituent had been added {e. g., ethyl alcohol, figs. 48 to 58). It thus seems to be a pathological condition and is characterized b.v marked malformation of the nucleus, manifesting itself in lobulation and by a breaking away of these lobules, so that what was formerly a single nucleus comes to consist of two, three, or as many as seven or eight apparently separate pieces.
 
 
The forms in which fragmentation presents itself are various, as maj' be seen by reference to figures 36 to 47, drawn from a 6-day growth from the stomach of a 5-day chick. The nucleus may be but moderately deformed, as in figure 49, where a small bud has become constricted off, or there may be two, three, or more lobes or appendages, as seen in figures 40 and 41. These small fragments are in all stages of constriction, ranging from a blunt, sessile protuberance to a small pedunculated mass, held sometimes by a mere thread, as in figure 46. Extremely irregular forms, as 37, are not infrequent, and completely separated portions, as in 36, 42, and 45, are quite often met with. Each fragment may or may not contain a nucleolus. In the smallest pieces it is absent. In some cases, as in figure 45, if the nucleolus happens to be caught in the constricting zone it may become separated, but this is a rare occurrence. Where the nucleus is lobulated the number of lobules usually exceeds the number of nucleolar portions. The culture shows other evidences of degeneration. The size of the nuclear portion seems here to bear no relationship to the size of the karyosome fragment, as it does in the multiple direct division of the nucleus described by Schultz (1915).
 
 
The extent to which this process of fragmentation may proceed is seen by reference to the fact that 66 per cent of the nuclei were malformed in some way, and 34 per cent were actually fragmented, in ten fields from the preparation from which figures 36 to 47 were drawn. There were no mitotic figures found in this preparation.
 
 
In no case was there found any evidence of division of the cell i)rotoplasm following nuclear fragmentation; on the contrary, a sort of syncytium was formed, in which the cytoplasm was filled with nuclear fragments of varying size. The l)icture jiresented by such a nuclear complex is markedly diff(>rent from that of the giant cell, among the points of differentiation being the widely varying size of the nuclei, their lobulation, and the presence of buds in process of separation from the main nuclear mass. Again, in fragmentation the cytojilasm does not increase, as in the case of the giant cell.
 
 
 
 
 
BIXICLEATE CELLS IX TISSUE CULTURES. 99
The entire absence of division of the cell protoplasm prevents this nuclear change from being regarded as a method of cell proliferation. Again, there is no evidence that such nuclear fragments ever reunite to form a spireme after the manner already described for the ordinary- type of amitotic nucleus; indeed, mitotic figures are absent from such preparations — a fact which seems to indicate that the conditions which bring about fragmentation also prevent karyokinesis.
 
 
The differences which fragmentation presents as compared with the usual form of direct nuclear division may be briefly summarized as follows: The nucleus is of irregular contour, multilobulated, and breaks up into a number of small, unequalsized parts, which fre(iuenth' do not contain nucleoli; the nuclear parts remain small, indicating that they have little or no power of growth, for the total volume of the nuclear substance does not seem to be increased following division. There is no evidence of fusion of the fragments contained in a single mass of cytoplasm to form a single mitotic figure. Finally, the process is found in growths which are existing under abnormal conditions, such as the presence of toxins or a deficiency of oxygen, and such conditions act to prevent mitosis.
 
 
As contrasted with this we find, in the case of the ordinary binucleate or multinucleate cell, nuclear portions of regular contour, few in number (usually not more than two), of almost equal size, eaoh containing as a rule one or more nucleoli. These parts apparently possess the power of growth, for in size they are comparable with the nuclei of the mononucleate cell. The fragments of the "double" nucleus are also able to combine and form a single mitotic figure. These cells are found in normal cultures, in which mitotic figures are frequenth' to be seen.
 
 
Fragmentation is similar to the division which produces the ordinary binucleate cell in that the position of the centrosphere and mitochondria with relation to the nucleus is the same. In figures 48 to 58 these structures will be seen occupying the cleft, as in 55 and 58, or situated between the fragments, as in 50 and 54.
 
 
Nuclear forms of this character are not infrequently found in the htcrature. Glaser (1907) describes an analogous form of nuclear fragmentation which occurs in the degenerating food ova of Fasciolaria lidipa. This he regards as "pathological amitosis" as distinguished from phj'siological amitosis. Child (^1907c, p. 288) speaks of "degenerative amitosis" in starving planarians, stating that these forms "differ in appearance from the amitoses in regenerating tissues;" again (1907c, p. 173) he finds that "nuclear fragmentation is a frequent accompaniment of degeneration."
On the whole, therefore, judging from the prevaiUng views of authors, and from the conditions obtaining in the cultures in which it occurs, it seems reasonable to regard nuclear fragmentation as an evidence of degeneration. These final changes are, perhaps, to be looked upon as an active reaction of the nucleus to unfavorable conditions of its environment, as, for instance, the presence of toxins due to katabolism, or chemical change in the media, or injurious material added to the media, as alcohol, or to deficiency in food or oxygen.
 
 
In this connection it is interesting to note that Lewis (1911) and Miller and Reed (1912) demonstrated that the presence of toxins caused an increase in the nimiber of lobes of the neutropliilic leucocyte in blood of the human subject and also in that of the guinea pig and rabbit. The}' looked upon this increase as a physiological reaction on the part of the leucocyte. WTierry (1913) found that amoebae
 
 
 
100 BINICI-KATE CKI,I,S I\ TISSUE CTLTrRES.
 
 
grown ill oxygen-iioor media showed di^'ision of the nucleus without cleavage of the cell protoplasm, and ^^'ieman (1910) ex])resses the view that lack of oxygen may be a cause of a similar nuclear fragmentation in the material which he examined. Again, Holmes (1914) noted such a fragmentation in tissue cultures kept a week or more without changing the medium; when, however, the medium w^as changed freciuently there was no indication of such nuclear change. Fragmentation was acc()iiii)aiiied by other evidences of degeneration. Here, too, lack of oxygen may be the underlying cause, and the increased nuclear surface due to the change in form and multiple division of the nucleus maj' represent the effort, on the part of the cell, to secure an increased respiratory area.
 
 
The mechanics of nuclear fragmentation is no less complicated than that of true nuclear amitosis; indeed, it is probable that new and obscure factors bring about a change in nuclear outline and division of its substance. The activity of the centrosphere and mitochondria may be regarded as similar to that found in the true form of nuclear amitosis, since their relation to the nucleus is the same.
 
 
Nuclear forms somewhat resembling those just described, but simpler in character, are occasionally seen in apparently normal tissue cultures; e. g., those shown in figures 11, 12, and 13. Similar forms have been described in embryonic tis.sue developing normally, as figures 76, 8 a and c of Child (1904) and some of the figures of ^Nlaximow^ (1908). They appear to be examples of sporadic and simple fragmentation. The fate of these buds is obscure, but is probably degeneration.
 
 
SUMMARY.
 
 
The following general conclusions, based upon tlu^ results of the for(>going investigations, have been reached:
Bl NUCLEATE CELL.
 
 
Incidence: In 20 preparations the binucleate cells made up 0.9 per cent of the total cells apjjearing in the new growth. They were more abundant in membranes growing from the heart than in growths from any other tissue, and in cultures of hearts of 5 days of age than in those from older cardiac tissue. They were also more abundant in new growths from cultures of the .second day than in those of the first; this suggests that some, at least, f)f these cells have arisen in the new growth rather than in the original piece, with subsequent migration into the new- growth.
 
 
The proportion of cells containing amitotic (constricted) nuclei to the total number of cells was 1 to 835; that of amitotic nuclei to bii)nrtite nuclei was 1 to 7.5, and that of amitotic to mitotic nuclei was 1 to 3.4.
 
 
Origin: The paired nuclei of binucleate cells in ti.ssue cultures arise l)y direct division of the nucleus, or nuclear amitosis, without division of the cyto])lasni. This occurred in perfectly normal cells.
 
 
Constriction of the nuclear membrane, from one or both sides, which seems to be associated with a karyoplasmic streaming away from the nuclear equator, was the only mechanism observed in direct nuclear fission, and in this process an activity of the centrosphere and mitochondria, coml)ined with elongation of the nucleus, api)eared to be the ))rincipal factors. The centrosphere does not divide, nor do the cent rosom(>s .s{>parate.
 
 
 
 
 
BIXrCLEATK CELLS IN TISHIE CTI-TrKES. 101
The process of nuclear amitosis is slow, excepting the final stage, whicli is rapid. There seems to be a critical jjoiiit in nuclear constriction; before this point is reached the nucleus maj' return to its original form, but after it has been passed the cleavage of the nucleus proceeds rapidly, and results in two separate nuclear parts.
 
 
Division of the nucleolus is not an essential of amitotic nuclear division ; it may, however, be concerned with the size of the nuclear fragments.
 
 
There is no evidence of a form of nuclear amitosis that depends upon the formation of an intranuclear membrane which subsecjuently splits. Such a structure is simulated by the apposed surfaces of the nuclear membranes of the parts of the nucleus of a binucleate cell, when they are in close contact. Sometimes, also, nucleoli, mitochondria, and inbending of the nuclear wall may resemble such a membrane.
 
 
Fate: There is no evidence that direct division of tlw; cytoplasm follows direct division of the nucleus: thus amitosis is not a method of com])lete cell divi.sion, but is to be looked upon as a change in form of a healthy nucleus.
 
 
The regular process of mitosis may occur in binucleate cells. During this process the chromatin material from both nuclear portions is merged into one equatorial plate of chromosomes, the spiremes, which begin to arise separately in the two nuclear parts, joining together to form the chromosomes. Furthermore, this is the only kind of cell division which was found to occur in binucleate cells: they either divide by mitosis or remain as they are, without fission of the protoplasm.
 
 
The separate parts of the double nucleus have no reproductive independence (though they may have metabohc independence), and act as a unit in mitosis. Hence the reproductive capacity of the bipartite and monopartite nucleus is the same.
 
 
Mitosis occurred as frequently in the binucleate as in the mononucleate cells.
 
 
Nuclear fusion, without mitosis, has not been found to occur.',
GENERAL.
 
 
Mitutiis occurs in a nucleus irrespective of its shape; thus the spireme was found in nuclei of rounded form, in those presenting equatorial constriction, and in those divided into two portions.
 
 
Chromosome hypothesis: Nuclear amitosis is not incompatible with theories of inheritance which assume that the chromosome is the bearer of hereditary characters.
 
 
Giant cells: The binucleate cell seems to be the first stage in the formation of the giant cell, which probably arises by a repetition of nuclear amitosis. This conception does not include the formation of the foreign-body giant cell.
 
 
Nuclear fragmentation was found to occur where conditions for life were not favorable, and was thus a form of degenerative change. Fission of a health}- nucleus (amitosis) must thus be distinguished from fission of an unhealthy nucleus (fragmentation).
 
 
Karyosomes of the cells examined were irregular in shape, underwent continuous change in morphology, size, number, and position, and were made up of numerf)us closely packed masses of gel, each with a core of greater density.
 
 
The centrosphere in the cells examined was a sUghtly concentrated gel containing a centrosome (usually paired). Its border is irregular, and this undergoes continu
 
 
 
102 BINICLKATE CKI.l.S IN TISSIK CILTURES.
 
 
ous change in outline and appears to l)e intimately associated witli adjacent mitochondria.
 
 
Vital dyes: Gentian violet did not i)r()vc to be a true vital dye. While it stained the intranuclear bodies and nuclear membrane, its action was toxic and coagulative, and the cells six>edily degenerated.
 
 
Janus green, in low dilutions, was found to stain mitochondria specifically, but its action was destructive, causing speedy death of the cell, with dissolution of the mitochondria.
 
 
Embryonic cells: Many, at least, of the facts obtained from observation of cells in tissue cultures may be applied to the interpretation of similar cells developing normally in the embryo.
 
 
In conclusion I wish to record my indebtedness to M. R. and W. H. Lewis for the loan of their splendid collection of fixed preparations, and for their valued guidance; also to Dr. F. R. Lillie for the courtesy of a room at the :Marine Biological Laboratory at Woods Hole, where some of the work was carried on.
 
 
 
 
 
LITERATURE CITED.
 
 
 
 
 
Cahrel, Alexis, 1913. Coiitributkins to the study of the mechanism of the growth of connective tissue. Jour. Exper. Med., Lancaster, Pa., xviii, 2S7-299.
 
 
Cahy, L. R. 1909. The life history of Diplodiscus temporatus Stafford. Zool. Jahrb., Jena, Abt. f. Anat., xxviii, 595-659.
 
 
Child, C. M. 1904. Amitosis in Mouiczia. Anat. Anz., Jena, xxv, 545-558.
 
 
. 1906. The development of germ cells from differentiated somatic cells in Moniezia. Anat. .\nz., Jena, XXIX, 592-597.
 
 
. 1907a. Studies ou the relation l>etwcen amitosis and
mitosis. I. Developmentof the ovaries and oogenesis in Moniezia. Biol. Bull., Woods Holl, xii, S9114.
 
 
. 19076 Studies on the relation between amitosis and
mitosis. II. Development of the testes and spermatogenesis in Moniezia. Biol. Bull., Woods Holl, XII, 175-224.
 
 
• •. 1907c. Amitosis as a factor in normal and regulatory
growth. Anat. .\nz., Jena, xx.\, 271-297.
 
 
. 1907i/. Studies on the relation between amitosis and
mitosis. III. Maturation, fertilization, and cleavage in Moniezia. Biol. Bull., Woods Holl, xiii, 138-160.
 
 
. 1907c. Studies on the relation between amitosis and
mitosis. IV. Nuclear division in the somatic structures of the proglottid of Moniezia. V General discussion and conclusions concerning anu'tosis and
. mitosis in Moniezia. Biol. Bull., Woods Holl, xiii,
165-184.
 
 
. 1911. The method of cell-division in Moniezia.
 
 
Biol. Bull., Woods Hole, .xxi, 280-290.
 
 
Churchman, J. W., and D. G. Russell, 1914. The effect of gentian violet on protozoa and on growing adult tissue. Proc. Soc. Exper. Biol, and Med., N. Y., XI, 120-124.
 
 
Conklin, E. G. 1903. Amitosis in the egg follicle cells of the cricket. Amer. Natural, Boston, xxxvii, 667-675.
 
 
CoWDRY, E. V. 1914. The comparative distribution of mitochondria in spinal ganglion cells of vertebrates. -■Vmer. Jour. Anat., Phila., xvii, 1-29.
 
 
. 1915. The structure of chromophilc cells of the
nen-ous system. Contributions to Embryologj-, No. 11, Carnegie Institution of Washington Pub. No. 224.
 
 
Deineka, D. 1912. Dcr Netzapparat von Golgi in einigen Epithel- und Bindegewebszellen wahrcnd der Ruhe und wiihrend der Teilung derselben. Anat. Anz., Jena, xli, 289-309.
 
 
Klemming, W. 1892. Entwicklung und Stand der Kenntnisse ubcr Amitose. Anat. Hefte, 2. Abt., Wiesb., Ti, 37-82.
 
 
. 1893. Morphologie der Zelle und ihrer Teilungsere
scheinungen. Anat. Hefte, 2. Abt., Wiesb., iii, 24131.
 
 
GL.\SEn, O. C. 1907. Pathological amitosis in the food-ova of Fasciolaria. Biol. Bull., Woods Holl, xm, 1-4.
 
 
HAR.MAN, Maky T. 1913. Method of cell-division in the sex cells of T;Enia feniacformis. Jour. Morph., Phila., XXIV, 205-243.
 
 
Hauuison, RossG. 1913. The life of tissues outride the organism from the embryological standpoint. Trans. Congress .\mcr. Physic, and Surg., N. Haven, ix, 63-76.
 
 
Holmes. S. J. 1914. The behavior of the epidermis of amphibians when cultivated outside the body. Jour. Exper. Zool., Phila., xvii, 281-295.
 
 
Howaud, W. T., and O. T. Schultz. 1911. Studies in the biologj- of tumor cells. Monographs Rockefeller Instit. Med. Research, No. 2.
 
 
Kaiii'ow, W. 1904. [Untersuchungen ilber direktc Zellteilung.] Inaug.-Diss. Moskau, 1904. Rrf. in Jahresber. u. d. Fortschr. d. .Vnat. (Schwalbe), Jena, n. F. x, 42.
 
 
 
 
 
Kite, C. L. 1913. .Studies on the physical properties or protoplasm. I. The physical properties of the protoplasm of- certain animal and plant cells. Amer. Jour. Physiol., Boston, xxxii. 146-164. , and R. Chambers, Jr. 1912. Vital staining of chromosomes and the function and structure of the nucleus. .Science, N. Y., n. s., xxxvi, 6.39-6-11. Lambert, R. A. 1912a. Variations in the character of growth in tissue cultures. .\nat. Rec, Phila., vi, 91-108.
 
 
. 1912b. The production of foreign-body giant cells
in vitro. Jour. Exper. Med., Lancaster, Pa., xv, 510-515.
 
 
Lewis, M. R. 1911. The blood picture in tuberculosis. Johns Hopkins Hosp. BulL.'Balt., xxii, 428-434.
 
 
. 1915. Rhjthmical contraction of the skeletal muscle
tissue obserA'cd in tissue cultures. .\mer. Jour. Physiol., Boston, xxxviii, 153-161.
 
 
Lewis, M. R., and W. H. Lewis. 1911. The growth of embryonic chick tissues in artificial media, agar and bouillon. Johns Hopkins Hosp. Bull., Bait., xxii, 120-127.
 
 
, . 1912a. The cultivation of sympathetic
nerves from the intestine of chick embryos in saline solutions. -\nat. Rec, Phila., vi, 7-17.
 
 
, . 19126. The cultivation of chick tissues in
media of known chemical constitution. Anat. Rec., Phila., VI, 207-211.
 
 
1912c. Membrane formations from tissues transplanted into artificial media. Anat. Rec., Phila., VI, 195-205.
 
 
. 1914. Mitochondtia in tissue culture. .Science, N. Y., n. s., xxxix, 330-333.
 
 
1915. Mitochondria (and other cytoplasmic structures) in tissue cultures. ,\mer. Jour. Anat., Phila., XVII. 339-101.
 
 
LoEB, J.vCQVES. 1906. The dynaniics of living matter. New York, The MacMillan Co.
 
 
M.OCIMOW, A. 1908. Ueber Amitose in den embryo
 
 
nalcn Geweben bei Silugetieren. Anat. Anz., Jena, XXXIII, 89-98.
 
 
Meves, Fn. 1891. Ueber .imitotische Kerntcilung in den Spermatogonien des Salamanders und Verhalten der Attraktionssphore bei derselben. Anat. Anz., Jena, VI, 626-639.
 
 
Miller, J. A., and Margaret A. Reed. 1912. Studies of the leucocytes in pulmonary tuberculosis and pneumonia. Arch. Intern. Med., Chicago, ix, 609i-636. Also in Trans. Amer. Climat. .\sso., Phila., 1911, xxvii, 192-223.
 
 
Patterson, J. T. 1908. Amitosis in the pigeon's egg. Aiiat. Anz., Jena, xxxii, 117-125.
 
 
R-\TH, O. voM. 1891. L^eber die Bedeutung der amitoiiscben Kerntheilung im Hoden. Zool. Anz., Leipz., xrv, 331, .342, 3,55.
 
 
. 1895. Ueber den fcinercn Bau der Drusenzellen des
Kopfes von Anilocra me<literranea Leach im Speciellcn und die .-Vmitoscnfrage im Allgemcinen Zcitfclu'. f. wisscnsch. Zool., Leipz., lx, 1-89.
 
 
RiCH-SROS, A. 1909. On the method of cell division in Taenia. Biol. Bull., Woods Hole, xvii, .309-326.
 
 
. 1911. The method of cell division in the development of the female sex organs of Moniezia. Biol. Bull., Woods Hole, xx, 123-178.
 
 
Robertson, T. B. 1909. Note on the chemical mechanics of ccll-di%-ision. Arch. f. Entwicklungsmech. d. Organ., Leipz., xxvii, 29-34.
 
 
-. 1911. Further remarks on the chemical mechanics
of cell-division. Arch. f. Entwicklungsmech. d. Organ., Leipz., x.xxii, 308-313.
 
 
. 1913. Further explanatory remarks concerning the
chemical mechanics of cell division. .-Vrcb. f. Entwicklngsmech. d. Organ., Leipz., xxxv, 692-707.
 
 
 
 
 
103
 
 
 
104
 
 
 
BINICLEATE CELLS L\ IISSIK CTLTrUES.
 
 
 
 
 
KiBAScHKix, \V. 1905. L'cbcr doppelte uiid ijolyiiiorplio Kernc iu Tritonblastomeren. Arch. f. mikr. Aiiat., Bonn. Lxvi. 485-500.
 
 
Hi ssELL. D. G. 1914. The effect of gentian violet on protozoa and on tissues growing in vitro, with especial reference to the nucleus. Jour. Expcr. Med.. Lancaster. Pa., XX. 545-553.
 
 
ScHiLT/. O. T. 1915. Promitosis in tumor cells. Jour. Med. Research. Boston, xxxii (n. s.. xxvir). 257-270.
 
 
Whkhky. Wm. B. 191.'}. Studies on the biology of an amoeba of the Limax group. Vahlkampfia sp. No. I. Arch, f. Protisteuk.. Jena. xxxi. 77-94.
 
 
WiEM.tN. H. L. 1910. A study in the germ cells of Leptinotarsa •^ignaticoUis. Jour. Morph.. Phila.. xxi. 135-21(5.
 
 
 
 
 
\Vii.so.v, E. B. 1900. The cell in development and inheritance. 2 ed. New York. The Macmillan Co., 1900.
 
 
YoiNG. R. T. 1913. The histogenesis of the reproductive organs of Ta?nia pisiformis. Zool. Jahrb.. Jena. .\bt. f. Anat.. xxxv. 355 414.
 
 
ZiKULER, H. E. 1S91. Die biologische Bedeutung der aniitotischen (direkten) Kernteilung im Tierreicli. Biol. C'entralbl.. Leipz.. xi. 372-389.
 
 
. u. O. vo.M Rath. 1^91. Die amitotisehe Kernteilung
bei den .\tthropoden. Biol. Centralbl.. Leipz.. xi.
 
 
 
 
 
EXPLANATION OF PLATES.
 
 
Platk I.
 
 
Fig. 1. .\iea of now growth from Xo. 42, S : 12 : 14 (Lewis (•oUcotion). In the field are six binucleate (a) and one quadrinucleate (b) cells. The material was heart from a chirk which had been incubated for 6 days; growth was of 48 hours' duration, in Locke (0.5 per cent dextrose), fixation by osmic-acid vapor, and staining with iron hematoxylin. (<) shows two young daughter cells, the product of a recent mitosis. The guide-line from (a) terminates in the centrosphere of a binucleate cell. Retouched photograph. X465.
 
 
Fig. 2. Elongated nucleus with bilateral coastriction — the beginning of direct bilateral nuclear fission. The nucleolus is also apparently dividing. This figure, and also Nos. 3, 4, ."), 11, 12, 13, 14, 1.5, 16, 17, 18, and 22 are from No. 14, 9 : 1 : 15 (Lewis). Heart from (3-day chick; grown in Locke (0.5 per cent dextrose) with a little yolk; fixed on third day of growth in Zenker; .stained with Mallory's connective ti.s.sue stain. On account of the technique the cytoplasmic details are not represented. This and following drawings, except figures 24 to 35, were outlined by camera lucida. X 1,012.
 
 
Fig. 3. Elongated nucleus almost comiiletely dividcil; final .stage of direct bilateral nuclear fi.ssion. X 1,012.
 
 
Fig. 4. Nuclear fi.ssion compk>ted; nuclear parts divided and lying in contact. Xl,012.
 
 
Fig. .5. Direct unilateral nuclear fission; initial stage. XI, 012.
 
 
Fig. (). Direct unilateral nuclear fission ; final stage; cell of connective-tissue type; nuclear parts connected only by tlie merest filament; centrosphere between nuclear sacs; mitochondria streaming across the narrow connecting strand. Drawn from preparation No. 2 (Lewis); 7-day chick heart grown for 5 days in Locke (I per cent dextrose); osmic-acid vapor and iron hematoxyUn. X 1,032.
 
 
Fig. 7. Nuclear fi.ssion completed; giowth from heart membrane; cell similar to that .shown in figure 4, but prepared to show the centrosphere and mitochondria; the single centrosphere contains two centrosomes; its position, opposite the line of contact of the two nuclear portions (below and to the right ), is characteristic. No. 42 (Lewis) (see fig. 1). X 1,032.
 
 
I'late II.
 
 
Fig. 8. Final stage of direct bilateral nuclear fission in a cell of connective-tissue type; shows the somewhat unequal nuclear parts joined by a very slender thread, apparently the attenuated nuclear membrane; overlying this are several strands of mitochondria, a similar relationship to that of figure 6; the larger nuclear sac contains two nucleoli; the smaller but one. A large centrosphere, from which many mitochondria radiate, is conspicuous. The entire cell is very thin and shows mitochondria streaming out into the processes. The morphology and arrangement of the mitochondria is characteristic for the connectivetissue cell growing in vitro, at periods other than mitosis. No. 17, 24 : 11 : 14 (Lewis). 6-day chick stomach; Locke (1 per cent dextrose); 3-day growth; osmic-acid vapor and iron hematoxylin. Xl,012.
 
 
l"iG. !t. .\ binucleate cell from heart membrane; the two parts are somewhat separated, and lying between them a single centrosphere and-mitochondria are to be seen; the latter resemble cocci or short bacilli and show the characteristic radia arrangement about the centrosphere. This type of mitochondria is found in cells of membranes growing from chick hearts. No. 42 (Lewis) (see fig. 1). X 1.012.
 
 
Fig. 10. Nucleus of distorted form in cell of connective-tissue type found in a culture growing in a weak alcoholic medium. The nucleoli in this preparation show as aggregations of granules; this appearance of the nucleoli in connective-tissue celLs stained in this way is found when differentiation with iron alum is carried too far. Mitochondria are apparently uninjured. No. 23, 24 : 11 : 14 (Lewis). 6-day chick .stomach grown in Locke (1 per cent dextrose) to which ethyl alcohol had been added to make approximately 1 per cent; 3-day culture; osmic-acid vapor and iron hematoxylin. XI, 01 2.
 
 
Figs. 11, 12. 13. These figures show a simple degree of nuclear fragmentation. They were found in a culture in which the cells were otherwise aiiparently normal. In figure 11 the nucleus is constricted at one end; the larger portion contains two nucleoh of unequal size and irregular shape. In figures 12 and 13 the constriction is farther advanced. No. 14, Lewis (see fig. 2). XI, 012.
 
 
Fig. 14. Prophase of mitosis. Nuclear membrane and nucleoli are disappearing and skein is forming; cell not yet roundeil. No. 14, Lewis (see fig. 2). This, and the four figures which follow it, represent the process of mitosis in a mononucleate cell. All drawn from the same preparation. X 1,012..
 
 
Fig. 15. Metaphase. Cell rounded and compact; processes drawn in; cytoplasm granular and stains very densely with hematoxylin; definite spindle with equatorial plate of chromosomes. X 1,012.
 
 
Fig. 16. Anaphase. The chromosomes have separated and the remains of the spindle may be seen as faintly defined streaks conn^f-ting the two aggregations of chromosomes; cytoplasm still densely gianular and darkly staining, the entire cell contracted; cell-processes small and thread-hke. X 1,012.
 
 
Fig. 17. Early telopha,se. Chromosomes less clearly marked, the chromatin masses breaking up. No evidence of nuclear membranes is to be seen. The cytoplasm is dividing, as shown by constriction about the equator. Markedly granular and darkly staining protoplasm. X 1,012.
 
 
 
 
 
106 BINUCLEATE CELI^S IX TiaSUE CULTURES.
 
 
Fig.
 
 
IS. I-iite telophase. The two (hiUKhter coll.-; are .seen, separated and spread out thinly; protoplasm stains much more lightly; nuclei well defined and contain coarsely granular du-omatin. In each daughter nucleus the beginning of a nucleolus is to be seen; but is very small compared with the size of this body at maturity. X 1,012.
 
 
Fig.
 
 
19. Late prophase of mitosis, in cell probably of connective-tissue type; two centrospheres at opposite poles; nuclear membrane has disappeared; spireme well marked; mitochondria short and thick. Xo. 42, Lewis (.see fig. 1). Figures 20 and 21 are from the same preparation. Xl,032.
 
 
Fig.
 
 
20. Early prophase in a nucleus showing beginning direct unilateral fission; skein forming, nucleoU disappearing; centrosphere .<till single, situated in tlie fi.ssurc; mitochondria becoming .shorter and thicker, and are intermediate in these respects between those .seen in figure S and those of figures 19 and 21 ; nuclear membrane has almost disappeared. Cell of connective-ti.ssue type. X 1,032.
 
 
Fig.
 
 
21 . Late prophase in a nucleus undergoing diioct unilateral fission. Skein has formed and nuclear membrane has disappeared; one centrosphere is to be seen in the cleft, and there is some indication of a second on the opposite side of the nucleus, in the area devoid of mitochondria. X 1 ,032.
 
 
Fig.
 
 
22. Prophase in a binucleate cell. Early stage. Skein is forming; membrane and nucleoli are disappearing. Some chromatin has become segregated in the area of contact between the two parts. The method of fi.\ation and staining does not permit of the centrosphere and mitochondria being seen. Xo. 14 Lewis (see fig. 2). X 1,012.
 
 
Fig. 23. Prophase in a binucleate cell of connective-tissue type; somewhat later stage than figure 22. In each nuclear portion there has been formed simultaneously a skein and tlie nuclear membrane hiis disappeared. The chromatin material of the combined double nucleus will form a single equatorial plate of chromosomes, as in figure 67. Only one centrosphere, containing two centrosomes, is .seen in the preparation, situated at one extremity of the fusing nucleus, it having come from the interval between the nuclear parts. It is thus probable that the spindle will form parallel with the long axis of the fusing nucleus. Mitochondria are short and thick. Xo. 18, 21:2: 14, Lewis. 7-day chick heart, grown for 2 days in Locke (0.25 per cent dextro.se); osmic-acid vapor and iron hematoxylin. X 1,032.
 
 
Pl.\te III.
 
 
Figs. 24-35. \ series of diawings from a living connective-tissue cell made at 15-niinute intervals for 2j hours. The nucleus at the start was elongated and notched at one side. It was seen to take various forms, and ended as two separate nuclear parts. The series thus shows direct nuclear fi.ssion. It will be seen that the centrosphere is contained within the unilateral cleft, and when the nucleus ultimately divides the centrosphere is situated between the parts of the nucleus; mitochondria stream across the interval separating these two parts. The nucleolar bodies undergo interesting changes. The nuclear outlines, position of nucleoli and centrosphere, the cell outlines, and principal features of the cj-toplasm were sketched in freehand from direct ob-siervat ion of the living cell. The drawings were afterwards retouched by reference to fixed preparations. Small circles represent fat globules, and short threads mitochondria, f. figure 24, marks the centrosphere. 5-day chick heart; 57 hours' cultivation, from Xo. 7, 9 : 1 : 15, in Locke (0.5 per cent dextrose) with extract of chick embryo. X about 900.
 
 
Figs. 36-47. Fragmenting nuclei showing probable effect on form of nucleus of prolonged growth in unchanged media; outlines of nuclei very irregular, each has a number of lobes; in some cases separation of these lobes has taken place. Culture shows other evidences of degeneration. X'o division of the cj-toplasm following division of the nucleus was observed. Drawn from various cells selected from No. 23, 12 : 1 : 15 (Lewis). .5-day chick .stomach in Locke (0.5 per cent dextrose). Zenker; Mallory connective-tissue stain. Culture grown for 6 days in the .same media. Xl,012.
 
 
Figs. 48-58. A collection of nuclei of irregular form, grown in media containing alcoliol; centro.spheres are sketched in to show (heir characteristic relationship. Saiiic prepunition us figure 10. XI, 500.
 
 
Fkj. 59. A regular paired nucleas from the same jireparation as (luit from which the series 48-58 was drawn. Some of the nuclei have escaped di.stortion. X 1,500.
 
 
Plate IV.
 
 
Flos. 60-70. .\ .series of camera-lucida drawings from a single living binucleate cell of the connective-tLssuc type, which was observed continuously for 8 hours. .\t the beginning there were two separate nuclear parts, with one centrosphere; the parts combined to form a single mitotic figure, and the successive stages of mitosLs are seen in figures 60 to 70. The ultimate result is two separate mononuclear cells, each containing a single centrosphere. The .series brings out the fact that the parts of the "double" nucleus arc not indepenilent so far as their re])roductive capacity is concerned, bvit in cell division they combine and act as a single nucleus, c, figure 60, represents the centrosphere. 7-day chick heart, grown for 19 hours when the observation commenced. Locke (1 per cent dextrose) with extract of chick embrvo. Culture of March 15, 1915. X 1,500.
 
 
 
CONTRIBUTIONS TO EMBRYOLOGY
 
 
 
Volume V, No. 14
 
 
 
 
 
Published by the Carnegie Institution of Washington Washington, 1917
 
 
 
CARNEGIE INSTITUTION OF WASHINGTON
Publication No. 225
 
 
 
PRESS or OIBSON BROTHERS WASHINGTON
 
 
 
CONTRIBUTIONS TO EMBRYOLOGY, NO. 14
 
 
 
THE DEVELOPMENT OF THE CEREBRO-SPINAL SPACES IN PIG AND IN MAN
 
 
 
By Lewis H. Weed
 
 
 
CONTENTS.
 
 
 
 
 
I. Introductory 7
II. Review of literature 8
The comparative anatomy of the meninges 11
Literature on the development of the
mammalian meningeal spaces 13
III. Methods of investigation 15
rV. Injections and replacements in the cerebrospinal system 20
Results of replacements in the ventricular
system of true solutions 20
The results of injections of true solutions 25 Results of injections of nitrate of silver. . 27
The injection of india ink 29
V. Undeacribed structures in roof of the fourth
ventricle 30
An undescribed area in the superior portion of the roof of the fourth ventricle 31 The area membranacea superior in the
pig embryo 31
The area membranacea sujjerior in the
human embryo 36
The area membranacea superior in
other animals 40
General consideration of the area
membranacea superior 41
An undescribed area in the inferior portion of the roof of the fourth ventricle 43
The area membranacea inferior in
the pig embryo 44
The area membranacea inferior in the
human embryo 47
General consideration of the area
membranacea inferior 50
VI. Passage of fluid tlirough roof of the fourth
ventricle 53
The accumulation of injection-masses in
the superior membranous area 53
The sites of fluid passage through the
roof of the fourth ventricle 54
 
 
 
VI. Passage of fluid through roof of the fourth
ventricle — Continued. Factors concerned in the experimental
fluid passage 57
The passage of silver nitrate and india
ink through the membranous areas
in the roof of the fourth ventricle. . . 60 Relation of the ependymal differentiation
to the passage of fluid 61
VII. General histological difl'erentiation of the
cerebro-spinal spaces 63
The periaxial mesenchyme 63
The formation of the araclinoidea 64
The circulation of fluid through the
subarachnoid spaces 70
VIII. A consideration of the embryonic pia mater. 71 The general histologj' of the pia mater. . 72 The relation of the pia mater to the fluid
channels 73
The adhesion of the pia mater to the cerebral tissue 75
IX . The development of the cranial dura mater . 75 The general process of the formation of
cramal dura 76
The subdural space and the mesothelial
lining of the dura 85
The competency of the early dura as a
cellular membrane 87
X. The return of cerebro-spinal fluid to the
venous system 88
XI. The chorioid plexuses and the elaboration
of cerebro-spinal fluid 91
The development of the chorioid plexuses 91 The glycogen content of the chorioid
plexuses 93
XII. Perivascular spaces in the embryo 95
XIII. The perineural spaces in the pig embrj-o. . . 97
XIV. General summary 101
XV. Conclusions 107
XVI. Bibliography 109
XVII. Explanation of plates Ill
5
 
 
 
THE DEVELOPMENT OF THE CEREBRO-SPLNAL SPACES IN PIG AND IN MAN.
 
 
 
 
 
By Lewis H. Weed.
 
 
 
 
 
I. INTRODUCTORY.
 
 
Probably no field in embryology has been less explored than that relating to the meninges. Our knowledge of the transformation of the perimedulla
 
 
ry mesenchyme into the three fully developed membranes about the cerebro-spinal axis has been largely of a crude sort, with gross generalities based on inexact or incomplete evidence. The present work was undertaken in the hope that by a study of the various stages in the development of the cerebro-spinal spaces there might be gained some knowledge which would afford a basis for a conception of this dynamic metamorphosis.
 
 
Many of the problems centering around the development of the meningeal spaces have recently been expounded by Cushing^^) . * Not only do we lack knowledge as to the method of differentiation of the primitive mesenchyme, but we know little about the establishment of the circulation of the cerebro-spinal fluid. When do the chorioid plexuses begin to secrete? When does the venous absorption of the fluid take place? When does the perivascular system begin to remove waste products from the cerebral tissue? And also, what factors play a part in the formation of the subarachnoid and subdural spaces?
These questions, some of which it is hoped the present study will answer, relate to the field of physiological anatomy. Consideration of the subject, however, serves to convince one that they must be investigated coincidently with the stages of morphological differentiation; for it may readily be conceived that the physiological use of the meningeal spaces may precede any morphological differentiation of the three membranes, nor indeed is it unlikely that one of the active causative factors in the metamorphosis concerns this filling of the mesenchyme about the nervous system with fluid.
 
 
This study, therefore, has been anatomical, but with a broader scope than purely morphological studies would have afforded. Not only has it dealt with the morphological differentiations about the nervous system, but throughout the investigation the relationship of these structures to the possible presence of cerebrospinal fluid has been considered. As the problem developed it was projected more and more into the difficult realm of "tissue spaces." Interest in these spaces largely concerned their physiology, but many points of correspondence between structure and function were found.
 
 
In some measure this work is a development of an earher study of some of the anatomical and physiological problems of the cerebro-spinal fluid, carried out in the laboratory of Dr. Harvej^ Cushing at the Harvard Medical School.
 
 
* The figures in parentheses refer to the bibliography at the end of this paper.
 
 
 
II. REVIEW OF LITERATURE.
 
 
In order fully to understand the problems which confront one in the study of the embr3'onic cerebro-spinal spaces, a comprehension of the stage to which investigations have brought our knowledge of these fluid-pathways in the adult is necessar}'. It is with this purpose that the adult relationships are here considered. The inclusion of this material may be pardoned, for it will be seen that unanimity of opinion has bj' no means existed in regard to any of the problems concerned in the circulation of the cerebro-spinal fluid.
 
 
Modern anatomical knowledge of the meninges dates from the work of Axel Key and Gustav Retzius'29). These Swedish investigators, in their excellent monograph published in 1875, first conclusively demonstrated the anatomical continuity of the spinal and cerebral subarachnoid spaces. But for years after their publication appeared, a physiological continuity between the subdural and subarachnoid spaces was argued for bj' many observers, notably by HilK^*). Gradually, however, workers in this field have reached the opinion that the subarachnoid spaces (the interrupted but continuous channels between arachnoidea and pia) are functionally the channels for the cerebro-spinal fluid. Between the intra-leptomeningeal and the subdural spaces no anatomical connection exists; physiologically there may be some mode of fluid-passage. Thus Hill' 24) states that either by filtration or through actual foramina fluid passes readilj- from one space to the other. Quincke '*^, from observations on animals, somewhat similarly premised a connection between the two spaces, but only in the direction from subdural to subarachnoid. His experiments, based on the results of the injection of cinnabar granules, are open to criticism as indicating a normal passage-way for the fluid; for, as he has recorded, an intense phagocytosis of practically all of his granules occurred. More modern conceptions of the subdural space treat it as a space anatomically closed, lined externally by a polygonal mesothelium. Less error is introduced if it be regarded as analogous in many respects to well-known serous cavities rather than as an essential portion of the pathway for the cerebro-spinal fluid.
 
 
The question of the absorption or escape of cerebro-spinal fluid from the subarachnoid space has claimed the attention of many workers. Since the original conception that the meningeal coverings were actually serous cavities, anatomical investigations have furni.'^hed man}- new views. Key and Retzius, by spinal subarachnoid injection of gelatine masses colored with Berlin blue, demonstrated an apparent pa.ssage of the injection fluid into the great cerebral venous sinuses through the Pacchionian granulations (die Arachnoidzotten). Their observations were made on a cadaver and the injections carried out under fairly low pressures (about 60 mm. of mercury). A le.sser drainage of the fluid into the lymphatics was also shown.
 
 
Since the view advanced by Key and Retzius of the absorption of cerebrospinal fluid, the general trend has been away from the idea of an absorption into the venous sinuses. Quincke's''*^) observations, made on lower animals after the subarachnoifl introduction of cinnabar granules, really offer some substantiation of this theory, but the failure to find the great Pacchionian granulations in infants and in the lower animals caused many workers to reject utterly the conception of the Pacchionian granulations as the functionally active mechanism for the fluid escape.
 
 
Physiological evidence, however, advanced by Hill^24) fj-Q^i intraspinous injection of methylene blue, indicated that the major escape of the cerebro-spinal fluid was into the venous sinuses of the dura, while a slow and minor absorption took place along the lymphatic channels. Ziegler(57j^ with potassium ferrocyanide introduced into the cerebro-spinal space, Ukewise found that the venous absorption was much greater and more rapid than the lymphatic. Reiner and Schnitzler *^> with the same agent detected the ferrocyanide in the jugular blood-stream after injection. With oUve oil these investigators found a similar venous absorption, but with a slowing of the venous blood-stream. Lewandowsky'^^', also using ferrocj'anide. found this salt in the urine within 30 minutes after its subarachnoid injection. Spina 52'^ from observations on freshly killed and hving animals, presented somewhat similar evidence of a major venous and lesser lymphatic absorption. Gushing ^ suggested a valve-Uke mechanism of escape of the fluid, his hypothesis being based on the findings after the introduction of mercurj' into the meningeal spaces.
 
 
Several theories concerning the absorption of cerebro-spinal fluid into the bloodvascular system have more recenth' been offered. Mott '**', from a studj' of dilated perivascular and permeuronal spaces, has advanced the idea of fluid-escape by way of the perivascular system into the cerebral capillaries. Dandy and Blackfan^'"^, from an analysis of their evidence, consider that the chief drainage of the fluid is into the capillaries of the pia-arachnoid. Opposed to this conception of a major drainage of cerebro-spinal fluid into the l)lood-vascular system is the view championed by Cathelin'^', that the Ij'mphatic drainage is the chief method of fluid-escape. Cathelin's contention of a veritable circulation of the fluid has not received support from other workers.
 
 
Thus it will be seen that since the work of Key and Retzius the trend of opinion has been away from the view that the Pacchionian granulations carry the cerebrospmal fluid into the venous sinuses.
 
 
In the earlier investigation' ^^ carried out in the Harvard IMedical School the problems of this fluid absorption were attacked in a somewhat different manner than by previous workers. True solutions of potassium ferrocyanide and ironammonium citrate, such as have been used in the present investigation, were injected into the spinal subarachnoid space under pressures but shghtly above the normal. The animals (dogs, cats, and monkej-s) were kept imder anesthesia during the period of injection, which was usualh- continued for several hours. Complete filling of the subarachnoid channels was secured bj- this technique, pro\-ided the injections were continued for a sufficient length of time. At the conclusion of the experiment the foreign solution was precipitated in situ and blocks were carried through for histological purposes.
 
 
Many of the anatomical findings in this work carried out as outlined are of interest in the present problem. The complete correspondence of the spinal and cerebral subarachnoid spaces as demonstrated by Key and Retzius was amply verified. The normal return of the cerebro-spinal fluid to the general circulation by way of the arachnoidal villi into the great dural sinuses was demonstrated. These viUi are projections of the arachnoidea through the dural wall, prolonged directly beneath the vascular endothelium of the venous sinuses. Furthermore, columns of arachnoid cells were found, normally affording fluid channels in the dura. In addition to the major escape of cerebro-spinal fluid into the sinuses a lesser drainage was alsi) demonstrated, slower than the primary drainage, out along certain of the emergent nerves into the lymjihatic system. No evidence whatsoever was obtained in support of any of the theories of a drainage of cerebro-spinal fluid into either the leptomeningeal or cerebral capillaries, nor could an anatomical valve-like mechanism along the great sagittal sinus be demonstrated. The process of escape of cerebrospinal fluid from the arachnoid villus unto the great sinus appeared to be a simple one of filtration or of diffusion.
 
 
Another of the problems concerning cerebro-spinal fluid, which has been of interest to anatomists and phj'siologists, is the source of the fluid. Haller^^D and Magendie'^^ to whom the greatest credit for work on this subject must be given, believed it to be the product of the leptomeninges. Faivre^^^} {^ 1853 and Luschka^''*) in 1855 were the first to suggest the chorioid plexuses as the elaborators of this circumambient medium. Since then the view has been generall}- accepted that these villous structures do give origin to the fluid, but the early evidence was based wholly on the glandular character of the plexus. Cappelletti^^) and Pettit and Girardi'*^) offered more definite proof of this relationship by the introduction of pharmacological agents which affected the rate of production of the fluid. These latter authors recorded definite histological changes in the cells of the plexus when influenced by these drugs, indicating, in conjunction with the changed rate of production of the fluid, an undoubted relationship of the chorioid plexus to the fluid elaboration. Since these early investigations many observers — Findlay^^^^, jMeek(37)^ Mott^'*^', Pellizzi''*2), Hworostuchin'26)^ and others — have studied the histology of the chorioid plexus with reference to its function as an elaborator of the cerebro-spinal fluid.
 
 
In addition to the elaboration of the fluid by the chorioid plexuses, increments are furnished bj' the nervous tissue itself. This elimination from the nervous system occurs bj' waj' of the j^erivascular spaces. In the previous work referred to'55j it was found possible to inject the entire perivascular system by continuing a physiological injection of the spinal subarachnoid space, and subsequently causing an extreme cerebral anemia. By this procedure an injection of the system to its termination about the cerebral capillaries and nerve-cells could be secured. From this and other evidence the view was advanced that the cerebro-spinal fluid was derived from a dual source — in part from the perivascular system and in greater part from the chorioid plexuses. This view had already been advanced, but on rather insufficient grounds, by Mestrezat'^s* and by Plaut, Rehm, and Schottmuller (*'*>, Recently Frazier^^^) has signified his acceptance of this conception of the source of the fluid.
 
 
Such, then, is the basis for our present understanding of the meninges, in regard to their characteristic morphology and particularly their functional relationship to the cerebro-spinal fluid. Without a consideration of the circumambient fluid morphological studies of these membranes would be incomplete, for in order to understand the meninges knowledge concerning the cerebro-spinal fluid is necessary.
 
 
THE COMPARATIVE ANATOMY OF THE MENINGES.
 
 
Sterzi's^) has pubUshed a comprehensive report of the comparative anatomy of the spinal meninges. From his studies he has advanced hypotheses, supported by observations on a Hmited number of fetuses, regarding the development of the human meninges. On account of the interest of this subject in relation to the present discussion a brief summarj' of Sterzi's work will be here included.
 
 
In the acrania there is no special envelope of the central nervous system, but rather a fibrous sheath corresponding to the meninges of higher forms. This fibrous sheath is largely made up of circular fibers, except in the median ventral line, where there occurs a ventral hgament of longitudinal fibers. In cyclostomes, however, there is found a single "primitive meninx" — vascular and composed of white and elastic fibrils coursing in a longitudinal direction. Some of these fibrils traverse the perimeningeal spaces (filled with star-like cells, with some fatty tissue) and are attached to the inner surface of the vertebrae. This same general plan of a single "primitive meninx" is hkewise found in fishes (elasmobranchs, teleosts, etc.); the membrane here is often pigmented and follows closely the external architecture of the spinal cord. The perimeningeal space is filled by mucus in elasmobranchs, but in teleosts this is replaced by fat. For the most part there are found dorsal and ventral ligaments and two lateral ligaments.
 
 
The next stage in the development of a more complete form of spinal covering is found in the urodele amphibia. A "primitive meninx," formed of two layers, often artificially separated from each other, replaces the simpler meninx of C5'clostomes and fishes. Of the two layers in this membrane the external is thin and free from pigment; the inner, strongly pigmented, adheres to the spinal cord. The meninx is perforated by the denticulate Ugaments.
 
 
In amphibia (Anura) Sterzi found the first evidence of a "secondary meninx," corresponding to the pia-arachnoid. Surrounding this membrane, but separated from it, is the dura, thin and transparent; between the two meninges is the intradural (subdural) space. The dura lies in the peridural space. The spinal prolongations of the endohonphatic canals he in the dorsal part of the peridural space. Both the dura and the "secondarj^ menmx" continue outward along the roots of the spinal nerves and along the filum terminale. Embryologically the perimedullary mesenchyme is differentiated into these two meninges in the Anura.
 
 
This arrangement of the two meninges in Anura is followed out in reptiles. The dura, thin as in the amphibia, is covered by endothehum and is vascular. The "secondary meninx" possesses laterally the denticulated hgaments and ventrally the ventral hgament. Both the peridural and intradural spaces are very small.
 
 
 
Likewise in birds Sterzi was able to differentiate only two meninges — the dura and the "secondary meuinx." These membranes are quite similar to those of reptiles. The " secondary meninx " has acquired three layers — an outer endothelial covering, a middle vascular layer, and an inner membrane closely adhering to the cord. This is a distinct approach to the three meninges of mammals. An intradural (subdural) space covered by endothelium can be easily made out. The development of these avian meninges concerns a differentiation of the perimedulla
 
 
ry mesenchyme.
 
 
The arachnoid, according to Sterzi, first appears as a definite membrane in mammals (marsupials and placentals). In marsupials this arachnoid has become well differentiated and the pia mater possesses denticulated and ventral Ugaments. A transformation of the extradural portion of the denticulated ligaments unites the dura to the endorachis. In perissodactyla the differentiation of the three meninges (particularly of the arachnoid) is incomplete. The arachnoid is separated from the pia mater by a pecuUar tissue wliich contains numerous lymphatic lakes, forming the intra-arachnoid spaces. No intradural (subdural) space is apparent, due to the approximation of dura and arachnoid. The subdural space is clothed by endothelial cells; these can not be made out in the intra-arachnoid spaces. The dura is surrounded by a fatty pad.
 
 
According to Sterzi the augmentation of the intra-arachnoid (subarachnoid) space is the distinguishing characteristic of the meninges of carnivora. This increase takes place at the expense of the peridural space.
 
 
As Sterzi developed the knowledge of the comparative anatomy of the lower forms — of the transition from the primitive meninx of the cyclostomes to the three membranes of mammals — the possible correlation of this analogj'^ to the embryological development in mammals became apparent. He extended his observations to human beings and to human fetuses. His findings will be detailed in the following section.
 
 
Farrar'i*'', in a short discussion of the development of the meninges of the chick, finds in early stages three laminse about the spinal cord, "the middle one of which alone still presents the primitive features of the mesoblastic-sheath." The inner layer, close to the medullary tissue, is highly vascular; in the outer zone "the connective-tissue elements arc assuming elongated forms and crowding together with long axes parallel, giving a very close mesh with long but extremely narrow spaces, in contradistinction to the loose irregular reticulum of the pia-arachnoid." The outer lamina becomes dura mater, while the inner two zones are considered together as the embryonic pia-arachnoid. Farrar defines the pia-arachnoid as developmentally a single membrane consisting of a loose reticulum, at the outer and inner borders of which limiting membranes are formed.
 
 
 
LITERATURE ON THE DEVELOPMENT OF THE MAMMALIAN MENINGEAL SPACES.
 
The development of the meningeal spaces in mammals has not been studied extensivel}', and the literature in regard to it is quite meager. Only a very few workers have touched upon the subject except casually. Reford"*^', working in the Anatomical Laboratory of the Johns Hopkins University, studied the development of these spaces bj^ the method of injection with india ink. His work unfortunatelj' has never been published, but it has been rather extensively referred to bj' Sabin *3' in 1912 and by Cushing'^^ in 1914. Their sununaries of this work are here included. Miss Sabin thus speaks of it:
"In a study of the arachnoid made bj' the injection method in the Anatomical Laboratory of the Johns Hopkins University by L. L. Reford, and as j^et unpublished, it has been shown that the thinning out of the mesenchjine around the central nervous sj-stem is not haphazard, but that injections of the same stage give the same pattern, and that the form of the arachnoid space changes as the brain develops. That is to say, the arachnoid space has as definite a form as the ccelom, and it never connects with the IjTnphatics."
Cushing(^) gives the following summary:
"It was thought that an investigation of the cerebro-spinal spaces in the embryo would most hkely shed Ught on the subject, and some unpublished studies in this direction were undertaken in 1904 and 1905 by Lewis L. Reford in ^Mall's laboratory' in Baltimore. In living pig embryos of various stages low spiaal india-ink injections were made either into the wide central canal or into the subarachnoid space, and the embryos were subsequently cleared. It appeared from the course taken by the injection mass that the full development of the spinal arachnoid preceded that of the intracranial spaces, the impression being gained that the separation of the primitive meninx into its layers occurred later over the cerebral vertex than in the basilar portion of the chamber. Still, I never felt quit« convinced that the failure of injection of the meninges over the surface of the hemispheres in many of Reford's specimens was not due to the floating up of the brain against its envelopes by the introduction of the injection mass from below. Howe\'er this may be, it was nevertheless apparent that a venous injection of the body of the embryo was often produced, and the impression was gained that a communication existed between the basal subarachnoid spaces and the precursors of the sinusoidal veins of the cranial chamber which empty into the jugulars. If due to an artifact from a vascular ruptm-e, at all events the conununication always occurred at the same point. Reford, moreover, in agreement with Cruveilhier, Reichert, and KolUker, came to doubt the existence of the foraminal opening described by Magendie, beUeving that the opening was an artifact and that the fluid escaped by seepage through a persistent membrane."
It is regrettable that Reford's studj' has not been pubUshed, as it represents the onlj^ attempt to solve the problenas of the development of the cerebro-spinal space by the method of injection. As stated in subsequent sections of this communication, his apparent failure to control pressures of injection and to use only granular suspensions is unfortunate.
 
 
In a study of the development of the blood-vessels of the human brain, ^Mall^^^e) noted the ease with which an extravasation into the embryonic arachnoid spaces could be brought about by increasing the pressure in a venous injection. In a specimen of 46 mm. an arterial injection with aqueous prussian-blue resulted in a complete subarachnoid spread, due to rupture of the vessels as they perforated the nervous tissue. In general, it was found that this rule held: an arterial extravasation always took place from the perforating capillaries, while a similar venous rupture occurred in the veins themselves.
 
 
Mall made similar observations on living pig embryos from 30 to 80 mm. in length, with analogous results. But when, in these embryo
 
 
s, the arachnoid spaces were completely filled by an intraventricular injection of india ink, no passage of the granular injection into the veins or sinuses occurred. The ventricular injection flowed into the extraventricular spaces "through the medial opening of the fourth ventricle." From the spinal cord the ink extended for a short distance along the main trunks of the spinal nerves. In the larger embryos (above 50 mm.) the ink usually gushed from the mouth, reaching it by way of the Eustachian tube. Using, in the pig embryo, the heart as the mechanism for injecting the ink, extravasation from the cerebral vessels in the arachnoid spaces occurred.
 
 
In one human specimen of 90 mm., Mall found both the arachnoid spaces and the cerebral ventricles filled with india ink after an arterial injection of that suspension. He states: "The injection passes through the medial opening into the fourth ventricle (Magendie), and apparently the ventricles are injected through this opening from the arachnoid."
To His'25) and to KolUker*^!^ belongs the credit of first having established on a firm basis the development of all the meninges in man from mesenchyme. This perimedulla
 
 
rj' layer of mesenchyme Salvi'^o^ called the "primitive meninx" — a term now used extensively in comparative anatomy. The primitive meninx divides into two layers, the outer forming the dura and the inner the pia-arachnoid. Sterzi(53), working on the development of the human spinal meninges, advanced a \dew similar to that of KoUiker. The perimedullary mesenchjTne (the "primitive meninx") divides into two portions, one hugging the inner surfaces of the vertebra? and the other adhering to the cord. This inner layer of the perimedullarj^ mesenchyme, according to Sterzi, should properly be termed the "primitive meninx," as it divides subsequently into dura and the "secondary meninx," which in turn forms both arachnoid and pia. The denticulate hgaments develop in the "primitive meninx." The dura and arachnoid in human embryos are modeled up to a certain point on the cord; then, with the augmentation of the subarachnoid space, they follow the outline of the vertebral canal.
 
 
His'25) has given information regarding the development of the meninges, with particular reference to the formation of the subarachnoid space. He affirms the mesenchymal origin of all of the cerebro-spinal membranes. His describes the first differentiation of mesenchyme to form the meninges as consisting of two zones of condensation, the outer being closely associated with the developing perichondrium of the vertebral column and the inner facing upon the cord. Between these two zones of condensation the subarachnoid space develops, posterior and anterior spaces first appearing, with later fusion laterally. These appearances were met with in chicks of 10 to 12 days' incubation. Quite soon after this process of spacedevelopment a separation occurs which gives rise to a complete subarachnoid space. Later the splitting-off of dura from the vertebral periosteum takes place.
 
 
III. METHODS OF INVESTIGATION.
 
 
In the study of any problem dealing with the development of fluid-spaces within the body, the method of investigation must of necessity be such as to offer exceptional opportunities for control. In the present work several well-known and generally accepted anatomical procedures were naturally suggested, such as injection of the spaces about the central nervous system, reconstruction from serial sections, or merely study of the various stages by means of serial sections.
 
 
It was ascertained early in the investigation that by injection and serial sections without reconstruction the necessary stages in the process of meningeal differentiation could be estabUshed. In regard actually to the physiological aspects of the problem more reUance was placed on the results of injection than on any histological differentiation, for, as explained above, considerations of the pathway and of the flow of the cerebro-spinal fluid were deemed most important. No method of injection, however, holds out much promise in such a problem unless it can be applied, under conditions approximating the normal, wnthin the spaces about the nervous system. The greatest objection to reliance upon injections in this problem is in relation to pressures. From the very nature of the case it wall be reahzed that any ordinary injection into the embryonic central canal or perispinal space must result in an extraordinary increase in the normal tension of the fluid. This objection applies to any method employed, whether that of a simple syringe and needle, the glass tube and bulb devised by Knower'^', or a glass capUlary-tube contrivance.
 
 
The erroneous conclusions drawTi by investigators from the emploj^ment of excessive pressures of injection are nowhere more strikingly illustrated than in studies of the circulation of the cerebro-spinal fluid. Many such examples were recently brought forward in a critical review •''5' published in connection with a study of the fluid. In the embryo, with structures and membranes still of very little tensile strength, the consequences of a disregard for the pressures of injection are even more disastrous.
 
 
A second criterion for the study of fluid-pathways in the body is necessarily the type of injection mass. Not only should attention be paid to the pressures involved, but the peculiarities of the particular body-fluid concerned must be considered. Adopting for this work on the embryo the same standards followed in the previous investigation on the adult, true solutions were used in place of the customary granular suspensions. Emulsions and viscous solutions were not emploj'ed because of their obvious disadvantages in studying the passage through membranes. India ink and process black (in which carbon granules are the particulate matter) were also used, but only for comparison with the standard true solution, as the likelihood of the insoluble granules being phagocyted wathin the period of experimentation or of being caught mechanicalh'^ in tissue meshes appeared a priori to be too great.
 
 
In any study of fluid-pathways in the body, not only must the injection fluid be a true solution, but it must also be one which is not attracted to particular cells (as with many stains). Likewise, colloid stains (such as the benzidene group) could not be employed, because of the fact that certain cells (macrophages, as described by Evans’s phagocyte the small colloidal particles. In addition, the true solution must be readily precipitated as an insoluble salt, capable of remainhig unchanged in histological technique. After trying many salts in long-continued injections into the adult cerebro-spmal spaces, it was found that solutions of potassium ferrocyanide and iron-ammonium citrate in equal parts were admirably adaj^ted to the purposes of the experiment. By the addition of a mineral acid (preferably hydrochloric) ferric ferrocyanide could be precipitated. This prussian-blue is insoluble in the routine technique and is readily identified in sections. After mounting in damar or balsam the blue granules can be observed unchanged for several months, but after a year there is some deterioration in the specimen, due to a conversion of the blue into indefinite greens.
 
 
 
 
 
Text-figure 1. — .Schematic sketxjh of mechanism used (or replacing ventricular and spinal fluid of an embno with a foreien solution or suspension. The system is here shown in balance, the difference in fluid-level in reservoirs and needles representing the hydrostatic pressure necessary' to overcome the capillary resistance of tubes and needles. The stands holding the injecting needles may be moved about without altering the balance of the system. As one reserv'oir is raised, the other is lowered in a corresponding degree.
 
 
 
 
 
 
In regard to these two major factors in the employment of injections (pressure and true solution) it was found necessary to devise a method of experimentation which would satisfy the requirements of the problem. Solutions of the ferrocyanide and of the citrate were non-toxic within the central nervous sy.stem and afforded an excellent histological means for following the fluid-pathways. It was hoped at first that a simple "replacement" type of injection might be employed, as in the adult animals. In this procedure a given amount of fluid was withdrawn from the subarachnoid spaces and immediately replaced by an equal quantity of the injection fluid. The method was successfully tried on fetal cats of consideralile size, but was impracticable on small embryos. After such a replacement the animals were allowed to Uve for varying periods of time (up to 3 hours) and then killed.
 
 
It was soon ascertained that the essential circulation of the cerebro-spinal fluid was established in pig cml^ryos of less than 30 mm. in crown-rump measurement. Hence the ordinary method of replacement had to be discarded for some more delicate system. With the realization that a simultaneous withdrawal and introduction in a living embryo would be far preferable to a two-stage procedure, the extremely simple apparatus pictured in text-figures 1 and 2 was employed. This device consists of two glass tubes of uniform and like bores, suspended from above by a string running over a pulley. To the tapering lower ends of these reservoirs are attached rubber tubes which connect the reservoirs to two needles. These needles are held at the same level by two metal brackets which can be moved at will on a level glass plate.
 
 
 
 
 
Text-figure 2. — Diagrammatic representation of the method of rcplacinR the cerebro-spinal fluid in a living embryo
 
 
. The spinal needle is inserted into the central canal of the spinal cord, while the cerebral needle is introduced into one of the cerebral ventricles. The canal of the spinal cord and the cerebral ventricles are represented by the interrupted lines. The foreign fluids are introduced by the spinal needle and withdrawn by the cranial.
 
 
 
 
 
The apparatus is employed as follows : Both tubular reservoirs are filled up to the point where the fluid is just ready to fall from the needle in a drop. This point is easily obtained by fiUing the reservoirs slightly in excess and allowing this excess fluid to run off from the needle. With the system thus in balance the needles he in the same horizontal plane and can be moved without altering the balance of the solutions. The injection is made by inserting one needle into the central canal of the spinal cord and the other into one of the lateral ventricles; then as the reservoir connecting with the spinal needle is raised the other is lowered, so that an amount of fluid equal to that introduced into the spinal canal is withdrawn from the cerebral ventricles. In this way the whole contents of the cerebral ventricles and central canal of the spinal cord can be slowlj' withdrawn without increasing the pressure in the central nervous sj^stem. The initial pressure necessary to secure this flow is only that required to overcome the capillary resistance of the medullarj'-canal system. In practically all cases this can be accomplished by using a positive pressure of less than 60 mm. of water (associated with a negative pressure of the same degree) .
 
 
 
In the present study the above procedure was the routine method of injection employed. Pig embryos, brought from the abattoir, contained in the uterus, were found to be wholly satisfactory material. If not permitted to cool excessively in transit the embryos lived for at least two hours in a 38° incubator. On being received at the laboratory a section of the uterine wall contaming the placenta was excised, with the embryo left connected by the umbilical structures. As soon as the technical preparations for injection were completed the amnion was opened and the embryo placed upon a padded block at the proper level. The first needle was then inserted into the easily discernible central canal of the spinal cord and the second into the left cerebral ventricle or into the mesencephalic ventricle. By elevation of the reservoir connected with the first needle the cerebro-spmal fluid was replaced by the injection solution. As soon as the replacement was complete the needles were withdrawal and the embryo and its uterine portion replaced in the incubator. The heart of the embryo could be easily obser\^ed in the smaller forms and served as the index of a continued circulation.
 
 
The incubation of the embryos was continued for varying periods of time, but it was soon ascertained that a period of over 30 minutes generally resulted in a complete spread of the injection solution. For comparison the period of incubation was lengthened and shortened, but the best results were usually obtained with a 45-minute incubation after the replacement.
 
 
Injections of the necessary true solutions were made, in the routine experiment, with a 1 per cent concentration of potassium ferrocyanide and iron-ammonium citrate in distilled water. By a 1 per cent solution is meant a salt concentration of this amount (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm.; water, 100 c.c). The resultant true solution should be practically isotonic with the body-fluids. In this way any injurious consequences due to hypertonic or hypotonic solutions were apparently overcome. The factors of osmosis and diffusion also had to be considered in this connection.
 
 
Other concentrations of the so-called "ferrocyanide mixture" were used, but only for the sake of comparison or for the purpose of investigating some particular phase of the problem. The results obtained by the use of these concentrations were not relied upon as affording standards for the normal pathway of the cerebro-spinal fluid.
 
 
In addition to the replacement type of injection, many observations were carried out on pig embryos, with a simple syringe-injection of the ferrocyanide solution into the central canal of the spinal cord or into the cerebral ventricles. It proved to be a very simple matter to regulate the pressures by this method, and three arbitrary standards (mild, moderate, and strong) were found to be of value in a comparison of the extent of the spread obtained by replacement and by injection.
 
 
The prussian-blue reaction (formation of ferric ferrocyanide) was obtained in these experiments by fixing the whole embryo in an agent containing hydrochloric acid. For histological study the best results were obtained by immersing the specimen from 1 to 10 minutes in a 10 per cent formaldehyde solution containing 1 per cent hydrochloric acid. After this primary procedure, during which the ferrocyanide was precipitated, the embryo was transferred to Bouin's fluid (saturated aqueous picric acid, 75; formaldehyde 40 per cent, 20; glacial acetic acid, 5). The specimens were allowed to fix over night and were then dehydrated in graded alcohols. From 30 per cent alcohol, use was made of 4 per cent changes up to 60 per cent; and from this point to absolute the changes were by 2 per cent gradations.
 
 
In addition to the technique outlined above, Carnoy's solution and 10 per cent formol were employed. The Carnoy fluid, containing acid (absolute alcohol, 60; chloroform, 30; glacial acetic acid, 10; hydrochloric acid, 1) proved to be of particular service in the study of specimens cleared by the Spalteholz method; histologically, however, it has not been as valuable as Bouin's fluid.
 
 
Besides the ferrocyanide solution, two other injection masses were constantly employed. Solutions of silver nitrate in concentrations of 0.5 per cent were injected into the central canal of the spinal cord and into the cerebral ventricles. This method, with reduction of the silver salt in the sunlight, gives very pleasing preparations. It is, however, subject to obvious limitations. The intraspinous toxicity of the silver, together with its action as a precipitant of albuminous substances, renders its use unsatisfactory in replacement experiments. Furthermore, it reacts apparently with any protein tissue, irrespective of the true function of that tissue (as, for example, its coagulation of the lining ependyma of the ventricles).
 
 
India ink, the other substance employed, is of extreme value in anatomical studies. Because of the suspension of carbon granules it possesses the disadvantages already commented upon for the study of any true pathway of fluid. It has been of service, however, in the present work in showing marked differences in spread from that of true solutions and in furnishing information in regard to fluid passage through a membrane.
 
 
This investigation has been carried out on the basic idea of correlating the physiological spread of the embryonic cerebro-spinal fluid with the gradual transformation of the perimedullary mesenchyme into the three fully formed meninges. This has necessitated a histological study of the embryo
 
 
. Pigs for the most part were the animals used, but the findings have all been verified by a study of the same regions in the human embryos in possession of the Carnegie Institution of Washington. In addition, certain structural characters have Ukewise been identified in sections of chick, rabbit, and cat embryos.
 
 
It was early apparent that the material to be of value must be free from any great shrinkage about the central nervous system. Comparative freedom from this artifact was obtained by fixing the embryo alive in Bouin's fluid and dehydrating by 2 and 4 per cent gradations of alcohol. The material was chiefly cut in paraffin after being embedded by means of xjdol.
 
 
The methods of investigation outhned in the foregoing paragraphs have been followed throughout the major portion of the work. In many minor instances other procedures not commented upon have been employed ; these wiU be detailed in appropriate subdivisions of this paper.
 
 
IV. INJECTIONS AND REPLACEMENTS IN THE CEREBRO-SPINAL SYSTEM.
 
 
RESULTS OF REPLACEMENTS IN THE VENTRICULAR SYSTEM OF TRUE SOLUTIONS.
 
 
The results of experiments carried out on embr3'o pigs by the technical procedures outlined in the previous section will be detailed here. The study was made on this animal because of the facility with which it could be obtained living and in good condition and also because it exhibits the characteristic meningeal anatomy of all mammals. The chick could not be used in this investigation on account of the dissimilarity between the avian and the mammaUan menmges.
 
 
The chief problem concerned here was the actual physiological extent of the cerebro-spinal spaces. This apparently could be ascertained by the replacement of cerebro-spinal fluid by the ferrocj^anide mass. But there was also to be considered the passage of fluid from the ventricles out into the periaxial* spaces, corresponding exactly to a similar passage in the adult.
 
 
If into the central canal of the spinal cord of a hving pig embryo of 9 mm., cro\\Ti to rump measurement, an injection of the ferrocyanide solution be made under very mild syringe-pressure, the ventricles can be fairly well filled without rupture of any element. Incubation of this experimental embryo with its circulation continuing almost unabated for an hour should cause a further spread of the fluid throughout the normal canals. If at the end of this time the whole embryo is fixed in an acid medium the ferrocyanide will be precipitated in situ.
 
 
Such a specimen, subsequently cleared by the Spalteholz method, is represented in figure l.f In this drawing the spread of the injection solution is clearly sho\\Ti. Running upward from the point of introduction, wholly within the central canal of the spinal cord, it reaches the bulbar region and extends outward into the large fourth ventricle, appearing as a dense collection of the prussian-blue. Cephalad from this region it spreads in diminishing intensity until it is finally lacking in the diencephalon.
 
 
The injected solution, then, in spite of the unavoidable increase in the normal intramedullary pressure, is contained only within the medullary-canal system (central canal of spinal cord and cerebral ventricles). There is no evidence of any spread outwards, either from the third or fourth ventricle.
 
 
In the next stage of meningeal development the replacement method can be used, as the embryo is no longer too small for its employment. In figure 2 is represented an embryo of 13 mm., in which the circulation continued for 90 minutes after the replacement. The same general picture shown in figure 1 results. The whole medullary-canal system is filled with the precipitated prussian-blue, which is densest in the region of the fourth ventricle. The roof of the ventricle, however, shows a striking difference from that of the ventricle in the embryo of 9 mm. Just posterior to the cerebellar lip is a regular oval, which is covered from within by a dense collection of prussian-blue granules, causing it to stand out in clear contrast to the thinner and more evenly distributed blue lining of the remainder of the roof. This oval area is comparatively large and comprises a portion of the superior or anterior half of the ventricular roof. This area, differentiated from the remainder of the rhombencephalic roof, is clearly shown in figure 2, a drawing of a cleared specimen of this stage.
 
 
• ThrouRhout this paper the term "periaxial " has been used in the seii90 of "arouud the central nervous system " or "around the ccrebro-Bpinal axis."
fThroughout this work the rcforeDoe "figure " 1, etc., refers to plate illustrations; the word "lext-figure" refers to the illustrations inserted in the text.
 
 
 
 
With the exception of this strikingly dense area in the rhombic roof, the uajection spread in an embryo of 13 mm., subjected to replacement of the cerebro-spinal fluid by the ferrocyanide, differs in no way from that in the embrj'^o of 9 mm. Careful inspection of figure 2 is convincing that the spread still remains within the medullary canals, with no extension of the fluid into the spaces outside of the cerebrospinal axis. It seems justifiable, then, to speak of the cerebro-spinal spaces at this stage of development as being onlj^ mtramedullary in type, with no indication as yet of a meningeal fluid cushion (corresponding to the adult subaraclmoid space).
 
 
With the use of larger embryos, however, for the medullary replacement with ferrocyanide and citrate, the picture gradually changes. The first indication of a more advanced stage of development is obtained in embryos whose length exceeds 14 mm. Figure 3, of a pig embryo of 14.5 mm., is included here as representing this further extension of the injection fluid. The cerebro-spinal fluid of this specimen was replaced, by the compensating mechanism, bj' a solution of potassium ferrocj'anide and iron-ammonium citrate. The embryo was then kept alive (as judged b)^ the heart-beat) for a period of one hour. At the end of this time it was fixed in an acid medium and subsequently cleared in oil of wintergreen after careful dehydration.
 
 
The essential differences between an embryo of this stage and one of the stage represented in figure 2 concerns the spread of the injection fluid from the roof of the fourth ventricle. Both specimens show a complete filling of the intramedullary system (cerebral ventricles and central canal of the spmal cord) with the precipitated prussian-blue granules. The specimen of 13 mm. (fig. 2) is characterized bj' a dense oval collection of the prussian-blue on the upper and inner surface of the rhombic roof. In the specimen shown in figure 3, in contradistinction to this localized aggregation of granular matter, there is a deUcate extension of the injection fluid caudalwards from the roof of the fourth ventricle. This fusiform projection is here readily made out, lying beneath the skin over the ventricular roof and separated quite distinctly from the easily discernible line of the roof. This outward extension of the fluid has a fairly wide and deep origin from the upper portion of the roof, but tapers caudally to a sharp point with considerable rapidity.
 
 
At the stage of 14 mm. the roof of the fourth ventricle shows the small depression which marks the formation of the chorioid plexuses. With this depression occurrmg transversely the relation of the external surface of the embryo to the ventricular roof necessarily alters somewhat in this region. The chorioidal depression of the roof graduallj' becomes separated from the skin; and it is into this area between the skin and the ventricular ependyma that the first spread from the cerebral ventricles occurs. At this stage, illustrated in figure 3, the injection is intramedullary in type, with but sUght extension into the pericerebral tissues.
 
 
The pericerebral spread may be made out in nearly all replacements in embryos of 14 mm., but in a few cases the injection has remained intramedullary in type. In embryos of 16 mm. the spread into the pericerebral tissues is invariably found. Often, with this extension of the replacement solution outside the ventricles, the oval area noted in the stage of 13 mm. persists. (This phenomenon is especially well shown in a simple injection of silver nitrate, illustrated in figure 11.)
The next stage of importance in the development of the cerebro-spinal spaces is represented in figure 4, a drawing of a pig embryo of 18 mm. in which a typical intramedullary replacement of the cerebro-spinal fluid with a solution of potassium ferrocyanide and iron-ammonium citrate had been made. Here, with the exception of the region of the roof of the fourth ventricle, the replaced fluid is contained solely within the central canal of the spinal cord and within the cerebral ventricles. The roof region, however, exhibits a new phenomenon, which distinguishes it from the stage shown in figure 3. The chorioid plexus invagination has become strongly developed, dividing the roof into two parts. These roof divisions have been termed superior and inferior, the former lying anteriorly and orally from the chorioid fold. The general surface outline is but little changed, due to the mesenchyme filling up the area between roof and skin. From two areas in the entire roof of the fourth ventricle the foreign fluid has escaped into the pericerebral tissue. These points of fluid passage he in the two divisions of the ventricular roof. The superior area of escape corresponds to the oval outlined by the prussian-blue in figure 2 and to the point of emergence of fluid shown in figure 3. The lower area of fluid escape is in the inferior half of the ventricular roof, where the ependymal lining and its supporting tissue are developing into a well-marked dorsal distension. This area corresponds to Blake's'3' caudal protrusion, though, as Heuser'23) has pointed out, the shape of the structure in the pig in no way resembles the "finger of a glove."
The extraventricular spread of the injection fluid in this specimen is considerably greater than in the pig embryo of 14 mm. (fig. 3). On the whole, however, the distribution of the replaced fluid is not extensive as compared with the adult relationship, where the central nervous axis is entirely surrounded by its subarachnoid cushion of cerebro-spinal fluid. From the superior area of fluid passage the replaced solution (as shown by the resultant precipitation of the prussian-blue) has passed both superiorly and inferiorly. In the median line, and extending laterally but sUghtly, a projection of the blue may be seen occu])ying a large portion of the extraventricular area formed from the chorioidal invagination. This area of fluid passage occupies at this stage about one-third of the total transverse diameter of the ventricular roof. From it the blue tapers caudally, diminishing in all directions. Above, the precipitate maj'^ be made out extending superiorly over the cerebellar lip. Its extension into the pericerebellar tissue is not marked; here again it tapers from the area of fluid passage, its midline prolongation stretching farthest anteriorly. This relationship is easily made out in figure 4, a frank lateral view of such an experimental rei)lacement. The granules which result from the introduced ferrocyanide solution are found only in the central canal of the spinal cord and not in any perispinal arrangement.
 
 
In the pig embryo of 18 mm., shown in figure 4, the replaced solution has been carried somewhat farther than in the embryo of 14 mm. (fig. 3). The chief point of differentiation lies in the fact that in the latter stages two areas have apparently become permeable to the intraventricular fluid, so that a larger periaxial spread has resulted. Then, too, the extension of the ferrocyanide solution from the superior area is considerablj^ greater, overlapping the cerebellar Up and filling in some degree the pericerebral tissue in the chorioidal invagination.
 
 
With a definite periaxial spread established for the cerebro-spinal fluid in pig embryos of 14 to 18 mm., it seemed not unreasonable to expect a gradual increase in the extent of the future subarachnoid distribution in more advanced stages. The earliest extension of the fluid into the peribulbar tissues occurred with the inception of the infolding of the ventricular roof to form the chorioid plexuses of the fourth ventricle. Its further extension, particularly its passages through a second area, occurred with the greater development of the chorioidal invagination (i. e., 18 mm. stage). A still more extensive pericerebral flow of the ferrocyanide and citrate is illustrated in figure 5. Here the cerebro-spinal fluid in a hving pig embryo of 19 mm. was replaced by the ferrocyanide solution. The embryo was kept alive for about an hour after the replacement and was then fixed in toto in an acid fixing medium, which caused the precipitation of the prussian-blue. On clearing subsequently by the Spalteholz method the spread of the solution was found to be somewhat more extensive than in the stage of 18 mm. (cf. figs. 4 and 5). In figure 5 the whole periaxial area over the roof of the fourth ventricle is shown to be completely filled by a dense aggregation of the prussian-blue granules. The separation of the two areas of fluid passage can not be made out in such a specimen. This dense periaxial extension ahnost completely covers the cerebellar Up, not onlj^ in the medial region but laterally to the limit of the ventricular roof. The injection precipitate lies directly beneath the skin in this area, but more posteriorly its separation from the skin becomes more marked. Tracing this dense periaxial injection posteriorly, it is seen (fig. 5) to end somewhat abruptly in the region of the cephaUc flexure. The Une of termination of the denser mass, to the ventral surface of the medulla, tapers somewhat anteriorly. This extraventricular spread is medial to the otic vesicle, but extends peripheraUy along the caudal cerebral nerves, reaching outward as far as the peripheral gangUa. The periaxial spread also closely covers the ventral surface of the medulla and extends in this plane around the pontine flexure for a short distance upwards along the basilar surface of the mid-brain.
 
 
Examined from its dorsal aspects, the superior portion of the spinal cord is found to be covered (in a perispinal relation) by a fine deposit of the prussian-blue. This is shown in figure 5. CaudaUy from the higher cer\'ical region there is no exidence indicating a further spread in the perispinal tissues. Such a spread from above dowTiward is wholly at variance wdth Reford's^^' conception of a development of the spinal meningeal spaces before the cerebral. The complete filling here of the central canal of the spinal cord and of the cerebral ventricles with the replaced fluid, with no evidence of a periaxial spread except in the region of the fourth ventricle, indicates that in the pig cinbrj'o the adult human relationship between the cerebral ventricles and the subarachnoid spaces endures. There is apparently in this embryo no evidence of the foramina of Bichat and of Mierzejewsky, a findmg in accord with the observations of Dandy and Blackfan'i*''.
 
 
In the slightly larger embryos the further extension of the embryonic extraventricular spaces progresses rapidly. Figure 6 represents such an extension in a pig embrj-o of 21 mm., in which the normal cerebro-spinal fluid was rei)laced by a dilute solution of potassium ferrocj'anide and iron-amiuonium citrate. In this specimen the central canal of the spinal cord and the cerebral ventricles are completely filled with the precipitated prussian-blue. But m addition there is almost a total filling of the periaxial spaces. Viewed laterally the densest aggregation of the blue granules is again in the region of the roof of the fourth ventricle. As in the embryo of 19 mm. (fig. 5), the whole extraventricular tissue posterior to this ventricular roof is filled with the granules precii)itated from the foreign solution. The spread from this region is similar to that in the previous specimen, except in its far greater extent. The granules maj'^ be traced caudalwards in the perispinal spaces to the point of injection. The arrangement of the precipitated material, both withm the central canal of the spinal cord and surrounding it in the perispinal relationship, is well shown in figure 7, a frank dorsal view of the same specimen represented m figure 6. The greater density of the perispinal granules in the upper region of the cord, as contrasted ^^■ith the granules in the thoracic region, is probablj^ of importance in indicating the direction of the flow from above dowaiwards. The increased amount of the injection fluid in the region about the pomt of insertion of the spinal needle.is in all likelihood due to a local spread from the needle, such as frequently occurs in a very limited area. The phenomenon may, however, be due to an actual increase in the size of the potential perispinal space, though observations upon other embiyos of the same stage of development argue against this view. The segmental outlming of the caudal portion of the perispinal space is to be noted in this figure.
 
 
The cephalic regions in the specimen of 21 mm. show a quite extensive spread (fig. 6), and there is the same general distribution of the granules about the medulla, as in the specimen shown in figure 5. The rhombencephalon is completely surrounded by the blue, the ventral sheet inclosing it tight]}-. Lateral!}' the prussianblue is shown in a dense mass, in intimate relation to the cranial nerves as thcj' join the brain-stem. The cerel^ellum is jiractically complete!}' covered by the precipitate; from the ventral portion of the pericerebellar granules the replaced solution (as evidenced by the granules of prussian-blue) spreads forward and surrounds a portion of the mid-brain. Only the ventral surface of the posterior half of the mid-ljrain is circumscriljed by the granules; anteriorly it is wholly surrounded by the i)eriaxial injection; more anteriorly tlic extension i.s iimited to the mesial structures, leaving unsurrounded the cerebral hemispheres, althougli creejnng between the hemisi^lieres and the mi(l-l)rain.
 
 
The pecuhar avoidance by the replacement fluid of the extreme dorsal half of the mid-brain is also to be made out in the dorsal view of the specimen (fig. 7).
 
 
 
The two lateral extensions from the ventral sheet of the injection granules approach on either side this mesencephalic eminence. The peculiar appearance of the injection spread caused by the chorioidal invagination of the roof of the fourth ventricle is also here illustrated.
 
 
In this specimen, then, of a pig embryo 21 mm. the periaxial spread is almost complete, the only areas not entirely surrounded being the aiiterior mesencephalon and the cerebral hemispheres. In an embryo but a few milhmeters larger this periaxial exten.sion of the solution is complete. The mesencephalon first becomes entireW covered by the jjrussian-blue precipitate, with later extension over the hemispheres. This complete periaxial injection occurs usually in replacements in embryo
 
 
s varying in length from 24 to 28 mm.
 
 
A specimen exhibiting a complete extension of the replaced solution around the central nervous system is shown in figure 8. This specimen was prepared by replacing the cerebro-spinal fluid in a Uving embryo of 26 mm. and then keeping the embryo alive for an hour. After fixation in an acid medium, dehydration, and clearing, the uijection was found to occupy the whole medullarj-canal system and also to surround completely the cerebro-spinal axis, as shown in the lateral view. The striking features of this stage are simiJar to those observed in the younger specimens — the dense accumulation of granular material in the region of the roof of the fourth ventricle, the surrounding of the central portion of the caudal cranial nerves, and the thin pericerebral covering by the replacement mass. In addition the specimen exhibits in the thoracic region an extension of the granular material laterally along each spinal nerve. An observation of this peculiarity reveals the prussian-blue extending outwards only as far as the gangha on the posterior roots.
 
 
The relationships, then, observed in an embryo pig of 26 mm. are those which exist in the adult; the cerebro-spinal axis contains cerebro-spinal fluid within its cerebral ventricles and within the central canal of the spinal cord, while in turn it is cornpletel}' surrounded by cerebro-spinal fluid within the subarachnoid space. Communication between the ventricles or intra-medullary sj'stem and the perispinal spaces occurs only in the region of the fourth ventricle. Here again the adult human relationship holds. The evidence, therefore, from a study of the fluid spread in a replacement experiment with the use of true solutions, indicates that in pig embryos of about 26 ram. an adult distribution of cerebro-spmal fluid occurs.
 
 
THE RESULTS OF INJECTIONS OF TRUE SOLUTIONS. In the preceding section there have been detailed the results of experiments on living pig embrj-os in which the cerebro-spinal fluid of both the central canal of the spmal cord and the cerebral ventricles has been replaced by a dilute solution of potassium ferrocyanide and iron-ammonium citrate. After the replacement, carried out so as to avoid anj- increase in the normal tension, the embryos were incubated for varj^ing periods of time so that the normal current of the fluid might cause an extension of the loreign solution. In the experiments which will be recorded in this section the same true solution was injected from an ordinary syringe and the salts immediately precipitated as prussian-blue. The purpose of these observations was solely to ascertain the effect of injections at pressures above the normal tension, so that the conclusions drawn from the replacement method might be more fully substantiated.
 
 
It was soon ascertained that the pressures caused by injections with a simple syringe could be fairly well controlled and that several degrees of tension might be employed. Thus it was found to be simple and serviceable to designate the injections as those made with mild, moderate, or strong syringe-pressure. Most of these injections were made into the central canal of the spinal cord, but occasionally into the perispinal spaces or cerebral ventricles. Injections under equivalent pressures in the central canal of the spinal cord or into the cerebral ventricles always gave corresponding results. It is necessary to record that the injections, even under strong pressure, were not carried to the point of macroscopic rupture.
 
 
The so-called mild syrmge-pressure, making use of solutions of potassium ferrocyanide and iron-ammonium citrate, resulted in extensions of the prussian-blue wholly similar to those obtained in the replacement experiments which were carried on for 30 minutes and over. This similarity indicates a complete filhng of the available cerebro-spinal system in the replacement method, for certainly (even in the mildest syringe injections) the intraventricular pressure must be excessively increased. Figure 1 shows a specimen under such conditions, with a marked thinning of the mjection mass in the region of the fore-brain. This finding is customarily present in the injections under mild pressure, due to the pushing upwards of an existent ventricular fluid.
 
 
"When moderate pressures are employed with the syringe the picture gradually changes. The essential difference in the results obtained by moderate syringe injection and by the replacement method lies in the greater extension of the foreign solution in the smaller embryos. Thus in figure 9 the spread of the injection precipitate in a pig embryo 16 mm. is shown to be about as extensive as that obtained by the replacement method in an embryo of 19 mm. (fig. 5). The extra ventricular distribution of the injected solution around the medulla, the extension (even more marked here) along the central roots of the caudal cranial nerves, and the localized perispinal spread are easily made out in this specimen of 16 mm.
 
 
This general rule applies to all of the results obtained with the use of syringepressures above the mildest. Dependent upon the degree of syringe-tension, the spread extends in simple ratio. Thus, by the use of moderate pressures of injection into the central canal of the spinal cord, a complete intramedullary and periaxial spread was secured in a pig embryo of 22 mm. somewhat earlier than the equivalent stage was obtained by the use of the replacement method.
 
 
With the highest syringe-pressures (insufficient, however, to cause macroscopic rupture) the same general tj'pe of injoctiou spread was obtained, bringing the more complete stages down into smaller and smaller embryo
 
 
s. Most of these embryos, however, on microscopic section showed obvious rupture of some part of the central nervous system.
 
 
 
 
 
INJECTIONS AND REPLACEMENTS IN THE CEREBRO-SPINAL SYSTEM. 27
The most important feature of these findings in the embryo pig injected with true sohitions under moderate pressures from a syringe concerns the fact that the extension of the injection coincides, except as to the size of the embryo, in everj* instance with that obtained by the replacement method. Thus similar and analogous spaces are filled by injections under syringe-pressures in small embryos and by the solution under normal tension in larger embryos. It must be assumed, then, that the pressure of injection is sufficient to dilate potential cerebro-spinal spaces which normally would not be concerned in the pathway taken bj' the cerebro-spinal fluid. No evidence of new or abnormal pathways for the fluid is afforded by the observations made with the increased pressure; these phenomena indicate great potential strength in the tissues which limit the immature cerebro-spinal spaces.
 
 
Injections with a simple s^Tinge may he made with such a degree of pressure that gross rupture of the tissues becomes apparent. In such an injection into the central canal of the spinal cord the infundibular region ordinarily ruptures in the smaller embryos (under 15 mm.), while in larger embryos rupture usually occurs into the subcutaneous tissues of the back of the neck over the fourth ventricle.
 
 
In discussing the effects of the introduction of solutions of ferrocyanide under pressures higher than normal into the central canal of the spinal cord, it may be appropriate to record observations made in the attempt to inject the cerebro-spinal spaces from the perispinal space. In embryos under 15 mm. in length it is quite difficult to make a perispinal injection. As the embryos exceed this measurement the injection becomes increasinglj' easy, but not until a length of 20 mm. is attained can it be made under the mild pressure advisable. These obser\'ations tend to substantiate the findings alreadj' recorded in both the intramedullary replacements and the injections under mild pressure.
 
 
RESULTS OF INJECTIONS OF NITRATE OF SILVER.
 
 
In a number of experiments a dilute solution (0.5 per cent) of nitrate of silver was injected into the central canal of the spinal cord and the salt then reduced in the sunhght. This solution, although a true one, is wholh- unsuited for the replacement type of injection, on account of its great toxicity and its power to coagulate protein. It was employed here onh' for the simple type of injection.
 
 
The results obtained by this intraspinous injection of solutions of nitrate of silver were of but little value in the determination of a pathway for the cerebrospinal fluid, but they vividly present certain aspects of the problem. Thus, m figure 11, a drawing of a specimen (pig) of 16 mm., the area through which fluid passes in the superior portion of the roof of the fourth ventricle is clearly outlined by a denser deposition of the silver. This specimen was prepared by introducing the solution of nitrate of silver into the central canal of the spinal cord under the so-called moderate sj-ringe-pressure. The dra\^Tng shows a shght, cone-shaped extraventricular spread of the injection fluid. This spread takes place solely from the superior area of f.uid passage, a result in accord with the finding that the solution of potassium ferrocyanide and iron-ammonium citrate passed first through the superior area. Of course it is realized that the precipitant action of the silver may have exerted a more potent action on the structures constituting the lower area of fluid passage.
 
 
Another interesting phenomenon of the injections of silver nitrate is shown in figure 12. The embryo
 
 
of 13 mm. here represented was injected under strong syringe-pressure with a sohition of silver nitrate into the central canal of the spinal cord. On subsequent reduction and clearing it was found that the excessive pressure had resulted in a comjilete intramedullary injection with a localized pedunculate spread into the tissues from the roof of the fourth ventricle. This bulbous extravasation into the extravcntricular tissue has not been observed in any specimens except those into which the solution of silver nitrate was injected. Such a spread is probably to be accounted for by an immediate coagulation of the surrounding tissue.
 
 
The extensive use of solutions of sUver nitrate as a means of demonstrating vascular channels naturally suggests a careful comparison of the results obtained from its use and those obtained from the employment of other available true solutions, in regard to the evidence afforded by the two methods of intraspinous injections. The chief objection to the use of silver nitrate, as has aheady been mentioned, is its power to coagulate protein. This is illustrated by many features of the specimen shown in figure 11 — ^by the sharp outhning of the area of fluid passage, the markings on the caudal process of the fourth ventricle, and the delimitation of the cerebellar hp. But much more marked are the evidences of this coagulative power as shown in figure 12, the pedunculated extraventricular spread, the transverse corrugation of the cerebellar hp (amountmg to circumscribed indentations), and the peculiar outlining of the roof attachment to the bulb. These phenomena obtained by the intraspinous injection of solutions of silver nitrate must be classed as artifacts. The difi"erent degrees of this corrosive action of the silver probably result from the varj^ing rates of reduction of the salt to the metal, a factor which is not easily regulated. The findings, therefore, with this method are worthless unless controlled.
 
 
Many embryos of varying sizes were injected with the silver nitrate. In the main these observations followed the course of development of the cerebro-spinal spaces as evidenced bj-^ the replacement experiments with the ferroc3'^anide. The injections required moderate pressure in the syringe in order to secure more than a local extension from the roof of the fourth ventricle, and to secure the same extent of spread it was generally necessary to use embryos a few millimeters larger than those required in the replacement experiments; but this is to be expected, m view of the i^robability of a constant precipitation of the albuminous tissues by the injection fluid.
 
 
Specimens prepared by the intraspinous injection of silver nitrate, then, afi"ord but little reliable evidence in this j)roblem except of a substantiative sort. The findings by this method indicate that (he i)erisi)inal and i^ericcrebral spaces, in pig embryos of 25 mm. and upward, could be filled by an injection of silver nitrate under moderate pressures into the central canal of the spinal cord. The point of passage of the fluid from the intramedullary to the periaxial system was in the region of the roof of the fourth ventricle.
 
 
THE INJECTION OF INDIA INK.
 
 
The objections to the use of any fluid of insoluble particles in suspension have already been discussed in considering the methods of injection which were possible for use in this study; but for comjiarisou Avith results obtained by more promising methods and to ascertain to what extent injections with India ink are reliable they will be further considered here. No granular substance other than India ink (carbon granules) was employed in this investigation. In every way this suspension possesses advantages over other possible masses — in its ease of preparation, in the small size of the granules, and the insolubihty in the reagents used for microscopic technique.
 
 
Suspensions of nidia ink (diluted from 4 to 10 times) were introduced first into the medulla
 
 
rj'-canal system of living pig embryos by the replacement method. In no case, however, even though the circulation of the embryo may have continued for 90 minutes, was there any evidence of an extension of the replaced mass outward into the periaxial spaces. The carbon granules remained wholly within the ventricles, a striking difference from the results obtained by the ferrocyanide replacements. It would appear, then, without the further evidence afforded by microscopic section of the specimens, that there is an existing mechanism which prevents the passage of the carbon granules from the fourth ventricle into the periaxial spaces. This finding was found to be constant ui all the living embryos subjected to the cerebro-spinal replacement.
 
 
Quite similar to these results by the replacement method are those from the injection of a suspension of india ink under mild syringe-pressure. In no instance, provided the pressure was maintained at a low enough degree, was there any passage of the granular material into the periaxial tissue. In embrj^os of over 30 mm., however, even with the lowest pressure, it becomes increasingly difficult to prevent a sudden spread into the periaxial spaces. The type of spread indicates a sudden release of some restraining agent and suggests a rupture of a membrane. This spread is usually local and takes place from the roof of the fourth ventricle.
 
 
With moderate and strong syringe-pressures, however, it is possible to secure a periaxial spread, but this is quite different from the distribution of the uijections by the use of ferrocyanide solutions. Figure 10 illustrates a specimen of a pig embryo of 21 mm. into whose central spinal canal india ink was injected under strong syringe-pressure. The resultant spread of the injection is easily discerned; the cerebral ventricles are quite filled with the carbon, while from the superior portion of the roof of the fourth ventricle a dense but localized periaxial spread is made out. This extraventricular extension of the ink is well defined; it stretches caudalwards for a slight distance, curving about the bulbous caudal portion of the ventricle and extending lateralwards but a short distance. The median portion of the cerebellar lip is covered by the granules. E\adeuces of the excessive pressure at which the injection was made are shown by the lines of mvasion of the spinal cord and mid-brain. A comparison of the spread of this injection mass with the extension of a ferrocyanide replacement in an embryo of the same size (21 mm.) is afforded by figures 10 and 6. With such a divergence in the results obtained by the two methods of approach it is not surprising that observations such as Reford's^"*^) fail to coincide with these findings. The unsuitabiHty of suspensions of granular material in the investigation of the cerebro-spinal sjiaces has been many times verified in this work.
 
 
In the further study of the course of the spread with injections of india mk it was found that, in pig embryos of approximately 22 mm. and over, a partial periaxial injection could be secured bj- plunging the syringe-needle into the perispinal spaces. The carbon granules could subsequentlj'^ be seen filling the perispinal spaces and also mounting upwards in jjartial pericerebral relationships, particularly around the medulla. This result was obtained by the use of strong syringe-pressures. Apparently the resistance to the spread of the ink in injections or replacements in the medullarj'-canal system occurs in the passage of the fluid from the roof of the fourth ventricle into the periaxial spaces. So far as is known, Reford^"*^) did not control his injection pressures. These results with the injection of india mk under strong pressures coincide with the idea of his observations afforded by the abstracts given by 8abin(*3) and Cushing^^'. Suspensions of india ink. then, injected under mild syringe-pressure or by the replacement method, offer no evidence, in the pig embryo, of a passage of the cerebro-spinal fluid into the j^eriaxial spaces. Only by employing pressures much above the normal tension can such evidence be obtained.
 
 
V. UNDESCRIBED STRUCTURES IN ROOF OF THE FOURTH VENTRICLE.
 
 
The results of the replacement of the existing cerebro-spinal fluid by a true solution of potassium ferrocyanide and iron-ammonium citrate in a living pig embryo indicated, as detailed in the foregoing section, that the fluid passed from the ventricular system into the periaxial tissues in the region of the roof of the fourth ventricle. This important transit of the fluid, agreeing with the established conception of the relationship in the adult, was first observed in an embryo pig of 14 mm. (fig. 3). At this stage the exudation of the replaced fluid occurred in one defined area, seemingly corresponding to the dense oval in a smaller embryo shown in figure 2.
 
 
Such a passage of fluid from ventricle to i)eriaxial tissue is necessarily a ])hysiological phenomenon, and it was in the hope of finding an anatomic basis for this phenomenon that the roof of the fourth ventricle was studied histologically. It was reahzed that failure to demonstrate anatomically differentiated structures would not vitiate the physiological observations, but that a correspondence between function and structure was most desirable. Hence observations were undertaken to determine, if pn.ssible, an area of histological differentiation in the roof of the fourth ventricle which might be concerned in the primary passage of fluid from the cerebral ventricles into the periaxial tissues. The investigation concerned first the examination of this region in pig embryos of 14 to 15 mm., at which stages the fluid passes from a single area. Subsequently, similar studies were undertaken in regard to the second, more inferior area (shown in figure 4). The results of these studies will be given here.
 
 
AN UNDESCRIBEX) AREA IN THE SUPERIOR PORTION OF THE ROOF OF THE FOURTH VENTRICLE.
 
 
THE AREA MEMBRANACEA SUPERIOR IN THE PIG EMBRYO.
 
 
Examination of the roof of the fourth ventricle in a pig embryo of 14 mm. revealed a peculiarly differentiated area in the superior portion. The general topography of this area is show7i in the rectangular area marked off in figure 32 — a median sagittal section from a pig embryo of this critical stage. In figure 33 this rectangular area is enlarged to show the morphologA' in greater detail.
 
 
In this figure the densely staining ependj'ma lining the fourth ventricle approaches from both sides. The superior portion of the ependyma ends abruptly, while the inferior Hne of th(? layer tapor.s more slowly. Between these two jjoints is an area having none of the characteristics of the ventricular lining at all other points. The comparatively smooth contour of the ependymal cells is replaced by an irregular cell-border. The pyknotic nuclei of the cells have been replaced by less densely staining, elongated, spindle-like nuclear bodies. The cell-layer lining the ventricle is here really only of a single cell in thickness, although blood-capillaries closely applied to it suggest a greater thickness. The mesenchj-nie lietween this layer and the peripheral epidermis is quite thin, but resembles in everj' way the mesenchyme in the immediate neighborhood.
 
 
There is, therefore, as pictured hi figiues 32 and 33, an area in the roof of the fourth ventricle which is morphologically dissimilar to the characteristic ependyma lining the cavity. Is this the result of some distortion in fixation or in the routine histological technique? Is it a constant finding and, if so, what is its historj'? Does it arise at a definite period and persist throughout intra-uterine life onlj' or through adult life also?
The question of the actual existence of this area, or of its being caused by technical manipulations, is one which must be answered. That this differentiated portion of the roof of the fourth ventricle is not an artifact is verified bj* the general history of its formation, by its invariable occurrence (not onh' in the pig but in other animals), and by its general histological appearance. Moreover, the physiological importance of this area undoubtedh' incUnes one completely from the possible explanation that it is due to an artifact. Xo single finding wholly excludes such a possibility; rather is one convinced, by many features, of its actual occurrence.
 
 
Considering the fact, then, that this differentiated structure in the roof of the fourth ventricle may be found in all embryo pigs at the stage of 14 mm., it becomes necessary to ascertain at what time in the development of the embryo it first appears and how it is formed. Obviously the most satisfactory' method is to trace the area through the lower stages and also through the older embryos. For the sake of greater clearness, however, a description of the area \s-ili be given from its first differentiation through its maximum transformation to its final disappearance — for the structure is only temporary.
 
 
In pig embryos of 8 mm. and less in crown-rump measurement, the roof of the fourth ventricle is fcfrmed of cells morphologically and tinctorially different from those lining other parts of the ventricular cavities. These cells are quite unlike the deeply staining ependjTiial cells, which can be so readily identified as the lining cells in older embryos. In this yoiuiger stage of 8 mm., the entire ventricular roof is composed of several layers of cells with round or somewhat oval nuclei and fairly abundant cytoplasm. The cell-boundaries are not well defined. The nuclei are not deeplj' tinged with hematoxylin. The chromatin material is sparse and irregularly distributed. Nucleoli are prominent. The cytoplasmic border hning the ventricular cavity is rough and ragged at times, often blending with the coaguUvted albumen of the cerebro-spinal fluid. Altogether, these lining cells bear a much greater resemblance to the epithelial cells than to the ependymal.
 
 
These characteristics of the lining cells of the roof of the fourth ventricle are shown in figures 24 and 25, from a pig embryo of 8 mm. The close association of the roof cells to the surface epithelium is easily made out in figure 25, as well as the general character of the lining cells.
 
 
At the stages of 8 mm. and under, in the pig embryo, the roof of the fourth ventricle is relatively'' quite large. In its whole extent it is formed of the peculiar lining cells described above. With the growth of the embryonic nervous sj^stem, the roof of the fourth ventricle is subjected to alterations in form and position; to some extent these changes influence the cells which line the cavity in the early stages.
 
 
In pig embryos between 8 and 12 mm. in length the roof of the fourth ventricle undergoes a change. The ependyma, which from comparison with later stages is regarded as typical, begins to encroach upon the epithelial-like cells which are so numerous in the 8 mm. stage (fig. 25). The area occupied by these cells diminishes, not only relatively but absolutely. It becomes smaller and the cells gradually change their character. These changes are shown in figures 26 and 27, from a pig embryo of 11 mm. Figure 26 gives the location, in a sagittal section near the midline of the area in figure 27, taken at a higher magnification.
 
 
In figure 27 the densely staining lips of ependymal and nerve cells are seen approaching each other. For a considerable space in the central portion of the photograph there is an area similar to that shown in figure 33. But considered in connection with figure 25 this area represents the epithelial-like cells of the roof of the fourth ventricle. This relationship is more clearly shown in figures 28 and 29, taken in a more lateral plane from the same embryo (11 mm.). Examination, however, of th(; area in figure 29 shows the epithelial-like cells again a])i)arent in the roof of the fourth ventricle.
 
 
The i^rocess of transformation, then, as shown in these photographs from an embryo i)ig of 11 mm., concerns a gradual encroachment upon the area of epithelial-like cells by the more densely staining and more closely packed ependymal cells. Gradually the epithelial-like cells in the central portion of the area lose their former character (fig. 27), while around the periphery, especially on the lateral sides, the epith('lial-lik(! ajjpearance persists (fig. 29).
 
 
On the lateral side of this area, just as the tyincal ependymal lining is al)out to become isolated (fig. 29), the epithelial-like lining cells form a several-celled layer.
 
 
 
The nuclei are poor in chromatin material and the cytoplasm somewhat small in amount. The inner cytoplasmic border hning the ventricle is in contrast, by its ragged outline, with adjacent smoother ependj^ma on both sides. At this stage of the pig embryo the characteristics of the epithehal-like cells are still to be made out, but a gradual transformation is becoming evident.
 
 
The metamorphosis becomes much more marked in the central portion of the area, as shown in figures 26 and 27. In these figures the whole central area seems to have lost some of its former character as an intact cell-laj'er. Closer examination, however, under higher power demonstrates that it still possesses an intact surface as a hning for the ventricle. Delicate cytoplasmic strands stretch in a continuous line across the whole area between the Ups of denser tj-pical ependjina. The nuclei in this difi'erentiated area are seeminglj' altered from their rounded form and have elongated almost into spindles. The inner cytoplasmic border is characteristically rough, with, small amounts of coagulated albumen adhering to the processes. The area, then, in its central portion, at the stage of 11 mm., has assumed the character of the stage of 14 mm. (fig. 32). On the periphery, however, the cells stiU resemble those of smaller stages (8 mm.).
 
 
From the pictures presented by the intermediate stages (figs. 27, 28, and 29) the differentiation goes on verj' rapidly, so that in the pig embryo of 13 mm. there is rarely any evidence of the epithelial-like cells. Figures 30 and 31 are photomicrographs of a sagittal section of an embryo pig of 13 mm.; here there are no evidences of the epitheUal-like cells. The whole area, pictured in figure 31 as sharply delimited from the tongues of tjijical ependyma above and below, has become well differentiated. The cell-character observed in figures 27 and 33 (elongated nuclei and scanty strands of protoplasm) has become very obvious. The ragged and roughened intraventricular border, the coagulated albumen, and the abrupt transition from the neighboring tj-pical ependyma are well shown in the photomicrographs of this specimen.
 
 
The differentiation of this area in the roof of the fourth ventricle of the pig embryo proceeds at a very rapid rate, so that within the growth of a few millimeters (from 8 to 13 or 14) a great histological change occurs. Figures 32 and 33, already described, show the extent of this metamorphosis in a pig embrj^o of 14 mm. The process, however, continues, modified possibh' bj- the changing of the roof of the fourth ventricle. For this roof structure is subjected to marked alteration in stages of 14 mm. and upwards, both by the lateral development of the chorioid ple.xuses and b}^ the readjustment of the cervical and pontine flexures. Its maximal differentiation may be said to appear at a stage of 18 mm.; this is maintained through several millimeters, until undergoing final retrogression.
 
 
This maximal change in the roof of the fourth ventricle is shown in figures 34, 35, 36, and 37. Several points of interest are brought out in these photomicrographs. Figure 35 represents an enlargement of the rectangular area in figure 34, taken from transverse sections of an embryo pig of 18 mm. The area is particularly well shown in this figure, in which, from the right, the typical ependyma, in a fairly smooth single-cell layer, approaches the differentiated cells in the central portion. On the left, too, similar typical ependyma is shown. In the central area, which has been repeatedly described, the elongated nuclei, the strands of protoplasm, and the ragged, iiTegular intraventricular surface are well presented. The photomicrograph has been reproduced to show the relation of this differentiated area to the various blood-channels in the supporting mesenchyme. Apparently the whole ventricular roof is, at this stage, a site for an extensive capillary plexus; from both sides, as shown in figure 35, vessels (one of great caliber) approach the central area of differentiation. Directly beneath this area smaller capillary channels can be made out, from which, apparently, a sUght extravasation of red blood-cells has occurred. Here, as in the greater part of the basilar pericerebral region, extravasation of the blood-cells is very frequent. This phenomenon has already been pointed out by INIaU'^"".
 
 
The large extent and the great differentiation of this pecuUar area in the roof of the fourth ventricle are well shown in figures 36 and 37, taken from a transverse section of a pig embryo of 18 mm. In the photomicrograph of higher magnification the two sharp tongues of typical ependyma are quite striking. Their abrupt termination in the wide, differentiated area has nowhere been more convincingly shown. The resemblance of these lining cells in the central area to the mesenchymal elements adjoining is here also seen. The most interesting of all the phenomena exhibited in this reproduction, however, is the attachment, apparently by precipitation, of the coagulated albumen of the cerebro-spinal fluid. This coagulation, in this specimen, deUmits the differentiated area in the roof of the fourth ventricle. The phenomenon is seemingly only an ampUfication of a similar attachment of small fragments of the albuminous precipitate shown in other figures.
 
 
Beyond the stage of 18 mm., which may be termed the maximal stage, the differentiated area in the roof of the fourth ventricle undergoes a regression. This is apparently due to the morphological alterations in this rhombic roof. The chorioid plexuses in embryos over 18 mm. long deeply invaginate the fourth ventricle, possibly drawing some of the true roof with them, but surely encroaching upon the mid-hne with their lateral tuftings. This growth tends to decrease the available extent of the differentiated area, but an even more potent factor is the rapid development of the cerebellum. The caudal growth of the cerebellar lip soon largely occupies or replaces the superior half of the roof. These two factors, the cerebellar growth and the enlargement of the chorioid plexuses, render the persistence of the differentiated area impossible, so that a regression or disappearance is to be expected.
 
 
With these considerations before us, the study of sectioned pig embryos of a greater length than 18 mm. becomes important. The process of disappearance, however, does not occur at once. Thus, in an embryo pig of 19 mm. (figs. 42 and 43) the differentiated area is as large and as characteristic as in the stage of 18 mm. This same aijpoarance and maintenance of size may be observed through the next several millimeters' growth, but in pig embryos of 23 mm. the chorioid plexus has usually developed to such an extent that a continuation of the former size becomes impossible. This is shown in figures 44 and 45. Figure 45, the enlarged squared area from figure 44, is a photomicrograph from a pig embryo of 23 mm. The differentiated area, duo to the factors favoring its regression, now appears in close proximit}^ to the chorioid plexus. It has more the appearance of a degenerating area at this stage than in any of the younger embryos, but it still shows a characteristic delimitation of both edges — on the one from the typical ventricular ependyma, and on the other from the differentiated ependj'ma of the chorioid plexus. The cytoplasmic strands of the area which forms the ventricular border do not show to advantage in the photomicrograph, but the .same ragged character with the covering of coagulum may be made out. The process of regression, mechanical as it perhaps is, has begun at this stage in the pig, and in the course of the next few millimeters' growth will become even more active.
 
 
With the encroaclmaent of the chorioid plexuses and the do\\Tiward growth of the cerebellar lip, the superior portion of the ventricular roof soon disappears, and is practically non-existent in embrj^os of 30 mm. and more in length. The differentiated area thus encroached upon from the sides and above becomes a mere vestige of its former size. Thus in a pig embryo
 
 
of 32 mm. (figs. 46 and 47) it appears as a very small break in tli(> lining continuity of the ventricular epend^Tna. Without the intermediate stages such a picture would undoubtedly be considered as an artificial erosion of the ependjinal lining of the ventricle, but when studied in connection with figure 45 the true vestigial character of the area becomes established.
 
 
The final fate of this differentiated area in the roof of the fourth ^•entricle is a complete disappearance, with the occupation of the region by chorioidal epithelium and cerebellum. In this study it was impossible to find traces of the differentiated areas in pig embryo
 
 
s of over 33 mm. in length; vestiges maj' persist, but so small as to present difficulties of decision. The persistence of such a differentiated vestige in rare instances would not be surprising; the transitory character of the area and the method of disappearance make this seem not unlikely.
 
 
This transitory area of differentiation in the roof of the fourth ventricle of the pig has not, so far as can be determined, been noted or described bj- any p^e^•ious author. His'25\ in a retouched photomicrograph of a sagittal section of a human embryo of 17 mm., reproduced the area as differentiated from the roof, but he has made no comment upon it. I have called this differentiated area in the superior portion of the rhombic roof ventricle the "area membranacoa superior ventriculi quarti." This terminology is based on the anatomical character of the area as a continuous membrane, but chiefly on its physiological significance. For, as will be shown in the succeeding section of this paper, the transit of embryonic cerebrospinal fluid from ventricle to periaxial tissue occurs in this area, which functions apparenth- as a physiological membrane. With such a physiological conception of the area, the ttim "area membranacea" seems most suitable, inasmuch as it also meets the anatomical requirements.
 
 
 
THE AREA MEMBRANACEA SUPERIOR IN THE HUMAN EMBRYO.
 
 
The finding of the diflferentiated area in the superior portion of the roof of the fourth ventricle in the embryo pig suggested the value of a study of the same region in the human embrj^o in the further solution of the problems underlying its occurrence. Hence this region in the roof of the fourth ventricle has been examined in the sectioned human embryos of the Department of Embryology of the Carnegie Institution of Washington. It was found that a similar area occurred in the human embr^'o of approximately the same age.
 
 
The study of the roof of the fourth ventricle is usually more difficult in the human embrj^o than in the pig. This is due to the fact that the roof of the fourth ventricle quickly suffers from poor fixation and dehydration — collapse or inversion of the whole structure being commonly met with. It is rarely possible, in the younger embryos, to secure the most satisfactory fixation, whereas in the pig these factors may be controlled as desired. Furthermore, the undue pressures to which the human OAaim is frequently subjected in abortion may cause crushing of the more delicate parts of the nervous sj^stem.
 
 
It is probably best, in the human embryo as in the pig, to trace the formation of the area membranacea superior ventriculi quarti from its beginning, through the various differentiations.
 
 
In a human embrj^o of 4 mm. (No. 836 of the Collection of the Carnegie Institution of Washington) the entire roof of the fourth ventricle is composed of cells with round or slightly oval nuclei and palely staining cytoplasm. The nuclei of the cells are poor in chromatin material as contrasted with the pyknotic character of the typical ependymal cells. The Uning tissue is of the thickness of several cells. The ventricular cytoplasmic border is fairly smooth at this stage. This characteristic ventricular lining is shown in figures 40 and 41, both taken from embryo No. 836. The whole picture is similar to that exhibited by the pig embryo of 8 mm. (figs. 24 and 25).
 
 
A similar accumulation of epitheUal-Uke cells is found in a human embryo of 7 mm. (No. 617 of the Carnegie collection). This is pictured in figures 48 and 49. The photomicrograph of higher magnification shows these poorly staining cells heaped up in a rather localized part of the ventricle, fairly sharply dolimitod from the adjoining ventricular lining. This accumulation of cells in the roof of the ventricle invariably occurs, and it must not be considered as being due to the distortion of the ventricular roof. The reason for the asymmetry of the rhombic roof shown in these figures hes in the fact that in this embryo, as in practically all the embryos of similar stages in this collection, some degree of distortion of the roof of the fourth ventricle is present. Photomicrograi)hs (figs. 50 and 51) taken more posteriorly (from embryo No. 617) give strong evidence of this distortion. They are reproduced not only to show the possible distortion, l)ut also to give a further picture of the lining of the ventricle, with its epithelial-like cells in several layers (fig. 51).
 
 
Similar accumulations of these epithelial-like cells are to be found in human embryos of 9 mm. Reproductions of a much fragmented specimen of this size (No. 721) are given in figures 52 and 53. In the latter figure the complete occupation of the ventricular roof by these cells is well illustrated. Moreover, the specimen shows the many-layered stage to a degree but seldom found. It is unfortunate that such a degree of fragmentation and distortion is found throughout this specimen.
 
 
Thus far, in human embryos up to and including 9 mm. in length, the roof of the fourth ventricle has shown the same architecture as appears in the pig. As will be recalled, the first evidence of a further differentiation of these cells in the pig embryo was found at a stage of 11 mm. (figs. 26 and 27). In one human embryo of this stage (No. 544) a distinct break in the roof of the fourth ventricle can be made out. This is shown in two photomicrographs (figs. 54 and 55). The picture in this cas3 is somewhat obscured bj' the shrinkage and distortion of the ventricular roof, but a distinct differentiation of the lining epithelium can be made out. On the caudal side of figure 55 considerable nervous tissue is seen. Just superior to this (toward the left) the Uning tissue is almost lacking, a few nuclei, only, preserving the contour of the ventricle. Above this area appears again the ventricular lining of many layers of cells. It has been quite difficult to interpret these findings. The area under discussion shows a rather tj'pical adherence to the coagulated albumen ; there is evidence of its extension also into the adjacent mesenchyme, a finding observed in no other similar stage. The caudal position of the opening, the character of the tissue approximating the ventricular cavity, and the presence of the albumen in large amount in the adjacent mesenchyme — all indicate that in great measure the pictures presented in this specimen are largelj^ artifacts. It seems most likely, though, that some differentiation of the tissue in this area has occurred.
 
 
In a human embryo of 14 mm.,* as in the pig of the same stage, the area membranacea superior has attained a great degree of differentiation. This is particularly well shown in figures 56 and 57, the latter being an enlargement of the squared area in the former. These photomicrographs are from embryo No. 144 of the collection of the Carnegie Institution of Washington. Figure 57 shows a characteristic which distinguishes the area membranacea from that of the pig, although in the later stages of the pig embryo (figs. 45 and 47) this feature is present. This concerns the marked decrease of cellular tissue in the membranous area. In figure 57 the deeply stainmg typical ependyma is shown approaching from below. These cells end abruptly at the border of the area membranacea; the ventricle in this area is lined by cells possessing small elongated nuclei and long cytoplasmic processes, which unite to form a ventricular lining. The oval nuclei along the ventricular border become more closely massed together in the superior portion of the area, but nowhere is there the same architecture as in the equivalent stage in the pig (fig. 33). A feature of the histological appearance of the membranous area in the pig embryo is also shown in figure 57; this is the marked adherence of the coagulated albumen of the cerebro-spinal fluid to the area membranacea superior.
 
 
The roof of the fourth ventricle in the human embryo is subjected to the same factors causing changes in the form and relationships which were commented upon in the pig; but these play little part until the chorioid plexuses become of sufficient size to divide the ventricle into a superior and inferior portion. In the human embryo
 
 
, as in the pig, the superior half of the ventricular roof is sacrificed to the greater growth of the cerebellum.
 
 
* Measured on the slide after mounting.
 
 
 
 
 
In human embryos of 17 mm., however, these factors have not begun to influence the membranous area. This is shown in figures 58 and 59, photomicrographs from embryo No. 57G. The section is somewhat to the side of the midline, but in the superior portion of the roof of the fourth ventricle the differentiated membranous area can be made out. The sharp delimitation of this area from the denser t3T5ical ependyma on both sides is quite apparent. The ragged character of the ventricular border, with its few elongated spindles, seems wholly in keeping with the transverse view of this area afforded by figure 37.
 
 
Embryo No. 576 exhibits one characteristic of the area membranacea superior very frequently seen in human embryos, but almost invariably absent in these stages in the pig. Along the lateral margins of the superior membranous area are dense borders of the many-layered epithelial-like cells which lined the ventricular roof in younger stages. This feature is well shown in figures 60 and 61, the latter figure being a higher magnification of the former. The cellular border of the superior area reaches transversely only through a few 15-micron sections, but it extends throughout the whole cephalo-caudal diameter of the area. It seems likely that this represents purely a survival of the epithelial-like cells in the younger embryos. In rarer instances the whole area membranacea superior may be surrounded bj^ such a border of many-layered cells, but even in these cases the superior and inferior margins are quite thin.
 
 
No apparent agencies favoring the disappearance of the superior membranous area m the roof of the fourth ventricle of the human embryo are apparent in stages up to the fetus. Thus, in human embryos of 18 mm. this differentiated area in the roof has reached its maximal differentiation. A section from an embryo of this size (embryo No. 409) is reproduced to show the distortion and its influence upon the topography of the area membranacea. The two photomicrographs (figs. 62 and 63) show the extreme collapse and distortion of the roof of the fourth ventricle. In the figure of higher power (No. 63) the membranous area appears facing posteriorly, due to the shrinkage; the proper leader runs to this area. It shows the differentiation from the adjoining tj'pical ei)endyma which is characteristic of the full}' developed area membranacea superior.
 
 
In a beautifully preserved and sectioned human embryo of 21 mm. (No. 460) in the collection of the Carnegie Institution of Washington the area membranacea superior appears as a sharply delimited area (figs. 64 and 65). Thes(> figures give a very good idea of the definiteness of th(> area when the fixation and dehydration approach the perfect. The tissue of this membranous area lining the ventricle here appears to be wholly lacking in an epithelial covering; the mesenchyme seems to serve as the ependymal lining. Study of this area, however, through different stages argues most strongly against such a view.
 
 
 
he process of regression of the area memhranacea sujjerior in the human embryo differs somewhat from that described in the pig. This alteration in the mode of disappearance is largely due to the fact that in the period of growth from 20 to 35 mm. the superior portion of the roof of the fourth ventricle in the human embryo is not sacrificed to the cerebellar lips; for in the human the cerebellum grows largely into the fourth ventricle, enlarging beneath the superior part of its roof. Thus, the attachment of this part of the roof is not greatly interfered with by the rapid development of the cerebellum. The total extent, then, of the superior portion of the roof is hardly altered in these stages in the human, while in the pig embryo the roof is shortened by its attachment to the inferior portion of the cerebellar hp, which retains its earUer characters. These differences in the relationship of the superior portion of the ventricular roof in human and pig embryos may be seen by comparison of figures 74 and 89.
 
 
Another factor which renders the mode of disappearance different in the two embryos concerns the greater tufting and development of the chorioid plexuses of the fourth ventricle in the pig. This greater size and complexity of the plexus causes an encroachment upon the roof structures which, in the pig embryo, seems of considerable importance in the t'nal closure.
 
 
In the human embryo, however, it has been found very difficult to explain the final disappearance of the superior membranous area on the same mechanical factors w^hich seemed so well to account for its transitory characters in the pig; but at approximately the same stage of growth the process of regression occurs in the human fetus. The area maintains a fair size in stages up to a length of 23 mm. Thus, in figures 89 and 90 (No. 453 of the Carnegie collection) a sagittal section from a human fetus of this size is illustrated. In the higher power (fig. 90) the superior membranous area is sho^\Ti, rather sharply dehmited on its superior border by the t}i:)ical, dense ventricular ependyma. Below, its edge is irregularly formed by the deeply staining ependyma over the invagination of the chorioid plexus. The cell-character of this area resembles that shown in the photomicrographs from the specimen of 21 mm. (figs. 64 and 65). There is left in the area no indication of the cellular architecture which characterized the original ventricular ependj-ma; the cells with their elongated cj'toplasmic processes here have the oval nuclei which are found almost invariabh' in this membranous area.
 
 
In the human fetus of 26 mm. (Xo. 1008 of the collection of the Carnegie Institution of Washington) there is but sHght evidence of a superior membranous area in the upper portion of the roof of the fourth ventricle. The evidence present in this specimen consists in a locaUzed thickening of the lining cells of the ventricle in the situation of the area in other stages. This thickening is illustrated in figures 91 and 92; it consists of several layers of epithehal-hke cells, similar in all respects to the many-layered border shown in figure 83. The picture is somewhat obscured by the vascular pl-^xus directly beneath the ventricular lining.
 
 
There is difficult)' in determining exactly when the last evidences of the superior membranous area in the roof of the fourth ventricle may be found. This is due to the likelihood of artifacts disturbing the character of the ventricular lining in human material, where the freshness and fixation of the specimen may not be ideal. In the larger specimens in the collection of the Carnegie Institution, which are well fixed and sectioned, the existence of the area membranacea superior could not be wholly verified. Thus, in specimen 405 (26 mm.) the presence of the area seemed probable though not definite. In another embiyo of this same size (So. 782) the existence of this area was still more questionable. In a larger embrj^o (30 mm.. No. 75) the presence or absence of the area could not be assured; many indications suggested its existence, but the resemblance to an artificially separated ependyma was strong. In all specimens of human embryos of over 30 mm. examined, no evidence of the area membranacea superior could be found. It appears likely, then, that the final disappearance of this differentiated area in the roof of the fourth ventricle occurs at a slightl}' earlier stage in the human embryo than in the pig.
 
 
 
The final disappearance of the area membranacea superior in the human embryo is not accompanied by the same ingrowth of typical ependyma that characterizes the process in the pig. There is a great tendency, in the human, as indicated in figure 92, for a replacement of the area by the same type of epithelial-like cell which comprised the whole ventricular roof in the earlier stages (fig. 41) and later formed lateral borders for the superior membranous area (fig. 83). Thus, in a human embryo of 24 mm. (No. G32 of the Carnegie collection) there is evidence of a very small membranous area surrounded by a border of epithelial-like cells. In a slightly larger specimen (No. 840, 24.8 mm.) the whole membranous area is occupied by the epithelial-like cells. The frequent association of these cells with the area indicates that in disappearing the area membranacea is probably replaced first by these cells, which in turn disappear, so that the whole roof is finally composed of the typical, densely staining ependyma.
 
 
THE AREA MEMBRANACEA SUPERIOR IN OTHER ANIMALS.
 
 
In order to ascertain whether the area membranacea superior existed in other animals examinations of serial sections of the rabbit, cat, sheep, and chick of suitable stages were made. All of these animals were found to possess a differentiated area in the roof of the fourth ventricle.
 
 
Opportunity was afforded for the study of serial sections of the head of a chick* of 121 hours' incubation. The head was carefully dehydrated and embedded by Dr. E. R. Clark, and was subsequently sectioned bj' Dr. C. R. Essick. The material was beautifully fixed and dehydrated, showing practically no evidence of shrinkage. Typical portions of the superior membranous area are reproduced in figures 66, 67, 68, and 69. Figure 67, taken near the crown of the embr3-o and representing the squared area in figure 66, shows the two dense masses of ependyma separated by the more lightly staining area membranacea. The cellular character of this differentiated zone resembles more the histological features of the similar afea in the pig than those of the human embryo. This resemblance is also to be seen in figure 69, taken more posteriorly than the two preceding figures. The dense ependynia approaching on both sides is sharply delimited at the edge of the hroad membranous area. This is composed of cells having elongated, chromarin-poor nuclei, and long cytoplasmic processes, which form the ventricular roof. The adlierence of the albuminous coagulum occurs here also.
 
 
 
* Tlii9 chiuk mciiDurcci 14 mm. in 40 par cout alcohol.
 
 
 
 
 
In the rabbit the occurrence of the superior membranous area was verified as in the other species studied. In a rabbit embryo of 13 mm. (series x in the embryological collection of this laboratorjO the area was well differentiated from the surrounding typical ependyma. The cells of the area resembled those of the adjacent mesenchyme. The ventricular surface was roughened bj- the projection of numerous protoplasmic processes. An albuminous coagulum was attached to the cells of the membranous zone.
 
 
One sheep embryo from the collection of this laboratory was also studied. The sections, although labeled as an embryo of 10 mm., resembled in every way a pig embryo of 18 mm. The area membranacea was easily identified in the roof of the fourth ventricle; it is similar in every respect to the same area in the pig and the human embrjo.
 
 
In a cat embryo of 10 mm. a smiJl but highly differentiated area membranacea superior was made out. The most striking feature in this specimen is the great adherence of the coagulated albumen to the cells of the area and the resemblance of these cells to the mesenchymal elements adjacent. The edges of this differentiated area are sharp and clear-cut.
 
 
No attempt was made to identify the area membranacea superior in other animals — as further suitable material was not immediateh- available. The chief study has been made on pig embryos and on human embryos. The occurrence of the area in the cat, sheep, and rabbit probably indicates its existence in aU mammals. The finding of such an area in the chick is also suggestive.
 
 
GENERAL CONSIDERATION OF THE AREA MEMBRANACEA SUPERIOR.
 
 
The occurrence of a definite area of differentiation in the superior portion of the roof of the fourth ventricle has been pointed out in preceding subdivisions of this paper. It has been described in detail in the pig embryo and in the human embryo
 
 
; it has been identified also in cat, sheep, rabbit, and chick embryos. It remains here to discuss the general characteristics of this area.
 
 
No description of such an area of differentiation in the ventricular roof has been found in the literature. It may be that the distortion of this structure in the course of the usual embryological technique has rendered its discovery IcvSS likely. His'-='', m his description of the ventricular roof, has not commented upon the occurrence of this membranous area, even though in a retouched photomicrograph of his fetus C-1 (a human specimen, of the beginnmg of the third month) the area membranacea superior can be made out. Likewise in his description of the pUca chorioidea he faJs to mention anj' differentiated areas in the roof, although plate I, in his "Die Entwickelung des menschUchen Rautenhims, von Ende des ersten bis zum Beginn des dritten Monats," shows a slight irregularity in the roof. Practically all of the contributions to the anatomy of the roof of the fourth ventricle deal with the lower half of the structure, with particular reference to the occurrence of the foramen of Magendie.
 
 
The general biological process involved in the formation of the area membranacea superior concerns a dilTerentiation of the epidermal elements which Une the ventricular cavity. This differentiation, both in human and in pig embryos, first begins with the occurrence in the ventricular roof of an area of epithelial-hke cells. These, in the course of enlargement of the roof, become more or less isolated in the superior portion of the structure, and then undergo a metamorphosis into the typical cells of the membranous area. They are characterized by oval or elongated nuclei (rather poor in chromatin as compared with the nuclei of the typical ependjonal elements) and by cytoplasmic strands (in which the cell-boundaries are very poorly marked) which compose the ventricular border. The ventricular surface in the area membranacea is more ragged and u-regular than where lined by typical ependjana. In many instances, as in figure 57, from a human embryo of 14 mm., this transformation has proceeded to such an extent that the epithehal character of the lining cells is almost wholly lost, and the ventricle seems, in this area, to be lined by mesenchyme. Study of the membranous area in many stages convinces one that such an hypothesis is untenable; in every case the ventricle must be considered as being lined by epidermal elements, no matter to what extent the process of differentiation has proceeded. There is no real evidence to support the view that the ependymal lining of the ventricle has been replaced by mesenchymal elements to form the area membranacea superior.
 
 
In general the area membranacea superior is a rounded oval. Its measurement is quite difficult except when fixation and dehydration have been excellent, because of the highly abnormal distortion of the ventricular roof which frequently occurs in the technically poor specimens. Measurements have been made in a considerable number of favorable specimens, both of human and pig embryos. With the history of this area in mind, it will be realized that the size of the structure necessarily varies with the length of the embryo, attaining its greatest dimensions at about the length of 18 or 20 mm. Herewith is a short table of the measurements taken.
 
 
Dimensions of area membranacea superior.
 
 
 
 
 
Species.
 
 
 
No. of specimen.
 
 
 
Length of embryo.
 
 
 
Width of area.
 
 
 
Length of area.
 
 
 
Species.
 
 
 
No. of specimen.
 
 
 
Length of embryo
 
 
.
 
 
 
Width of area.
 
 
 
Length of area.
 
 
 
Pig
 
 
98 107 144 119 106
 
 
mm.
 
 
12
13
14*
14
14
 
 
mm. 0.37 0.95 1.25 0.45 0.65
 
 
7iim. 0.5 0.4 1.1 0.6 0.85
 
 
Pig
 
 
1 mm.
 
 
121 1 16
576 1 17
108 18 (?)
45 18
84 1 22
 
 
mm. 0.6 1.5 0.8 0.9 0.8
 
 
mm.
 
 
0.48
0.9
0.8
0.4
0.7
 
 
Rabbit
Human
Pig
 
 
Human
Sheep
Pig
 
 
Chick
 
 
Pig
 
 
 
 
 
 
 
 
 
 
 
 
In a rough way, then, we may consider the area membranacea as an oval; in some cases the longitudinal diameter exceeds the lateral, and in others the reverse holds. The measurements given above were taken from mounted sections and are probably somewhat disturbed by the histological tcchnifjue which was followed.
 
 
 
* MuisurecI on alidc uftpr iiiouDting.
 
 
 
 
 
The borders of this oval area membranacea are usually fairly regular and smooth, but in some instances they are irregular, due to the fact that small extensions of the area run into the bordering ependyma. These extensions are more commonly met with at the stage when the area has reached its maximum size, as in figures 38 and 39, photomicrographs from an embryo pig of 19 mm. The higher power of these two photographs shows two areas in the smoother ependymal wall. These are extensions of the area membranacea, and within a section or two directly connect with the differentiated area. Both of these small spots on the circumference resemble technical errors; their ragged appearance, the relative excavation of their surface, and the intact ependymal borders w'ould seem to encourage such a view; but when considered in connection with the character of the whole area membranacea they assume a definite relationship in this regard. Other similar areas, rather rare in occurrence, are found separated entirely from the main area membranacea. These isolated areas are of the same size as those shown in figure 39. In significance and character they are probably identical with the larger area membranacea superior.
 
 
Most of the general features of the area membranacea superior have been commented upon in descriptions of the various stages of differentiation in both pig and human embryos. The characteristics most commonly observed concern the differentiated character of the cells of the area, the sharp borders of the typical ependyma, the ragged ventricular surface throughout the whole extent, and the peculiar adhesion of the albuminous coagulum from the embryonic cerebro-spinal fluid to the lining cells. The area membranacea superior should be considered, then, as a transitory focus of differentiation of the typical ependymal hning of the roof of the fourth ventricle.
 
 
AN UNDESCRIBED AREA IN THE INFERIOR PORTION OF THE ROOF OF THE FOURTH VENTRICLE.
 
 
With success attending the effort to find in the superior portion of the rhombic roof an anatomically differentiated area which would furnish a morphological basis for the jihysiological phenomenon of the extraventricular passage of the cerebrospinal fluid, attention was necessarily directed to the inferior portion of this roof (considering the whole roof structure to be divided by the chorioid plexuses). The spread of the replaced injection fluid (fig. 4) into the periaxial tissues through two points in the roof of the ventricle suggested a study of this stage (pig embryo
 
 
of 18 mm.) as the basis of the investigation. As a histologically differentiated area in this inferior portion of the roof is easily made out, the complete history of the area will be given chronologically. It has been termed the "area membranacea inferior ventricuU quarti," the terminology being based on the same physiological and anatomical features which led to its adoption in the case of the analogous area in the upper portion of the roof.
 
 
 
 
 
THE AREA MEMBRANACEA INFERIOR IN THE PIG EMBRYO.
 
 
The inferior portion of the fourth ventricle shows no evidence of a differentiation from the typical lining ependyma until the length of 15 mm. is reached. In this development consideration must be given to the factors concerned in the process. It will be recalled that in the younger embryos, both pig and human, up to and including a length of 9 mm. the whole roof of the ventricle is occupied by the epithehal-like cells. With rapid growth of the medulla and corresponding enlargement of the fourth ventricle the roof becomes elongated and widened. This process results in the isolation of the area composed originally of the ejiitheUal-Uke cells and the subsequent formation of the superior membranous area. The epithehal-like cells remain in the superior portion of the enlarged ventricular roof, while the whole inferior half is composed of the densely staining, typical ependyma. The division of the roof by the laterally developing chorioid plexuses becomes evident in pig embryos of 14 mm. At this stage the whole inferior portion shows a ventricular lining composed of the typical ependyma.
 
 
The first indication of a differentiation in this inferior half of the roof was found in a pig embryo of 15 mm. This is illustrated in figures 70 and 71. The sagittal section from which these photomicrographs were taken is near the mid-line of the embryo, as is indicated by the partial section of the central canal of the spinal cord (fig. 70). The division of the ventricular roof into two parts is also indicated in figure 70 by the invagination of the chorioid plexus. The squared area in the lower half is reproduced in figure 71 under higher magnification; here the first evidence of an ependjonal differentiation is observed. The dense Une of the typical ependjona appears from both sides, but in the center of this ventricular lining a small area of differentiation is seen. This area, isolated by the abruptly terminating pyknotic ependymal elements, is composed of two or three layers of less deeply staining cells. The nuclei are round, rather larger than those of the adjacent mesenchyme, and contain httle chromatin. The cytoplasm stains fairly well with eosin and is not scanty in amount. The cells resemble those epithelial-hke elements which so largely make up the ventricular roof in the earlier stages. No albumen is found near this point of differentiation, although the whole ventricular cavity is filled with the normal amount. In figure 70 the marked zone of the area membranacea superior may easily be seen.
 
 
After this initial indication of a differentiation in pig embryos, the further differentiation of the tissue proceeds but slowly until the length of 18 mm. is attained. Thus, in a similar specimen from an embryo pig of 18 mm. the area of differentiation is not greatly hicreased in size. This is shown in figures 72 and 73. In the higherpower figure (fig. 73) both the superior and inferior membranous areas can be made out by the attachment to these areas of the protein coagulum of the ventricular cerebro-spinal fluid.
 
 
In the higher-power figure (fig. 73) of the squared area from figure 72, the area membranacea inferior shows the same character as exhibited by the specimen of 15 mm. (fig. 71). The opening maintains the same approximation to the lateral lip of the medulla, but the area is larger and the histological character more nearlj' approaches the permanent feature of the tissue. The nuclei in this zone are paler than those of the adjoining ependymal elements and contain less chromatin. The cytoplasm is not scanty, nor is it very abundant in amount. The area is also characterized by the occurrence of the cells in a layer, two or three cells in thickness.
 
 
In view of the very slow differentiation of the area membranacea inferior in the growth of the embryo from 15 to 18 mm., the enormous enlargement of the region within the next few miUimeters' growth is very astonishing. This period, as has been jjointed out, is a critical one in the extension of the embryonic cerebro-spinal fluid from a ventricular to a periaxial relationship. Apjjarently, in the course of the embryo's growth during these next few millimeters the whole inferior roof of the ventricle undergoes a transformation and enlargement, so that the differentiated area membranacea comes to occupy practically the whole inferior half of the roof. This portion of the roof, persisting, enlarging, and suffering no extension of nervous t'ssue upon it, becomes the tela chorioidea inferior.
 
 
The raj)id differentiation of the whole iiiferior half of the roof of the fourth ventricle is a very interesting process. Apparently the typical ependymal elements, visible on both sides of the membranous area in figure 73, undergo a very rapid alteration, so that in the course of a few millimeters' growth the cubical lining of the ventricle is replaced by a low- type cell, with round or oval nuclei, staining much less densely than do the ependymal elements. The whole area membranacea rapidly becomes a membrane in the true sense of the word; it is a continuous, intact laj'^er of cells, generally only one cell in thickness, closing in the fourth ventricle from the chorioid plexus above and the bulbar lips on the sides.
 
 
The general characteristics of this transformation are seen in figures 74 and 75. These photomicrographs are taken from a sagittal section of a pig embryo of 23 mm. On one side of the sharply deUmited membrane shown in figure 75 is a tongue of nervous tissue of the medulla; on the other is the differentiated ependjina of the chorioid plexus; between these two structures stretches uninterruptedly the area membranacea inferior. The flattened cells of the membrane, with their oval nuclei and almost continuous cytoplasm, effectually close the whole ventricle. The photomicrograph also shows an interesting characteristic of this membranous area which is universally present in the larger forms; this is the relatively unsupported character of the membrane. The highly vascular mesenchyme posterior to the area has gradually developed, during growth, larger and larger interstices between the cytoplasmic processes. The phenomenon is not due to shrinkage, but is intimately connected with the formation of the future cisterna cerebello-medullaris. This phase of the mesenchymal differentiation will be more fully considered in an appropriate section of this paper. It will suffice here merely to record the lack of support of the membrane.
 
 
Another phenomenon of unportauce in the cerebro-spinal fluid relationships of this stage is shown in figure 75. In the mesenchjanal spaces directly beneath the membranous area there is a large amount of albuminous coagulum. This phenomenon does not occur to any appreciable extent in earlier stages or in other parts of the mesenchyme, except about the nervous system. The c1o.se association of the coagulum from the ventricular cerebro-spinal fluid with the inner border of the area membranacea (shown in figure 75 as a slight roughening of the border) is of very great significance in this connection. In one i)oint in the membranous area (fig. 75) the albumen can be traced almost without interruption from the ventricle into the wide spaces of the mesenchyme (cf. fig. 8). This observation strongly suggests that the embryonic cerebro-spinal fluid, which is rich in protein material, is passing, in this stage of embryonic growth, from the ventricle into the periaxial mesenchyme; and such an interpretation becomes established by the comparative findings in the embryo of the same stage in which a replacement of the cerebro-spinal fluid by the ferrocyanide solution had been effected. These comparable findings are surelj^ of the utmost importance for the final solution of the problems centering about the embryonic cerebro-spinal fluid.
 
 
In the later stages of development of the area membranacea inferior in the pig embryo the same structural relationships persist that are shown in figure 75. Figures 76 and 77 are photomicrographs taken from a sagittal section of a specimen of 32 mm. In the enlargement of the squared area, from the first of these figures, the continuity and completeness of the membrane are well established. The photograph shows well the flattened character of the cells comprising the membrane and its sharp differentiation from the nervous tissue and ependyma below and from the ependj'ma and chorioid plexus above. Most important in this case is the distribution of the albuminous coagulum. Within the ventricular cavity this appears in considerable amount, and in several places it is in close adhesion to the lining area membranacea. This albuminous precipitate may likewise be traced in some places apparently through the cellular membrane into the periaxial spaces. For here, as indicated in figure 75, the clotted albumen from the cerebro-spinal fluid apjjarently exists in large amounts in the space just posterior to the membrane — the future cisterna cerebello-medullaris. Delicate strands of mesenchyme are still observed running through the wide space, but in general the whole tissue has returned to the line of the future arachnoid. The relative lack of substantial support of the membrane is well brought out in figure 77. A characteristic feature of this membrane, which Blake'3^ has championed, and which is indicated in figures 76 and 77, is the posterior bulging of the roof — "the caudal process like the finger of a glove."
Another section from the same pig embryo, taken more laterally, is represented in figures 78 and 79. In the photomicrograi)h of higher power the flattened character of the lining cells, the intactness of the membrane in isolating the ventricular cavity, the unsupported freedom of the membrane, and the relation to the albumen coagulum on both sides are of |)articular interest.
 
 
The ultimate fate of the area membranacea inferior will not be more fully entered into until the early history of the similar area in the human embryo has been detailed. For in this connection the occurrence of the foramen of Magendie requires discussion, and it seems best to delay the further consideration of the present topic until the whole question can be reviewed.
 
 
 
THE AREA MEMBRANACEA INFERIOR IN THE HUHMN EMBRYO.
 
 
The same process in the formation of an area of differentiation in the inferior portion of the roof of the fourth ventricle may also be followed in the human embryo. Unfortunately, however, human omliryological material can rarely be subjected to the immediate fixation and preservation which j'ield excellent histological results in the more plentiful specimens. It does not seem strange, therefore, that the determination of the exact stage at which an area of differentiation can be made out in the ventricular roof should be practicall}' impossible; for, in poor technical procedures, the roof of the fourth ventricle suffers almost more than does any other portion of the specimen.
 
 
In a human embryo of 13 mm. (No. 695 in the collection of the Carnegie Institution of Washington) there is slight evidence of a differentiation in the lower portion of the rhombic roof. The changing character of cells in this specimen is not marked, but as the central portion of this inferior roof is reached the ependymal cells seem to assume gradually a more cubical morphology. Associated with this change in shape, there is also a sUght loss of the deeply staining character of their nuclei. The whole differentiation, however, is slight and would be commented upon only from the conception of this area in the pig embryo.
 
 
The first definite evidence of differentiation in the inferior portion of the ventricular roof was found (specimen 390 in the Carnegie collection) in a human embryo of 15.5 mm. This initial differentiation occurs, then, in the human embryo of approximately the same length as in the pig. The specimen showed the same change in character of the hning ependj-ma as was found in the pig. The deeply staining ependymal elements are replaced in a limited central area in the inferior portion of the roof by cells with more elongated nuclei, poorer in chromatin, and resembling somewhat the epithehal-like cells which early filled the ventricular roof. These cells tend to compose a layer of more than one cell in thickness — a feature particularly noticeable in the peripheral portions.
 
 
The size of the area membranacea inferior observed in specimen 390 suggested that the earUest evidence was probabh' to be observed in somewhat smaller specimens. This could not, with the material at my disposal, be verified, but it is probablj^ safe to assume that the first signs of an ei)end\Tnal differentiation will be found in human embryo
 
 
s of about 15 mm. This time of appearance of the area in the human would coincide with its time of primarj* differentiation in the pig embryo
 
 
. In this limitation of the first appearance of the area membranacea inferior, the standard has been an unmistakable differentiation of ependyma and not an isolated change of a lining-cell or two which might have been the result of the technical procedure. Such a criterion was necessitated by the verj' marked changes in the ventricular borders observed in specimens in which distortion of the chorioidal roof had occurred.
 
 
The area membranacea inferior verj* rapidly increases in extent after the onset of the process of ependymal differentiation. This was hkewise observed in the pig embryo, although perhaps more stages could be made out. In a human embryo of 16 mm. (Xo. 406 of the collection of the Carnegie Institution) the area nierabranacea inferior is quite extensive, as is shown in figures 80 and 81. In the i)hotomicrograph under higher power (fig. 81) the denselj' stained ependj^ma approaches the membranous area (ami) as tongue-hke processes from above and below. These tips gradually lose their dense character and are prolonged as a delicate membrane, lining, in this localized area, the ventricular cavity. The nuclei of the cells here are not heavily laden with chromatin; they are oval and somewhat larger than the more densely packed nuclei of the typical ependj-mal element. Unfortunately, the middle portions of the membranous area in this specimen are surrounded bj'^ extravasated red blood-cells obscuring somewhat the structure (fig. 81). The process, though, of the differentiation of these ependymal elements into paler and larger epithelial-like ceUs is quite apparent.
 
 
As in the pig, the tendency of the differentiated ependymal cells forming the area membranacea inferior to lose in some degree their distinctive appearance and to approach in character the undifferentiated mesenchymal element is apparent in the human embryo very shortly after the original steps in the process of differentiation have occurred. Photomicrographs from two human embryos of 17 mm. have been included to show this phenomenon. Thus, in figure 88, an enlargement of the blocked area from figure 58, the area membranacea inferior (ami) is well defined. The sagittal section from which this photomicrograph was taken is from embryo No. 576, in the Carnegie collection. Above and below the dense fine of ependj-ma may be made out; this tapers quite abrupt!}', to be succeeded bj' the cells of the area membranacea inferior. These cells, products of ependymal differentiation, have lost much of their epithelial-like appearance; they now show rather small, oval or rounded nuclei, poor in chromatin. The cytoplasm of the cells is small in amount, but not disproportionate for the size of the nucleus. The ventricular border of these cells (fig. 88) exhibits a rather characteristic phenomenon, the adherence of a shght albuminous coaguluni. The fine processes of this coagulum fuse^ith the cytoplasmic borders of the cells and render these borders vague and indefinite. Beneath the cells of this inferior area small vascular channels maj'^ be made out. These tend to make the membrane appear denser than its cellular character warrants.
 
 
In another section from this same embryo (No. 576) the inferior membranous area is shown in relation to the tufted chorioid plexuses (figs. 82 and 83). In the reproduction under higher magnification (fig. 83) the ependj-mal lining may be traced caudalwards to a gradual fusion into the area membranacea inferior. From the rather high cubical cells in the immediate proximity to the plexuses the ependymal elements l^ecome reduced in size and in height, and then rather abruptly the pyknotic character of the ventricular lining is lost. This loss of the deeply staining character coincides with the sui)eri()r border of the area membranacea inferior {ami). The membrane of this area shows the same cell-character as already desorilied for this embryo. On the superior side of the plexuses (fig. 83) the lateral border of the area membranacea superior (ams) is shown composed of epithelial-like cells.
 
 
 
The apparent tendency of the cells composing the inferior membranous area to lose the epithelial-like character, as shown in the figures from embryo Xo. 576, is not an invariable phenomenon. Rather is an aggregation of epithelial-like cells met with in human embryos very commonly in this area, not onlj- in embr3'os of small size, but also in small fetuses. This phenomenon is illustrated in figures 84 and 85, reproductions of photomicrographs from a human embryo of 18 mm. (Xo. 409 in the collection of the Carnegie Institution). In figure 85 the total transverse extent of the area membranacea inferior (ami) is illustrated, with the villous chorioid plexuses appearing to the left. Although this membranous portion of the embryo has been distorted somewhat by the technical procedures to which the specimen was subjected, the cellular character of the membranous area is well indicated. The most striking feature, apart from the characteristic tinctorial differentiation from the typical ependymal elements, consists in the marked clumping of the cells in certain parts of the membrane. On one lateral extent the membrane is thickened into a bulbous swelling several cells in thickness. These cells have palely staining nuclei, poor in chromatin, with an oval or round form. In other places in the membrane smaller but no less characteristic clumps of similar cells maj' be made out. Between these cellular aggregations the membrane stretches in a continuous line with but few nuclei.
 
 
Analogous clumps of cells, \\-ith pale, rounded or oval nuclei, may be made out in figures 86 and 87, taken from a human embryo of 19 mm., X^o. 431 in the collection of the Carnegie Institution. Only a small portion of the membrane is reproduced in the figure under higher magnification, but a characteristic clump of epithelial-like cells (epc) is shown. These cells of the differentiated ependyma here again have oval and rounded nuclei, poor in chromatin, similar to those which have been pointed out manj^ times in the foregoing pages. A second broadened area in the inferior membrane is also showm in figure 87.
 
 
The further development of the area membranacea inferior proceeds in the human embryo in a manner very similar to that described for the pig. In the stages but shghtly above those already described the differentiation goes on slowly, but withui a few millimeters the cellular pictures resemble those given for the embryo of 17 mm. (figs. 82, S3, and 88). The cellular clumps which appeared quite frequently in the embryos under 20 mm. have not been found in the larger forms. Thus, in an embryo of 23 mm. (Xo. 453 in the collection of the Carnegie Institution) the inferior membranous area {ami) appears as an extensive membrane comprising almost wholly the inferior portion of the chorioidal roof. The membrane is here of a single cell in thickness; these cells are rather small, with oval nuclei, simulating in some measure those of the surrounding mesenchj-me. The most mteresting phase of the membranous area at this stage of 23 mm. concerns its completed cellular differentiation and its rather slow increase in size.
 
 
Wholly similar pictures of the inferior membranous area of the roof of the fourth ventricle are afforded by a human fetus of 26 mm. (figs. 91 and 92). These photomicrographs were taken from embryo X'o. 1008 in the collection of the Carnegie Institution. In this specimen (fig. 92) the fourth ventricle seems almost to lack a lining of oj^endymal (epidermal) elements in the area membranacea inferior (ami). The cells of this area are small, inconspicuous ua their distinctions from the underlying mesenchjme. The whole character resembles that of the superior area membranacea shown in figure 57.
 
 
The appearances exhibited by the inferior membranous area in the stages above 26 mm. are modified in great part by the development of the great cisterna cerebellomedullaris. As in the pig, the breaking-down of mesenchyme to form this cistern results finally in the almost total isolation of the inferior membranous area. The cistern is fairly rapidlj'' formed when once the process begins, and so in an embryo of 35 mm. (No. 199 in the Carnegie collection) the isolated character of the area membranacea inferior (a?ni) may be easily made out. This is shown in figure 94, an enlargement of the blocked area in figure 93. The general architecture of the membrane, particularly its intact character, appears in this photomicrograph, but its finer structure is obscured by the albuminous coagula which adhere on both surfaces. The cell structure of the area membranacea resembles closely that described in the embryos already pictured.
 
 
Discussion of the final disposition of the area membranacea inferior will be undertaken in the following subdivision of this paper, in order that the findings in the pig and in the human embryo may be correlated.
 
 
GENERAL CONSIDERATION OF THE AREA MEMBRANACEA INFERIOR.
 
 
The ependymal lining of the caudal portion of the roof of the fourth ventricle undergoes a process of differentiation which results in the formation of the area membranacea inferior. This transformation has been observed in pig and human embryo
 
 
s; in both, the first definite evidence of the cellular change has been observed in specimens of 15 mm. The essential phases of the process are identical in the two embryos. The tendency of the deeply staining typical ependymal elements is to lose their highly pyknotic character; the nuclei become poorer in chromatin and the cytoplasm somewhat more abundant. In the first stages of the metamorphosis the lining cells come to assume epithelial-like appearances, but in the final change the nuclei become small oval bodies, poor in chromatin, resembling to some degree the nuclei of the adjoining undiff'erentiated mesenchyme. In the human embryo, a tendency for the epitheUal-like characters to persist in isolated cellular aggregations is apparent.
 
 
After the initial process of differentiation has begun, the area membranacea inferior increases rapidly in extent and the differentiated cells which characterize it come to occupy the greater portion of the caudal part of the chorioidal roof. In the somewhat later stages the area membranacea is almost wholly unsupported by other tissues, due to the development of the cisterna cerebello-medulla
 
 
ris. As soon as the cistern forms, the area membranacea serves as practically the sole dividing membrane between the ventricular system and the future subarachnoid spaces.
 
 
 
The ultimate fate of this area membranacea inferior is necessarily involved in the distribution of the tela chorioidea inferior. Likewise it necessitates a discussion of the possible formation of the so-called foramen of Magendic and its mode of origin from the "caudal process" of Blake. It is proposed to discuss briefly some of these questions in the hope that some phases of the problem may be brought forth.
 
 
It must be clearlj' understood that the questions of the ultimate fate of this area membranacea inferior probably differ considerably in the different species of mammals. In the horse and in the pig the absence of the medial foramen (]\Iagendie) is fairly well established, but in man its existence .seems to rest on equally firm grounds. While, ))rimarily, this investigation has not been concerned with the possible existence of the foramen of IMagendie, the question has been presented manj' times in regard to the pig and human embryos examined.
 
 
As far as can be determined, no descriptive study of the development and differentiation of the inferior portion of the rhombic roof has been published. Heuser's'23' studies on the form of the cerebral ventricles of the pig have afforded a very good conception of the gradually changing relationships in this region. Hess(22j has devoted attention to the histological appearances of the inferior roof in the embryo. One of his interesting obseivations concerns the caudal portion of the rhombic roof in a fetal cat of 10 cm., where he noticed a very sudden interruption in the epithelial lining of the ventricle, with a complete closing by a fibrous net. This description by Hess is the only comment upon the histological appearance of the ventricular roof that has been found. His^^s^ pictures, without comment, in a retouched jihotomicrograj^h, a differentiated area in the proper situation in his fetus C-1 (beginning of the third month) .
 
 
The many writers in embryology have commented upon the roof of the fourth ventricle. Minot^'*<>>, in 1892, stated regarding it:
"Several writers have thought that the membrane was broken through at several points, but it probably is really continuous throughout life. The fourth ventricle is to be regarded, then, as an expansion of the central canal permanently bounded by the original medullary walls."
Kollman(32)^ on the other hand, advances the view that during the third month the rhombic roof is broken down to form the foramen of Magendie and the two foramina of Luschka. Streeter^S'*), in his chapter on the development of the nervous system in the Keibel-i\Iall Handbook of Embryology, advances a similar view. The majority of investigators to-daj' incline to the beUef that the roof of the fourth ventricle in man is perforated to form the median foramen of ]Magendie.
 
 
jjpgg(22) has advanced a conception of the foramen of Magendie that is supported by numerous observations. To test KolUker's statement that the fourth ventricle remained closed during human embryonic life. Hess sectioned the region in human fetuses, new-born infants, and in adults. The lengths of the fetuses cut were as follows: 7, 12.5, 15, 16, and 17 cm. In the 47 cases the roof showed a medial opening (IMagendie), except in one case, in which it was closed by a "thin pial membrane." Hess's conception of the process of formation of this membrane was that in earh' embn'ological life the rhombic roof was bordered by a regular, meshed tissue. Later the small meshes in this tissue fused to form the larger foramen of Magendie.
 
 
Blake's(2) hypothesis of the formation of the medial foramen has been quite extensively quoted in the more recent publications on this subject. In a study of the chorioidal roof Blake found a caudal bulging of the inferior velum; this outpouching became more and more extensive in the older embryos.. In man this pouch became sheared off at its neck, leaving the foramen of IMagendie.
 
 
In addition to the few studies referred to above, there have been in the past 25 years a great mmiber of articles (notablj' those of Wilder^^s) and Cannieu^^^) offering evidence that this median foramen of the fourth ventricle is an existent, functional opening. Into this literature it is not proposed to go in the present communication; it may be stated that in the larger part the views presented have been in favor of the consideration of the true occurrence of the foramen of IMagendie.
 
 
The material on which this study is based has been purely embryological in type, so that no relialile data regarding the foramen of IMagendie could be obtained. But even in the largest fetuses examined, there was no evidence which indicated a breaking-down or a shearing-off of the inferior roof of the fourth ventricle. In the largest human fetus at my disposal, in which the histological material was good enough to permit an accurate examination of the chorioidal roof (embryo No. 448, 52 mm. in the Carnegie collection) the area membranacea inferior appeared as an intact membrane supported only be a few pial cells. In the pig the material at hand has been such that accurate study of the roof could be made in specimens up to 20 cm. ; in all of these later fetal pigs the roof has been wholly without foramina. If, however, in these larger stages the histological procedures have not been of the best, ruptures and other artificial separations are very frequently found.
 
 
The area membranacea inferior, then, may be regarded as a region of ependymal differentiation. Whether it persists as an intact membrane or undergoes, in certain animals, a perforation to form a foramen of Magendie can not be here answered; this study has been concerned solely with the embrj^ology of the cerebro-spinal spaces, and it affords no evidence in favor of or against the existence of such a foramen. Nor has any study been made of the two foramina of Luschka, the two openings from the lateral recesses of the fourth ventricle into the subarachnoid spaces. It can Ije stated, however, that these foramina arc not in existence at the time of estal)lishment of the circulation of the ccrebro-si)inal fluid. This phenomenon, as recorded in the previous section, occurs in pig embryos of 26 mm.; at this time the lateral reces.ses are anatomically and physiologically closed.
 
 
 
VI. PASSAGE OF FLUID THROUGH ROOF OF THE FOURTH VENTRICLE.
 
 
On pages 20 to 'SO is a description of the passage of a true solution, substituted without increase in pressure for the embryonic cerebro-spinal fluid, through the roof of the fourth ventricle into the extravcntricular or periaxial spaces. This extension of fluid occurred in two localized areas, one in the superior half and the other in the inferior half of the rhombic roof. Histological study of these regions revealed a localized differentiation of the ependyraa, both in the upper and lower halves of the ventricular roof. It becomes necessary, then, to correlate, if possible, the areas of this fluid-passage to the anatomical differentiations pointed out.
 
 
THE ACCUMULATION OF INJECTION-MASSES IN THE SUPERIOR MEMBRANOUS AREA.
 
 
It has already been recorded that the first evidence of a change in the reaction to a replacement injection occurred in an embryo about 13 mm. long (fig. 2). This stage was characterized by a dense collection of the precipitated granules in a definite area in the roof of the fourth ventricle. At this stage also the area membranacea superior is well differentiated (fig. 31). That the site of the granular accumulation is this membranous area is easily proved by an inspection of figure 117, which represents an enlargement of the squared area in figure 116. In the low power photomicrograph the prussian-blue granules are not represented, but are found scattered through the ventricles, with a definite collection in the posterior region of the fourth ventricle. Under a higher magnification (fig. 117) the blue can be traced in but small quantity along the normal ependj'mal lining (shown to the left in the figure), but as soon as the differentiated area (area membranacea superior) is reached the granular material is heaped up in a dense mass, which extends as a thickened pad into the ventricle.
 
 
The same phenomenon of the accumulation of the injection fluid in the superior membranous area is shown in figures 112 and 113, the second photomicrograph representing the area outlined in the first, but reproduced under much higher magnification. In this specimen (an embryo pig) a dilute solution of silver nitrate was injected into the central canal of the spinal cord. On histological examination the accumulation of the silver also shown in figure 11 was found. Thus, in figure 113 the ventricular epithelium can be made out in the upper right-hand corner, while below (in the area membranacea superior) the silver is densely accumulated.
 
 
The explanation of this phenomenon of accumulation in the superior membranous area is not whoUj' clear. It occurs only in stages in which the histological differentiation of the ventricular roof has proceeded to some degree and in stages where the fluid-passage into the periaxial tissues is not wholly unobstructed. This aggregation of the precipitated granules of prussian-blue and of the reduced silver in a locaUzed area certainly suggests a phj'sical exi^lanation, as in these cases the physical laws of precipitation and reduction must hold. The many figures of the superior membranous area of the ventricular roof show that in the stage under consideration the cell-outlines projecting into the ventricles are rough and ragged as contrasted with the smoother and more regular surface of the adjoining ependyma.
 
 
 
Could not these roughened, irregular cell-surfaces become the site of the first and most extreme precipitation of the prussian-bluc and of the reduction of the silver? Certainly they would serve much more efficiently as the foreign substances about which precipitation would occur in greatest amount. This phj^sical explanation finds man}' arguments for its suj^ijort in these studies.
 
 
Another explanation of the phenomenon concerns the normal flow of the fluid and the relation of the direction of this flow to the roof of the fourth ventricle. As has already been emphasized, it is difficult to assume that there is any marked production of cerebro-spinal fluid before the periaxial spread occurs. Such an assumption would argue against the development of any special current toward the roof of the fourth ventricle in any stage smaller than that represented in figure 3, and would vitiate the explanation of the occurrence of the granular accumulation shown in figure 2 (a pig embryo of 13 mm.). In the later stages (16 mm., cj. fig. 11) this explanation would probably suffice for the phenomenon exhibited.
 
 
THE SITE:5 of FLUID PASSAGE THROUGH THE ROOF OF THE FOURTH VENTRICLE.
 
 
With consideration of the evidence presented as to the accumulation of the precipitates of the injected fluid about the area membranacea superior during certain stages in the development of the cerebro-spinal spaces, it would seem that the same area must be concerned in the passage of fluid from the ventricular cavities into the periaxial tissues. This view receives support from the reproduction of a cleared specimen (fig. 11) in which an injection of silver nitrate had been made into the central canal of the spinal cord. The pressure employed was great enough to force the fluid into the periaxial spaces, but the resultant picture clearly showed the oval outline of the area membranacea superior.
 
 
The study of the passage of fluid from the ventricular to the extraventricular spaces can best be made by simple histological serial sections. In these observations pig embryos in which the cerebro-spinal fluid had been replaced by the compensating device, supplying a true solution of potassium ferrocyanide and iron-ammonium citrate, were sectioned and examined with reference to the sites of fluid passage. The results of these studies are given here in order that the whole question of the connection of the cerebral ventricles with the subarachnoid spaces may be discussed.
 
 
In the stage represented by figure 3 (in which fluid passes from one area in the roof of the fourth ventricle into the extraventricular tissues) histological sections show that the point of fluid ])assage is localized and concerns solely the area membranacea superior. The replaced fluid (as demonstrated by the subsequent precipitation of the prussian-blue) passes through this entire membranous area into the adjoining mesenchyme. The process is wholly confined to this area; the adjoining ependyma is entirely imj)ervious to the ferrocyanide. This phenomenon of passage of the replaced fluid through the sujjerior membranous urea is well shown in figures 14, 18, and 23.
 
 
The distriltution of the minute granules of ])russian-l)lue in the cells of the superior membranous area is of iini)ortance in any discussion of the passage of fluid through a membrane; for this area (in the superior portion of the roof of the embryonic fourth ventricle) must be considered as a memljrane permeable in certain degrees to the fluids bathing it. That the area mcml:)ranacea is intact and does not contain stomata or other minute foramina has been demonstrated histologically. Further evidence of the entire lack of intercellular stomata is afforded by the distribution of the prussian-blue granules precipitated in situ after the replacement of the cerebro-spinal fluid by the ferrocyanide solution.
 
 
Figure 14 is a reproduction of the superior area from a transverse section of a pig embrj^o in which the routine replacement had been made. The position of the area is shown by the squared outline in figure 13. On both sides the impermeable ependyma is seen, with granules of the blue adhering to the ventricular border of the cells, but not penetrating them at all. To the left of the drawing the few ependymal cells possess, beneath their central border, a chain of the granules which have entered from the abrupt edge of the area membranacea. In the cellular border between the two hps of the ependyma, the area membranacea superior, the passage of the replaced fluid is easily made out by the resultant blue granules. The area is roughly deUmited by a ventricular collection of the blue granules. Examination of these cells shows that the prussian-blue is present within the cytoplasm, avoiding the nuclei with perfect precision. Some of the cells are rounded and almost free from the granules; others, particularly those whose cj'^toplasm is elongated, are completely filled with the granules, the nuclei standing out in a blue granular cytoplasm.
 
 
The question of the passage of the fluid between the cells must also be answered by the histological evidence. In the same drawing (fig. 14) in one or two places there are indications of a slight stream of granules between the cells of the area membranacea superior. This apparent transit of the fluid through intercellular passages is particularly clear in the small areas where the cellular cytoplasm is relatively free from the granular deposits. But upon careful examination of these areas under oil immersion it is always apparent that the adjoining cytoplasm is also mvolved in the granular precipitation, indicating that the cells, although almost free from the deposit, are also engaged in the process of the fluid passage. Compared to the whole area of fluid transit, the points indicative of a passage through possible intercellular stigmata are almost negUgible. It seems not unlikely that the outlining of canals between cells may be a physical phenomenon, as in most cases no cellular borders (as demonstrated bj' the precipitated granules) can be made out. These pecuUarities of fluid passage may be seen in figures 14, 18, and 23.
 
 
Consideration of all the evidence afforded by histological examinations of the essential character of the area membranacea superior and of the passage of fluid through it incUnes one inevitably to the belief that this area functionates as a cellular membrane. The fluid passes through it as through any permeable liv-ing membrane. Histologically the passage is for the most part through the cj-toplasm of the cells, but occasionally an intercellular course is suggested. Both processes are wholly compatible with the accepted view of a cellular membrane de\dsed for the passage of fluid through it.
 
 
 
 
The same phenomenon of the passage of fluid from the fourth ventricle into the periaxial spaces is beautifullj' illustrated in figure 23. This drawing is from a transverse section of a pig embryo (23 mm. in length) in a stage when the superior membranous area is rapidly being encroached upon by the developing cerebellum and by the caudal chorioid plexuses. Between the deeplj' staining epondj-mal cells on cither side the membranous area is densely outlined by the deposition of the granules of prussian-blue in the cytoplasm of the cells of the area membranacea superior. The avoidance of the nuclei of these cells by the ferrocyanide is well demonstrated in this reproduction, as is also the impenetrability of the ependymal cells. In a specimen of this nature the question of the passage of the injection fluid through possible intercellular foramina loses its significance; for the drawing shows clearly the importance of considering the entire area membranacea as a functioning whole — a permeable, Uvmg, cellular membrane.
 
 
It has been shown in a foregoing section of this memoir that histologically the area membranacea superior decreases to an almost neghgible remains in specimens of embryo pigs over 30 mm. long. This same rule apparently holds for its functional importance, as determined by the relative and absolute amount of prussian-blue granules deposited in the cells of the superior membrane. This decrease in the functional imi;)ortance may be inferred from figure 47, a photomicrograph from a pig embryo of 32 mm. Apparently the size of the membrane determines in large measure the amount of the replaced fluid which passes through it.
 
 
Thus far we have been concerned solely with the passage of fluid through the area membranacea superior. In the earUer stages of from 14 to 23 mm. the importance of the superior membrane functionally is great, but in the later stages the inferior membrane assumes far greater significance. This is demonstrated not only by the structural history of the two areas, but by the functional index afforded by the replacement of the cerebro-spinal fluid by a foreign solution.
 
 
In the foregoing section the first evidence of any histological differentiation in the inferior portion of the roof of the fourth ventricle was shown to occur in pig embryos of 15 mm. in length. From this stage upwards (figs. 4, 5, etc.) a portion of the inferior roof allows fluid to pass through it. The exact point of fluid iiassage is the localized ependymal differentiation forming the area membranacea inferior. This relationship is easily verified by reference to figure 18. In this drawing of a median sagittal section of a pig embryo the two localized pomts of fluid passage into the periaxial tissue are readily identified; they are quite Umited in comparison to the extent of the periaxial spread.
 
 
Figure IG represents the inferior membranous area of the roof of the fourth ventricle from a i)ig embryo of .similar size (18 mm.). The histological character of the inferior area is well shown in this drawing. It will be seen that, except in small areas, the histological differentiation of the ependyma has not proceeded to any great extent; the fluid from the ventricular cavity (as traced by the precipitated granules) closely follows the points of greatest cellular differentiation. There is no possibility of an intfri)r('tation of the findings concerned with the existence of intercellular stomata; the passage of fluid is here again to be looked upon as a transit through a cellular membrane.
 
 
The same general i)heiioinena of the pa.ssagc of fluid through a localized area (the area membranacea inferior, hi the caudal portion of the roof of the fourth ventricle) that have been observed in the superior portion of the roof are shown in figure 18. Cliief among these phenomena is the careful avoidance bj- the precipitated granules of the ependymal lining of the ventricles and the adherence of the granules to the Uning walls at the points of fluid passage. The ependymal lining, except in the two areas of differentiation, is everywhere impenetrable to the solution of the ferrocyanide.
 
 
As the size of the embryo increases the functional importance of this more caudal area becomes much greater (c/. figs. 3, 4, 5, and 76). The whole caudal half of the fourth ventricle becomes an area of ependj-mal differentiation and of fluid passage. It serves everywhere as a complete diff'using membrane, un))roken by the occurrence of stomata. Through this whole membrane the replaced solutions of potassium ferrocyanide and iron-ammonium citrate pass with apparent ease, as demonstrated by the precipitated granules of prussian-blue (fig. 18). From stages of 24 mm. and over, the lower membranous area is the onlj' one of significance in the total fluid passage.
 
 
The areas, therefore, through which the replaced solution of potassium ferrocyanide and iron-ammonium citrate jiassed, in the experimental pig embryos, are the two areas of histological differentiation in the roof of the fourth ventricle — the areae membranacese superior et inferior. There is no evidence whatsoever of any other point of escape of the fluid from the ventricular system into the periaxial spaces. The precipitated prussian-blue does not penetrate any of the lining cells of the ventricle except in the two areas under consideration. Nor is any evidence afforded by histological study of the escape of ventricular fluid through the described foramina of Bichat and of ^Mierzejewsky.
 
 
FACTORS CONCERNED IN THE EXPERIMENTAL FLUID PASSAGE.
 
 
It becomes necessary to discuss the question of the passage of the replaced fluid through the two cellular membranes in order to ascertain to what extent the results obtamed by the method may be relied upon. Naturally in such questions the factors concerned in the normal transit of body-fluid through such structures must be considered.
 
 
Probably the most essential element in obtaining rehable results in any injection is the control of the pressure at which the foreign fluid or mass is introduced. This matter has been fully discussed in the resume of the methods employed; it is suflRcient to reaffirm here that, in these observations, the normal cerebro-spinal tension has not been disturbed because of the use of a compensatory replacement. Other experiments, carried out under increasing pressures of injection, have been made, in order to compare the results with those furnished by the replacementmethod.
 
 
 
Consideration must next be given to the factors of diffusion, filtration, and osmosis in the passage of fluid through the roof of the fourth ventricle. The third factor, however, may be largely excluded, owing to the fact that the solutions of potassium ferrocyanide and iron-ammonium citrate employed were for the most part practically isotonic with the body-fluids. Furthermore, the use of hypertonic solutions apparently gave no difi"crent results (except in the increased density of the resultant precipitate) from those obtained by the employment of the isotonic solutions. Finally, it was found to be of service to use hypotonic replacement solutions in order to obtain very sUght precipitates; in these experiments also the spread of the replaced ferrocyanide solution was similar to the standard result afforded by the isotonic solution. These observations with varying concentrations of the foreign solutions replacing the cerebro-spinal fluid serve to indicate that osmosis plays but little part in the passage of fluid through the roof structures of the fourth ventricle. Undoubtedly the factor of osmosis can not be ignored in any consideration of the passage of fluid through a cellular membrane, but it seems unhkely that with solutions of practically the same salt-content it should be of great importance.
 
 
The influence of diffusion in this passage of the solution of the ferrocyanide and citrate from the cerebral ventricles into the extraventricular space is probably great. The whole plan of the experiment concerns the introduction of salts foreign to the body-fluids, even though in analogous concentrations. It seems not unhkely that as soon as the replacement of the existent cerebro-spinal fluid is effected the ferrocyanide and citrate must immediately begin to diffuse out into the periaxial tissues and the normal salts return to the ventricles. Probably, however, this same phenomenon plays a normal role in the human body. Jacobson's^^?) extensive and important studies on the chemistry of cerebro-spinal fluid have shown that the ventricular cerebro-spinal fluid is not identical with the subarachnoid fluid. The differences in the two fluids are probably to be accounted for by the fact that the ventricular fluid represents the pure elaboration of the chorioid plexuses, whereas the lumbar subarachnoid fluid is composed not only of the products of the chorioid plexuses but also of the fluids from the perivascular system. In this transference of the ventricular fluid to the subarachnoid space diffusion may play some part, the relative importance of which can hardly be estimated.
 
 
But will diffusion alone accoimt for the passage of the experimental fluid in the ventricle through two well-defined areas into the periaxial tissues? Will diffusion account for the varying extent of the injection in different stages of embryonic develojjment? There are several arguments against according diffusion a maximal imjjortance in the process. In the first i)lace, an injection of the solution of the ferrocyanide under mild syringe-pressure will give a spread similar in every re.spect to thos(; obtained by the replacement method. This indicates that the course taken by the two solutions is not necessarily the result of diffusion, but rather of the capabilities of the tissues for fluid-spread; and similarly the jiassage of this true solution through the roof areas need not be solely a diffusion process, but may be accounted for by the true flow of the fluid in this direction. Again, in the stages represented in figure 2 one would expect as extensive a spread of the replacing solution into the periaxial tissue were diffusion the active force in the movement of the fluid. Instead of such a periaxial spread the injection fluid remains wholly within the ventricular system, indicating that other forces than that of diffusion play an active role, in the more advanced stages, in the movement of the fluid. Finally, if diffusion is to be considered the sole agent in the distribution of the replacing fluid, why does not the ferrocyanide penetrate all the cellular structures lining the ventricular cavity? Surely it would he expected that diffusion between the body-fluids and the ferrocyanide solution would occur in each ependymal cell — a phenomenon observed only in the cells comprising the ventricular surfaces of the membranous areas of the rhombic roof.
 
 
While acknowledging that diffusion and osmosis may play important parts in the process of the passage of fluid from the fourth ventricle into the periaxial tissues, it seems apparent that some other factor or factors must be the determining agent or agents. It is not unlikely that the formation of cerebro-spinal fluid by the cells of the chorioid j)lexus may cause, in the replacement experiments, further passage of fluid into the extraventricular regions. Such an elaboration of fluid, with the ventricles filled with the experimental solution, would result in an increase in the normal ventricular tension. If this be the real explanation, the passage of the fluid into the extraventricular spaces would result in part from the increase in pressure on one (the ventricular) side of the membrane. The process, then, would be one of filtration through the membrane from the point of higher to that of lower pressure. This explanation best seems to cover the results obtained by the replacement method, and is supported by the histological examination of the developing chorioid plexuses and by many other features which are dealt with in other sections of the paper. This view is also strongly supported by the results of injections under mild syringe-pressure .
 
 
On the basis that the passage of fluid from the fourth ventricle into the periaxial tissues is in large measure a process of membrane filtration, the phenomenon of the fluid transit of the replaced solutions may be taken as a real index of the circulation and distribution of the cerebro-spinal fluid. It may be assumed, therefore, that the resulting distribution of the prussian-blue granules represents the course and extent of the fluid channels of the embryonic cerebro-spinal fluid.
 
 
The discussion of the fluid passage outward from the cerebral ventricles into the subarachnoid spaces has thus far been concerned with the processes involved for the transit of the true solutions of the salts. There is, however, an undoubted passage outward, as has already been indicated in a foregoing section, of the protein content of the normal cerebro-spinal fluid. This occurs in specimens in which a trul}- definitive membrane, intact throughout, can be seen inclosing the chorioidal roof. The explanations which suffice for the passage outward of the true solutions will not serve for this phenomenon.
 
 
The cells of the body probably are equipped to handle colloidal solutions in several ways, but two methods seem possible as explanatory of the problem at hand.
 
 
 
In the first place, it is conceivable that the cells in the differentiated areae membranaceae could phagocyte the colloidal albuminous particles of the ventricular fluid and excrete them into the subarachnoid spaces on the other side of the membrane; but it does not seem probable that this explanation is correct. Much more likely is it that the colloidal masses may follow the same laws of fluid-passage as the true solutions. But in such a passage through a cellular membrane the rate of passage will be much slower with the colloid.
 
 
These two theories regarding the passage of the albumen of the ventricular cerebro-spinal fluid into the subarachnoid spaces are not based on any findings presented in this article, but are ventured as being in keeping with current physiological explanations of such phenomena. On the basis of the second hypothesis, the failure of granular material to pass through the cellular membrane of the chorioidal roof must be explained as being due to the inability of the cells to handle the foreign material except in sizes which could be absorbed. The fact that the original unit was not phagocyted or passed through the membrane probably depended on the size of the molecule and the specific character of the lining-cells.
 
 
THE PASSAGE OF SILVER NITRATE AND INDIA INK THROUGH THE MEMBRANOUS AREAS IN THE ROOF OF THE FOURTH VENTRICLE.
 
 
Thus far in the discussion of the passage of the experimental fluids through the ventricular roof, true solutions of potassium ferrocyanide and iron-ammonium citrate only have been considered. This solution, as has been pointed out in this and in a previous article^^^)^ jg non-to.xic and is not taken up by the cells. With the dilute solutions (0.25 to 0.5 per cent) of silver nitrate, a far different problem is presented. Replacement experiments with this salt are rendered impossible bj- its intraspinous toxicity and by its precipitating action upon protein; but beautiful preparations may be made by this method by the simple injection with a syringe into the central canal of the spinal cord.
 
 
With mild syringe-pressure the result of such an injection with silver nitrate is in all cases a simple ventricular spread, with no extension into the ])eriaxial tissues. This general rule holds in all stages in which the central canal can be definitely entered without causing a spread into the j^erispinal tissues. This failure of the spread to extend into the periaxial tissues under mild pressure is undoubtedly due to the coagulating effect of the silver, which renders further passage of the fluid impo.ssible. The reduced silver collects about the superior membranous area in the roof of the fourth ventricle, outlining it distinctly. This phenomenon is illustrated in figure 115 (a transverse section of a pig emljryo of 19 mm.). At this stage the rejjlacement of cerebro-spinal fluid by a ferrocyanide solution results in a quite extensive spread {cf. fig. 5).
 
 
With increa.sed pressures of injection the silver may be pushed into the periaxial tissue through the roof structures of the fourth ventricle. The transit of the injection-mjuss occurs in the area meml^ranacea sujjcrior in practically all cases (c/. fig. 12). The inferior membranous ar(>a, in the earlier stages, is almost invariably impermeable to the silver (unless the injection-jiressure i.s extreme). When the superior area is examined after such an injection under high pressure the silver is found deposited throughout the cells of the area, extending only a short distance into the adjacent tissue. This feature of the injection is pictured in figure 113. In these injections the high pressure undoubtedlj' suffices to force the silver through the coagulated area membranacea. Its coagulating effect on the ependjina is almost equally marked, but the point of least resistance is apparently in the membranous area, allowing the fluid to pass through it.
 
 
Replacements of the cerebro-spinal fluid with diluted solutions of mdia ink within the medullary-canal system of small pig embryos never result in any extension of the granules into the periaxial tissues, for under the normal tension in the ventricles of the pig the arese membranaceaj are impermeable to the passage of granular material. After such a replacement the carbon masses maj' be found everywhere throughout the ventricles, but not in the periaxial tissues. However, india ink may be forced into the periaxial tissues by the use of high pressures of injection, as shown in figure 10. In this s])ecimen of a pig embryo (21 mm. m length) the periaxial spread occurred solely from the superior membranous area. This is analogous to the results obtained with silver nitrate, shown in figure 12. Without doubt in the earlier stages the superior area is much more permeable than the inferior. Histological examination of these specimens after an injection of india ink under high pressure reveals that the carbon granules gain the extraventricular space onh' through the area membranacea superior; some cells in this area are crowded with the granules, but for the most part extensive intercellular stomata have been made. The whole process must be viewed as a result of the excessive pressure of injection.
 
 
In the more advanced stages of the pig embryo (30 mm. and upwards) the pressure necessary to occasion an extraventricular spread of the india ink after intraspinous injection decreases somewhat, so that \\ith mild syringe-pressure a local periaxial sjn-ead from the fourth ventricle may be obtained from an injection into the central canal of the spinal cord. This is in accordance with the observation of MalU'^), who found that the injection flowed "through the medial opening of the fourth ventricle." The opening in these cases is in the area membranacea inferior, and in many instances subsequent examination showed rupture of the membrane with escape of the ink, even though the injection-pressure was moderate.
 
 
Taken as a whole, then, the findings are against the passage of solutions of silver nitrate or suspensions of india ink from the ventricles into the periaxial tissues, except when injected under pressures far above the normal intraventricular tension.
 
 
RELATION OF THE EPENDYMAL DIFFERENTIATION TO THE PASSAGE OF FLUID.
 
 
Under this heading it is jjroposed to discuss the relationship, if any, existing between the stages of differentiation of the ependjona of the roof of the fourth ventricle and the prssage of fluid through the two membranous areas. The discussion must necessarily be of a temporal character, with an attempt to consider possible factors in the process.
 
 
 
 
 
The most important question in this connection is whether the ependymal differentiation is necessary for the passage of fluid through it. In tlie \ng embryo of 13 mm. the area membranacea superior has reached a stage of marked differentiation (fig. 31), but at this same stage (fig. 2) there is no evidence of any pas.sage of fluid through the roof of the fourth ventricle into the periaxial tissue, only an outlining of the oval membranous area. Here, then, the histological differentiation has definitely preceded the assumption of function on the part of the area membranacea superior. The passage of fluid through the lower area occurs at a relatively earlier stage than it does through the superior opening. The first evidence of differentiation of the inferior roof of the fourth ventricle was observed in pig embryos of 15 mm. in length. At 18 mm., even though the process of differentiation was far from complete, some of the replaced fluid was able to pass through the lower area (figs. 4, 16, and 18).
 
 
A consideration of these observations leads to the assumption that some histological differentiation of the ependj^ma is necessary for the extraventricular passage of the replaced fluid. In the case of the superior area the differentiation occurs at a considerable developmental interval before fluid passes through it; in regard to the inferior area the assumption of function occurs at a somewhat earlier period in its differentiation. This slight difference between the two areas may possiblj' be explained on the basis that as soon as the stage of 14 mm. is attained (by the pig embryo) a greater amount of cerebro-spinal fluid is produced than can be cared for by the more slowly enlarging ventricular cavities. As soon as this disproportion occurs the excess of fluid is poured into the periaxial tissues through the already differentiated area membranacea superior; therefore, when the inferior area first shows evidence of formation there is still this excess of fluid in the ventricles. The fluid apparently avails itself almost at once of the new opening and its functional existence becomes immediate. It is apparent, moreover, that the capacity of the membranous areas for the passage of fluid is considerably in excess of the demands made upon them, and furthermore, that the provision for the passage of increasing amounts of fluid is completed before the demand arises.
 
 
In the passage of fluid from the ventricles into the mesenchyme, there is one other factor which has not as yet been considered. This concerns the iKitentiality of the adjacent mesenchyme to afford channels for the fluid poured into it. Were resistance offered to the flow of solutions through the mesenchymal tissue spaces, fluid could escape from the ventricles in only very small amounts, if at all ; as soon, however, as easily traversed fluid channels became established, the cerebro-spinal fluid could readily escape through the two membranous areas. The question as to what part the embr\'onic cerebro-spinal fluid i)lays in the further development of the meningeal si)aees also arises in this connection. It is at i)resent impossible to assign to any one of these factors a specific role in the passage of fluid from the fourth ventricle into the jieriaxial spaces, but it is important to consider them as possible determining agents. The evidence all indicates that the rate of production of the embryonic cerebro-spinal fluid is the most important factor, by far, in the extraventricular escape of the fluid.
 
 
 
 
 
VII. GENERAL HISTOLOGICAL DIFFERENTIATION OF THE CEREBRO-SPINAL SPACES.
 
 
The general problems concerned in the formation of the meninges and of the spaces inclosed within them deal with the gradual adaptation of a primitive undifferentiated mesenchyme to the anatomical and physiological requirements of the adult. Originally the meninges were held to be derived from the same epidermal infolding which gave origin to the central nervous system; then, with increasing knowledge of the structure, the dura alone was said to be a product of the middle germ-laj'er; and finally, by the researches of His^^s) and of K6lliker,'3i) the mesenchymal origin of the three meninges was established. The general process of the differentiation and the stages in this transformation have not been reported in great detail; here, too, the investigations must have an outlook for physiological anatomj^ as well as for pure morpholog}'.
 
 
It may be well to comment briefly on the relationships of the three meninges found in adult mammals. The dura is well estabUshed as the fibrous-tissue envelope of the leptomeninges and the central nervous system. But there is a tendency to regard the arachnoid and pia mater as constituting one structure — the leptomeninges or "pia-arachnoid," in the terminology of Middlemass and Robertson'^"). This difference of opinion in regard to the two inner meninges is due to their structural and intimate relationships. The arachnoid may well be a.ssumed to be a single membrane, worthy of being regarded as a single structure if one considers only its outer continuous membrane as the essential structure. But the inner surface of this membrane sends processes inward to fuse with the pia mater, which is so closely applied to the ner\-ous tissue. These processes divide the subarachnoid space (mcluded between arachnoid and pia) into the well-known meshes in which the cerebro-spinal fluid circulates. From the standpoint of these channels (the subarachnoid spaces) the arachnoid constitutes the parietal and the pia the visceral layer. Thus the intimate structural unitj' of the two membranes seems, in the opinion of many investigators, to warrant their designation as a single membrane. This view, however, has been strongly opposed by Poirier and Charpy^*^', who considered the distinction of three meninges very essential. Hence, in considering the transformation of tissues in the embryo, regard must be had for the dura as a well-differentiated structure, and for the leptomeninges as units, but certainly to be regarded from the standpoint of the subarachnoid spaces. In tliis connection Sterzi's'^^') observations on the comparative anatomy of the meninges are of interest. It will be recalled that the dura in lower forms becomes well estabhshed before the leptomeninges emerge from a primitive mesenchyme.
 
 
THE PERIAXIAL MESENCHYME.
 
 
Surrounding the central nervous system in young embryos is a rather thick cushion of undifferentiated mesenchyme, similar m all respects to the undifferentiated tissue in other parts of the embiyo. But verj' soon in the course of development the nuclei in this mesenchyme increase along the clear marginal zone of the spinal cord ami ha.silar structures, forming the initial indication of the pia mater. This phenomenon is indicated somewhat in figure 40, a photomicrograph taken from a human embryo (Xo. 836) of 4 mm., the earliest stage here illustrated.
 
 
The next essential change in the great differentiation of the meninges concerns a blastemal condensation of this same mesenchymal tissue to form ultimately the bony covering of the central nervous sj'stem and a portion of the dura; but between these two zones of differentiation the mesenchyme remains for a time almost unaltered. A portion of this tissue will go to form the arachnoid membrane and the trabeculaD which mark off the subarachnoid spaces. This process in the formation of the arachnoid will be discussed here; the formation of the pia mater and dura will be detailed in succeeding divisions of the paper. The differentiation will be discussed as a general process, in regard to both human and pig embryos, for in no respect has any essential difference between the two been observed.
 
 
The general character of the periaxial mesenchyme may be commented upon here. The tissue is of a verj' loose and typical structure, forming a syncytial network of rather small mesh, but fragile. The nuclei of the cells are oval, with a definite chromatin content; the cytoj^lasm is largely devoted to the maintenance of long processes which connect with adjacent cells. Adhering to the cj^toplasmic processes are very tiny albuminous coagula, of such small amount as to be hardly noticeable; also in the meshes of the mesenclwme very small quantities of this albumen may be identified. These albuminous coagula undoubtedly represent the protein of the tissue fluids in the undifferentiated stages.
 
 
THE FORMATION OF THE ARACHNOIDEA.
 
 
A general consideration of the problems here involved will surely shed light on some of the various factors concerned. It must be noted that in its development this membrane proceeds from an undifferentiated but small-meshed mesenchyme into the adult structure which contains the relatively large cerebro-spinal channels. Then, too, the enlargement of the tissue meshes in certain places — as the future cistcrnae — must be enormous. Besides this necessary dilatation of the spaces in the periaxial m(>senchymc, the outer portion of the tissue must separate from the future dura and form the outer surface of the arachnoid membrane. Here the process must be one of tissue condensation and proliferation. A similar agencj' is involved in the growth of the mesothelial cells which cover the outer surface of the arachnoid and also the inner suliarachnoid spaces.
 
 
The g(>n<'ral process, then, in the formation of the arachnoid membrane concerns a thiiuiing and readjustment of the primitive mesenchyme in certain areas, while in others the process is reversed, the membrane reaching the adult form through proliferative and condi-nsing i)henonieiia. Such alterative processes must naturally result from the apj)lication of certain mechanical or vital agents in the growth of th(? embryo. Is the mere growth of the central nervous system sufficient to furnish these alterative agents, or must we likewise trace the corres])onding development of the bony coverings of the brain and spinal cord? X{>ith(>r factor seems relatively of great importance when comi)ared to the possible influence of the presence and circulation of cerchro-spinal fluid on this periaxial tissue. This seems to be the most important factor, an internally-modifying influence to which the periaxial mesenchyme is subjected in the formation of an arachnoid and its subarachnoid spaces. It will therefore be from this standpoint that the development of the spaces will be discussed; for, as has already been pointed out, the periaxial mesenchyme becomes a functionally active tissue for the circulation of the cerebrospinal fluid at a stage when difi"erentiation has not begim. On this basis, the lack of differentiation shown in the |)eriaxial mesenchyme in the stages before the ventricular cerebro-spinal fluid is poured into the mesenchyme in the neighborhood of the roof of the fourth ventricle is not surprising. The character of the periaxial mesenchyme in the early stages is reproduced in numerous photomicrographs (figs. 25, 49, 51 , and 53). The mesenchj'me is here characterized by a rather dense meshwork of cytoplasmic processes, interspersed b^' a considerable number of oval nuclei. The content of the interstices in albumen, as judged by the persisting coagula, is very small. This picture of the periaxial mesenchyme persists untU cerebro-spinal fluid is poured from the ventricle through the area membranacea superior.
 
 
As will be seen in figure 3, the first indication of an extraventricular spread of the replaced fluid in the ventricles occurred in a pig embryo
 
 
of 14 mm. At this stage the membranous area in the superior portion of the roof of the fourth ventricle has already' become well differentiated. The fluid from the ventricles, however, does not reach any considerable spread until after a length of 18 mm. is attained; the periaxial spread during this period of growth is wholh^ confined to the peribulbar tissues. It is quite important in this connection that the first obvious differentiation of the mesenchyme for the formation of the arachnoid .should appear during this period and should involve the peribulbar tissues.
 
 
The first change to be noted in the transformation of primitive mesenchjone into the future arachnoid is an obvious thinning of the structure with a decrease in the number of nuclei per unit-volume. This is made out in a photomicrograph (fig. 57) of a section from a human embryo 14 mm.* long, when contrasted with a similar mesenchymal area posterior to the ventricular roof (fig. 53). In the pig embryo this thinning of the mesenchyme is as obvious at this early stage.
 
 
The process of dilatation of the mesench^Tnal sj^aces at this stage hardly seems to concern a direct disruption of the syncj'tial strands, but resembles more the spreading of the cell-bodies by the introduction of more fluid into the tissue spaces. This process would certainly result in an ap])earance similar in everj' way to that represented by figures 35 and 57. It probably also concerns other factors, as, possibly, the growth of the whole embryo without a corresponding degree of mesenchjTnal proliferation.
 
 
In a human f>ml)ryo of 17 mm. (Xo. 576) evidences are apparent of such a thinning of the mesenchyme about the medulla. Thus, in figures 58 and 59, from
*This embryo measured 14 mm. on the slide.
 
 
 
 
 
this specimen, the cellular decrease can be made out both in the region of the roof of the fourth ventricle and around the basilar surface of the medulla. It wdll be noted that the differentiation (i. e., the thinning) about the roof has proceeded more rapidly than along the anterior bulbar surface. This is perhaps to be expected in view of the initial pouring-out of the cerebro-spinal fluid into the mesenchyme just posterior to the roof.
 
 
In this mesenchymal differentiation a slightly increased amount of albuminous coagulum may be noticed. The truth of this is made obvious by an examination of figure 61, a photomicrograph from a human embrj^o of 17 mm. The almost entire freedom of the mesenchyme from albuminous detritus is most noticeable at earlier stages.
 
 
As was pointed out in the description of the results of replacing the cerebrospinal fluid, a marked change in the rate of development of the cerebro-spinal spaces in the pig-embryo ensues just after attaining the length of 18 mm. Within the growth of 2 mm. the injection spreads completely down the spinal cord and about the basilar structures of the cerebral cavity. This rapid extension finds its analogous process in the equally rapid changes which may be traced in the periaxial mesenclnTTie. Thus, in figure 72, a photomicrograph from a sagittal section of a pig embryo of 18 mm., the whole nervous tissue appears surrounded by a very thin, Ughtly staining tissue; this is the periaxial mesenchyme, which is undergoing its rapid metamorphosis. It will be noticed in this figure that the posterior structures (rhombencephalon) are surrounded by a much less dense mesenchyme than are the anterior (mesencephalon). This relative differentiation between the bulbar tissue and that around the mid-brain is only of temporal character; the mesenchyme about the medulla, as has already been pointed out, begins to differentiate first, the differentiation of the mesenchyme about the other nervous structures following somewhat later.
 
 
Figure 73 is a photomicrograph of higher power, taken from the squared area in figure 72. It shows to what a surprising degree the mesenchymal differentiation has proceeded during a few millimeters' growth. Two striking features of the process are brought out in this reproduction. In the first place, many of the mesenchymal trabecular have apparently been broken down, sacrificed to a few larger remaining strands. The cells connected with the destroyed trabeculie appear to recede until one of the heavier surviving strands is met with, when they adhere and apparently aid in the future development of a permanent arachnoid trabecula. The second feature of importance in figure 73 concerns the large amount of allnimen seen in the periaxial space. There is here a much greater amount of albumen than is ever found in the periaxial mesenchyme before the differentia ting process which results in the future subarachnoid space has become definite. The occurrence of this large amount of albuminous coagulum is apparently related directly to the outflow of the cnibrj'onic cerebro-spinal fluid, for the embryonic fluid is very rich in protein material, a.s can be readily seen by the partial filling of the embryonic cerebral ventricles with the clotted albumen.
 
 
This process of the l)reaking-down of the mesenchymal spaces to form fewer and hirger spaces goes on very rapidly in pig embryos as the\' exceed the length of 18 mm. Thus, figure 75 (from a pig embryo of 2.3 mm.) shows a marked decrease in the mesenchymal elements al)out the medulla; the strands are becoming fewer in number, and the albumen-filled spaces are increasing rapidlj' in size, but decreasing in number. About the mcsencej)halon, however, the process has only just begun (also shown by fig. 74). In this photomicrograjih (fig. 75) the mesenchj^mal elements have broken down somewhat; the spaces are Ijecoming enlarged, and a fine albuminous coagulum fills the interstices between the mesenchymal processes. The whole picture conveys an excellent idea of the forces which convert the many-spaced mesenchyme into the much fewer cerebro-spinal channels.
 
 
This general plan of the formation of the larger subarachnoid canals reaches its maximum in the formation of the various cisternae for cerebro-spinal fluid. The process is probably best illustrated in the case of the cisterna magna, which persists in the posterior cerebello-bulbar angle. Figures 74 and 75, taken from an embryo pig 23 mm. long, give an idea of the initial formation of the cisterna cerebellomedullaris. The mesenchymal strands, as shown in figure 75, are already broken down in part, and are profusely covered with albuminous coagula. The process has not proceeded to any extent in this specimen of 23 mm., but in the course of the next 10 millimeters' growth extensive changes occur, as arc shown in figures 76 and 77, photomicrographs from an embryo of 32 mm. In the space outside the inferior membranous area the mesenchymal trabecula? have almost disappeared; the space — or cistern, as it should now properly be called — is almost completely filled with the clotted albumen. The mesenchjTne is seen running through this embryonic cistern as a few isolated strands, but most of the tissue appears now as a fairly definite membrane on the outer side of the space. This membrane will go to form the inner surface of the dura and the continuous outer layer of the arachnoidea, as it furnishes a visceral layer for the subdural space.
 
 
More laterally in this same specimen the formation of the cistern has progressed to an even greater extent. In figures 78 and 79 the total freedom of the lower portion of the cistern from trabecular strands is seen; above, the mesenchjTne still sweeps down as a supporting structure for the chorioid plexus. A definite differential hne of mesenchymal condensation indicates the future outer border of the arachnoid as it incloses the cisterna cerebello-medullaris. This general process of mesench^Tnal breaking-down, altering the original small spaces into the larger arachnoid channels, holds as the embryo develops into larger forms.
 
 
In addition to this formation of the subarachnoid spaces in the adult through the enlargement of the embrA-onic mesenchjTnal spaces, the perimedullary mesench\Tne undergoes in these same localities condensations which result ultimately in the formation of the arachnoid membrane and the trabeculse diAnding up the cavum subarachnoidealo. Mention has already been made of the adhesion of the cellbodies of the disrupted mesenchymal elements to the persistmg strands — the initial step apparenth- in the ultimate differentiation of the mesothehal cells which line these spaces. Gradually ^vith the increasing growth of the embryo these cells seemingly become arranged in definite columns covering the persisting arachnoidal trabeculae. At the same time a differentiation of these primitive mesenchj-mal elements occurs, the cells ultimately being transformed into the very low cuboidal mesothelium of the subarachnoid spaces. This differentiation begins first in the basilar portions of the cranium and spreads upward, in a way similar to the course of development of the cranium and of the enlargement of the pericerebral spaces.
 
 
While such a general process as outUned accounts for the formation of the arachnoidal trabeculae and the subarachnoid spaces, it has but little bearing on the development of the outer intact membrane of the arachnoidea. This portion of the arachnoidea (which might be termed the arachnoid membrane as distinguished from the arachnoid trabecule) first appears as a distinct line of mesenchymal condensation separating the mesenchyme into the primitive arachnoid and dura mater, as in figures 76 and 77, dmc. This rather thin zone of cellular density in reality represents not only the outer surface of the arachnoidea, but also the inner surface of the dura mater. At first these develop in close fusion with a later separation of the two membranes. With this cleavage of the two surfaces, the arachnoid membrane rapidly differentiates, forming an intact layer over the subarachnoid spaces. The cells covering the surface membrane seem to change gradually into the low cuboidal type, similar to those covering the arachnoidal trabeculae. The details of these processes may be most easily studied in the region of the cerebral hemispheres; in this situation the transformation of the tissues occurs at a later period than in the basilar regions, for the differentiation of this mesenchyme follows the general plan of development of the cartilaginous and bony cranium.
 
 
The greatest problem in connection with the development of an external arachnoid membrane naturally concerns the separation of this leptomeningeal tissue from the pachymeninx. In the solution of this particular problem gross dissections have been found of benefit. For this purpose, pig embryos of larger size were used, and attempts were made to ascertain at what stage of development a true anatomical separation of the two membranes occurred. It was found that in embryo
 
 
pigs of about 40 mm. the dura over the calvarium could be well separated from the arachnoid, but areas of unseparated tissue still persisted at this stage. This was also found to be true in pig embryos of 50 mm. ; on the inner surface of the dura at this stage a mesothelial cell pattern could be demonstrated, although areas of attachment to the arachnoid existed. However, the differentiation of the periaxial mesenchjTne into the adult arachnoid does not occur coincidently with the possibility of a forceful separation of the dura from the surface of the brain; ])ut before this separation of the pachymeninx can l)e made the mesechyme which will go to form the arachnoid must undergo some differentiation. This process invohcs a condensation or accumulation of mesenchymal elements directly in the secondary dural thickening; the cells, with oval nuclei, soon form a continuous membrane of two or three cells in thickness. Apparently soon after the cellular accumulation has been accom])lished, a separation of the dnni from the arachnoid may be made. In certain areas, varying greatly in size, there is still an intimate connection between dura and arachnoid. These connections are particularly prominent over the developing cerebral hemispheres, and it is with this differentiation in the formation of the arachnoid spaces that we will now deal.
 
 
In a human fetus of 76 mm. (No. 1134) the arachnoid was found to constitute, in the region about the great sagittal sinus, a cellular layer which adhered quite closely to the dura, even though a Une of difTerentiation between the two meninges could be made out. This adhesion could undoubtedly be separated, even bj' gross dissection, although the tendency to adhesion was stronger than the attachment of the pia to the cortex. From its cell-character and general histology' the arachnoid at this stage must be considered as a formed membrane, but in a primitive state.
 
 
A somewhat similar but more advanced stage in the formation of the arachnoid membrane is seen in a human fetus of 100 mm. (No. 928-E) and in a fetal pig of 114 mm. In both the arachnoid membrane is verj' cellular, adhering to the dura only along the superior longitudinal sinus and in certain isolated areas. The cells comprising the arachnoidea possess oval, rather large nuclei which stain palelj- with hematoxyhn. No typical arachnoidal trabeculae could be made out in specimens in this cortical region.
 
 
The cellular character of the arachnoid persists in the larger embryos and fetuses as a layer, several cells in thickness, constituting the outer arachnoid membrane. In a fetal pig 190 mm. in length the membrane was practically differentiated, its outer wall being covered by me.sotheUal cells with large nuclei lying about a small fibrous-tissue base. The arachnoid trabeculae were developed only in the larger sulci, where they appeared as typical cellular cords about a core of fibrous tissue. At this stage, too, the vessels traversing the arachnoid spaces were found covered with similar cells. These may now be justly termed the mesotheUal cells.
 
 
Quite similar stages of arachnoidal differentiation occur in human fetuses of 200 (No. 870) and of 240 mm. (No. 1131). The arachnoid has everj-where practically become adult in character, except for a further decrea.se in the number of the peripheral layers of mesothelial cells. The fibrous tissue underlying this covering membrane possesses, as in the adult, almost a minimum of support.
 
 
In certain areas, however, the differentiation of the mesenchA-me into the adult arachnoidea does not keep pace with the general process. In the present study this phenomenon of unequal develoj^ment was especially well shown in fetal pigs of 150 mm. and upwards. It concerns the development of arachnoid trabeculae in the cerebral sulci. As is well known, the arachnoid membrane bridges the cerebral fissures, wliile the pia follows the cerebral contour. In the fetal pigs of the stages specified above, certain furrows showed a typical adult relationship with the covering arachnoid membrane and lining pia, the intervening space being traversed bj' definite arachnoid trabeculae. Other of the sulci were filled with an almost emb^^•onic tj-pe of mesench}Tne — a loose meshwork of cytoplasmic processes containing rather small oval nuclei. The explanation of this embryonic type of tissue seems to be that it occurs in the newly developing sulci and that some time must elapse in this formation before the tissue fully differentiates into the adult arachnoid membrane. Strangely enough, a similar collection of an embryonic type of tissue is sometimes met with, in these stages, between the two hemispheres.
 
 
The general process, then, of the formation of the arachnoidea involves both a breaking-down (or thinning-out) of the mesenchymal spaces and a condensation of the cells. The first of these processes results in the transformation of the interstices of the periaxial mesenchyme into the larger subarachnoid spaces, divided off by arachnoid trabeculae; the second finds its final accomplishment in the development of the outer arachnoid membrane which, covered with mesothelial cells, forms the inner surface of the subdural space. The transformation begins in the basilar regions of the cranium and spreads upward over the hemispheres.
 
 
THE CIRCULATION OF FLUID THROUGH THE SUBARACHNOID SPACES.
 
 
In view of the processes of differentiation involved in the formation of the arachnoidea and the subarachnoid spaces, the circulation of fluid through this pecuhar membrane must be considered. It seems important to ascertain, if jjossible, the relationships between the beginning of the passage of the cerebro-spinal fluid and the onset of the histological changes.
 
 
The conceptions of the development of the circulation of the cerebro-spinal fluid which are presented in this conmiunication are dependent, in large measure, upon the results of the replacement of the fluid, in living embryos, by the ferrocyanide solution. Additional evidence was obtained from the identification of albuminous coagula in the periaxial tissues. The correlation of these findings with the development of the chorioid plexuses and with the results of injections under low I^ressures, from a syringe and so forth, gave evidence of their correctness.
 
 
The differentiation of the mesenchyme into arachnoid membrane may be said to keep pace with the establishment of the periaxial channels for the cerebro-spinal fluid. In the main, the passage of this fluid into the undifferentiated mesenchyme about the nervous system precedes the process of histological change. This phenomenon is shown in figure 14, from a pig embryo of 18 mm. The replaced fluid is seen passing out into the mesenchyni(^ through the two membranous areas in the roof of the fourth ventricle. The mesenchyme at this stage has already differentiated somewhat, but hardly in proportion to the length of time during which the fluid has been passing into the space.
 
 
There are several features of interest in the course of the fluid through the periaxial spaces. In sections of embryos in which the cerebro-spinal fluid has been replaced by a foreign solution the granules of the precipitated salts may be identified in the periaxial mesenchyme in situations corresponding exactly to the extent of th(! spread shown in the cleared specimens (figs. 1 to 9). The exact location of the prussian-blue granules is of importance in this connection, as the exact form and distribution of the periaxial spaces and their relation to the adult subarachnoid spaces may thus be determined.
 
 
Kxamiiiation of serial sections from an embryo in which the embryonic ventricular fluid has been replaced by the ferrocyanide will reveal, if the embryo exceeded 14 mm. in length, granules of prussian-blue in the peribulbar mesenchj'me (fig. 14). The granule.s are not found in any cell-bodies in this tissue; they are made out, in large measure, adhering to the mesenchymal cell-processes or lying free in the mesenchymal interstices. The granules do not penetrate the pia mater or the dura mater, a finding which will be discussed more fully in the sections dealing with these membranes. Everywhere the transit of fluid into the nervous tissue seems to be prohibited by the pia; in some areas, however, the outer condensation of mesenchyme to form the dura-periosteum has not j'et occurred. This is shown particularly well in the region of the roof of the fourth ventricle (fig. 18), where the epidermis offers the only barrier to the passage of fluid from the pericerebral spaces.
 
 
In the earlier stages in which the phenomenon of fluid passage about the central nervous system may be observed, the outer layer of the arachnoid is not at all differentiated. Here the barrier to the fluid is the blastemal condensation of mesenchyme (fig. 16). In the later stages, when the outer layer of the arachnoid is beginning to appear as a mesenchymal thickening, the fluid (as indicated by the precipitated prussian-blue) is confined strictly within the immature arachnoid membrane.
 
 
The course, then, of the fluid which has replaced the cerebro-spinal fluid in the embryo follows that of the aduh cerebro-spinal fluid (as shown by the resultant blue granules). It is everywhere contained within spaces which topographicaUj' and embryologically correspond to the subarachnoid spaces in the adult. The spread of the replaced solution from the embryonic ventricle into the peribulbar tissue is analogous in every way to the passage of cerebro-spinal fluid from the fourth ventricle of the adult into subarachnoid spaces.
 
 
VIII. A CONSIDERATION OF THE EMBRYONIC PIA MATER.
 
 
Our present conceptions of the embryolog}' of the pia mater are largely due to the work of His^^s) and of Kolliker^^D. who first firmly established the idea that this inner leptomeninx was mesodermal in origin. While generally accepted (Farrar'i^*). this view has not been widely referred to in the literature; but the absence from all embryologies of any information concerning the development of the meninges is quite striking and it does not seem strange, therefore, that our information regarding the pia mater has not advanced in keeping wnth our knowledge of the embryology of other structures of the body. In the present section of this conmiunication it is purposed to present merely a general consideration of the process by which the pia mater is formed and to point out some of its functional characteristics, especially in regard to the fluid channels.
 
 
The term pia mater is accepted throughout this article as designating solely the cellular membrane which adheres closely to the outer surface of the nervous sj-stem, but it is in direct connection with the arachnoidal trabecular which traverse the subarachnoid space Whether the two membranes should be considered together as the pia-arachnoid or as the leptomeninx is a question in regard to which there is some disagreement ; it will suffice to consider the pia as a separate membrane.
 
 
 
 
 
THE GENERAL HISTOLOGY OF THE PIA MATER.
 
 
The findings in this mvestigation are wholly in accord with the conclusions of His(25), of Kolliker('i), and of Farrar^^^), that the pia mater is derived from the middle germ-layer. In the earliest stages the mesenchymal elements may be made out adhering to the outer i)ortion of the primitive nervous system. In the course of growth these cells are grouped about the mantle zone of the spinal cord in a rather dense laj'er, two or possibly three cells in thickness, with the tyjjical oval nuclei of the mesench5Tnal elements. Certain stages of this process may be made out in the figures in this paper. Thus, in a human embryo of 4 mm. (No. 836 of the Carnegie collection) the mesenchymal elements form a definite layer around the neural axis (fig. 41). The nuclei are oval in shape, possessing a moderate amount of chromatin, and are found in a layer two cells in thickness. This membrane, with its fairly scant cytoplasm, is sharply differentiated by its existence between two layers, in one of which nuclei are wanting, and in the other somewhat widely separated — the mantle zone of the spinal cord and the periaxial mesenchyme.
 
 
This typical arrangement of the mesenchj'mal elements about the cerebro-spinal axis holds in almost unchanged form throughout the whole embryonic growth. Thus, about the nervous tissue in figures 48 and 52 (from human embryos of 7 and 9 mm., respectively) the same condensation of the mesenchymal elements to form the pia mater are made out. This ajipearance is so familiar that further description in the later stages seems needless, but certain characters of this embryonic arrangement seem to require comment.
 
 
The general appearance of the pial layer is greatly altered by the early formation of the capillary blood plexus about the nervous sj'stem. This plexus tends to render the pial tissue more cellular, on first microscopic examination, as the endothehal channels branch greatly outside of the nervous tissue in this mesenchymal pia. The general character of the pial layer, however, as a membrane with prominent nuclei and scanty protoplasm, is not altered at all by the vascular plexuses.
 
 
The ultimate fate of these undifferentiated mesenchjinal elements forming this initial \nti\ condensation is a gradual transformation of the cells into ver}' low cuboidal mesothelial elements constituting the adult pia. The transformation concerns not only the differentiation of the cells but also a rearrangement so that the original layer of two or more cells in thickness becomes finally of but a single cell in thickness. The jjrocess, in a way similar to the development of the subarachnoid spaces, begins in the basilar portions and spreads upward; the process, hence, may often be studied in a single suitable sjiecimen.
 
 
More imjjortant, for our consideration, is the i)eculiar relationship of the pia mater to the roof of the foiutli ventricle, and in particular to the two area^ membranaceje. In this situation, in place of the slight mesenchymal condensation which characterizes the jjia, and which ^Minot''*"^ pictures in his figure 114, the mesenchyme seems altered. The condensation to form the pia, which takes place in other situations about the true nervous tissue, has not here occurred. This absence of the typical pial arnuigenit-nt may be noted even in very small embryos — those in which the roof of the fourth ventricle is composed of the many-layered, epitheUal-like cells. This is well shown in a photomicrograjih (fi}^. 53) from an injected human embryo of 9 mm. (Xo. 721) of the Carnegie collection. Likewi.sc, in this region in a [)ig embryo of 8 mm. (fig. 25), the same absence of a real pial condensation may be made out. But this peculiarity of the pia is most striking at the period of maximal differentiation of the superior membranous area in the rhombic roof. In figures 37 and 43, photomicrographs from pig embryos of this stage, the mesenchymal condensation, augmented by some vascular endothelium, is shown in adhesion to the ependyma on both sides of the membranous area; but directly behind the differentiated cells of the area membranacea evidence of a condensation of mesenchyme is whollj' lacking, even though both specimens show vascular channels in close appro.ximation. Similarly, in a human embryo of 14 mm. (No. 144, Carnegie collection) a total lack of the true pial thickening is to be observed (fig. 57).
 
 
Quite similar is the failure of a pial thickening about the inferior membranous area. This can be made out in figures 83 and 87, from human embryos in which the process of differentiation of the area is proceeding. In later stages of the formation of the area membranacea inferior, the marked absence of a true pial condensation in the mesenchyme in this region is noted in figure 75 (a specimen from a fetal pig of 23 mm.) But this apparent failure to form the typical mesenchymal condensation of the pia mater in certain areas in the roof of the fourth ventricle must not be construed as indicating an absence of pia mater. Such does not seem to be the case here, for in the later stages of the formation of the cisterna cerebellomedulla
 
 
ris the area membranacea inferior is found entirely unsupported, except for a layer of mesenchymal cells. This is shown in figures 77 and 79, both taken from fetal pigs of 32 mm. This mesenchymal laj'er must be considered as pia mater apparently modified for a specific purpose.
 
 
The general process, then, of formation of the pia mater concerns a condensation of mesenchymal elements to form an embryonic membrane about the central nervous system. From its earliest beginning very slight modification is needed to reduce it finally to the histological character of the adult membrane. The general process holds, except in the regions of the area? membranaceic in the roof of the fourth ventricle; here, apparently, a modification of the pia for a specific purpose, involving an absence of the primary pial condensation, takes place.
 
 
THE RELATION OF THE PIA MATER TO THE FLUID CHANNELS.
 
 
The cerebro-spinal fluid in its normal pathways comes everywhere into contact with the pia mater, which serves as the inner retainer for the subarachnoid space; therefore the functional relation of this membrane to the fluid which bathes it becomes of interest. To some degree the results of the experiments recorded in the earlier portions of this paper throw light upon the relation of the pia mater to the circulating fluid. The most important question in this connection is naturally that dealing with the possible penetration of the normal fluid through this embryonic membrane. In this regard the findings in replacement experiments with ferrocyanide solution serve to elucidate the problem. These observations give no evidence of any penetration of the pia mater by the fluid. This is well brought out in figures 14 and 18. In every respect (as demonstrated by numerous experiments of this type in pig embryos of varying lengths) the pia mater is wholly impenetrable to true solutions of foreign salts when injected so that the normal tension is not altered. The whole subarachnoid sjiace may, in such an experiment, be filled with the prussianblue, but none of these granules are found within the cells of the pia mater or in any layer between these cells and the nervous system. Evidence that the fluid has bathed the outer pial cells is afforded by the adhesion of granules of prussian-blue to the outer cytoplasmic borders of the cells.
 
 
Likewise the cells comprising the embryonic pia have been found to be impenetrable to true solutions (ferrocj'anide) when injected under varying pressures from a syringe. In these cases, rupture of the roof of the fourth ventricle or of the infundibulum may be produced by great pressure, without causing any of the fluid to penetrate the intact layer of the pia mater. The same result is obtained when india ink is substituted for the true solution.
 
 
The pia mater, then, even in its embryonic form, serves as an efficient fluidbarrier. This is demonstrated, in regard to the adult pia mater, in the report^^^) of the observations made on adult cats, dogs, and monkej's. But the barrier which the pia offers to the entrance of fluid from without exists also for fluid coming in the reverse direction. This is shown by the well-known phenomenon of the so-called subpial extravasation, which occurs in blood vascular injections when the injections are continued for too long a time at too high a pressure. The perforating vessels in such cases rupture as they enter the nervous system, and the injection mass spreads extensivelj' beneath the pia, stripping it away from the nervous tissue. Of interest in this discussion is the fact that the injection mass in these extravasations does not rupture the pia, which seemingly is an equallj' efficient fluid barrier to pressure exerted on it from within. Similar subpial spreads of the injection fluids have been observed in the course of this work. These extravasations resulted from the rupture of the whole nervous tissue from within, particularly in the region of the infundibulum, when the inj(>ction was made into the ventricular system under excessive pressure. In this respect, too, the i)ia seems to be wholly efficient as a retainer for true solutions or for granular suspensions. It is realized that the embryonic pia mater will not resist the passage of fluids through it under the highest pressures afforded by the syringe, but the membrane serves as an efficient barrier for all pressures such as are employed in careful anatomic injections.
 
 
With this conception of the impenetrabihty of the pia mater to fluids under ordinary pressures, it does not seem strange that there is a variation in the process of formation of the pia mater in the region of the roof of the fourth ventricle. It has been shown in the foregoing paragraj^hs that the phenomenon of mesenchymal condensation which results in the formation of pia elsewhere does not occur in the region of the two area? membranacefe. In view of the passage of cerebro-spinal fluid through these two membranous areas, the pia mater must necessarily be altered in these places. For were it not adapted to the jjurpose of affording fluid passage the cerebro-spinal fluid would, in its course from the ventricle to the subarachnoid space, form a subpial extravasation. It would seem that this modification of the pia is designed to meet the particular need and function of this region.
 
 
THE ADHESION OF THE PIA MATER TO THE CEREBRAL TISSUE.
 
 
It is a well-known fact in embryology that tlie pia mater and the peria.\ial mesenchyme in poorly dehydrated specimens split away from nervous tissue, but in adult preparations, if the meninges and brain are dehydrated in a block, the separation of the tissues occurs between the dura and the arachnoid, or (in more exceptional instances) the dura and arachnoid come away, leaving the pial layer closely applied to the cortical tissue. It is quite difficult in any adult mammal to separate the pia from the brain tissue. Realization of these peculiarities in the degree of adhesion of the pia led to an attempt to ascertain what structures were involved in the attachment of this mesodermal layer to the epidermal nervous system. The results of this attempt add nothing to the ultimate solution of the problem, but are perhaps of sufficient interest to justify brief presentation.
 
 
Two theories in explanation of this adhesion of the pia immediately suggested themselves. One of these concernod a possible growth into the pia of neuroglial elements, causing an intimate association between the pia and the cerebral cortex. Our findings in reference to the neuroglial outgrowth in fetal pigs gave no reUable basis for the assumption. The second theory dealt with a diminution in the elasticity of the walls of the perforating blood-vessels which supply the nervous system. The early embryonic vessels, with walls comjjosed solelj- of endothelium, when subjected to the distortions of poor dehydration, might possibly offer less resistance to the separation, so that the pia would come awaj' from the nervous tissue. In the later stages, however, the thicker-walled perforating vessels w^ould naturally oppose such a cleavage, so that the pia would remain adhering to the cortical tissue. This theory is also purely an hj^Dothesis, although it does not seem unlikely, especially if one takes into account a possible connection of the pia with the perivascular system. In examining blocks of the meninges and brain tissue taken together it was found that the pia mater separated cleanly from the nervous tissue in fetal pigs 15 cm. in length. Beyond this stage the arachnoid might remain in adhesion to the dura, but in such cases there was always found a layer of cells on the outer side of the cortical tissue, constituting a true pia mater.
 
 
IX. THE DEVELOPMENT OF THE CRANIAL DURA MATER.
 
 
The dura mater, like the two other meninges, is derived from the mesenchj-me about the central nervous system. The researches of Sterzi'"' on the comparative anatomj" of the meninges furnish additional evidence for this conception in the higher mammals. The origin of the pachymeninx from the middle germ-layer is now well established. But there is lacking in the literature a comprehensive account of the formation of this fibrous envelope. The gross generaUties of the process are given in pari, liut there is an almost total absence of the more intimate details of the transformation. One of the most essential points in the process concerns the relationship of the dura to the bony coverings of the cerebro-spinal axis. Does the adult dura serve as the periosteum of the bony skull? In the standard text-books of anatomy the adult human dura is described as being composed of two layers. In the skull these layers split, to comprise the walls of the great venous smuses. The outer layer of the dura serves as the periosteum for the bony skull, but below the foramen magnum the two layers separate to inclose the epidural space. The outer dural layer in this spinal region adheres to the inner surface of the bony vertebral column, where it functions as the periosteum; the inner layer here becomes the spinal dura.
 
 
In this account of the adult dura mater there is indicated a very suggestive periosteal relationship which implies an embryological basis for the disposition of the two laj-ers of the membrane. It must be granted, however, that this division of the cranial pachymeninx into two layers is quite arbitrary; there is nothing in the general histology of the fibrous covering to suggest such a halving except its division about the sinuses and its spinal relationships.
 
 
THE GENERAL PROCESS OF THE FORMATION OF CRANIAL DURA.
 
 
The first evidence of the development of the pachymeninx is found in the basilar region of the skull, where the mesenchyme thickens, to form eventually the bony covering of the brain. This thickening of the mesenchymal elements results not only in the formation of the chondro-cranium, but also in the final formation of the bony skull and possibly its internal periosteum and dura. In the process of differentiation the condensation of mesenchyme in the early stages gives no index of the varied character of the resultant tissues, so that, in the first place, the study of the process was necessarily related to the more adult stages. In this paper, however, the whole history of the dura will be detailed chronologically, beginning with the earliest stages.
 
 
Bardeen(2) has given data on the first appearance of the mesenchymal condensations which go to form the blastemal phenomena in both the cranial and spinal regions. The blastemal vertebra? become fairly well differentiated in human embryos during the first month of intra-uterine growth. At the end of the first month, in the occijiital region, three fairly well-marked occii)ital myotomes may be made out; these afterwards disappear. "During the early part of the second month the membranous anlage of the skull becomes extensively developed. The roof of the cranial cavity is formed by a dense membranous layer, which fu'st becomes marked at the side of the head in embryos from 9 to 11 millimeters in length" (Bardeen).
 
 
These evidences of a primary mesenchymal condensation about the central nervous system are concerned in the problem of the differentiation of the dura only in so far as they indicate the onset of the process which will give rise to the bone and possibly the periosteum — a part of the dura about the cerebro-sjjinal axis. Gaupp^'^) has already pointed out that this cranial l)lastemal condensation gives rise to these adjacent Init wholly different structures. These cranial mesenchymal condensations persist in simple form until after the cerebro-spinal fluid begins to fill its extraventricular bed; then, within a short time, the tissue becomes transformed by the development within it of cartilage, so that in the human embrj^o the caudal half of the chondro-cranium forms a ring of cartilage about the posterior portion of the brain. On the inner side of this ring of cartilage the mesenchyme later shows a marked condensation in the midst of the rarefied perimedulla
 
 
ry tissues. In this layer the nuclei soon become fewer in number and the cytoplasmic structures fibrillar, the whole resulting ulthnately in the formation of the fibrous adult dura. The mesenchymal condensations in the regions of the skull, where membranous bone formation holds, go directly into a membrane of fibrous tissue, in the outer portions of which bone is laid do\\Ti. The details of these processes will now be taken up.
 
 
In figures 30 and 32, photomicrographs from pig embryos of 13 and 14 mm., respectively, the well-established vertebral differentiations and the now poorly differentiated base of the skull are shown. From this stage upward the mesenchymal condensation in the head region proceeds rapidly. Thus at a stage of 17 mm. in the human embryo (fig. 60) the ventral portion of the vertebral canal has become cartilaginous, while the base of the skull has also undergone the chondrogenous transformation in its mere posterior portions. But of especial interest in our problem is the line of mesenchjTnal condensation, which may now be traced whollj^ around the brain-stem and hemispheres (fig. 60). The nature of this condensation is well shown in figure 61, an enlargement of the squared area of figure 60. The mesenchymal nuclei have become closely packed and rather sharply differentiated from the looser mesenchj^me which in part goes to form the arachnoidea. Figure 59 similarlj' shows this condensation proceeding upward to the vault.
 
 
Examined in another plane, the process of mesenchymal condensation seems to proceed much more rapidly in the posterior than in the anterior region. This is brought out in a transverse section of a human embryo of 18 mm. (fig. 62). Here the condensation is much more extreme about the medulla and roof of the fourth ventricle than in the more anterior parts of the mesencephalon. The same general appearance, typical of this stage, may be made out in figures 56 and 57 from a human embryo of 14 mm.* (No. 144, Carnegie collection). In the slightly larger stages the process of mesenchj^mal condensation about the nervous system becomes rapidly more marked. This increase in the number of cells comprising the denser membrane is shown in figures 64 and 65, photomicrographs of embryo No. 460 (21 mm.).
 
 
The degree of condensation of the mesenchyme in the various stages of the human embryo is followed quite closely in the pig embryo. The comparative degree of differentiation coincides within a millimeter or two. Thus, in a section from a pig embryo of 19 mm. (fig. 38), the degree of condensation about the roof of the fourth ventricle is practically similar to that in human embryos of the same length.
 
 
The phenomena just commented upon represent the stages concerned in the formation merely of a cranial blastema and are related to the formation of the dura only so far as it is out of this mesenchymal condensation that the periosteal portion of the
 
^Measured on slide after sectioning.
 
pachymeninx may be derived. The degree of condensation referred to in the figure.s has been solely of the blastemal type, but in some of the specimens this simple condensation is seen only in the more cephalic portions of the cranium. Thus, in the figures (64 and 60) taken from embryo 460, the mesenchymal condensation is still of the simjile undifferentiated type, whereas in this same embryo the more caudal sections show a chondro-cranium which is well develojied. The i)rocess of formation of the cranial dura, then, is one which begins in the basilar j^ortions of the cranium and proceeds from these points into the region of the calvarium. In general, all of the phases of this transformation into dura may be found in one specimen of sufficient and suitable size, the basilar differentiation re]:)resenting the advanced stages, while the steps in the differentiation are found in the areas nearer the vertex.
 
 
It is quite difficult to decide exactly what importance the primary condensation of mesenchyme maintains in the formation of the dura, because, coincident with the chondrification of the blastema, there occurs another condensation which forms the line of division between the inner surface of the dura mater and the outer arachnoid membrane. The first evidence of this secondary perimedulla
 
 
ry condensation is found in pig embryos of about 17 mm. In these specimens, in the narrow space formed by the mesencephalic flexure, mesenchymal cells collect together in the form of a fairly definite membrane. After its primary beginning in this area, the narrow line of its thickening may be traced to the basis cranii in embryos a little larger. In slightly older stages this secondary fine of condensation is found to be fairly extensive throughout the area between the middle and posterior cranial chambers.
 
 
At a stage of 20 to 21 mm. the whole basilar portion of the cranium shows evidence of this secondary line of condensation lying between the pia mater and the cartilaginous skull. The condensation occurs in the outer portions of the loose tissue which, as shown in a foregoing section (No. vii) becomes the subarachnoid trabeculae. The line of condensation is not broad on section; it comprises a cell-layer from three to six cells in thickness. Between this cellular border and the cartilaginous skull the tissue rapidly differentiates (a process seemingly sj^nchronous with the develojiment of this membrane). This tissue, which maintains dural relationships, is far more cellular and compact than the original perimedulla
 
 
ry mesenchyme. Even without the rather dense line of division in the mesenchymal ti.ssue, the dural structure can be easily outlined by its characteristic apjiearanco.
 
 
The original dural condensation between the two wings of nervous tissue which unite in the me8encei)halic flexure can be traced in slightly later stages around into the tentorium cerebelli. This structure develops as a wholly similar mesenchymal thickening in the midst of the jK-rimeduliary mesenchyme. The tentorium consists in these embryos of 20 to 2") nun. of two thin lateral plates which widen at their cranial attaclunents into prismatic areas. These areas, which finally lodge, in the two layers f)f dura, the sinus transversus, arc characterized by the same dense tyjie of mesenchyme. The jx-ripheral edges of the prismatic jwrtion of the tentorium sjjreads caudalwards as a definite line of condensation. In the earlier stages this line becomes indefinite as it extends from its teiitori:il attachment, but finnllv a similar line of condensation about the whole posterior chamber may be made out. This lies within the area of the cartilaginous skull and bounds the subarachnoid spaces.
 
 
This same process of formation of dura holds for the formation of the basilar dura in the more anterior portions of the cranium. The appearance of the .secondary zone, narrow and rather dense, may be made out inclosing the more cellular mesenchyme w^hich extends to the cartilaginous skull. The same process also endures for the formation of the dura of the calvarium, but here the addition of tissue from the undifferentiated mesenchyme is undoubtedly very small in amount. This will be discussed in a later paragraph. The various stages in the formation of this secondary condensation which goes to form the major portion of the dura may be fairly well studied in any one embryo of suitable stage, because the process, as pointed out above, begins in the basilar portion of the cranium and extends upward. Likewise, the condensations directly beneath the region of the dorsal membrane are delayed as compared to those of the lateral regions.
 
 
Some of the phenomena shown in the formation of the dura mater are illustrated in figures 46, 76, 77, 78, 79, and 94. Throughout these figures the letters dmc refer to the Une of the secondary mesenchymal condensation, which borders internallj^ the dura and which goes to form the outer membrane of the arachnoidea.
 
 
In figure 46, a photomicrograph of a pig embryo of 32 mm., the dura mater (dmc) is shown as a somewhat condensed tissue separated a sUght distance from the chondro-cranium. On the basilar surface, the inner line of dural tissue is quite remote from the inner surface of the basioccipital. Tracing this line of condensation forward, it is soon seen to merge more closely with the perichondrium* of the basioccipital. More anteriorly it again leaves the occij^ital plate and after a brief interval it fuses with the temporal perichondrium. Continuing slightlj^ more anteriorly the dural process toward the mesencephalic angle maj' be made out; this appears as a doubled membrane at its basal attachment. In its further prolongations the dural surface is at times a distinct structure; at other times it is completely fused with the perichondrium.
 
 
Posteriorly, in this figure 46, the line of dural condensation (incorporated also with the outer arachnoidea) may be traced upward around the cisterna cerebeUomedulla
 
 
ris. This hne of condensation is seen to lose its definitive character as it curves inward toward the chorioid plexus of the fourth ventricle — a phenomenon shown particularly well in figures 77 and 79, taken from the same pig embryo of 32 mm. The dura in this termination maj' be said to be in its formative stage; but dorsally, over the mesencephalon, the inner surface of the dura again becomes a definite membrane, as shown in figures 76 and 78. In the latter figure it is shown inclosing a wide mesh of dural vessels, between the arachnoidal surface and the membranous skull. Anteriorly, again, it seems to lose its definite hne of condensation.
 
 
•The term "perichondi.L n" is used tliroiighout this paper to designate only the ver>' dense cellular line delimiting the edge of the cartilngo. This dense zone is composed of the nuclei of the cartilage, crowded together, and represents probably some phenomenon of the growth or resorption of the cartilage. In a much broader sense, the whole dural tissue, lying between the line of secondarj- condensation and the cartilaginous border, could be termed "perichondritmi," as it probably represents the sole internal membrane which could be stripped from the cartilage.
 
 
 
Quite similar pictures are obtained regarding the dura mater in the human embryo. The relationships of the dura to the cisterna cerebello-medulla
 
 
ris are shown in figure 94, a photomicrograph of a human fetus of 35 mm. (No. 199 in the collection of the Carnegie Institution). In this rejiroduction the line of secondary mesenchymal condensation (representing the outer membrane of the arachnoidea and the inner surface of the dura) becomes widely separated from the occipitale superius in its superior portion.
 
 
In a fetal pig of 8 cm. the same general arrangements of the dura mater could be made out. The inner surface of the dura was in places still fused with, the outer arachnoid membrane, but in other places the areas of attachment were lacking, so that a true separation of arachnoid from dura had taken place. Along the ])eripheral points of the tentorium the dura and arachnoid were still closelj' a{)plied to each other. The dura itself was of the same cellular, rather loose tissue, with a dense inner surface. In i)laces, as described in the younger stages, the dural tissue was incorporated with the definitive perichondrium over certain cartilages or even over parts of the same structure. In other places a definitive perichondrium may be wholly lacking; in these areas the indefinite cartilaginous border gradually merges into the dura. In still other situations an intermediate arrangement of dura and perichondrium exists, where the cartilage is bounded bj' a somewhat condensed but not fully developed perichondrium which is continuous with the dura. Everywhere in the membranous sutures between the cranial cartilages or bones, the dura bridges the gap as a loose, cellular tissue. Over the calvarium the dura appears solely as a dense, rather fibrous membrane which is incorporated with and serves as the inner periosteum. This dura over the hemispheres is continuous with the fibrous sutures of the cranial vault.
 
 
The findings in a fetal pig of 98 mm. were not dissimilar to those just recorded. The dura was everywhere quite well developed, a rather loose cellular tissue except over the hemispheres, where it showed a more fibrous character. In the region of the occipito-atlantoid ligament the dura was fused with the ligamentous tissue, while above (over the occipitale superius) the dura became a distinct, thick cellular layer. The structure of the tentorium was wholly similar to the occijiital dura. In the basis cranii there are areas in which the dura is wholly fused with the periosteum or jierichondrium; in other areas it bridges the sutures or exists as a definite membrane on the inner surface of a definite perichondrium.
 
 
The dura mater in a fetal pig of 15 cm. did not vary greatly from those larger stages already tlescribed. The tissue, however, had become somewhat more fibroiis. The prismatic attachment of the tentorium was no longer as large proportionately, but the dura lining the occipitale superius remained a thick bulbous swelling on the dorsal surface. Hut most striking of all the features in the specimen was the very dense fusion of the dura of the calvarium with the fibrous sutures of the cranium. No line of demarcation betwecm dura and fibrous suture could be made out; the two fibrous layers are anatomically one structure.
 
 
 
The falx cerebri forms in the pig and human embrj-o by a process similar to that of the inner portion of the dura mater. In the sulcus between the two cerebral hemispheres the mesenchyme remains undifferentiated until quite late; then there appears in the posterior portion a narrow zone of condensation which soon presents two lateral surfaces separated by a layer of rather loose cellular tissue, similar in all regards to the dural tissue already described. This zone of condensation spreads forward to comprise the whole falx. The double surfaces of this membrane finally separate into two parts, forming the outer surface of the arachnoidea and the inner surface of the falx. At the cranial attachment of the fabc, the loose tissue forms a prismatic base, containing the suj)erior sagittal sinus and spreading laterally over the denser dura of the calvarium. The whole appearance of this region, which will again be referred to, is that the falx has been added onto the dura of the vertex. Its time of initial appearance is later than that of the rest of the cranial dura and there is apparently no additional acceleration of development. Hence the dural tissue in the fabc cerebri presents, in appropriate stages, a more immature type of differentiation than does the adjoining dura.
 
 
The process of the formation of the dura is not wholly a simple one due to the relation of the adult dura to, or its function as, the inner periosteum of the skull. In the figures already referred to, the almost complete fusion in some areas of the inner Une of dural condensation with the perichondrium has been commented upon. In other situations definite separations of the inner dural surface from the perichondrium occurred; in still other regions no perichondrium could be made out as a definite membrane. These differences in relationships of the dural tissue to the line of the perichondrium can not at present be wholly explained, but some indication of the meaning of the process can be given.
 
 
Out of the original cranial blastema, as described by Gaupp'^i^\ there develops the cartilaginous and bony skull, the periosteum, and the dura. But the observ^ations recorded above indicate that by far the major portion of the dura is formed by a secondarj^ mesenchymal condensation, which was indicated bj' a thin zone of more condensed cells on its inner border. This inner zone ultimately separated to form the inner surface of the dura and the outer membrane of the arachnoidea. The tissue included between this inner line of condensation and the cranial wall gradually differentiated into a more condensed but still a loose cellular tissue and finally became a fibrous dura.
 
 
In all cases the dural tissue extends from the inner line of condensation to the cranial blastema, to the perichondrium, or to the cartilage of the skull. The presence of a definitive perichondrium can not at present be explained, but apparently the perichondrium is manufactured by the cells of the original cranial blastema and not by the dural tissue which lies in approximation to it. When a definite perichondrium is found, it seems quite uninfluenced by the dura; at other times a fusion of an indefinite perichondrium with the dura seems to occur. The fusion of the perichondrium with the dural tissue derived from the secondary mesenchjinal condensation may occur, so that the small outer portion of the dura may be derived from this laj^er. The findings, however, in this investigation, are against any addition of perichondrium to the dural tissue; histologically, a definitive perichondrium is a membrane entirely apart from the dural condensation.
 
 
Over the cerebral hemispheres the dura of the cranial vault offers more difficulties of study than does that of the basilar regions. With the formation of a blastcmal condensation over the whole vertex — an extension of the dorsal membrane to form the membranous skull — there occurs very quickly a condensation to form the dura. This condensation may be first detected as a continuation anteriorl)'of the leaflet of the tentorium cerebelli, which stretches forward from the prismatic zone of the tentorial attachment. This zone of condensation is wholly similar to the narrow line of the mesenchymal thickening which was found in the more basilar regions of the skull. This zone of condensation occurs just wathin the cranial blastema and may be traced upward over the mesencephalon and laterally around the rapidly enlarging hemispheres. As the distance from the cerebellar attachment increases, the zone tends to approach the blastema, except in those regions in which the precursors of the dural veins occur. In such a situation this inner dural zone swings inward to encompass the vessels. Between this inner line of the dura (representing also the outer surface of the arachnoid) the same rather loose cellular tissue exists.
 
 
From the fabc cerebri a zone of dural condensation in the mesenchyme spreads laterally also; this gradually may be traced anteriorly and laterally until fusion with the similar lines of condensation from the basis cranii and the prismatic zone of the tentorium are reached. The condensation connected with the falx cerebri, however, is not an extensive process, the greater part of the hemispheres being covered by the development from the basis cranii and from the tentorium. It must be understood, however, that there is no active migration of this line of condensation, for the whole process is a development in situ. The appearance of an active extension is derived solely from the study of various stages and the increased area of condensation appears as an increment which has developed at the terminal points of the previous condensation.
 
 
The amount of dural tissue delimited in the mesenchyme by the secondary zone of condensation is not great in the region of the vertex. It is a thin layer which fuses to the inner surface of the cranial blastema. At the stage of this fusion the blastema has become somewhat fibrous and it constitutes the membranous skull. In this fibrous tissue (the union of the blastema and the dura) bone is deposited, but only in the outer layers. The phenomenon is easily studied in any suitable stage, for the sutures between the flat cranial bones remain incorporated with the inner memliranf!— the dura which includes the periosteum. Hence, over the cranial vault, the dura and periosteum become incorporated as a single membrane; this serves as the membranous .skull, into the outer layer of which bone is deposited.
 
 
In the basis cranii, as soon as ossification of the cartilaginous skull takes place, tin; dura becomes inc()r|)orated as the periosteum in a manner similar to that which takes i)lace in the cranial vault. While no definite relationship of dura to the peri-chondrium could bo made out in the earlier stages, the later function of the dura as the inner cranial periosteum is Cjuite obvious. Thus the adult relationships of the dura are obtained. But it is quite difficult to decide to what extent the dura (or internal cranial periosteum) is derived from the primary cranial blastema. It seems probable that this blastemal condensation, in its final resolution into bone, may contribute, in the form of a periosteal element, somewhat to the formation of the dura. Such an addition is verj' difficult of verification; certainly the greater part of the dura is derived by the secondary condensation from the perimedullary mesenchyme.
 
 
Before giving details of the fibrosis of the dura, it may perhaps be interesting to point out a peculiarity of the primarj' cranial blastema, which does not seem to be connected directly with the formation of the dura. This concerns the tendency of the membranous skull to form more than one layer in its original zone of condensation. In certain areas, as in figure 64, from a human embrj^o of 21 mm., the dorsal membrane is .shown split into two layers. Somewhat similar to this is the occurrence of two zones in the cranial blastema of a pig embryo of 23 mm. (figs. 22 and 101). Inthis latter figure a less cellular outer layer and a more cellular inner layer are seen. Neither of these have particular significance in the formation of the meninges, although the inner layer in early stages actively functions as a fluid retainer.
 
 
The question of the development of fibrous tissue in the dura mater in the course of its development requires consideration here. This phase of the problem concerning the formation of the pachymeninx has been followed, in this stud}', in the dura of the vertex about the sinus sagittalis superior. The tissue was removed in blocks, including the meninges and cortex cerebri, and was then sectioned in the coronal plane. For the most part the deposition of fibrous tissue was studied in sections stained with hematoxyhn and eosin; the findings were controlled by treating other sections from the same blocks with IMallory's connective-tissue stain. In this way the general histogenesis of the dural tissue could be well investigated.
 
 
Sections from such a block from a human fetus of 76 mm. (Xo. 1134, Carnegie collection) showed the dura to be composed of fibrous tissue everywhere except in the region of the great sagittal sinus. About this sinus an immature, almost embryonic, tj'pe of loose mj-xomatous tissue was observed. The fibrous tissue comprising the dura elsewhere is of a quite cellular, somewhat immature type of white connective tissue, with a considerable number of true fibrils. A wholly similar picture is found in a section, stained by Mallory's method, of a block from a fetal pig of SO mm. (fig. 104). Unfortunately the cellular character of the fibrous dura is not brought out, but the photomicrograph shows well the avoidance of the lateral walls of the sinus by the process of fibrosis. The more embryonic type of tissue in the region between the hemispheres is also well presented.
 
 
The dura mater of a human fetus of 100 mm. (No. 928-E, Carnegie collection) possesses fewer nuclei in a given area than does the dura from the specimen of 76 mm. (No. 1134). The tissue is fibrous, except in the immediate region of the sinus sagittaUs superior; but interspersed among the connective-tissue fibrils are many stellate or spindle-like nuclei, greatly exceeding in number the nuclei found in the dense dura of the adult. Bone is being laid down in the outer portion of this dura where it merges \\dth the membranous skull. The lateral walls of the great sinus are still free from fibrillar depositions. A somewhat analogous picture is afforded b}' a photomicrograph of a specimen stained after IMallory's method, from a fetal pig of the same length (fig. 105). In this specimen the outer portion of the dura, incorporated into a part of the membranous skull, is quite dense with the fibrous tissue; about the superior sinus, however, the decrease in the amount of fibrous tissue is very striking. The falx is beginning to exhibit a fair degree of fibrillar structure; it forms a definite division between the two hemispheres.
 
 
In the larger fetuses, above 100 mm. in length, the process of formation in the dura of denser and denser connective tissue proceeds rather slowly. It is realized, however, that this fibrous transformation in fetuses of 10 cm. is veiy extensive, the region about the sinuses alone remaining comparatively free from the development of the fibrils. The chief difference between the dura of this stage and the dura of the adult is a greater number of cell-nuclei in the fetal membrane. It is well, then, to consider the cellular character of the fibrous membrane and the region about the sinuses in the larger stages.
 
 
In a human fetus of 125 mm. (No. 900-H) the dura is quite fibrous, but still contains an increased number of the stellate and spindle forms of nuclei; likewise, about the superior sinus the tissue is an immature m30iomatous structure, fairly free from connective-tissue fibrils. This increased number of nuclei in the dural tissue holds also for human fetuses of 165 mm. (as in No. 745), but seems slightly decreased as compared with the smaller specimens. The lateral wall of the great sagittal sinus in this stage possesses distinct bands of white fibrils, but the tissue is much looser and more cellular than the fibrous dura over the hemispheres. These phenomena may be made out in similar stages of the fetal pig, as shown in figure 106, a photomicrograph from a specimen of 17 cm. In this specimen, treated by Mallory's stain, the superior longitudinal sinus is shown surrounded b}' a clear zone in which the deeply staining fibrils are comparatively few in number. On each lateral wall of the venous channels distinct fibrous bands may be made out, lying in the looser, more immature tissue. The lower portion of the falx has assumed quite an adult character.
 
 
Gradually the conversion of the tissue about the cerebral sinuses into the adult structure progresses. Thus, in both human and pig fetuses of 20 cm. length, the dura mater has acquired practically all of its adult features. Everywhere over the cereljral cortex the dura is characterized ])y dense layers of interlacing strands of white fibrous tissue, but the number of nuclei in the.se bundles may still be slightly greater than in the adult structure. In the more posterior regions, at this stage of 20 cm., the lateral walls of the sinus sagittalis superior are found to be completely occupied by the white fibrous tissue; in the anterior portion of the sinus much thinner tissue, resembling myxomatous structure, appears, as shown in figure 107. But in this specimen the invasion of the area about the great venous channel by fibrils has begun; isolated bundles may be made out everywhere in the lateral walls of the sinus. This freedom from connective-tissue formation does not persist, however, and the area is gradually invaded by the continued growth of the fibrils. The avoidance of the region about the sinuses by the connective-tissue resolution will be further commented on in the following subdivision of this paper.
 
 
The dura, then, develops probably first in connection with the mesenchjinal condensation which ultimately forms the bony skull and a portion of the dura (the cranial periosteum). It first becomes apparent, as a structural unit, as a more cellular layer differentiated, by a secondary condensation, out of the peria.xial mesenchyme. As the chondrogenous stage is approached it becomes differentiated as a distinct layer, maintaining varying relationships with the inner perichondrium of certain of the cranial bones. At a stage of 40 mm. m the fetal pig, the dura of the vertex may be dissected out as a distinct, somewhat fibrous laj'er. The process of fibrous-tissue transformation, however, is slow; the dura until late in fetal life shows an increased number of nuclei, as does any young connective tissue. The invasion of the region about the superior longitudinal sinus by connective-tissue fibrils is much more tardj' than is the transformation over the hemispheres.
 
 
THE SUBDURAL SPACE A>4D THE MESOTHELIAL LINING OF THE DURA.
 
 
The subdural space (cavum subdurale) has been the subject of controversy in regard to its role in the pathway of the cerebro-spinal fluid. Before the work of Key and Retzius^^s) ^\^q yj^^^y ^.f^g ]^q.](\ ^)jat the cerebro-spinal fluid occupied the subarachnoid space in the spinal cord, but that in the cranium the subdural space afforded an analogous pathway. This conception was largely due to the fact that, in dissection on fresh material, the dura and arachnoid in the spinal region are found to be in approximation; in the cranium the greater adhesion, bj' trabeculse, of the arachnoidea to the pia renders the freeing of the dura from the leptomeninges the simplest line of cleavage. This view was entirely disproved by the beautiful injections of Key and Retzius, who demonstrated the anatomical and physiological continuity of the subarachnoid spaces.
 
 
With the introduction of this latter view by Key and Retzius the conception of the subdural space naturally changed. These Swedish investigators demonstrated a typical mesothehal cell-lining on the inner surface of the dura, as shown by the method of silver reductions. Without an intimate connection with the true cerebro-spinal fluid, the subdural space has come to be looked upon as somewhat analogous to the serous cavities of the body. Quincke^-*^', after a subdural injection of cinnabar granules, ascertained that communications existed between the subdural and subarachnoid spaces, but only in the direction from subdural to subarachnoid. Leonard HilK^^), from the results of physiological experiments, assumed that fluid passed from the subdural to the subarachnoid space, and in the reverse direction, with great ease. The more recent investigations, however, lend evidence to the view that in the normal animal with undisturbed intracranial pressure relations the two spaces are physiologically as well as anatomically separate. The current impression that the subdural space is in manj' respects a serous cavity will probably finally have greatest support; intimate connections with the lymphatic system are, however, entirely lacking in the dura.
 
 
The development of the subdural space must necessarily follow the develoj)ment of the dura. It has been mentioned that in fetal pigs of 50 mm. the dura can be freed from the arachnoid by gross dissection, but that at this stage many areas of adhesion between the two membranes exist. Such an observation has considerable bearing on the subdural space. For in the development of this space two processes must proceed far enough to permit the separation of the dura and arachnoidea by the capillary layer of fluid. The first of these processes, in order of probable importance, concerns the condensation of mesenchj'mal cells to form the outer membrane of the arachnoidea. This thickening and resolution into a true membrane takes place in close approximation to the inner surface of the dura. The second factor concerns the covering of this inner surface of the dura with mesotheUal cells.
 
 
The lining of the subdural space by mesotheUal cells can be readily demonstrated on the inner surface of the dura by silver reductions, but the outer membrane of the arachnoid does not permit of a similar technique. This technical failure in regard to the outer arachnoid surface is probably to be accounted for by the dissimilarity in cell-structure in the two situations. Similar difficulties have been encountered by other observers.
 
 
In order, then, to ascertain, if possible, at what stage a really adult subdural space could be demonstrated, the inner surface of the dura from fetal pigs of various lengths was subjected to treatment with silver nitrate. After the reduction had taken place to a sufficient degree, the whole dura was washed with distilled water, stained with hematoxylin, and cleared in glycerin. The i)ictures afforded by this method were quite satisfactory, and the technical procedure was so simple and reliable that considerable faith could be placed in the absence of the intercellular reduction lines.
 
 
The smallest fetal pig in which a typical mesothelial cell-j)attern could lie demonstrated on the inner surface of the cranial dura was one of 50 mm. In this specimen the inner surface of the dura was not uniformly covered with the mesothelial cells; certain ragged areas seemed to represent the points of adhesion of the arachnoid to the dura. Figure 108 is a reproduction of a drawing made from one of the areas in this specimen where a mesothelial cell pattern could bo seen. The drawing shows many of the characteristics of mesothelial ci'll-i)ut terns of other parts of the body. The irregularities in the cell-borders, the frefiuent accumulations of the reduced silver in the cellular angles, and thegeneral cellular pattern are quite typical ; but the variation in the size of the cells, as shown in figure 108, is also somewhat different from the usual finding in the adult, where there is considerable constancy in the size of the cells. About half the cells in this fetal pig of 50 mm. are diminutive in size; the smallest are hardly a third the size of the largest. Transitions between the smallest and largest cells are al.so shown in this figure. It i.s difficult to a.scertain whether these smaller cells represent young elements which have not yet reached their maximal growth; no evidence of cellular division, as evidenced by mitotic figures, has been observed, although in this connection it must be granted that the cleared si^ecimens are hardly the most favorable. Und()ul)tedly this explanation of the smaller cells would seem to be the true one, but there is little proof for the view, except their absence from the adult dura and their disappearance in larger specimens.
 
 
This disappearance of the smaller mesothelial cells is not rapid, but is seemingly delayed over into the larger fetuses; thus, in figure 109, a similar preparation from a fetal pig of 75 mm., corresponding smaller cells arc outhned. On account of the absence from the field of the drawing of the larger elements, these cells do not appear relatively as great in number as in the preceding figure. Likewise, in figure 110 every gradation in cell-size is shown, in a specimen made in the same manner from a dura of a fetal pig of 90 mm.
 
 
Very slowly in the course of growth of the fetus the cells Uning the inner surface of the dura reach their standard size and compose the mesothehal surface, with very little variation in size. The process, however, is apparently very tard}', even though the fetus at 16 cm. shows an inner surface to the dura which is largely composed of standard cells (fig. HI); but even in this figure, from a relatively large fetus, the standard size of the cells has not been attained, for a few cells of small size appear in the drawing. In other respects the whole pattern, in general appearance, resembles closely the adult.
 
 
It seems most fair to assume that the occurrence of a true mesothelial cellpattern on the inner surface of the dura represents the initial estabUshment of a subdural space. On this basis the subdural space may be said to occur in fetal pigs 50 mm. in length; in the present investigation it has been found impossible to demonstrate the existence of the mesothehal cell-pattern in fetuses smaller than 50 mm. The separation of the dura, possible bj'^ gross dissection in pig fetuses of 40 mm., suggests that the space may be found at a sUghtly earUer stage than that in which the mesothehal cells have been demonstrated.
 
 
Anatomically the subdural space in pig fetuses resembles m everj' particular the adult space in cats and dogs; this was described in a paper^^S) pubhshed in 1914. In the large pig fetuses injections of solutions of potassium ferrocyanide and ironammonium citrate were made into the spinal subarachnoid space. After precipitating the foreign salts as prussian-blue, the injection is found to be wholly within the subarachnoid spaces, both in the spinal and cranial regions; the subdural space is absolutely free from any evidence of connection with the subarachnoid space. These findings wholly accord with the opinion concerning the adult subdural space which has been repeatedly expressed.
 
 
THE COMPETENCY OF THE EARLY DURA AS A CELLULAR MEMBRANE.
 
 
During the stage when the condensation of mesenchyme to form the cranial blastema is prono'inced the spread of the cerebro-spinal fluid becomes more and more extensive. In these stages, when the pig embryo measures from 16 to 25 m m. approximately, the outer membrane of the arachnoid is not 3'et formed, the arachnoid spaces extending from pial to blastemal condensation. WTien in these embryos the cerebro-spinal fluid is replaced by the ferrocyanide solution and the embryo kept alive for some time, the course of the injection may be traced to varying extents throughout the periaxial tissue. To this spread of the injection fluid (a true solution, during the progress of the experiment), however, the blastemal condensation of mesenchjme opposes an absolute barrier. This pecuUarity of the early condensation may be readily seen in figures 16 and 18. At this stage in development the blastemal thickening may be said to play the role of the outer membrane of the arachnoidea or of the inner surface of the dura.
 
 
This feature of the blastema as an impenetrable membrane — an absolute barrier to the passage of fluid — is found also to endure during injections of the ferrocyanide solution under pressures sufficient to rupture other parts of the central nervous system. Similarly, the early blastemal condensation resists the inflow of the other injections used (india ink and silver nitrate) under similar pressure conditions. In later stages the injection solutions, from ventricular or subarachnoid mtroduction, do not reach the dura. This is due to the development of an outer membrane of the arachnoidea and the formation of the subdural space. The arachnoid membrane when formed does not permit fluid to pass outward into the subdural space; but the competency of the early blastemal condensation in the mesenchyme affords a very good example of the perfect function of a tissue as a fluidbarrier.
 
 
An interesting phase of the competency of the secondary mesenchymal condensation (forming dura and outer membrane of arachnoid) may be seen in the region of the cisterna cerebello-medullaris. Here, as shown in figure 77, the zone of secondary condensation, while complete below, does not remain definitive above as the mesenchyme sweeps inward to the chorioid plexuses. At such a stage of 32 mm. in the pig, a replacement experiment would show no penetration of this secondary dural condensation by the foreign solution, where the condensation made a definitive membrane; above, however, in the region of the plexuses, a limited penetration by the introduced fluid could be made out.
 
 
X. THE RETURN OF CEREBRO SPINAL FLUID TO THE VENOUS SYSTEM. The question of the exact mode of return of the cerebro-spinal fluid to the general circulation has interested many investigators. It has occasioned a large amount of work, with the presentation of several hypotheses. Key and Iletzius<2»), from the results of injections of colored gelatin into the spinal subarachnoid space, held that the cerebro-spinal fluid returned through Paccliionian granulations into the great dural sinuses. Other workers, following Key and Retzius, were dissatisfied with this theory, because of the apparent lack of these granulations in infants and in the lower animals. Cathclin(6), with but little evidence, hypothecated an absorption of the fluid by way of the perineural sheaths into the lymphatic system, although the physiological findings of Ziegler(57j^ Reiner and Schnitzler(<«), Leonard 11111*2^), and others made it necessary to consider a direct absorption into the blood system. Cushing(9)
 
 
 
THE RETURN OF CEREBRO-SPINAL FLUID TO THE VENOUS SYSTEM. 89
premised the drainage of fluid into the great sinuses through a valve-Hke mechanism. Dandy and Blackfan'"' .-suggested its absorption l)y the capillaries of the pia-arachnoid — an untenable hypothesis in view of the work of Kadyi'^Sj, Shroeder van der Kolk(5i), Ekker(i*>, Adamkiewicz^*), and others. Still another conception of the process has been advanced by Mott('**), namely, that the absorption of cerebro-spinal fluid isone of the functionsof the cerebral capillaries. Ina previous investigation'^), making use of a method similar to the one here employed in the rej)lacement experiments, evidence was presented indicating the drainage of cerebro-spinal directly into the great dural sinuses through arachnoid villi. These structures represent an invasion of arachnoid tissue through the lateral wall of the sinuses.
 
 
In view of the findings in adult laboratory animals, interest naturally turned, during the course of this work, to the process of drainage of the embryonic cerebrospmal fluid. The evidence afforded by the replacement experiments )i\ith the ferrocj'anide solution indicated that in pig embryo
 
 
s of over 20 mm. cerebro-spinal fluid circulated throughout most of the periaxial tissue, and that in embryos of about 26 mm. the periaxial distribution was complete, the relations of the fluid at this stage becoming adult. With this evidence before us, the question of the drainage of the fluid became important. \^ as the absorption process similar to the normal adult procedure, or was it entirely lacking, the production of the fluid being balanced by the growth of the nervous sj'stem and its meningeal spaces?
The question of the absorption of cerebro-spinal fluid was approached in the embryo
 
 
in a similar manner to that employed in the adult animal. The problems incurred by the use of abnormal intracranial pressureswere eliminated by the method of replacing, without disturbing the normal tension, the embryonic cerebro-spinal fluid with the ferrocyanide solution. The embryo was then kept aUve and was finally fixed in a preservative which would precipitate the replaced fluid as prussianblue. This procedure was carried out in many embryos of varying lengths and the specimens were subsequently stained in serial sections.
 
 
The smallest embryo in which any evidence of absorption of the fluid from the periaxial tissue was obtained was a pig embryo, 23 mm. in length. In this specimen granules of prussian-blue could be traced through the mesenchymal spaces (arachnoidal) to the inner wall of the sinus transversus. The sinus is well differentiated at this stage in the human embryo of 21 mm., as demonstrated by Streeter^**). The wall of the sinus in this pig embryo was quite thin, the mesenchvTne lending the endothelium but Uttle support. The prussian-blue granules could be traced directly through the endothehnl wall of the sinus, and a few were identified lying free in the lumen. The conditions of the observations, permitting a flow of venous blood through the sinus, undoubtedly accounted for the fact that but few of the granules were found Ijing free in the sinus. This passage of the replaced fluid into the lateral sinus is portrayed in figure 21, taken from the pig embrj-o of 27 mm.
 
 
The same process of drainage of cerebro-spinal fluid may be obser^-ed in pig embryos more than 23 mm. in length. In all but one particular it corresponds exactly to the process observed in adult laboratory animals. There is the same lack of absorption on the part of the cerebral veins and embryonic capillary plexuses. In the adult, however, the process is not diffuse, but is confined to the arachnoidal villi, while in the embryo a considerable extent of the inner wall of the sinus lying in the mesenchymal tissue, which is breaking down to form the arachnoidal spaces, serves as a site for the fluid passage. In these earlier stages the sinus transversus functions as the chief sinus of absorption. This is probably to be explained by the primary basilar spread of the replaced cerebro-spinal fluid and also by the fact that the true sinus sagittalis superior is a much later addendum. In the human embryo, according to Streeterf^^), it is found in stages of over 50 mm.
 
 
The absorption of cerebro-spinal fluid in the embryo seems to follow the directing agencies which operate in the adult. Increase in the pressure employed in the injection of true solutions results in the drainage of more of the fluid, as determined by subsequent microscopical examination. This suggests that the process is determined by factors other than that of difTusion; it seems most likely that here, too, the process is one of filtration, with a possible distension of the cellular membrane, so that intercellular spaces are opened. The histological picture of the sinus waU, however, undoubtedlj^ gives the impression that the fluid has passed almost solely through the cytoplasm of the endoth(>lial cells and likewise through the layer of supporting mesenchyme. These findings are in accord with observations on the adult.
 
 
With dilute suspensions of india ink as the injection mass, the results are quite different in regard to the passage of the material into the sinus. Replacement experiments making use of this suspension of particulate matter yield no evidence, as the carbon granules do not leave the ventricular system. Likewise, simple injections of the suspension into either the central canal of the spinal cord or into the perispinal spaces furnish no information unless the syringe-pressure be high. In this case the carbon granules may be traced into the sinus transversus, which is apparently the point of least resistance. Because of the obscuring of the picture by the carbon it can not be determined histologically whether the granules pass into the sinus in the same manner as does a true solution, or whether the passage is effected by numerous small ruptures of the tissue. The impression gained from our study would incline one toward the latter view.
 
 
If the injection of india ink be made under very great pressure from a syringe, the segmental veins may be filled with the carbon. This filling is always subsequent to its flow into the sinus transversus. But in no case was an evidence of a flow into lymphatic channels observed.
 
 
The process of drainage of the cerebro-s])inal fluid into the venous system of fetuses will not l)e (U^tailed here. This undoubtcdl}' concerns a study of the formation of arachnoidal villi and of the differentiation of the lateral walls of the superior sagittal sinus, the best site for this study. The material at hand is not suited for this investigation, so that postponement is necessary.
 
 
 
XI. THE CHORIOID PLEXUSES AND THE ELABORATION OF CEREBRO-SPINAL FLUID.
 
 
With the realization that at a definite period in embryonic Ufe, cerebro-spinal fluid passes from the cerebral ventricles into the periaxial spaces, it seemed desirable to ascertain what relationship existed between the developing chorioid jilexuses and the elaboration of the fluid; for with the extension of the fluid into the periaxial tissue it becomes obvious that the balance between the development of the intraventricular fluid and the volume of the ventricles is destroyed and that more fluid is being elaborated than can be contained within the medullary-canal sjstem. This relationship between the ventricular volume and the production of cerebro-spinal fluid has been described at some length in a preceding section of this communication.
 
 
The determination, then, of the exact role plaj-ed by the chorioid plexuses in the further extension of the fluid into the periaxial tissue appeared to be of importance, for it could be conceived that the embryonic ependymal cells might be capable of elaborating the excess of fluid. With this purpose in mind the chorioid plexuses were investigated from morphological and cytological standpoints, in the hope that some index might be afforded as to the assumption of function on the part of the developing chorioid plexuses. These methods of study were apphed solely to the chorioid plexuses of pig embryo
 
 
s, for it is from them alone that evidence of the period of extraventricular extension of the cerebro-spinal fluid has been obtained.
 
 
THE DEVELOPMENT OF THE CHORIOID PLEXUSES.
 
 
The development of the chorioid plexuses is so well understood that only a ven,brief outline will be given here. The general scheme of origin of these glandular structures concerns a gradual histological differentiation in certain localities of the ventricular ependyma. The ependyma of the roof of the fourth ventricle thickens along the transverse invagination (phca chorioidea) and then gradually becomes tufted in villous projections into the ventricle, following the ingrowth of a capillar^' plexus and supporting mesenchyme. This general process of differentiation occurs at first along the lateral portions of the phca; the central portion of the ependj-ma remains unaffected by the villi even when the tufts have become quite well differentiated (fig. 23).
 
 
Quite similar to this process of development of the plexus chorioideus of the fourth ventricle is the differentiation of the other plexuses. The plexus of the third ventricle develops as an infolding of the tela chorioidea of the roof. In every case the process holds of ependymal invagination and subsequent vascularization and suspension by mesenchymal ingrowth.
 
 
The histological differentiation of the ependjTnal cells into the glandular tj-pe of plexus, as first determined by Luschka''^^ and Faivre^^^)^ jg hardlj- satisfactory as an index of the production of fluid, as the secretory phenomena of the adult cells have not as yet bten completely established. The researches of Pettit and Girard(^\ dealing with the correlation of histological changes in the chorioidal cells and their functional state, first furnished reliable evidence that these cells give rise to cerebrospinal fluid. Since the publication of their investigations in 1900 many workers — Meek(37), Findlay("), Pellizzi(''2^, Mott^^D, Hworostuchin(26), Engel(>2), and othershave been concerned with this problem and have established on fairly definite bases the relationship of the plexuses to the production of the fluid. The histological appearances of the secretory cells, however, does not rest on incontrovertible ground, as has been stated in a previous paper^^^).
 
 
The process of diff'erentiation of the ependymal cells which form the glandular elements of the chorioid plexuses occurs with the invagmation and tufting of these structures. The various stages of transformation from the low type of cubical epithelium constituting the ependymal layer are shown in various figures in this paper. The nuclei of these cells assume basilar positions and the outer zones of the C3-toplasm become granular with their greater height. The process is rather a slow one, as might be expected from the fact that the whole villus is gradually enlarging and becoming more and more tufted.
 
 
The histological differentiation of the plexuses need hardly concern us here, except as an index of the assumption of function. The final completion of this change into the adult morphology occurs at a much later stage of development than our evidence indicates for the establishment of a cerebro-spinal circulation. It becomes obvious, then, that the final liistological changes are not necessary for the process of elaboration of the fluid. This assumption seems warranted also.b}'' the fact that the embryonic fluid contains much more albuminous material than does the adult fluid.
 
 
The time of appearance of the chorioid plexuses in relation to the extraventricular spread of the fluid would surely seem to offer undoubted evidence in regard to the first elaboration of the fluid by the plexuses. It has been shown that in pig embryos over 14 mm. in length the replaced solution in the cerebro-spinal system spreads from the roof of the fourth ventricle into the periaxial tissues. This extraventricular extension occurs practically simultaneously with the first indications, in the pig embryo, of the formation of the chorioid plexuses of the fourth ventricle. Thus, in a pig embryo of 14 mm., the primitive thickening and tufting of the ependyma of the roof of the fourth ventricle may be observed (fig. 32). In earlier stages no definite evidence of this developmental process is found.
 
 
From the first indication of a developing chorioid plexus in a pig embryo of 14 mm., the growth of the tufts progresses rapidly, so that at 18 nmi. the process is well advanced. In embryos of 20 mm. and over the tufts of the plexuses in the fourth ventricle are quite marked, as shown in figures 22, 44, 40, and 92.
 
 
The chorioid plexuses of the third and lateral ventricles api)ear at a somewhat later stage than do those in the more caudal v(>ntricle. Thus the first indication of their ai)i)earance in pig eml)ryos is found in si)ecimens measuring 19 mm. in length. This coincides well with the further extension of the replaced fluid in specimens of 19 mm. and over. The definite differentiation of these plexuses, however, does not actually take place until the embryo reaches a length of 23 mm.— a fact suggestive of some relationshif) to the complete periaxial spread found in embryos of this measurement.
 
 
 
Considered, then, as a whole, there seems to be a very definite relationship between the developing chorioid plexuses and the periaxial spread of the embryonic cerebro-spinal fluid; for immediately after the first appearance of chorioidal tufting in the roof of the fourth ventricle (at 14 mm.) the replaced injection sjjread appears in the periaxial tissue (fig. 3). This extraventricular spread does not become marked until a length of 19 mm. is attained (fig. 5) — a factor in accord with the elaboration of the villi in the chorioid plexus of the fourth ventricle. The periaxial spread remains localized in the rhombencephalic region until the 20 mm. stage is attained, when it rapidly becomes pericerebral and perispinal (figs. 6 and 7). This coincides with the first indications of the chorioid plexuses in the more cephalic ventricles. But the further spread is here delayed (as in the stages between 14 and 19 mm.) until a length of at least 24 mm. is reached — which is perhaps of importance in the further development of the cerebral plexuses and the greater elaboration of the cerebro-spinal fluid. Thus it seems possible to conclude that coincident with the first appearance of the chorioid plexuses a more rapid production of cerebro-spinal fluid occurs, necessitating the passage of the fluid into the periaxial tissues.
 
 
THE GLYCOGEN CONTENT OF THE CHORIOID PLEXUSES.
 
 
In the hope that some cytological method might afford direct and incontrovertible evidence of the time of the assumption of function by the chorioid plexuses, stains demonstrating the intracellular presence of glycogen were applied to these structures. The quantitj' of the starch in the chorioid plexuses of rat and mouse embryos, as shown bj' Goldmann, suggested that this substance might be associated with the early elaboration of the cerebro-spinal fluid. Furthermore, the presence in the adult fluid of a definite reducing bodj', demonstrated by XawTatschi to be dextrose, added some weight to the hope that a definite conclusion might thus be afforded.
 
 
Several important studies concerning the presence of glycogen in the cells of the embryonic and fetal chorioid plexuses have been made. Creighton W found that the glycogen of the chorioid plexus was verj^ abundant about the middle of embryonic life, while von Loeper concluded that the great content in the cells of the fetal plexus was characteristic. Goldmann^^o) found large quantities of gh'cogen in the plexus in rats and mice, not only in embryonic life but also in animals from two to three weeks old. In the adult plexuses the cells contained no trace of glycogen.
 
 
The observations here included were made after fixing the chorioid plexuses of various pig embryos in absolute alcohol and staining the sections (cut either from celloidin or paraffin blocks) by Best's carmine method. This technique is similar to that employed by Goldmann. The staining reaction is such that a very striking differentiation of the glycogen occurs, but the shrinkage of the embryonic tissue in the fixation in absolute alcohol is a disadvantage. In these obsers'^ations the plexuses from the fourth and lateral ventricles were used.
 
 
As shown in the table on page 94, glycogen could be identified in the cells of the chorioid plexuses in pig embryos varying in length from 28 to 155 mm.
 
 
Below the first measurement no glycogen was demonstrated bj' the method employed; above the higher limit in only one instance (series No. 41) was glyocgen found. This finding of a limited period in the embryonic hfe of a pig during which glycogen occurs in the cells of the chorioid plexuses does not coincide with Goldmann's observations on the rat and mouse. Furthermore, it was found here that in stages up to 100 mm. the glycogen was practically generally distributed throughout all the cells of the chorioid jilexus, occurring with gn^it regularity in every villus and cell. This general distribution was not found in the plexuses of embiyos over 110 mm. in length; in these more advanced stages the cells containing starch occurred in clumps, giving a localized distribution. In the stages under 100 mm. the glycogen was present in very large amount, as estimated histologically. As the stages advanced the quantity of glycogen decreased rapidly. This great amount of starch was present in the same stages in wliich the general distribution of the cells occurred.
 
 
Occurrence of glycogen in the chorioid plexuses of embryo pigs.
 
 
 
 
 
C.P.
 
 
series,
No.
 
 
 
C.R.
 
 
naeasure,
mm.
 
 
 
Glycogen.
 
 
 
Globular forms of glycogen.
 
 
 
Plaques of glycogen.
 
 
 
Amount of glycogen.
 
 
 
Distribution
of glycogen
throughout
plexus.
 
 
 
Intracellular distribution of glycogen.
 
 
 
16
3
13
12
14
6
9
8
4
1
42
17
15
20
10
18
27
39
25
32
40
41
24
19
23
11
21
22
26
5
 
 
18
23
28
33
36
39
40
55
66
80
90
100
105
118
132
155
155
155
158
160
163
170
173
185
195
209
213
223
244
260
 
 
 
 
 
 
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present . .
 
 
Present
Present
Present
Present
Present
Absent
 
 
Present
Present
Present
Absent
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
 
 
Present
Present
Present
Present
Present
Present
Absent
Present
Present
Absent
Present
Absent
Absent
Absent
Absent
Absent
 
 
Great
Great
Great
Great
Great
Great
Great
Great
Great
Great
Small
Small
SmaU
SmaU
Small
SmaU
 
 
General
Localized
General
General
General
General
General
General
General
Genearl
General
Localized
General
LocaUzed
Localized
Localized
 
 
Basilar. Basilar. Basilar. Basilar. Basilar. Basilar. General. General. General. General. Basilar. General. General. General. General. General.
 
 
 
 
 
 
Absent
 
 
 
 
Present
Absent
 
 
Present
 
 
Absent
 
 
Small
 
 
Localized
 
 
General.
 
 
 
Absent
 
 
 
Absent
 
 
 
 
Absent
 
 
 
Absent
 
 
 
 
Absent
 
 
 
 
 
 
Cioldmann'20) pictures the glycogen as occurring throughout the cells of the chorioid i)lexus in the form of globules of larger or smaller size. Some of these globules may be .seen even in the s\n'rounding cerebro-sjiinaj fluiil. This general intracellular disposition was observed in this series in specimens measuring 6G mm. and over (fig. 95). Below this measurement the glycogen occurred practicallj'
 
 
 
PERIVASCULAR SPACES IN THE EMBRYO. 95
entirely in the basilar portion of the cell, central to the nucleus. Furthermore, in the stages between 30 and 00 mm. the glj'cogen globules were present in but small numbers and the glycogen was found in crescentic plaques (fig. 96). This formation of definite plaques is ai)parently to be ascribed to the fusion of the globules when the amount of glycogen becomes extreme. As far as is known tliis plaque formation with glycogen has not previously been noted; in one of Goldmann's figures the fusion of some of the globules has apparently taken place.
 
 
The table on page 94 records the findings in these observations.
 
 
The occurrence of glycogen in the cells of the chorioid plexus only during a certain portion of embryonic life is, as shown l)y the foregoing table, a fairly definite phenomenon, but there is surelj' no indication that this temporary presence of the animal starch bears any relation to the assumption of function on the part of the chorioid plexuses. The evidence afforded by the extraventricular flow of the replaced fluid, with the apparent relationship of the developing chorioid plexuses to the periaxial extension of the fluid, argues strongly against such an assumption.
 
 
XII. PERIVASCULAR SPACES IN THE EMBRYO.
 
 
In 1865 His' using a puncture injection, found that each nerve-cell existed in a so-called space. These pericellular spaces connected, as demonstrated by the flow of the injection mass, with an extensive perivascular network, more complex in its gray matter than in the white. In all of His's cases continuation of the injection led to a peripheral spread toward the pia, both in the spinal medulla and in the brain.
 
 
]Mott('*^\ working on the brains of animals in which an experimental cerebral anemia had been produced by ligation of the head arteries, found the perivascular spaces enormously dilated and the perineuronal spaces Likewise verj' evident. Direct connections between the perivascular and perineuronal spaces are pictured in Mott's communication.
 
 
The deduction which ^Nlott made from his findings, regarding the possible absorption of cerebro-spinal fluid by the cerebral capillary bed from this perivascular and perineuronal system, was discussed by the present author in a paper two years ago(55). It was there shown that, with the use of true solutions as the injection (potassium ferrocyanide and iron-ammonium citrate), the whole perivascular sj'stem could be filled. This injection of the spaces, however, occurred only when the pressure conditions within the cranial cavity were such that the subarachnoid pressure exceeded the vascular tension. This reversion of the pressure relations was accomplished by maintaining at normal the subarachnoid pressure with the injection fluid, and occasioning a simultaneous and complete vascular anemia. Under the routine conditions of injection (with undisturbed pressure relations) no injection of the perivascular system from the subarachnoid space resulted. It was found impossible to mject the perivascular system, using granular suspensions as the injection mass, without employing pressures far above the normal.
 
 
From these results here recorded briefly, the belief was expressed in tliis former paper that each nerve-cell was surrounded by a capillary space which drained along the perivascular channels into the subarachnoid spaces. Probably this sj'^stem represents a mechanism for accessory tissue drainage comparable ph3^siologically to the lymphatic channels of the other parts of the body.
 
 
In view of these findings in the adult mammal it seemed desirable to ascertain at what period of intra-uterine life such function was acquired. It also seemed not unhkely that information of interest might be acquired from the embryonic intramedullary circulation which would amplify our knowledge of this sj^stem in the adult. It was thought that there might be a correlation between the production of the perivascular fluid and the enlargement of the subarachnoid channels, similar to the evident connection between the chorioidal invagination and the extraventricular spread of the fluid.
 
 
Experiments to demonstrate possible perivascular and perineiu-onal spaces were first attempted on rather large fetuses (pig), as follows: The spinal meninges were exposed in a fetus in which the heart was still beating vigorously. Into the spinal subarachnoid space was introduced a needle connected with a small reservoir, containing the injection solution (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm. ; water, 100 c.c). The reservoir was then adjusted so that a pressure of 160 mm. of water was maintained in the subarachnoid sjiace. The arteries and veins in the neck of the fetus were then severed, and the subarachnoid pressure maintained at its former level. At the end of 20 minutes the head was placed in a fixative containing 1 per cent hj^drochloric acid.
 
 
This procedure, as outlined above, in the adult laboratory mammal, usually resulted in a complete injection of the perivascular system. In the embryo, however, the procedure was uniformly unsuccessful. The injection solution, as shown subsequently by the precipitated prussian-blue, rarely ascended over a centimeter above the point of injection. This indicated that the existent cerebro-spinal fluid was not replaced by the injection solution, and that the failure to demonstrate the perivascular system was to be explained on this basis, if the system were functional at this stage. Attempts were then made to replace the subarachnoid fluid with the injection solution before the cerebral anemia occurred. These attempts likewise m^t with failure, because of the impossibility of keeping the heart beating for any length of time in the larger pig fetuses. Other attem]its were also made to demonstrate these channels, in larger pig embryos, by means of a procedure which in the adult gave at times good injections of these intracortical canals. This method differed from the method first employed only in the maintenance of a high pressure (100 mm. Hg) in the spinal subarachnoid si)aces. It likewise met with failure, due ai)|)arently to the same causes which occasioned its failure in the adult: the high sul)arachnoid pressure o})erated chiefly to compress the cerebral and spinal tissues, rendering the injection of the i)erivascular spaces impossible.
 
 
The same procedures were attempted in smaller pig embryos (15 to GO mm.). The method usually successful in demonstrating the spaces (subarachnoid pressure slightly above normal, with subsequent cerel)ral anemia) failed, ajjparently because the cranial cavity at these stages is in no sense a rigid closed box. as in the adult.
 
 
Any method of service in the adult — which must have in consideration the physical character of the skull as a closed box — was here necessarih' doomed to failure.
 
 
Together with these technical failures to demonstrate a perivascular system, it must be borne in mind that these are merely failures to demonstrate the existence of the perivascular system in the pig embryo. The system wall probably be demonstrated as soon as a suitable technique is devised. The spaces are very likely present soon after the capillary plexus invades the nervous system, but the observation in many histological preparations of the spaces around the cerebral vessels must not be considered as offering proof of their existence, because of the likehhood of shrinkage influencing the picture. It is interesting, how^ever, to note that elasticity of the cerebral tissues seems greatest along the course of the blood-vessels, for here the phenomenon of shrinkage is most frequently observed. The existence of the perivascular and perineuronal spaces, probably of only capillary thickness, must remain — in the embryo as in the adult — a subject of physiological demonstration; histological e\adence, except with proper physiological regard, is of no value.
 
 
The early development and function of such a system as the perivascular and perineuronal canals afford seems most likely from the standpoint of pure speculation. It is not improbable that fluid is poured from this system into the embryonic subarachnoid space at a period soon after the capillary plexus invades the cerebrum. There is no evidence, however, from the observations recorded in foregoing paragraphs, that adequate subarachnoid channels are afforded until the pig embryo reaches a length of about 25 mm. The hypothesis of Essick^^^) regarding the damming of the perivascular fluid as the cause of the two cava corporis striati is of extreme interest in this connection. It remains, however, for future work to afford real evidence in regard to the embryonic perivascular system.
 
 
XIII. THE PERINEURAL SPACES IN THE PIG EMBRYO.
 
 
The question of the existence of potential or actively functional spaces around the peripheral nerves is of great interest, partly because of the possible relation of these spaces to the developing lymphatic system, and also on account of the anatomical evidence of the possible existence of such spaces.
 
 
It is realized that before much dependence can be placed on any theory regarding these potential spaces around the cerebro-spinal nerves, the possibility of their being purely artifacts must be dealt with. The methods of demonstration, in the adult, in the hands of the earliest workers were such as to favor the production of artifacts. As far as can be ascertained, Cotugno('), dealing with the ner\-us ischiadicus, was the first to conceive of these possible spaces. His method of demonstration consisted in filling the spinal subarachnoid space with mercurj' (in a cadaver placed in the erect posture) . Globules of the mercurj^ were subsequently found about the sciatic nerve in what then became the perineural spaces.
 
 
Modern anatomical interest in these spaces was aroused by the remarkable injections of Key and Retzius^^s). These investigators, by means of gelatin injections into the spinal subarachnoid space, were able to demonstrate perineural spaces around the cranial nerves, especially around the optic pair. Their results, however, are open to criticism, because of the excessive pressures employed ("not over 60 millimeters of mercury") and because the injections were made in fresh cadavers kept warm for periods of 10 or more hours.
 
 
Some of the difficulties concerned in the problems of the perineural spaces were cleared up in a studj'^^^) of the cerebro-spinal circulation published in 1914. In this work injections of true solutions (similar to those used in the present study) were introduced into the spinal subarachnoid space in living cats and dogs, under pressures but sUghtly exceeding the normal intraspinal tension. These injections were continued for several hours, and the course of the injection fluid was then estabhshed bj'' precipitating the solution in situ. By means of this procedure, which it was beUeved approached the physiological, the perineural spaces around the cranial nerves could be demonstrated. In these adult laboratory mammals the cerebral nerves without exception showed prussian-blue granules in a perineural relation, extending outward along the nerves bej^ond the termination of the dural cuff. This extension of the injection mass outward was more striking around the first two cranial nerves than about any of the others. Thus, the olfactorj'^ nerves uniformlj'^ showed perineural deposits beyond the cribriform plate, extending downwards into the nasal epitheUura, while the optic nerves were surrounded by the granules in the inf ravaginal sheath, which spreads out over the posterior surface of the eyeball. The caudal cranial nerves were likewise characterized by extensive perineural injections.
 
 
These findings were interpreted as e\'idencing a true perineural space, probably of only capillary thickness, which could be injected by filling the cerebro-spinal spaces with a demonstrable true solution. As far as could be made out under the microscope, they had no appreciable existence except when fiUed with the precipitated true solution. These spaces were not filled in the early moments of the injections under low pressures, and could be demonstrated only when the injection had been continued for several hours.
 
 
The perineural spaces are quite different from the spaces surrounding the spinal ganglia and the gangUa of the cranial nerves. These ganglia he in the true subarachnoid space, wath the dura investing the arachnoid membrane. Distal to the ganglion the dura ends upon each nerve. In the injection under low pressure with the ferrocyanide the cranial and spinal ganglia were all surrounded bj^ the precipitated salts; the cranial nerves .showed extensive perineural injections, whereas the spinal nerves rarely showed a true perineural injection, and then only of hmited extent.
 
 
The existence of perineural spaces in the embryo, however, has been under dispute. The larger nerv^es in sectioned embryos almost invariably show spaces about them, either a complete separation of the surrounding mesenchyme or a partial dilatation of the mesenchymal interstices. Sabint'*^), in 1902, noted that in pcrispinal injections with inflia ink the spinal nerves could be outlined by the carbon granules, but in no case did such an injection run into true lymphatic channels. No evidence was afforded by her work of any lymphatic channels arising from these apparent perineural channels.
 
 
 
In the course of this investigation of the cerebro-spinal spaces interest naturally turned to the perineural spaces. In the typical experiments (a replacement of the embryonic cerebro-spinal fluid with a demonstrable true solution in the living embryo), there was evidence of a spread of the replaced solution around the cranial nerves. Because of the procedure u.sed (merely a filling of the ventricles and central canal of the spinal cord) no evidence of a perineural spread occurred until the foreign solution passed into the periaxial tissues. Here the spread chiefly involved the caudal cranial nerves curving around the lateral surface of the medulla in fanshaped processes (figs. 5, 6, 8, and 9). The spread, however, was not extensive. In figure 8 a similar slight spread along the spinal nerves is to be made out. Closer study of these cleared specimens, and examination of the same and of similarly injected embryo
 
 
s after serial sectioning, convinces one that the apparent perineural spread in these cases extends around the sensory ganglia and not further toward the periphery. In no case, cither in the caudal portion of the cranial or in the spinal region, has the replaced injection fluid passed the blastemal condensation of mesenchyme. This finding is well shown by the distribution of the injection fluid in figures 9, 16, and 18.
 
 
The optic nerves, however, jjossessing gangha in the retina, usually show, in the typical replacements in the living embryo, a partial or complete surrounding of the nerves by the prccij)itated pru.ssian-bluc. An incomplete example of this — more typical, according to these observations, than a total circumvention — is given in figures 19 and 20. The higher-power reproduction of this field is very interesting. It shows in the central portion the fiber bundles comprising the optic nerve, surrounded by mesenchyme and the developing ocular muscles. In the region between the ners'e and the muscles is an undiflferentiated mesenchyme which is characterized by a crescent of the precipitated granules of prussian-blue. The non-penetration of the surrounding tissue by the ferrocyanide is verj- well brought out in this drawing. The prussian-blue has reached its position about the nerve by extension from the pericerebral spaces; actually it has still the same distribution as noted in figure 8 above. The adult dura will completelj- surround the optic nerve in its whole extent; the subarachnoid space will likewise extend unbroken to the posterior surface of the eyeball. Hence it must be assumed that in this case the perineural space does not extend beyond the peripheral ganglion. With regard to the olfactory nerves, no evidence of a perineural spread was obtained in specimens of pig embryos up to 45 mm. in length.
 
 
It seems obvious, then, that in the embryo pig true solutions, when substituted for the cerebro-spinal fluid, do not extend peripherally along the nerves any further than does the dura in the adult. The replaced fluid (if, as appears most likely, it indicates the true circulation of the cerebro-spinal fluid) extends only through the future subarachnoid space. Such a conclusion is best supported bj' the observations. The only discreparcv between the findings in the pig embryo and those in the adult with the same method lies in the fact that in the adult the cranial nerves showed a much more extensive permeural injection. This seeming discrepancy may be accounted for in two ways. In the first place, the experimental replacement in the embryo pigs lasted at most one hour (due to the fact that the embryo's heart frequently ceased beating at the end of this time), while in the adult cat or dog they were continued for several hours; and it was only in the long-continued experiments in the adult that the extensive perineural injections were obtained. On this basis it seems more than likely that the communications between peripheral perineural spaces and the subarachnoid space are very small and that diffusion must account for the slow filhng of the peripheral system. The second explanation seems undoubtedly to concern the time of development of these perineural spaces in the embryo. It may be that the spaces are morphologically non-existent until late in fetal life; in that case, of course, it is not strange that they have not been filled wdth the injection fluid.
 
 
From the observations recorded above it is quite apparent that in the typical experiment in which the normal cerebro-spinal tension is not increased no evidence of the perineural space, as injected by Miss Sabin, has been adduced. However, the possibility of mjecting these spinal spaces as was done by Miss Sabin is easily demonstrated. The injections may be made with ease, either with . granular suspensions or with true solutions. Success invariably attends such an injection into the perispinal tissues. The injection solutions easily run out around each nerve, more readily, apparently, in the younger embryo than in the older. It is not clear whether this difference is due to the fact that in younger embryos the resistance is greater to the perispinal flow and less peripherally, or merely to the fact that a greater amount of fluid must be introduced in order to attain the same result. Careful repetition of these observations has led to the conclusion that such a demonstration of the spinal perineural spaces results from excessive pressures of injection. WTienever the pressure exerted by the injection is but slightly above the normal, or does not exceed the normal (as in replacements), the perineural spaces are not injected around the spinal nerves. Miss Sabm's conclusions from her results, that no connection exists between the spaces and the lymphatic system, seem to be wholly substantiated by these observations.
 
 
The apparent perineural spaces around the embryonic nerv^es must be looked upon as artifacts. In tissue carefully fixed, dehydrated, and embedded, there is no real evidence of these spaces. Theh- size apparently varies with the care observed in the histological technique.
 
 
 
XIV. GENERAL SUMMARY.
 
 
In the foregoing sections of this communication some of the problems concerned with the embryology of the cerebro-spinal spaces have been discussed and observ^ations have been presented in the hope that a better conception of the processes might obtain. It is purposed to present here briefly the results of these obser^^ations and to attempt to correlate the findings so far as is possible; and in this, as in the detailed reports in the preceding pages, the relationship of the jihysiological processes concerned will be referred to the morphological changes in the developing embryo.
 
 
As a means of studying the physiological extent of the embryonic cerebro-spinal spaces, a method of replacing the medullary fluid with a foreign solution was devised. The procedure consisted in substituting, in the hving embryo, a solution of potassium ferrocyanide and iron-ammonium citrate for the cerebro-spinal fluid. The embryos were then kept alive, for periods of about an hour, by placing them with the attached placenta; in an incubator at 38°. At the end of this time, which varied in the many experiments, the whole embryo was fixed in a medium containing hydrochloric acid, thereby precipitating an insoluble prussian-blue. Specimens prepared in this manner were studied after sectioning or after clearing by the Spalteholz method.
 
 
Pig embryos, subjected to such experimental replacements, exhibited only an intraventricular retention of the foreign solution until after a stage of 14 mm. was attained. In the earliest specimens, embryos of about 9 mm., there was no characteristic distribution of the foreign solution, except that it remained within the medullary-canal system. In stages of about 13 mm. the replaced fluid also was retained within the cerebral ventricles, but in these specimens a dense accumulation of the precipitated prussian-blue may be made out in a distinct oval in the superior portion of the rhombic roof. This granular aggregation occurs against a histological differentiated area in the roof of the fourth ventricle — an area which represents apparently the more epithelial-Uke elements of the earher roof-plate. This area must be considered solely as a differentiation of the epidermal lining of the medullary-canal system.
 
 
In hving pig embryos of 14 mm. and over, the result of the routine replacement of the ventricular cerebro-spinal fluid was a sUght extraventricular spread into the tissues posterior to the rhombic roof. The passage of this foreign solution outward occurred through the same area of ependjTnal differentiation, outlined bj' the collection of granules against its inner surface in the previous stage. The extraventricular spread remains definitely locahzed to a ver}' small conical area which does not rapidly increase in size.
 
 
The factors which cause this initial flow into the pericerebral spaces are of interest. It follows that in the growth of the embryo the production of the intraventricular and intraspinal cerebro-spinal fluid must necessarily keep pace with the increasing size of the cerebral ventricles. It is also necessarj' for the occurrence of an extraventricular spread of the fluid that the production of the fluid within the ventricles must exceed the amount required to keep the medullary-canal sj'stem filled. From our knowledge of the elaboration of the adult cerebro-spinal fluid, it is impossible to conceive of the production of a true cerebro-spinal fluid in the perimcdullary mesenchyme. Such a view would be a reversion to the old hypothesis of Haller, who regarded the leptomeninges as the elaborators of the fluid. Likewise, the passage of the replaced foreign solution into the extraventricular spaces would render such a hjiDothesis untenable.
 
 
Hence, it becomes incumbent to regard such an extraventricular spread of the experimental solution as an mdication that the production of the cerebro-spinal fluid within the cerebral ventricles exceeds the capacity of the ventricles to care for the fluid. This argues strongly that the process of elaboration of the fluid in these pig embryos of 14 mm. is no longer sluggish, but that an active production, sufficient to cause a sUght extraventricular flow during the observation, is now taking place. This acceleration of the flow is not great, but it represents a marked change in the relationship of the process of fluid elaboration to the increasing volume of the ventricles.
 
 
It seemed desirable to endeavor to correlate this extraventricular spread of the experimental fluid with the morphology of some intraventricular structure at this critical stage of 14 mm. m the pig embryo
 
 
. The first evidences of villous tufting in the chorioid plexus of the fourth ventricles were found to occur at this stage in the pig. Other studies of this plexus, particularly those which concerned the occurrence of glycogen in these glandular cells, were found to offer no additional evidence of value in regard to the onset of function in these structures. The correspondence between the initial tufting of the ependyma to form the rhombic chorioid plexuses and the initial extraventricular spread must be regarded as of the utmost importance. It would appear most Ukely that as soon as the chorioid tufts occurred an increased production of cerebro-spinal fluid took place, necessitating an extraventricular expulsion of the excess of fluid. Such a view receives the utmost support from these recorded observations; it is in keeping with the best conceptions of the processes of production of cerebro-spinal fluid in adult mammals.
 
 
With the initial pericerebral extension of the experimental fluid occurrmg in pig embryos of about 14 nmti., the further extension of this spread did not occur until after a length of 18 mm. was attained. At this stage the replaced foreign solution passed from the fourth ventricle through two areas in the roof-plate. The chorioid plexuses now have divided the roof into two portions; from each, fluid escaped. The superior area of fluid passage is the same which was concerned in the mitial outpouring of the ventricular fluid. The inferior area, like the superior, is an area of ependymal differentiation, of which the first evidence may be made out in pig and human embryos of 15 mm. This differentiation consists in the transformation of the densely staining ependymal elements into cells with larger nuclei, poor in chromatin, and with more abundant cytoplasm.
 
 
After the functional employment of the two membranous areas is established at about 18 mm. in the pig, the further pericerebral spri.'ad of the replaced solution occurs very rapidly. The peribulbar tissues are filled with the fluid and from this region extensions occur downward into perispinal spaces and upward into the more basilar pericerebral spaces. Thus, the spinal spaces must be considered as develop>ing physiologically from above, and not from below upward, as Reford found. The complete filling of these perispinal spaces is found in pig embryos of 21 mm. At this stage the pericerebral spaces are filled, with the exception of those around the superior portion of the midbrain and about the cerebral hemispheres.
 
 
The final filling of all the periaxial spaces occurred in pig embryos of about 26 mm. This phenomenon may be taken to indicate the estabUshment of the true cerebro-spinal relationships of the adult, for in this case there is an intraventricular production of the fluid and an extraventricular spread. Likewise, the fluid returns to the venous system in embryos of over 23 mm., and this escape of the fluid from its periaxial bed is, as in the adult, directly into the venous sinuses of the dura mater.
 
 
The rapidity of the further extension of the replaced solution after the stage of 18 mm. is passed is apparently due to a second marked acceleration in the rate of production of the ventricular cerebro-spinal fluid. As in the first instance, this increased elaboration seems connected intimately with the formation of the chorioid plexuses of the third and lateral ventricles. As soon as these tufts develop, the cerebro-spinal fluid is produced in amounts which far exceed the quantities for which the more slowlj' enlarging ventricles can provide.
 
 
The histories of the two areae membranaceae of the fourth ventricle are dissimilar. Both are areas apparently differentiated from the normal lining ependyma for a specific functional purpose — the passage of fluid from the ventricles into the future subarachnoid spaces. The superior membranous area reaches its maximum functional importance in the stages of 18 to 20 mm. in the pig and also in the human embryo and from these stages on it slowly regresses. The final obUteration of the area, if it do not persist as an occasional small remnant, is due to the increasing growth of the cerebellum and the enlargement of the chorioid plexuses of the fourth ventricle. On the other hand, the inferior membranous area continues to increase both in size and functional importance after its initial differentiation from the ependyma; it finally occupies the greater portion of the velum chorioidea inferior. These observations can not solve the interesting question of a perforation of the inferior velum to form the foramen of Magendie.
 
 
Of the factors which influence the passage of fluid outward into the periaxial spaces, it must be reahzed that probably there is difference in this regard between the true solutions of the salts and the colloidal suspensions. For the true solutions (as in the experimental replacements) diffusion probably plays some role; but that this is not the sole factor is shown by the failure of the fluid to pass through the membrane in the stages under 14 mm. The findings of the granules of prussian-blue within the cytoplasm of the cells of this membrane mdicates that the fluid passage is similar in every way to that through a true membrane. There is also a possible site of fluid passage between the cells of this membrane. But, surely, the most important factor in this process is one of filtration of the fluid from the point of higher pressure to one of lower. This is mdicated by all of the findings : that the mcreased production of the fluid or the increased mtra ventricular pressure (whether due to normal or experimental agencies) causes a marked extraventricular spread seems firmly established. For the colloidal suspensions (particularly the protein of the normal ventricular fluid) a slower process of diffusion and filtration seems the probable agencj^ for passing the ventricular colloids into the subarachnoid spaces.
 
 
That the results obtained by the method of replacement were not solely due to diffusion, but represent a fiUing of the physiological extent of the cerebro-spinal spaces, has been shown in many ways, but probably the chief argument against such a view is that whollj^ similar extensions of the foreign solution may be obtained by injections under mild pressures from a syrmge; with increasing pressures these injections show the same type of spread, but always in a smaller embryo than the replacement method demonstrates as the standard for a given stage of the extension. The results recorded in the foregoing pages indicate also that suspensions (India ink) and true solutions (when powerful precipitants) are valuable only for affording comparisons in problems concerning the normal processes of absorption.
 
 
Of interest in any discussion of the results of injections into the perispinal spaces or into the spinal central canal are the findings in regard to the perineural spaces. It is possible to inject such spaces around each of the segmental nerves, but only when the pressures of injections are extreme. In no case, however, were such injections found to enter the lymphatic system — a finding in accord with the observations of Reford and Sabm. The physiological importance of these spaces in the adult is probably great, but the same methods of demonstration (with carefully controlled pressures) which suffice in the adult are unavailing in the embryo.
 
 
The origin of the three meninges from the perimedullary mesenchjTue is well established. His, Kolliker, Sterzi, Farrar, and others have placed this conception on a very firm basis. Most of the investigators have been concerned with the differentiation of the spmal meninges, while the observations here reported have been concerned solely with the cranial portion of these membranes. In general, the same phenomena in the transformation of the primitive periaxial mesenchyme as recorded by these earlier workers may be found in the cranium. The division of the primitive mesenchyme by a secondary condensation, a view advanced chiefly by Salvi, seems well supported. The findings in the cranium are in accord with this concej^tion; the outer portion of this primitive meninx becomes the dura mater, the inner forms both the pia and arachnoid. The processes in the formation of the arachnoid are, however, quite diversified and concern both the formation of the subarachnoid spaces and the outer membrane of the arachnoid.
 
 
Out of the rather loose-meshed periaxial mesenchyme, the subarachnoid spaces develop. The process concerns the transformation of the small " tissue spaces " of this mesenchyme into the larger subarachnoid channels, which are interrupted by the well-known arachnoid trabecule. Well-marked stages in this metamorphosis, which begins in the basis cranii, can be made out. The first appearance of a differentiation is seen in the gradual increase in the size of the mesenchj'mal mesh. This is closely as.sociated with an increased amount of an albuminous coagulum which in a measure fills the larger interstices. Following this initial dilatation of the spaces occurs a breaking-down of some of the syncytial strands; these ruptured mesenchymal processes then retract and adhere to the persisting trabeculue. The process continues with the formation of larger channels in this mesodermal tissue, with also the formation of the permanent arachnoidal trabeculae. Throughout these larger spaces, in the smaller fetuses, the coagula of protein material are everyT\-here found, the remains apparently of the albuminous portion of the circumambient fluid.
 
 
In the formation of the various cisternse, particularly the great cistema cerebellomedullaris, the process of the dilatation and confluence of the original mesenchymal spaces reaches its maximum. Here the breaking-down of the original sjTicytial strands proceeds to such an extent that very few of the strands remain to persist through hfe.
 
 
Such a process of the enlargement of mesenchj^ual spaces to form the larger subarachnoid spaces, as described in some measure by His for the spinal meninges, is apparently intimately connected with the circulation through these spaces of the embryonic cerebro-spinal fluid. The fluid flows everj- where through the spaces, as evidenced by the replacement experiments and by the increased content in albumen, before the process of enlargement of the mesenchymal spaces begins. It seems most likely that this circulation of the fluid acts as the causative agent in initiating and probably also in completing the enlargement of the "tissue spaces." The great content of albumen in the embryonic cerebro-spinal fluid has greatly facihtated the investigation, as the presence of the coagula from this protein has permitted the absolute exclusion of artifacts in the process of the tissue-dilatation.
 
 
This mechanism of enlargement of the tissue spaces finds its analogue in the formation of the anterior chamber of the eye and in the perilymphatic spaces of the ear (Streeter). In both these situations, as in the meningeal spaces, there are special body-fluids, more or less characteristic in their physical and chemical characters, obviouslj' subserving specialized functions. In both the eye and cranium, the absorption of the fluids is by way of special organs, directly into venous sinuses; in both, the origin of the specialized fluid is from epidermal organs; this fluid is at fijst poured into epidermal spaces and then subsequently into mesodermal spaces (subarachnoid space and anterior chamber of the eye). Thus, m these situations, the characteristic fluids have certain definite channels through rather larger spaces, connected finally with the venous system, and only indirectly with the Ij-mphatic system.
 
 
In no sense must the cerebro-spinal circulation be taken as a portion of the lymphatic system. Increasing knowledge of the cerebro-spinal fluid, of its physiology and chemistry, and of its pathway, have separated it permanently from any connection with the lymph of the lymphatic system, variable though that be. No longer may the meningeal spaces be compared to serous cavities, except possibly in the case of the subdural space, and this space is really a space apart from the true cerebro-spinal or subarachnoid spaces. Quite s i milarly, in place of the many varjdng channels in the dura and to a lesser extent in the leptomeninges, which older writers considered lyonphatic in nature, our increasing knowledge has caused the introduction of specialized arachnoidal cell-chains running throughout the pachymeninx. Unquestionably, the cerebro-spinal fluid i)ossesses its own peculiar and characteristic pathway, analogous in no way to the lymphatic vessels of other tissues.
 
 
The outer continuous membrane of the arachnoidea forms as a mesenchymal condensation, at first in common with the inner surface of the dura mater, but soon separated from it by the subdural space. The very low cubical mesothelium which covers the arachnoid membrane on both surfaces and also invests the arachnoid trabeculae differentiates apparently from the original mesenchymal elements in the periaxial tissues.
 
 
One of the most interesting features of this study has been the relation of the various mesenchymal condensations to the foreign true solution which was introduced into the medullary-canal system. This fluid circulated throughout the periaxial spaces which enlarge to form the subarachnoid channels, but it never penetrated the primary blastema which served as a primitive dura, nor did it ever invade the pial cells which so closelj^ adhere to the nervous tissue; hkewise, as soon as the secondary mesenchymal condensation dividing the dura from the arachnoid spaces appeared, this condensation was impervious to the true solution. No evidence of any penetration, as might be expected as due to diffusion, could be made out.
 
 
This summarj' has been included in order that some correlation between the topics discussed separately in the foregoing sections might be made. No attempt has been made here to present the findings in abstract form; these have been summarized at the end of each division of this communication.
 
 
 
 
 
XV. CONCLUSIONS.
 
 
Based on the observations recorded in the foregoing sections, the following conclusions seem warranted:
(1) During the earlj' part of the growth of the pig embryo there is no extraventricular spread of the cerebro-spinal fluid. The first extension of the ventricular fluid into the periaxial tissues occurs in pig embryos of 14 mm.; the adult relationship of the ventricular and meningeal cerebro-spinal fluid is established in pig embryos of about 26 mm.
 
 
(2) The ventricular cerebro-spinal fluid escapes into the periaxial tissues through two areas of ependjmal differentiation in the roof of the fourth ventricle. Both of these areas differentiate at a shghtly earlier period than that at which they function actively. The area membranacea superior undergoes a gradual regression and obliteration due to the changing form of the rhombic roof; the area membranacea inferior gradually occupies the major portion of the velum chorioidea inferior.
 
 
(3) The embryonic cerebro-spinal fluid, as evidenced by the replacement with a true solution, spreads from the ventricles into the mesenchymal tissue about the central nervous system. It docs not penetrate the cranial or vertebral blastemal condensations, nor does it invade the pial cellular layer.
 
 
(4) The subarachnoid spaces arise by a process of breaking-down of the perimedullary mesenchj'mal sj'ncytium and a dilatation of the existent mesench3Tnal spaces. This phenomenon of the enlargement of the mesenchymal spaces is associated with the presence in the spaces of an increased amount of albumen. The process occurs at a period shghtly later than that at which the initial flow of the cerebro-spinal fluid into the spaces is recorded.
 
 
(5) The dura mater, arachnoid, and pia mater develop out of the perimedulla
 
 
ry mesenchj'me. The arachnoid trabeculae are left by the breaking-down of the original mesenchymal strands, while the outer arachnoid membrane is formed, together with the inner surface of the dura, by a separate mesenchymal condensation. The dura develops between this secondary Une of condensation and the embryonic skull.
 
 
(6) There is indicated a very close relationship between the tufting of the chorioid plexuses of the fourth ventricle and the first extraventricular spread of the cerebro-spinal fluid.
 
 
(7) By means of the method of replacement it is possible to demonstrate perineural spaces as far out along the nerve trunks as the peripheral gangUa. The extensive injections of the perineural spaces along the segmental nerves are not obtained by the method of replacement.
 
 
The work, of which this paper forms the report, was done in the Anatomical Laboratory- of the Johns Hopkins Medical School. It was largely due to aid received from the Department of embryo
 
 
logj^ of the Carnegie Institution of Washington that the completion and scope of this paper were possible. The wTiter gladly acknowledges his indebtedness to the Carnegie Institution. January, 1916.
 
 
 
 
 
BIBLIOGRAPHY.
 
 
 
 
 
Ad AMKiEwicz, Albert. Die Blutgef &S8e des menschlichen Rilckenmarkes. I Theil: Die Gefassc der RQckenmarks-Substanz. II Theil: Die Gefassc der Rlickcnmarksobcrflache. Sitjungshcrichte der k. Akad. d. Wiss. in Wien. Math.-naturw. CI., 1H81, Lxxxrv, pt. 3. 469; 1882, Lxxxv, pt. 3, 101.
 
 
Bardeen, C. R. The development of the skeleton and of the connective tissues. In Kcibel and Mall, Manual Fluman Embryology, Philadelphia and Ix>ndon, 1910, I, 292.
 
 
BuAKE, J. A. The roof and lateral recesses of the fourth ventricle considered morphologically and embryologically. Jour. Comp. Neurol., Granville, Ohio, 1900, X, 79.
 
 
Canniec, a. Contribution & I'dtude de la voOve du quatrifeme ventriculc chez Ics mammifires; le trou de Magendie. Jour, de Mfed. de Bordeaux, 1897, xxvii, 547.
 
 
Cappelletti, L. L'^coulemcnt de liquide cerebrospinal par la fistule c^phalo-rachidienne en conditions normales ct sous I'influence de quelques m6dicaments. Arch. Ital. de Biol., Turin, 1900, xxxv, pt. 2. 463.
 
 
. L'efflusso del liquido cerebro-spinale dalla
fistola cefalo-rachidiana in condizioni normali e sotto I'influenza di alcuni farmaei. .Vtti dell' / ccademia d. Scienze Med. e Nat. di Fcrrara, 1900, ^xxiv, 85.
 
 
Cathelin, F. La circulation du liquide ctphalorachidien. Paris, J. B. Bailli^re, 1912.
 
 
CoTL'GNO [CoTUNNius], DoMENico. De ischiade nervosa commentarius. Viennae, R. GriifTcr, 1770.
 
 
Creighton, C. Microscopic researches on the formative properties of glycogen. London, \. & C. Black, 1896-99.
 
 
CusniNO, H. Physiologische und anatomische Beobachtungen iiber den Einfluss von Hirnkompres-sion auf den intracraniellen Kreislauf und ilber einigc hiermit verwandtc Erscheinungcn. Mittcilungen a. d. Grenzgeb. d. Med. und Chir., Jena, 1902, ix, 773.
 
 
. Some experimental and clinical observations
concerning states of increased intracranial tension. Am. Jour. Med. Sciences, Philadelphia and New York, 1902, cxxi\-, 375.
 
 
. Studies on cerebro-spinal fluid: No. 1. Introductory. Jour. Med. Research, Boston, 1914, xxxi (n. s. XXVI), 1.
 
 
Dandv, W. E., and K. D. Buckfan. An experimental and clinical study of internal hydrocephalus. Jour. Am. Mod. Assoc, Chicago, 1913, Lxi, 2216.
 
 
. Internal hydrocephalus: An experimental, clinical and pathological study. Aiii. Joiir. Dis. Children, Chicago, 1914, viii, 406.
 
 
Ekker, H. De cerebri et meduUte spinalis systemate vasorura capillari. Trajecti ad Rhenum, 1853.
 
 
Kngel, E. A. Uelx^r die Secretionscrscheinungen in den Zellen der plexus chorioidei des Mcnschen. Arch. f. Zellforschung, Leipzig, 1909, ii. 191.
 
 
EssicK, C. R. Transitory' cavities in the corpus striatum of the human embryo. Contributions to Embrj-ologj-, 1915, II, No. 6, 95. (Carnegie Institution of Washington, Pub. No. 222.)
Evans, H. McL. The macrophages of mammals, Am. Jour. Physiol., Baltimore, 1915, xxxvii, 1.
 
 
Faivre, J.-J.-A.-E. Des granulationes miningiennes. Thfee de Paris, 1853. (No. 142, vol. 540.)
Farrar, C. B. The embryonic pia. .\m. Jour. Insanity, Baltimore, 190& 7 LXiii, 295.
 
 
FiNDLAT, J. W. The choroid plexuses of the lateral ventricles of the brain — their histolog}-, normal and pathological. Brain, London and New York, 1899, xxu, 161.
 
 
 
 
 
18. Frazier, C. H. The cerebro-spinal fluid in health and
disease. Jour. Am. Med. Assoc, Chicago, 1915, LXIV, 1119.
 
 
19. Gaupp, E. Die Entwickelung des Kopfskelettes. In
Hertwig's Lehrbuch der Entwickelungsgeschichte d. Menschcn u. d. W'irbcltierc. Jena, 1906, K od., 692.
 
 
20. Goldmann, E. E. V'italfarbung am Zentralnerven
systcm. Berlin, G. Reimcr, 1913.
 
 
21. Haller, a. Elementa physiologix corp^jris humanis.
 
 
Lausanne, 1757-78.
 
 
22. Hess, Carl. Das Foramen Magendii und die Oeffnung
en an den Reces.iu.s laterales des \*ierten Ventrikels. Morphol. Jahrbuch, Leipzig, 1885, x, 578.
 
 
23. Hecser, C. H. The development of the cerebral ven
tricles in the pig. Am. Jour. Anat., Philadelphia, 1913-14, XV, 215.
 
 
24. Hill, L. Physiology- and pathology of the cerebral
circulation. London, J. & A. Churchill, 1896.
 
 
25. His, W. Ueber ein perivasculiires canalsystem in den
nervoseu Ccntralorganen. Leipzig, Engelmann,
1865. . Die Hfiute und Hdhlen des Korpers; ein aca
demisches Programm. Basel, Schweighauser, 1865. . Die Entwickelung des mcnschlichcn Rauten
hirns, von Ende des ersten bis zum Beginn des
dritten Monats. Leipzig, Hirzel, 1890. . Die Hiiute und Hohlen des Korpers. Arch. f.
 
 
Anatomie u. Physiol., Anat. .\bth., 1903, 368. . Die Entwickelung des menschlichen Gehims.
 
 
Leipzig, Hirzel, 1904.
 
 
26. HwoROSTUCHiN, W. Zur Frage uber den Bau des Plexus
chorioideus. Arch. f. mikr. Anat., Bonn, 1911, Lxxvii, 232.
 
 
27. Jacobson, C. Unpublished.
 
 
28. Kadvi, Heinrich. Ueber die Blutgefisse des mensch
lichen RQckenmarkes. Lemberg, Gubrj-nowicz 4 Schmidt, 1889.
 
 
29. Key, E. A. H., and G. Retzius. .\natomie des Nerven
systems und des Bindegewebcs. Stockholm, Samson & Wallin, 1875-76.
 
 
30. Knower, H. McE. a new and sensitive method of inject
ing the vessels of small embrjos, etc., under the microscope. Anat. Record, Philadelphia, 1908, ii, 207.
 
 
31. KoLLlKER, A. Entwiekelungsgeschichte des Mcnschen
und der hoheren Thiere. 2ed. Leipzig, Engelmann, 1879.
 
 
. Gnmdriss der Entwiekelungsgeschichte des
Mcnschen. I^eipzig, Engelmann, 18S0.
 
 
32. KoLLMANN, J. Lehrbuch der Entwiekelungsgeschichte
des Menschen. Jena, Fischer, 1898.
 
 
33. Lewandowskt, M. Zur Lehre von der Cerebrospinal
flQssigkcit. Zeitachr. f. klin. Med., Berlin, 1900, XL, 480.
 
 
34. vov LrscHKA, H. Die .\dergeflechte des menschlichen
Gehimes. Berlin, G. Reimer, 1855.
 
 
35. Magen-die, Francois. Recherches sur le liquide csfephalo
rachidien. Paris, 1825.
 
 
. Recherches philosophiqucs et cUniques sur le
liquide c^phalo-rachidien ou c6r6bro-spinal. Paris, 1842.
 
 
36. Mall, F. P. On the development of the blood-vessels
of the brain in the human embryo. Am. Jour. Anat., Baltimore, 1904, rv, 1.
 
 
37. Meek, Walter J. A study of the choroid plexus. Jour.
 
 
Comp. Neurol, and Psychol., 1907, xvii. No. 3, 286.
 
 
38. Mestrez.\t, W. Le liquide c^phalo-rachidien. Paris,
Maloine, 1912. Thise de Montpellier, 1911, No. 17. .39. MiDDLEM.tss, James, and W. F. Robertson. Pathologyof the nervous system in relation to mental diseases. Part III. Morbid conditions of the dura mater. Edinburgh Med. Jour. 1895, xl, pt. 2, 704.
 
 
MiNOT, C. S. Human enibr>'olog>'. New York, W.
 
 
Wood A Co.. 1892. MoTT. F. W. The OIiver-.Shari)cy lccture.s on the cerc
bro-spinal fluid. Liincct. London. 1910, ii, 1 ; 79. Peluzzi. B. Experinientelle histologische Untersuch
ungcn ilbcr die Plexus chorioidei (Adcrgeflechte).
 
 
Foli.i neuro-biologica, Haarlem, 1911. v, No. 4, 305. Pf.ttit. a., and J. Giiuhd. Sur la fonetion .s4or6toire
ct la morphologie des plexus chorioldes des ventri
lulcs latfraux du Syst^nie ncrveux central. Arch.
 
 
d'anat. micr., Paris, 1902, v. 214. PuLT, F., O. Kehm, and H Schott.mOller. Leitfadcn
zur Untcrsuchung der ZerebrospinalflOssigkeit.
 
 
Jena, Fischer, 1913. PoiBiER, P., and A. Charpy. Traits d'antomie huniaine.
 
 
Pari,'*, BattaiUe, 1892-1902. Qci-vcKE, H. Zur Physiologie der Cerebrospinalflussig
keit. Arch, f . Anat., Physiol, u. wissensch. Med. (Du
Bois-Reymond), Leipzig, 1872, 153. Refobd, L. L. Unpublished. Work referred to by
Sabin (49) and Gushing (9). Reiner, M., and J. Schnitzler. Ueber die Abflu.sswege
des Liquor cerebrospinalis. Centralblatt f. Physiol.,
Leipzig and Wien, 1894, viii, 6H4. . Zur Lehre vom Hirndruck. Wiener klin. Wo
chenschr.. 1895, viii, 371. Sabi.n, F. R. Development of lymphatic sj-stcm. In
Keibel and Mall's Manual Human Embryology,
London 4 Philadelphia, 1912, ii, 709. Salvi, G. Histogenise et structure des meninges.
 
 
Thise de Pari.", 1898. . Quoted in Poiricr et Charpy. Traitt d'Ana
tomie humaine. Paris, 1901, tome iii, 113. ScHROEDER VAN DER KoLK, J. L. C. Bau und Func
tionen der Medulla spinalis und oblongata. Aus
dem Holliindischen iihertragen von Dr. Fr. W.
 
 
Theile. Braunschweig, Vieweg & Sohn, 1859. Spin.i, a. Expcrimenteller Beitrag zur Kenntniss der
Hyperamic des Gehirns. Wiener med. Blatter,
1898, XXI, 17; 247. . Experimentelle Untcrsuchungen iiber die Bil
dung des Liquor cerebrospinalis. Arch. f. d.
 
 
gesamte Physiol., Bonn, 1899, Lxxvi, 204.
 
 
 
 
 
52. Spina. A. Uebcr den Einfluss des hohen Blutdrucks auf
die Neubildung des CerebrospinalflQssigkeit. Arch, f. d. gcsamtc Physiol., Bonn, 1900, i.xxx, 370.
 
 
. Untersuchungen uber die Resorption des Liquor
bei normulcm und crhohtem intracraniellem Drucke. .Vrch. f. d. gesamte Physiol., Bonn, 1900-1901, Lxxxiii, 120; 415.
 
 
53. .'^TERZi, G. Recherches sur I'anatomie comparfee et sur
rontogent-se des m6ninges. Arch. Ital. de Biol., 1902, XXXVII, 257.
 
 
. Ricerche inforno all' anatomia comparata ed
air ontogenesi dcllc mcningi, e considerazioni suUa filogenesi. Atti del R. Instituto Veneto di scienze, letterc ed .\rti, 1900-1901, LX, parte ii, 1101.
 
 
54. Streeter, G. L. Development of the nervous system.
 
 
/n Keibel and Mall, Manual Human Embryology, London & Philadelphia, 1912, ii, 1.
 
 
. The development of the venous sinuses of the
dura mater in the human embrjo. Am. Jour, of Anat., Philadelphia, 1915, xviii, 145.
 
 
55. Weed. L. H. Studies on cerebro-spinal fluid. No. II:
The theories of drainage of cerebro-spinal fluid with
an analysis of the methods of investigation. Jour.
 
 
Med. Research, Boston, 1914, xxxi (n. s. xxvi), 21. . Studies on cerebro-spinal fluid. No. Ill: The
pathways of escape from the subarachnoid spaces,
with particular reference to the arachnoid Wlli.
 
 
Jour. Med. Research, Boston, 1914, xxxi (n. a.
 
 
XX VI), 51. . Studies on cerebro-spinal fluid. No. IV: The
dual source of cerebro-spinal fluid. Jour. Med.
 
 
Research. Boston, 1914. xxxi (n. s. xxvi). 93.
 
 
56. Wilder, B. G. Notes on the Foramina of Mageudie in
man and the eat. Jour. Nerv. and Ment. Dis., New
York, 1886, xiii, 206. . The metapore (foramen of Magendie) in man
and an orang. Med. News, Philadelphia, 1893,
LXiii 439. . Meninges. Ref. Handbuch Med. Sciences, New
York, 1893, ix (suppl.), 606.
 
 
57. ZiEOLER, P. Ueber die Mechanik des normalen und
pathologischen Hirndruckes. Arch. f. klin. Chirurgie, Berlin, 1896, Liii, 75.
 
 
 
 
 
EXPLANATION OF PLATES.
 
 
KEY FOR FIGURE-LEGENDS.
 
 
ami, area membranacea inferior. dmc, dura mater cerebri (inner surface, in pme, pia mater cerebri.
 
 
amt, area membranacea .superior. approximation with arachnoid). j»p6, precipitated prussian-blue.
 
 
cbl, cranial blastema. epe, epithelial-like cells linini; ventricle. pun, reduced silver nitrate.
 
 
cent, cisterna cerebello-mcduUari.i. epe, ependyma. s<u, subarachnoid spaces.
 
 
chp, plexus chorioiduus. 4"^, vcntriculus quarlus. tir, sinus trans\ersu3.
 
 
Plate I.
 
 
Fig. 1. Drawing of a pig embryo of 9 mm., into the spinal central canal of which an injection of 0.5 per cent solution of potassium ferrocyanide and iron-ammonium citrate was made under very mild sjiinge-prcssure. The embr>-o was fixed in Camoy's fluid to which 1 per cent hydrochloric acid had been added. The specimen was carefully dehydrated and cleared by the Spalteholz method. The resultant precipitate of prussian-blue is found w holly within the central canal of the spinal cord and within the cerebral ventricles. Enlargement, 11 diameters.
 
 
Via. 2. Drawing of a pig embryo of 13 mm., in which the cerebro-spinal fluid was replacc<l by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. The embrj-o was kept alive for 90 minutes after this replacement and was then fixed in 10 per cent formol containing 1 per cent hydrochloric acid, .\fter dehydration the specimen was cleared by the Spalteholz method. The occiurence of a definite oval, outlined by the denser mass of the granules, in the roof of the fourth ventricle, is characteristic of this stage. Enlargement, 9 diameters
Fig. 3. Drawing of a pig embrj-o of 14.5 mm. ir which the cerebro-spinal fluid was likewise replaced by the ferrocyanide solution, .\fter the repLncement, the embryo
 
 
was kept alive for 60 minutes; it was fixed in Camoy's fluid (with 1 per cent hydrochloric acid added) and after dehydration it was cleared by the SpaltehoU method. The earliest indications of a peria.idal spread of the replaced fluid from the roof of the fourth ventricle is here shown. Enlargement, 8 diameters.
 
 
Plate II.
 
 
Fig.
 
 
4. Drawing of a pig embr>'0 of 18 mm., in which a typical replacement of the spinal fluid had been made. The animal was kept alive for 45 minute." and was then fixed, dehydrated, and cleared in the usual manner. The extra ventricular spread of the replaced fluid from two are;is in the roof of the fourth ventricle is well illustrated. Enlargement, 9 diameters.
 
 
Fig.
 
 
5. Drawing of a pig embryo
 
 
of 19 mm., in which likewise a typical replacement of the cerebro-spinal fluid by the ferrocyanide solution had been made. After this procedure, the embryo
 
 
was kept aUve for 55 minutes and was then carried through the routine technique for the Spalteholz method. The ftirther pericerebral spread of the replaced fluid is recorded. Enlargement, 8 diameters.
 
 
Pl.^te III.
 
 
Fig.
 
 
6. A frank lateral drawing of a pig embryo of 21 mm. The typical replacement of the embryonic cerebro-spinal fluid by the ferrocyanide solution was effected in this embryo
 
 
and it was then kept aUve for 45 minutes. At the end of this time the embryo was fi.xed in an acid fluid, dehydrated, ana cleared. The almost complete periaxial spread of the replaced fluid is indicated by the precipitated granules. Enlargement, 7.6 diameters.
 
 
Fia. 7. A dorsal view of the embryo illustrated in fig. 6. The perispinal spread of the replaced fluid is well shown. Enlargement, 7.8 diameters.
 
 
Plate IV.
 
 
Fia. 8. Drawing of a pig embryo of 26 mm. in which the t j-pical replacement of the cerebro-spinal fluid has been made. After the introduction of the ferrocyanide solution the embryo was kept alive for one hour; at the end of this time it was fixed in an acid solution, subsequently dehydrated, and cleared in oil of wintergreen. The specimen shows a complete periaxial spread of the replaced fluid, as evidenced by the precipitated granules, in addition to a total filling of the intramedullary system. Enlargement, 6.5 diameters.
 
 
FiQ. 9. Drawing of a pig embrj-o of 16 m.m., in which the central canal of the spinal cord was injected with the ferrocyanide so.u'ion under moderate syringe-pressure. After fixation in an acid mediiun the embn,-owaa dehydrated and cleared by the Spalteholz method. The extraventricular snread in the peribulbar region Ls easily made out. Enlargement, 9 diameters.
 
 
 
Plate V.
 
 
Fig.
 
 
10. Drawing of a pig embryo of 21 mm., in which an injection of diluted india ink was made into the central canal of the spinal cord. The preasure employed was the highest obtainable from the syringe, yet below the tension causing rupture. The specimen, after injection, was fixed, dehydrated, and cleared. The slight extent of the periaxial spread of the carbon granules can be easily seen. Enlargement, 7 diameters.
 
 
KiG. U. Drawing of a pig cmbrj-o of 16 mm., in which an injection (under moderate syringe-pressure) of 0.5 per cent solution of silver nitrate was made into the central canal of the spinal cord. The silver was reduced in the sunlight, the erabrj-o then fixed. After dehydration, the embryo was cleared in benzol and oil of wintergreen. Enlargement, 7. .5 diameters.
 
 
TiQ. 12. Drawing of a pig embryo of 13 mm.; into the central canal of the spinal cord a dilute solution of nitrate of ."ilver was injected under strong syringe-pressure. Reduction of the silver was accomphshed by exposure to sunlight; the embryo was then fixed, dehydrated, and cleared. Enlargement, diameters.
 
 
Plate VI.
 
 
Fig. Vi. I'hotomicrograph of transverse section of a pig embryo of IS mm. Specimen obtained from an embryo in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and ironammonium citrate. After this replacement the embryo was kept aUve for 65 minutes. The resultant priL'Ssian-blue precipitate is not included in this photomicrograph. Enlargement, 13 diameters.
 
 
Fig.
 
 
14. Drawing of blocked area in fig. 13, under higher magnification and including the resultant precipitate of prussian-blue. The typical ependymal cells (epc) lining the fourth ventricle are shown on either side; between them occurs the area membranacea superior {ams). The transit of the replacement fluid through the membranous area and the spread through the adjacent mesenchyme are illustrated. Enlargement, 245 diameters.
 
 
I'lo. 13 rhotomicrograph of transverse section from embryo pig illustrated in fig. 13. Section taken from more caudal plane than that given in the former figure. The prussian-blue spread is not illustrated. Enlargement, 10 diameters.
 
 
Fig.
 
 
16. Drawing, under higher magnification, of the rectangular area in fig. 15. The passage of the replaced solution, as shown by the resultant precipitate of prussian-blue, through tlie area membranacea inferior (ami) is here illustrated. The extension of the replaced fluid through the adjacent mesenchyme and the nonpenetration of the solution into the condensed mesenchyme are shown. Enlargement, 140 diameters.
 
 
Flo. 17. Photomicrograph of sagittal section of a pig embryo of 18 mm. Specimen obtained from an embrj-o in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. After this replacement the animal was kept alive for 45 minutes. Fixed for 5 minutes in 10 per cent formol containing 1 per cent hydrochloric acid; then over night in modified Boiiin's solution (saturated aqueous solution of picric acid 75, formaldi-hyde 10, glacial acetic arid 10). Dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraffin. Serial sections, stained by hematoxylin and eosin. The resultant precipitate of prussian-blue has not been reproduced in the photomicrograph. Enlargement, 8 diameters.
 
 
Fig. 18. Drawing of blocked area in fig. 17 under higher magnification. The granules of prussian-blue are here represented by the blue stenciling. The transit of the fluid, as shown by the granules, into the periaxial mesenchyme through the two membranous areas {arm and ami) in the roof of the fourth ventricle are well shown. Enlargement, 35 diameters.
 
 
Plate VII.
 
 
Fig.
 
 
19. Photomicrograph from a sagittal section of a fetal pig of 27 mm. The cerebro-spinal fluid in this specimen was replaced by a 1 per cent solution of potas.siuni ferrocyanide and iron-ammonium citrate; the fetus was kept alive for 40 minutes; fixed in 10 per cent formol containing 1 per cent hydrochloric acid for 15 minutes; then over night in modified Benin's solution; dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraflTm. The prussian-blue granules are not represented in this photomicrograph. Enlargement, 8 diameters.
 
 
Fig.
 
 
20. Drawing of squared area in fig. 19. The center of the field is occupied by the optic ner^-e; around it the developing extrinsic optic muscles are shown. The precipitate of prussian-blue occurs in the perineural mesenchyme. Enlargement, 190 diameters.
 
 
Fig.
 
 
21. I'hotomicrograph of rectangular area in fig. 19. The passage of the ferrocyanide solution into the sinus transvcrsus {sir) is represented by the precipitated blue granules. Enlargement, 133 diameters.
 
 
Via. 22. Pliotomicrograph of a transverse section of a pig enibrj-o of 23 mm. The cerebro-spinal fluid was replaced in this embr>'o with a 1 per cent solution of potassium fenocyanide and iron-ammonium citrate. The embryo was kept alive for .50 minutes and was then fixed over night in 10 per cent formol containing 1 [ler cent hydrochloric acid. The granules of pnissian-blue are not shown in this reproduction. Enlargement, 13 diameters.
 
 
Via. 23. Drawing of squared area in fig. 22. The area membranacea superior (rima) is shown, sunounded on either side by tufts of the chorioid plexus (rhp) .ind the typical ventricular ependyma. The transit of the solution is shown, as represented by the resultant granules, through the areji, with the subsequent spreail into the periaxial mesenchyme. Enlargement, 125 diameters.
 
 
 
 
 
Plate VIII.
 
 
Fig.
 
 
24. PhotomicroKraph of a transverse section of a piR embryo
 
 
of 8 mm. Fixed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent grades of alcohol, embedded in xylol-parafiin. Enlargement, 30 diameters.
 
 
Fig.
 
 
25. Photomicrograph, retouched, of the blocked area in fig. 24. The character of the cells (epc) composing the roof of the fourth ventricle Uve) is showTi in this reproduction. Enlargement, 16.5 diameters.
 
 
Fig.
 
 
26. Photomicrograph of a sagittal section from a pig cinbrj-o of 11 mm. Fi.xed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent gracles of alcohol, embedded in xylol-parafhn. Enlargement, 11 diameters.
 
 
Fig.
 
 
27. Photomicrograph of the blocked area in fig. 26. The area membranacea superior (ams) in the roof of the fourth ventricle is shown sharply delimited from the two processes of typical ependyma {epe). Enlargement, 67 diameters.
 
 
Fig.
 
 
28. Photomicrograph of a more lateral section of the pig embryo
 
 
of 11 mm. given in fig. 26. Enlargement, 11 diameters.
 
 
Fig. 29. Photomicrograph, under higher magnification, of the blocked area in fig. 28. The lateral border of the area membranacea superior {ams) of the roof of the fourth ventricle is given. Enlargement, 50 diameters.
 
 
Fig.
 
 
30. Photomicrograph of a sagittal section from a pig embryo of 13 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 8 diameters.
 
 
Fia. 31. Photomicrograph, under higher magnification, of the squared area in fig. 30. The reproduction comprises a sagittal section of the area membranacea superior (ams) of the roof of the fourth ventricle. Enlargement, 67 diameters.
 
 
Fig. 32. Photomicrograph of a sagittal section of a pig embryo of 14 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 11 diameters.
 
 
Fig.
 
 
33. Photomicrograph of the blocked area in fig. 32 under higher magnification. The area membranacea superior (ami) in the roof of the fourth ventricle is reproduced. Enlargement, 75 diameters.
 
 
Plate IX.
 
 
Fig.
 
 
34. Photomicrograph of a transverse section of a pig embrj-o of 18 mm. Fixed in Camoy's fluid (6:3:1), deh3'drated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffin. Enlargement, 13 diameters.
 
 
Fig.
 
 
35. Photomicrograph, under higher magnification, of the blocked area in fig. 34. The area membranacea superior (atns) is here given, flanked on either side by typical ependyma (ept). Enlargement, 170 diameters.
 
 
Fig. 36. Photomicrograph of a transverse section of a pig embrj-o of 18 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 13 diameters.
 
 
Fig.
 
 
37. Photomicrograph of rectangular area outlined in fig. 36. The extent of the area membranacea superior (ams), with its adherent coagulum of albuminous material, is well differentiated from the adjacent typical ventricular ependyma (epe). Enlargement, 100 di.imeters.
 
 
Fig.
 
 
38. Photomicrograph of a transverse section of a pig embryo of 19 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 13 diameters.
 
 
Fig. 39. Photomicrograph, under higher power, of the rectangular area in fig. 38. \ small break in the integrity of the lining ependyma of the roof of the fourth ventricle, representing the irregular boundarj- of the area membranacea superior (ams), is given. Enlargement, 290 diameters.
 
 
Fig. 40. Photomicrograph of a transverse section of a human embryo of 4 mm. (No. 836 of collection of Carnegie Institution of Washington). Enlargement, 33 diameters.
 
 
Fig. 41. Photomicrograph, retouched, of the blocked area in fig. 40. The epithelial-like cells (epe) composing the roof of the fourth ventricle (4^) are here shown separated from the denser nervous tissue. Enlargement , 100 diameters.
 
 
Plate X.
 
 
FlQ. 42. Photomicrograph of transverse section of pig embryo of 19 mm. Fixed over night in modified Bouin's solution, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in x>-lol-paraffin. Enlargement, 13 diameters.
 
 
Fig.
 
 
43. Photomicrograph of squared area in figure 42, under higher magnification. The area membranacea superior (ams) with the attached coagulum of albumen is reproduced. Enlargement, 115 diameters.
 
 
Fig.
 
 
44. Photomicrograph of sagittal section of pig embryo of 23 mm. Fuced in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 5 diameters.
 
 
Fig.
 
 
45. Photomicrograph, under higher magnification, of squared area in fig. 44. The area membranacea superior (atns) is here shown, delimited by the cells of the chorioid plexus (chp) on one side and by the further ependymal prolongation (epc) of the cerebellar lip. Enlargement, 88 diameters.
 
 
Fig. 46. Photomicrograph of sagittal section of pig embryo
 
 
of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Certain portions of the dura mater (dmc) areindicat^. Enlargement, 5 diameters.
 
 
Fig. 47. Photomicrograph of blocked area in fig. 46, under higher magnification. The small remaining area membranacea superior (ams) is quite surrounded by encroaching ependyma in the chorioidal folds. Enlargement, 88 diametere.
 
 
Fig. 48. Photomicrograph of transverse section of human embryo of 7 nun. (No. 617 of the collection of the Carnegie
Institution of Washington). Enlargement, 10 diameters. Fiu. 49. Photomicrograph of squared area in fig. 48, under higher magnification. The epithehal-like cells (epc)
composing the roof of the fourth ventricle at this stage are well shown. Enlargement, 100 diameters. Fig.
 
 
50. Photomicrograph of transverse section of human embryo of 7 mm. (No. 617 in the Caniegie Institution of
Washington). Enlargement, 10 diameters. Fig.
 
 
51. Photomicrograph of blocked area in fiu;. 50. The marked invagination of the roof of the fourth \entricle
(4if ) with the lining of epithelial-like cells (epc) is given. Enlargement, 33 diameters. Fig. 52. Photomicrograph of transverse section of human embryo of 9 mm. (No. 721 in the collection of the Carnegie
Institution of Washington). Enlargement, 10 diameters. Fig.
 
 
53. Photomicrograph of squared area outlined in fig. 52. The pale, large cells {epc) comprising the roof of the
fourth ventricle characterize the reproduction. Enlargement, 50 diameters. Fig.
 
 
54. Photomicrograph of sagittal section of human embryo of 11 mm. (No. 544 in the collection of the Caniegie
Institution of Washington). Enlargement, 6 diameters. Fig. 55. Photomicrograph of blocked area in fig. 54. The apparent break in the continuity of the roof of the fourth
ventricle with exudation of the ventricular albumen into the mesenchyme is brought out. Enlargement,
.50 diameters.
 
 
Plate XI.
 
 
I-'iG. 56. Photomicrograph of sagittal section of human embryo of 14 mm. measured on the slide (No. 144 of the collection of the Carnegie Institution of Washington). Enlargement, 8 diameters. Fig.
 
 
57. Photomicrograph, under higher magnification, of blocked area in fig. 50. The greater part of the ventricular
wall shown is composed of the area membranacea superior (ams), bounded below by typical ventricular
ependyma {epe). Enlargement, 67 diameters. Fig. 58. Photomicrograph of sagittal section of human embryo of 17 ram. (No. |576 of the collection of the Carnegie
Institution of Washington). Enlargement, 10 diameters. Fig. 59. Photomicrograph of rectangular area in fig. 58, showing the area membranacea superior {ams) of the roof of
the fourth ventricle. Enlargement, 50 diametere. Fig.
 
 
60. Photomicrograph of sagittal section of human embryo of 17 mm. (No. 576 of the collection of the Carnegie
Institution of Washington). Enlargement, 7 diameters. Fig. 61 . Photomicrograph of tlie blocked area in fig. 60 under higher magnification. The aggregation of epithehal-like
cells (epc) on the lateral border of the area membranacea superior is here portrayed. Enlargement,
67 diameters. Fig.
 
 
62. Photomicrograph of transverse section of human embryo of 18 mm. (No. 409 of the collection of the Carnegie
Institution of Washington). Enlargement, 7 diameters. Fig.
 
 
63. Photomicrograph, under liigher power, of squared field in fig. 62. The peculiar inversion of the roof of the
fourth ventricle (/fVe) indicated in fig. 62, has resulted in a marked dislocation of the area membranacea
superior (nm«), shown in this figure. Enlargement, 75 diameters.
 
 
Plate XII.
 
 
Fig.
 
 
64. Photomicrograph, retouched, of a transverse section of a human embryo of 21 mm. (No. 460 of the collection of the Carnegie Institution of Washington). The field taken consists of a portion of the fourth ventricle with the lining of typical ependyma (cjk) on either side. The area membranacea superior {ams) is shown between the two lips of ependyma. Enlargement, 33 diameters.
 
 
Fig.
 
 
65. Photomicrograph, retouched, of a similar section to that given in fig. 64, but taken from a more anterior plane from the same embryo. The field shown is analogous in everj' way to that in the preceding figure.
 
 
Fig.
 
 
66. Photomicrograph of a transverse section of an embryo chick of 121 hours' incubation. Fixed in Bouin's solution. Enlargement, 15 diameters.
 
 
Fig.
 
 
67. Itetouched photomicrograph, under higher magnification, of the blocked area in fig. 66. The area memhranacoH superior (ams) is here given, delimited 8hari)ly from the lips of ependyma {cpc) which line the rtmf of the fourth ventricle. Enlargement, 133 diameters.
 
 
Fig.
 
 
68. Photomicrograph of a more caudal section from the same embryo as portrayed in fig. 66. Enlargement, 15 diameters.
 
 
Fig.
 
 
69. Retouched photomicrograph, under higher magnification, of the blocked area in fig. 68. The area membranacea superior {am») is shown at the point of its greatest transverse diameter. ICnlargement, 88 diameters.
 
 
Flo. 70. Photomicrograph of a migittul section of a pig embryo of 15 mm. Fixed in modlfie<l Bouin's solution, dehydratiil by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement , S diameters.
 
 
Fig.
 
 
71. I'hutomicrograph, under higher magnification, of blocked area in fig. 70. The earliest evidence of the area membranacea inferior (ami) in the roof of the fourth ventricle is here shown. I'nlargement, 12.'i diametere.
 
 
Fig. 72. Photomicrograph of sagittal section of a pig embryo of 18 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 11 diameters.
 
 
Fig. 73. Photomicrograph, under higher power, of the rectangular area outlined in fig. 72. The enlarging area membranucea inferior (ami) is shown in the midst of the typical lining ependyma of the roof. Enlargement, 100 diameters.
 
 
Fig. 74. Photomierogruph of sagittal section of a pig embryo of 23 mm. Fixed in modified Bouin's solution, dehydrated by 2 :\iid 4 per cent grades of alcohol, and embedded in xylol-paraflSn. Enlargement, 6 diameters.
 
 
Fig. 75. Photomicrograph of blocked area in fig. 74. The area membranacea inferior {ami) is, at this stage, quite extensive, as shown in the reproduction; the early stages in the development of the cistema cerebellomedullaria may also be seen. Enlargement, 75 diameters.
 
 
Fig.
 
 
76. Photomicrograph of sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 7 diameters.
 
 
Fig. 77. Photomicrograph, under higher magnification, of blocked area in fig. 76. The unsupported character of the area menibranarca inferior and the formation of the cistema cerebello-medullaris is here reproduced. Enlargement, 67 diameters.
 
 
Plate XIV.
 
 
Fig. 78. Photomicrograph of a sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xj'lol-parafiin. Enlargement, 7 diametera.
 
 
Fig.
 
 
79. Photomicrograph of the blocked area in fig. 78, under higher magnification. The intact area membranacea inferior (ami), unsupported by any mass of tissue, is shown separating the ventricular cavity from the developing cisterna ccrebello-medullaris. Enlargement, 67 diameters.
 
 
Fig. 80. Photomicrograph of a sagittal section of a human embrj-o of 16 mm. (No. 406 in the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.
 
 
Fig. 81. Photomicrograph of the area outlined in fig. 80, but under liigher magnification. An early stage in the differentiation of the area membranacea inferior (ami) is given. Enlargement, 50 diameters.
 
 
Fig. 82. Photomicrograph of a sagittal section of a human embryo of 17 mm. (No. 576 in the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.
 
 
Fig. 83. Photomicrograph, under higher power, of the area blocked in fig. 82. The chorioid plexuses of the fourth ventricle Ue in the central portion of the field; above is the thick cell-layer on the lateral side of the area membranacea suf)erior (ams), while below the upper limit of the area membranacea inferior (ami) appears. Enlargement, 67 diameters.
 
 
Fig. 84. Photomicrograph of a transverse section of a human embryo of 18 mm. (No. 409 in the collection of the Carnegie Institution of Washington). Enlargement, 5 diameters.
 
 
Fig.
 
 
85. Photomicrograph of the blocked area in fig. 84. The cellular character, and especially the clumping of cells, of the area membranacea inferior (ami) is shown. Enlargement, 25 diameters.
 
 
Fig. 86. Photomicrograph of a sagittal section of a human embryo of 19 mm. (No. 431 in the collection of the Cam^e Institution of Washington). Enlargement, 5 diameters.
 
 
Fig.
 
 
87. Photomicrograph of the blocked area outUned in fig. 86. The area membranacea inferior (ami) appears separating the fourth ventricle from the developing cistema cerebello-medullaris. Enlargement, 25 diameters.
 
 
Plate XV.
 
 
Fig. 88. Photomicrograph from a sagittal section of a human embryo of 17 mm. (No. 576 of the collection of the Carnegie Institution of Washington), representmg an enlargement of the second blocked area in fig. 58. The area membranacea inferior (ami) appears sharply delimited from the adjoining tj^pical ependjina. Enlargement, 67 diameters.
 
 
Fig. 89. Photomicrograph of a sagittal section of a human embrj-o of 23 mm. (No. 453 of the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.
 
 
Fig. 90. Photomicrograph of the blocked area in fig. 89. The area membranacea superior (ams) appears in the stage of closure, while the area membranacea inferior (ami) is becoming well differentiated from the tj-pical ependyma lining the other portions of the fourth ventricle. Enlargement, 26 diameters.
 
 
Fig. 91. Photomicrograph of a sagittal section of a human embrjo of 26 mm. (No. 1008 of the collection of the Carnegie Institution of Washington). Enlargement, 4.5 diameters.
 
 
Fig. 92. Photomicrograph, under higher magnification, of the blocked area in fig. 91 . The area membranacea superior has been almost completely closed by the dense ependyma of the superior half of the roof of the fourth ventricle, while the inferior area (ami) has become a membrane lacking wholly the character of ependyma. Enlargement, 23 diameters.
 
 
Fig.
 
 
93. Photomicrograph of a sagittal section of a human embrj-o of 35 mm. (No. 199 of the collection of the Carnegie Institution of Washington). Enlargement, 3 diameters.
 
 
Fig. 94. Photomicrograph, under higher powers, of the blocked areas in fig. 93. The formation of the cistema cerebello-medullaris is shown in relation to the ventricular roof. EnLirgemcnt, 23 diameters.
 
 
Fig.
 
 
95. Drawing of cells of the chorioid plexus from the lateral ventricles of a fetal pig of 132 mm. The specimen waa fixed in absolute alcohol, and stained by Best's carmine stain for glycogen. The glycogen occurs in the form of globules within the epithelial cells. Enlargement, 950 diameters.
 
 
Fig.
 
 
96. Drawings of the cells of the chorioid plexus from the lateral ventricles of a fetal pig of 36 mm. The specimen was fixed in absolute alcohol and stained by Best's carmine method. The glycogen appears in the epitheUal eclls in the form of basilar plaques. Enlargement, 950 diameters.
 
 
Plate XVI.
 
 
Fig.
 
 
97. PhotomicroRTaph of a transverse section of a pig embryo of 10 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent f;radcs of alcohol, and embedded in xylol-paraffin. Enlargement, 10 diameters.
 
 
Fig.
 
 
98. Photomicrograph, under higher magnification, of the blocked area in fig. 97. The double condensations of mesenchyme to form pia mater (pmc) and cerebral blastema (cbl) appear separated by a region of mesenchyme which is breaking down. This central area of mesenchyme, with the marked albumen-content, is to become the arachnoid spaces. Enlargement, 133 diameters.
 
 
Fig. 99. Photomicrograph of a transverse section of a pig embryo of 20 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffm. Enlargement, 10 diameters.
 
 
FlO. 100. Photomicrograph, under higlior powers, of the blocked areas in fig. 99. The relations of the pial condpn.sation ipmc) of mesenchyme to tlie nervous system, as well as the infiltration of tlie arachnoid mesenchyme (sas) with albumen, is reproduced. Enlargement, 133 diameters.
 
 
Fig.
 
 
101. Photomicrograph, under higher magnification, of the blocked area in fig. 22. The reproduction is included here to show the double condensjition {cht) of mesenchyme which goes to form ultimately bone and possibly a portion of the dura. Enlargement, 132 diameters.
 
 
Fig.
 
 
102. Photomicrograph of a transverse section of a pig embryo of 18 mm. The embryo was one in which the cerebro-spinal fluid was replaced by the ferrocyanide solution. Subsequently the embryo was fixed in 10 per cent formol containing 1 per cent hydrochloric acid for a few minutes to precipitate the prussianblue. It was then transferred to modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. The granules of prussian-blue are not represented in this figure. Enlargement, 10 diameters.
 
 
Fig.
 
 
103. Photomicrograph of the squared area in fig. 102. The relation of the thinning mesenchyme in the arachnoid areas to the caudal cranial nerves is shown. The granules of prussian-blue, scattered through the area of thin mesenchyme (sas), are not reproduced. Enlargement, 40 diameters.
 
 
Fig. 104. Photomicrograph of a coronal section of a tissue block which includes the meninges and cerebral cortex in the region of the sinus sagittalis superior. The block was obtained from a fetal pig of 80 mm., fixed in Zenker's fluid, and stained, after embedding in celloidin, by Mallory's technique for coimective tissue. Enlargement, 27 diameters.
 
 
Plate XVII.
 
 
Fig.
 
 
105. Photomicrograph of a coronal section of a tissue block including cerebral cortex and meninges in the region
of the sinus sagittalis superior. The block was obtained from a fetal pig of 10 cm., fixed in Zenker's fluid,
and stained by Mallory's technique for connective tissue. Enlargement, 13 diameters. Fig.
 
 
106. Photomicrograph of a coronal section, similar to that in figs. 104 and 105, except in that it was obtained from
a fetal pig of 17 cm. The same technical procedures employed in the other specimens were used in this.
 
 
Enlargement, 27 diameters. Fig. 107. Photomicrograph of a similar section to those of the foregoing figures. The specimen was obtained from a
fetal pig of 20 cm. and was treated in the manner outlined above. Enlargement, 20 diameters. Fig. 108. Drawing of the cell pattern from the inner surface of the dura mater of a fetal pig of .5 cm. The specimen
was prepared by the reduction of a dilute solution of silver nitrate in sunlight. The preparation was
subsequently stained by hematoxylin. Enlargement, 190 diameters. Fig.
 
 
109. Drawing of a preparation, similar to that of fig. 108, but obtained from the inner surface of the dura mater
of a fetal pig of 7.5 mm. Enlargement, 28.5 diameters. FlO. 110. Drawing of a preparation, similar to those of figs. 108 and 109, obtained from the inner surface of the dura
mater of a fetal pig of 90 mm. Enlargement, 28.5 diameters. Fig.
 
 
111. Drawing of a preparation from the inner surface of the dura mater of a fetal pig of 16 cm. The specimen
waa made in the same manner as outlined in fig. 108. Enlargement, 285 diameters. Fig.
 
 
112. Photomicrograph of a sagittal section of a pig embryo of 17 mm. An injection of an 0.5 per cent solution of
nitrate of silver was made into the central canal of the spinal cord; the silver was reduced in sunlight
and the embrjo fixed in formalin. Enlargement, 13 diameters. Fig. 113. Photomicrograph, under higher powers, of the blocked areas in fig. 112. The accumulation of the reduced
silver (p»n) again.st the area membranacea superior is rc|)rcsented in black. Enlargement, 117 diameters. Fig.
 
 
114. Photomicrograph of a transverse section of a pig embryo of 19 mm. An injection of 0.5 per cent solution of
silver nitrate was made into the central canal of this embryo
 
 
and the silver immediately reduced in
sunlight. The embryo waa fixed in formalin, carefully dehydrated, and embedded in xylol-peiraflBn.
 
 
Enlargement, 10 diameters. Fig.
 
 
115. Photomicrograph, under higher magnification, of the blocked area in fig. 114. The collection of reduced
silver {psn) against the cells at the inferior end of the area membranacea superior is illustrated. Enlargement, 100 diameters. Fig.
 
 
116. Photomicrograph of a tranverse section of a pig enibr\-o of 16 mm. The central canal of the spinal cord of
this embryo was injected with a 1 per cent ferrocyanide and citrate solution under mild syringe-pressure;
the enibrj'o was then fixed in 10 per cent formol containing 1 i>er cent hydrochloric acid. Enlargement,
10 diameters. Fig.
 
 
117. Photoniicrogrnph of the blocked area in fig. 116, under higher magnification. The accumulation of the
precipitated injection fluid against the area membranacea superior is represented in black. A slight
extravcntriculai Bprciul of the fluid, which is found in this as in all embryos of this stage, can not be
miule out in the reproduction. Enlargement, 07 di.amuterH.

Latest revision as of 16:45, 5 December 2019

Embryology - 24 Apr 2024    Facebook link Pinterest link Twitter link  Expand to Translate  
Google Translate - select your language from the list shown below (this will open a new external page)

العربية | català | 中文 | 中國傳統的 | français | Deutsche | עִברִית | हिंदी | bahasa Indonesia | italiano | 日本語 | 한국어 | မြန်မာ | Pilipino | Polskie | português | ਪੰਜਾਬੀ ਦੇ | Română | русский | Español | Swahili | Svensk | ไทย | Türkçe | اردو | ייִדיש | Tiếng Việt    These external translations are automated and may not be accurate. (More? About Translations)

Weed LH. The development of the cerebro-spinal spaces in pig and in man. (1917) Contrib. Embryol., Carnegie Inst. Wash., 5, No. 14 .

Online Editor  
Mark Hill.jpg
This historic 1917 paper by Weed describes the development of the cerebro-spinal spaces.



Modern Notes:

Neural Links: ectoderm | neural | neural crest | ventricular | sensory | Stage 22 | gliogenesis | neural fetal | Medicine Lecture - Neural | Lecture - Ectoderm | Lecture - Neural Crest | Lab - Early Neural | neural abnormalities | folic acid | iodine deficiency | Fetal Alcohol Syndrome | neural postnatal | neural examination | Histology | Historic Neural | Category:Neural
Historic Disclaimer - information about historic embryology pages 
Mark Hill.jpg
Pages where the terms "Historic" (textbooks, papers, people, recommendations) appear on this site, and sections within pages where this disclaimer appears, indicate that the content and scientific understanding are specific to the time of publication. This means that while some scientific descriptions are still accurate, the terminology and interpretation of the developmental mechanisms reflect the understanding at the time of original publication and those of the preceding periods, these terms, interpretations and recommendations may not reflect our current scientific understanding.     (More? Embryology History | Historic Embryology Papers)

The Development of the Cerebro-Spinal Spaces in Pig and in Man

Contributions to Embryology


By Lewis H. Weed

Volume V, No. 14

I. Introductory

Probably no field in embryology has been less explored than that relating to the meninges. Our knowledge of the transformation of the perimedulla


ry mesenchyme into the three fully developed membranes about the cerebro-spinal axis has been largely of a crude sort, with gross generalities based on inexact or incomplete evidence. The present work was undertaken in the hope that by a study of the various stages in the development of the cerebro-spinal spaces there might be gained some knowledge which would afford a basis for a conception of this dynamic metamorphosis.


Many of the problems centering around the development of the meningeal spaces have recently been expounded by Cushing^^) . * Not only do we lack knowledge as to the method of differentiation of the primitive mesenchyme, but we know little about the establishment of the circulation of the cerebro-spinal fluid. When do the chorioid plexuses begin to secrete? When does the venous absorption of the fluid take place? When does the perivascular system begin to remove waste products from the cerebral tissue? And also, what factors play a part in the formation of the subarachnoid and subdural spaces? These questions, some of which it is hoped the present study will answer, relate to the field of physiological anatomy. Consideration of the subject, however, serves to convince one that they must be investigated coincidently with the stages of morphological differentiation; for it may readily be conceived that the physiological use of the meningeal spaces may precede any morphological differentiation of the three membranes, nor indeed is it unlikely that one of the active causative factors in the metamorphosis concerns this filling of the mesenchyme about the nervous system with fluid.


This study, therefore, has been anatomical, but with a broader scope than purely morphological studies would have afforded. Not only has it dealt with the morphological differentiations about the nervous system, but throughout the investigation the relationship of these structures to the possible presence of cerebrospinal fluid has been considered. As the problem developed it was projected more and more into the difficult realm of "tissue spaces." Interest in these spaces largely concerned their physiology, but many points of correspondence between structure and function were found.


In some measure this work is a development of an earher study of some of the anatomical and physiological problems of the cerebro-spinal fluid, carried out in the laboratory of Dr. Harvej^ Cushing at the Harvard Medical School.


  • The figures in parentheses refer to the bibliography at the end of this paper.


II. Review of Literature

In order fully to understand the problems which confront one in the study of the embr3'onic cerebro-spinal spaces, a comprehension of the stage to which investigations have brought our knowledge of these fluid-pathways in the adult is necessar}'. It is with this purpose that the adult relationships are here considered. The inclusion of this material may be pardoned, for it will be seen that unanimity of opinion has bj' no means existed in regard to any of the problems concerned in the circulation of the cerebro-spinal fluid.


Modern anatomical knowledge of the meninges dates from the work of Axel Key and Gustav Retzius'29). These Swedish investigators, in their excellent monograph published in 1875, first conclusively demonstrated the anatomical continuity of the spinal and cerebral subarachnoid spaces. But for years after their publication appeared, a physiological continuity between the subdural and subarachnoid spaces was argued for bj' many observers, notably by HilK^*). Gradually, however, workers in this field have reached the opinion that the subarachnoid spaces (the interrupted but continuous channels between arachnoidea and pia) are functionally the channels for the cerebro-spinal fluid. Between the intra-leptomeningeal and the subdural spaces no anatomical connection exists; physiologically there may be some mode of fluid-passage. Thus Hill' 24) states that either by filtration or through actual foramina fluid passes readilj- from one space to the other. Quincke '*^, from observations on animals, somewhat similarly premised a connection between the two spaces, but only in the direction from subdural to subarachnoid. His experiments, based on the results of the injection of cinnabar granules, are open to criticism as indicating a normal passage-way for the fluid; for, as he has recorded, an intense phagocytosis of practically all of his granules occurred. More modern conceptions of the subdural space treat it as a space anatomically closed, lined externally by a polygonal mesothelium. Less error is introduced if it be regarded as analogous in many respects to well-known serous cavities rather than as an essential portion of the pathway for the cerebro-spinal fluid.


The question of the absorption or escape of cerebro-spinal fluid from the subarachnoid space has claimed the attention of many workers. Since the original conception that the meningeal coverings were actually serous cavities, anatomical investigations have furnished many new views. Key and Retzius, by spinal subarachnoid injection of gelatine masses colored with Berlin blue, demonstrated an apparent passage of the injection fluid into the great cerebral venous sinuses through the Pacchionian granulations (die Arachnoidzotten). Their observations were made on a cadaver and the injections carried out under fairly low pressures (about 60 mm. of mercury). A lesser drainage of the fluid into the lymphatics was also shown.


Since the view advanced by Key and Retzius of the absorption of cerebrospinal fluid, the general trend has been away from the idea of an absorption into the venous sinuses. Quincke's observations, made on lower animals after the subarachnoid introduction of cinnabar granules, really offer some substantiation of this theory, but the failure to find the great Pacchionian granulations in infants and in the lower animals caused many workers to reject utterly the conception of the Pacchionian granulations as the functionally active mechanism for the fluid escape.


Physiological evidence, however, advanced by Hill^24) fj-Q^i intraspinous injection of methylene blue, indicated that the major escape of the cerebro-spinal fluid was into the venous sinuses of the dura, while a slow and minor absorption took place along the lymphatic channels. Ziegler(57j^ with potassium ferrocyanide introduced into the cerebro-spinal space, Ukewise found that the venous absorption was much greater and more rapid than the lymphatic. Reiner and Schnitzler with the same agent detected the ferrocyanide in the jugular blood-stream after injection. With oUve oil these investigators found a similar venous absorption, but with a slowing of the venous blood-stream. Lewandowsky also using ferrocyanide. found this salt in the urine within 30 minutes after its subarachnoid injection. Spina 52'^ from observations on freshly killed and hving animals, presented somewhat similar evidence of a major venous and lesser lymphatic absorption. Gushing suggested a valve-Uke mechanism of escape of the fluid, his hypothesis being based on the findings after the introduction of mercurj' into the meningeal spaces.


Several theories concerning the absorption of cerebro-spinal fluid into the bloodvascular system have more recenth' been offered. Mott '**', from a studj' of dilated perivascular and permeuronal spaces, has advanced the idea of fluid-escape by way of the perivascular system into the cerebral capillaries. Dandy and Blackfan^'"^, from an analysis of their evidence, consider that the chief drainage of the fluid is into the capillaries of the pia-arachnoid. Opposed to this conception of a major drainage of cerebro-spinal fluid into the l)lood-vascular system is the view championed by Cathelin'^', that the Ij'mphatic drainage is the chief method of fluid-escape. Cathelin's contention of a veritable circulation of the fluid has not received support from other workers.


Thus it will be seen that since the work of Key and Retzius the trend of opinion has been away from the view that the Pacchionian granulations carry the cerebrospmal fluid into the venous sinuses.


In the earlier investigation carried out in the Harvard Medical School the problems of this fluid absorption were attacked in a somewhat different manner than by previous workers. True solutions of potassium ferrocyanide and ironammonium citrate, such as have been used in the present investigation, were injected into the spinal subarachnoid space under pressures but shghtly above the normal. The animals (dogs, cats, and monkej-s) were kept imder anesthesia during the period of injection, which was usually continued for several hours. Complete filling of the subarachnoid channels was secured by this technique, provided the injections were continued for a sufficient length of time. At the conclusion of the experiment the foreign solution was precipitated in situ and blocks were carried through for histological purposes.


Many of the anatomical findings in this work carried out as outlined are of interest in the present problem. The complete correspondence of the spinal and cerebral subarachnoid spaces as demonstrated by Key and Retzius was amply verified. The normal return of the cerebro-spinal fluid to the general circulation by way of the arachnoidal villi into the great dural sinuses was demonstrated. These viUi are projections of the arachnoidea through the dural wall, prolonged directly beneath the vascular endothelium of the venous sinuses. Furthermore, columns of arachnoid cells were found, normally affording fluid channels in the dura. In addition to the major escape of cerebro-spinal fluid into the sinuses a lesser drainage was alsi) demonstrated, slower than the primary drainage, out along certain of the emergent nerves into the lymjihatic system. No evidence whatsoever was obtained in support of any of the theories of a drainage of cerebro-spinal fluid into either the leptomeningeal or cerebral capillaries, nor could an anatomical valve-like mechanism along the great sagittal sinus be demonstrated. The process of escape of cerebrospinal fluid from the arachnoid villus unto the great sinus appeared to be a simple one of filtration or of diffusion.


Another of the problems concerning cerebro-spinal fluid, which has been of interest to anatomists and phj'siologists, is the source of the fluid. Haller^^D and Magendie'^^ to whom the greatest credit for work on this subject must be given, believed it to be the product of the leptomeninges. Faivre^^^} {^ 1853 and Luschka^*) in 1855 were the first to suggest the chorioid plexuses as the elaborators of this circumambient medium. Since then the view has been generall}- accepted that these villous structures do give origin to the fluid, but the early evidence was based wholly on the glandular character of the plexus. Cappelletti^^) and Pettit and Girardi) offered more definite proof of this relationship by the introduction of pharmacological agents which affected the rate of production of the fluid. These latter authors recorded definite histological changes in the cells of the plexus when influenced by these drugs, indicating, in conjunction with the changed rate of production of the fluid, an undoubted relationship of the chorioid plexus to the fluid elaboration. Since these early investigations many observers — Findlay, Meek(37)^ Mott^'*^', Pellizzi*2), Hworostuchin'26)^ and others — have studied the histology of the chorioid plexus with reference to its function as an elaborator of the cerebro-spinal fluid.


In addition to the elaboration of the fluid by the chorioid plexuses, increments are furnished bj' the nervous tissue itself. This elimination from the nervous system occurs bj' waj' of the j^erivascular spaces. In the previous work referred to'55j it was found possible to inject the entire perivascular system by continuing a physiological injection of the spinal subarachnoid space, and subsequently causing an extreme cerebral anemia. By this procedure an injection of the system to its termination about the cerebral capillaries and nerve-cells could be secured. From this and other evidence the view was advanced that the cerebro-spinal fluid was derived from a dual source — in part from the perivascular system and in greater part from the chorioid plexuses. This view had already been advanced, but on rather insufficient grounds, by Mestrezat'^s* and by Plaut, Rehm, and Schottmuller (, Recently Frazier) has signified his acceptance of this conception of the source of the fluid.


Such, then, is the basis for our present understanding of the meninges, in regard to their characteristic morphology and particularly their functional relationship to the cerebro-spinal fluid. Without a consideration of the circumambient fluid morphological studies of these membranes would be incomplete, for in order to understand the meninges knowledge concerning the cerebro-spinal fluid is necessary.


The Comparative Anatomy of the Meninges

Sterzi's has pubUshed a comprehensive report of the comparative anatomy of the spinal meninges. From his studies he has advanced hypotheses, supported by observations on a Hmited number of fetuses, regarding the development of the human meninges. On account of the interest of this subject in relation to the present discussion a brief summarj' of Sterzi's work will be here included.

e In the acrania there is no special envelope of the central nervous system, but rather a fibrous sheath corresponding to the meninges of higher forms. This fibrous sheath is largely made up of circular fibers, except in the median ventral line, where there occurs a ventral hgament of longitudinal fibers. In cyclostomes, however, there is found a single "primitive meninx" — vascular and composed of white and elastic fibrils coursing in a longitudinal direction. Some of these fibrils traverse the perimeningeal spaces (filled with star-like cells, with some fatty tissue) and are attached to the inner surface of the vertebrae. This same general plan of a single "primitive meninx" is hkewise found in fishes (elasmobranchs, teleosts, etc.); the membrane here is often pigmented and follows closely the external architecture of the spinal cord. The perimeningeal space is filled by mucus in elasmobranchs, but in teleosts this is replaced by fat. For the most part there are found dorsal and ventral ligaments and two lateral ligaments.


The next stage in the development of a more complete form of spinal covering is found in the urodele amphibia. A "primitive meninx," formed of two layers, often artificially separated from each other, replaces the simpler meninx of C5'clostomes and fishes. Of the two layers in this membrane the external is thin and free from pigment; the inner, strongly pigmented, adheres to the spinal cord. The meninx is perforated by the denticulate Ugaments.


In amphibia (Anura) Sterzi found the first evidence of a "secondary meninx," corresponding to the pia-arachnoid. Surrounding this membrane, but separated from it, is the dura, thin and transparent; between the two meninges is the intradural (subdural) space. The dura lies in the peridural space. The spinal prolongations of the endohonphatic canals he in the dorsal part of the peridural space. Both the dura and the "secondarj^ menmx" continue outward along the roots of the spinal nerves and along the filum terminale. Embryologically the perimedullary mesenchyme is differentiated into these two meninges in the Anura.


This arrangement of the two meninges in Anura is followed out in reptiles. The dura, thin as in the amphibia, is covered by endothehum and is vascular. The "secondary meninx" possesses laterally the denticulated hgaments and ventrally the ventral hgament. Both the peridural and intradural spaces are very small.


Likewise in birds Sterzi was able to differentiate only two meninges — the dura and the "secondary meuinx." These membranes are quite similar to those of reptiles. The " secondary meninx " has acquired three layers — an outer endothelial covering, a middle vascular layer, and an inner membrane closely adhering to the cord. This is a distinct approach to the three meninges of mammals. An intradural (subdural) space covered by endothelium can be easily made out. The development of these avian meninges concerns a differentiation of the perimedullary mesenchyme.


The arachnoid, according to Sterzi, first appears as a definite membrane in mammals (marsupials and placentals). In marsupials this arachnoid has become well differentiated and the pia mater possesses denticulated and ventral Ugaments. A transformation of the extradural portion of the denticulated ligaments unites the dura to the endorachis. In perissodactyla the differentiation of the three meninges (particularly of the arachnoid) is incomplete. The arachnoid is separated from the pia mater by a pecuUar tissue wliich contains numerous lymphatic lakes, forming the intra-arachnoid spaces. No intradural (subdural) space is apparent, due to the approximation of dura and arachnoid. The subdural space is clothed by endothelial cells; these can not be made out in the intra-arachnoid spaces. The dura is surrounded by a fatty pad.


According to Sterzi the augmentation of the intra-arachnoid (subarachnoid) space is the distinguishing characteristic of the meninges of carnivora. This increase takes place at the expense of the peridural space.


As Sterzi developed the knowledge of the comparative anatomy of the lower forms — of the transition from the primitive meninx of the cyclostomes to the three membranes of mammals — the possible correlation of this analogj'^ to the embryological development in mammals became apparent. He extended his observations to human beings and to human fetuses. His findings will be detailed in the following section.


Farrar'i, in a short discussion of the development of the meninges of the chick, finds in early stages three laminse about the spinal cord, "the middle one of which alone still presents the primitive features of the mesoblastic-sheath." The inner layer, close to the medullary tissue, is highly vascular; in the outer zone "the connective-tissue elements arc assuming elongated forms and crowding together with long axes parallel, giving a very close mesh with long but extremely narrow spaces, in contradistinction to the loose irregular reticulum of the pia-arachnoid." The outer lamina becomes dura mater, while the inner two zones are considered together as the embryonic pia-arachnoid. Farrar defines the pia-arachnoid as developmentally a single membrane consisting of a loose reticulum, at the outer and inner borders of which limiting membranes are formed.


Literature On The Development Of The Mammalian Meningeal Spaces

The development of the meningeal spaces in mammals has not been studied extensivel}', and the literature in regard to it is quite meager. Only a very few workers have touched upon the subject except casually. Reford"*^', working in the Anatomical Laboratory of the Johns Hopkins University, studied the development of these spaces bj^ the method of injection with india ink. His work unfortunatelj' has never been published, but it has been rather extensively referred to bj' Sabin *3' in 1912 and by Cushing'^^ in 1914. Their sununaries of this work are here included. Miss Sabin thus speaks of it: "In a study of the arachnoid made bj' the injection method in the Anatomical Laboratory of the Johns Hopkins University by L. L. Reford, and as j^et unpublished, it has been shown that the thinning out of the mesenchjine around the central nervous sj-stem is not haphazard, but that injections of the same stage give the same pattern, and that the form of the arachnoid space changes as the brain develops. That is to say, the arachnoid space has as definite a form as the coelom, and it never connects with the IjTnphatics."

Cushing() gives the following summary:

"It was thought that an investigation of the cerebro-spinal spaces in the embryo would most hkely shed Ught on the subject, and some unpublished studies in this direction were undertaken in 1904 and 1905 by Lewis L. Reford in ^Mall's laboratory' in Baltimore. In living pig embryos of various stages low spiaal india-ink injections were made either into the wide central canal or into the subarachnoid space, and the embryos were subsequently cleared. It appeared from the course taken by the injection mass that the full development of the spinal arachnoid preceded that of the intracranial spaces, the impression being gained that the separation of the primitive meninx into its layers occurred later over the cerebral vertex than in the basilar portion of the chamber. Still, I never felt quit« convinced that the failure of injection of the meninges over the surface of the hemispheres in many of Reford's specimens was not due to the floating up of the brain against its envelopes by the introduction of the injection mass from below. Howe\'er this may be, it was nevertheless apparent that a venous injection of the body of the embryo was often produced, and the impression was gained that a communication existed between the basal subarachnoid spaces and the precursors of the sinusoidal veins of the cranial chamber which empty into the jugulars. If due to an artifact from a vascular ruptm-e, at all events the conununication always occurred at the same point. Reford, moreover, in agreement with Cruveilhier, Reichert, and KolUker, came to doubt the existence of the foraminal opening described by Magendie, beUeving that the opening was an artifact and that the fluid escaped by seepage through a persistent membrane."

It is regrettable that Reford's study has not been pubUshed, as it represents the only attempt to solve the problenas of the development of the cerebro-spinal space by the method of injection. As stated in subsequent sections of this communication, his apparent failure to control pressures of injection and to use only granular suspensions is unfortunate.


In a study of the development of the blood-vessels of the human brain, (Mall) noted the ease with which an extravasation into the embryonic arachnoid spaces could be brought about by increasing the pressure in a venous injection. In a specimen of 46 mm. an arterial injection with aqueous prussian-blue resulted in a complete subarachnoid spread, due to rupture of the vessels as they perforated the nervous tissue. In general, it was found that this rule held: an arterial extravasation always took place from the perforating capillaries, while a similar venous rupture occurred in the veins themselves.


Mall made similar observations on living pig embryos from 30 to 80 mm. in length, with analogous results. But when, in these embryos, the arachnoid spaces were completely filled by an intraventricular injection of india ink, no passage of the granular injection into the veins or sinuses occurred. The ventricular injection flowed into the extraventricular spaces "through the medial opening of the fourth ventricle." From the spinal cord the ink extended for a short distance along the main trunks of the spinal nerves. In the larger embryos (above 50 mm.) the ink usually gushed from the mouth, reaching it by way of the Eustachian tube. Using, in the pig embryo, the heart as the mechanism for injecting the ink, extravasation from the cerebral vessels in the arachnoid spaces occurred.


In one human specimen of 90 mm., Mall found both the arachnoid spaces and the cerebral ventricles filled with india ink after an arterial injection of that suspension. He states: "The injection passes through the medial opening into the fourth ventricle (Magendie), and apparently the ventricles are injected through this opening from the arachnoid." To His'25) and to Kolliker belongs the credit of first having established on a firm basis the development of all the meninges in man from mesenchyme. This perimedullay' layer of mesenchyme Salvi'^o^ called the "primitive meninx" — a term now used extensively in comparative anatomy. The primitive meninx divides into two layers, the outer forming the dura and the inner the pia-arachnoid. Sterzi(53), working on the development of the human spinal meninges, advanced a view similar to that of KoUiker. The perimedullary mesenchjTne (the "primitive meninx") divides into two portions, one hugging the inner surfaces of the vertebra? and the other adhering to the cord. This inner layer of the perimedullary mesenchyme, according to Sterzi, should properly be termed the "primitive meninx," as it divides subsequently into dura and the "secondary meninx," which in turn forms both arachnoid and pia. The denticulate hgaments develop in the "primitive meninx." The dura and arachnoid in human embryos are modeled up to a certain point on the cord; then, with the augmentation of the subarachnoid space, they follow the outline of the vertebral canal.


His'25) has given information regarding the development of the meninges, with particular reference to the formation of the subarachnoid space. He affirms the mesenchymal origin of all of the cerebro-spinal membranes. His describes the first differentiation of mesenchyme to form the meninges as consisting of two zones of condensation, the outer being closely associated with the developing perichondrium of the vertebral column and the inner facing upon the cord. Between these two zones of condensation the subarachnoid space develops, posterior and anterior spaces first appearing, with later fusion laterally. These appearances were met with in chicks of 10 to 12 days' incubation. Quite soon after this process of spacedevelopment a separation occurs which gives rise to a complete subarachnoid space. Later the splitting-off of dura from the vertebral periosteum takes place.


III. Methods Of Investigation

In the study of any problem dealing with the development of fluid-spaces within the body, the method of investigation must of necessity be such as to offer exceptional opportunities for control. In the present work several well-known and generally accepted anatomical procedures were naturally suggested, such as injection of the spaces about the central nervous system, reconstruction from serial sections, or merely study of the various stages by means of serial sections.


It was ascertained early in the investigation that by injection and serial sections without reconstruction the necessary stages in the process of meningeal differentiation could be estabUshed. In regard actually to the physiological aspects of the problem more reUance was placed on the results of injection than on any histological differentiation, for, as explained above, considerations of the pathway and of the flow of the cerebro-spinal fluid were deemed most important. No method of injection, however, holds out much promise in such a problem unless it can be applied, under conditions approximating the normal, wnthin the spaces about the nervous system. The greatest objection to reliance upon injections in this problem is in relation to pressures. From the very nature of the case it wall be reahzed that any ordinary injection into the embryonic central canal or perispinal space must result in an extraordinary increase in the normal tension of the fluid. This objection applies to any method employed, whether that of a simple syringe and needle, the glass tube and bulb devised by Knower, or a glass capUlary-tube contrivance.


The erroneous conclusions drawTi by investigators from the emploj^ment of excessive pressures of injection are nowhere more strikingly illustrated than in studies of the circulation of the cerebro-spinal fluid. Many such examples were recently brought forward in a critical review 5' published in connection with a study of the fluid. In the embryo, with structures and membranes still of very little tensile strength, the consequences of a disregard for the pressures of injection are even more disastrous.


A second criterion for the study of fluid-pathways in the body is necessarily the type of injection mass. Not only should attention be paid to the pressures involved, but the peculiarities of the particular body-fluid concerned must be considered. Adopting for this work on the embryo the same standards followed in the previous investigation on the adult, true solutions were used in place of the customary granular suspensions. Emulsions and viscous solutions were not emploj'ed because of their obvious disadvantages in studying the passage through membranes. India ink and process black (in which carbon granules are the particulate matter) were also used, but only for comparison with the standard true solution, as the likelihood of the insoluble granules being phagocyted wathin the period of experimentation or of being caught mechanicalh'^ in tissue meshes appeared a priori to be too great.


In any study of fluid-pathways in the body, not only must the injection fluid be a true solution, but it must also be one which is not attracted to particular cells (as with many stains). Likewise, colloid stains (such as the benzidene group) could not be employed, because of the fact that certain cells (macrophages, as described by Evans’s phagocyte the small colloidal particles. In addition, the true solution must be readily precipitated as an insoluble salt, capable of remainhig unchanged in histological technique. After trying many salts in long-continued injections into the adult cerebro-spmal spaces, it was found that solutions of potassium ferrocyanide and iron-ammonium citrate in equal parts were admirably adaj^ted to the purposes of the experiment. By the addition of a mineral acid (preferably hydrochloric) ferric ferrocyanide could be precipitated. This prussian-blue is insoluble in the routine technique and is readily identified in sections. After mounting in damar or balsam the blue granules can be observed unchanged for several months, but after a year there is some deterioration in the specimen, due to a conversion of the blue into indefinite greens.


Text-figure 1. — .Schematic sketch of mechanism used (or replacing ventricular and spinal fluid of an embno with a foreien solution or suspension. The system is here shown in balance, the difference in fluid-level in reservoirs and needles representing the hydrostatic pressure necessary' to overcome the capillary resistance of tubes and needles. The stands holding the injecting needles may be moved about without altering the balance of the system. As one reserv'oir is raised, the other is lowered in a corresponding degree.


In regard to these two major factors in the employment of injections (pressure and true solution) it was found necessary to devise a method of experimentation which would satisfy the requirements of the problem. Solutions of the ferrocyanide and of the citrate were non-toxic within the central nervous sy.stem and afforded an excellent histological means for following the fluid-pathways. It was hoped at first that a simple "replacement" type of injection might be employed, as in the adult animals. In this procedure a given amount of fluid was withdrawn from the subarachnoid spaces and immediately replaced by an equal quantity of the injection fluid. The method was successfully tried on fetal cats of consideralile size, but was impracticable on small embryos. After such a replacement the animals were allowed to Uve for varying periods of time (up to 3 hours) and then killed.


It was soon ascertained that the essential circulation of the cerebro-spinal fluid was established in pig cml^ryos of less than 30 mm. in crown-rump measurement. Hence the ordinary method of replacement had to be discarded for some more delicate system. With the realization that a simultaneous withdrawal and introduction in a living embryo would be far preferable to a two-stage procedure, the extremely simple apparatus pictured in text-figures 1 and 2 was employed. This device consists of two glass tubes of uniform and like bores, suspended from above by a string running over a pulley. To the tapering lower ends of these reservoirs are attached rubber tubes which connect the reservoirs to two needles. These needles are held at the same level by two metal brackets which can be moved at will on a level glass plate.


Text-figure 2. — Diagrammatic representation of the method of rcplacinR the cerebro-spinal fluid in a living embryo. The spinal needle is inserted into the central canal of the spinal cord, while the cerebral needle is introduced into one of the cerebral ventricles. The canal of the spinal cord and the cerebral ventricles are represented by the interrupted lines. The foreign fluids are introduced by the spinal needle and withdrawn by the cranial.


The apparatus is employed as follows : Both tubular reservoirs are filled up to the point where the fluid is just ready to fall from the needle in a drop. This point is easily obtained by fiUing the reservoirs slightly in excess and allowing this excess fluid to run off from the needle. With the system thus in balance the needles he in the same horizontal plane and can be moved without altering the balance of the solutions. The injection is made by inserting one needle into the central canal of the spinal cord and the other into one of the lateral ventricles; then as the reservoir connecting with the spinal needle is raised the other is lowered, so that an amount of fluid equal to that introduced into the spinal canal is withdrawn from the cerebral ventricles. In this way the whole contents of the cerebral ventricles and central canal of the spinal cord can be slowlj' withdrawn without increasing the pressure in the central nervous sj^stem. The initial pressure necessary to secure this flow is only that required to overcome the capillary resistance of the medullarj'-canal system. In practically all cases this can be accomplished by using a positive pressure of less than 60 mm. of water (associated with a negative pressure of the same degree) .


In the present study the above procedure was the routine method of injection employed. Pig embryos, brought from the abattoir, contained in the uterus, were found to be wholly satisfactory material. If not permitted to cool excessively in transit the embryos lived for at least two hours in a 38° incubator. On being received at the laboratory a section of the uterine wall contaming the placenta was excised, with the embryo left connected by the umbilical structures. As soon as the technical preparations for injection were completed the amnion was opened and the embryo placed upon a padded block at the proper level. The first needle was then inserted into the easily discernible central canal of the spinal cord and the second into the left cerebral ventricle or into the mesencephalic ventricle. By elevation of the reservoir connected with the first needle the cerebro-spmal fluid was replaced by the injection solution. As soon as the replacement was complete the needles were withdrawal and the embryo and its uterine portion replaced in the incubator. The heart of the embryo could be easily obser\^ed in the smaller forms and served as the index of a continued circulation.


The incubation of the embryos was continued for varying periods of time, but it was soon ascertained that a period of over 30 minutes generally resulted in a complete spread of the injection solution. For comparison the period of incubation was lengthened and shortened, but the best results were usually obtained with a 45-minute incubation after the replacement.


Injections of the necessary true solutions were made, in the routine experiment, with a 1 per cent concentration of potassium ferrocyanide and iron-ammonium citrate in distilled water. By a 1 per cent solution is meant a salt concentration of this amount (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm.; water, 100 c.c). The resultant true solution should be practically isotonic with the body-fluids. In this way any injurious consequences due to hypertonic or hypotonic solutions were apparently overcome. The factors of osmosis and diffusion also had to be considered in this connection.


Other concentrations of the so-called "ferrocyanide mixture" were used, but only for the sake of comparison or for the purpose of investigating some particular phase of the problem. The results obtained by the use of these concentrations were not relied upon as affording standards for the normal pathway of the cerebro-spinal fluid.


In addition to the replacement type of injection, many observations were carried out on pig embryos, with a simple syringe-injection of the ferrocyanide solution into the central canal of the spinal cord or into the cerebral ventricles. It proved to be a very simple matter to regulate the pressures by this method, and three arbitrary standards (mild, moderate, and strong) were found to be of value in a comparison of the extent of the spread obtained by replacement and by injection.


The prussian-blue reaction (formation of ferric ferrocyanide) was obtained in these experiments by fixing the whole embryo in an agent containing hydrochloric acid. For histological study the best results were obtained by immersing the specimen from 1 to 10 minutes in a 10 per cent formaldehyde solution containing 1 per cent hydrochloric acid. After this primary procedure, during which the ferrocyanide was precipitated, the embryo was transferred to Bouin's fluid (saturated aqueous picric acid, 75; formaldehyde 40 per cent, 20; glacial acetic acid, 5). The specimens were allowed to fix over night and were then dehydrated in graded alcohols. From 30 per cent alcohol, use was made of 4 per cent changes up to 60 per cent; and from this point to absolute the changes were by 2 per cent gradations.


In addition to the technique outlined above, Carnoy's solution and 10 per cent formol were employed. The Carnoy fluid, containing acid (absolute alcohol, 60; chloroform, 30; glacial acetic acid, 10; hydrochloric acid, 1) proved to be of particular service in the study of specimens cleared by the Spalteholz method; histologically, however, it has not been as valuable as Bouin's fluid.


Besides the ferrocyanide solution, two other injection masses were constantly employed. Solutions of silver nitrate in concentrations of 0.5 per cent were injected into the central canal of the spinal cord and into the cerebral ventricles. This method, with reduction of the silver salt in the sunlight, gives very pleasing preparations. It is, however, subject to obvious limitations. The intraspinous toxicity of the silver, together with its action as a precipitant of albuminous substances, renders its use unsatisfactory in replacement experiments. Furthermore, it reacts apparently with any protein tissue, irrespective of the true function of that tissue (as, for example, its coagulation of the lining ependyma of the ventricles).


India ink, the other substance employed, is of extreme value in anatomical studies. Because of the suspension of carbon granules it possesses the disadvantages already commented upon for the study of any true pathway of fluid. It has been of service, however, in the present work in showing marked differences in spread from that of true solutions and in furnishing information in regard to fluid passage through a membrane.


This investigation has been carried out on the basic idea of correlating the physiological spread of the embryonic cerebro-spinal fluid with the gradual transformation of the perimedullary mesenchyme into the three fully formed meninges. This has necessitated a histological study of the embryo. Pigs for the most part were the animals used, but the findings have all been verified by a study of the same regions in the human embryos in possession of the Carnegie Institution of Washington. In addition, certain structural characters have likewise been identified in sections of chick, rabbit, and cat embryos.


It was early apparent that the material to be of value must be free from any great shrinkage about the central nervous system. Comparative freedom from this artifact was obtained by fixing the embryo alive in Bouin's fluid and dehydrating by 2 and 4 per cent gradations of alcohol. The material was chiefly cut in paraffin after being embedded by means of xjdol.


The methods of investigation outhned in the foregoing paragraphs have been followed throughout the major portion of the work. In many minor instances other procedures not commented upon have been employed ; these wiU be detailed in appropriate subdivisions of this paper.


IV. Injections And Replacements In The Cerebro-Spinal System

Results Of Replacements In The Ventricular System Of True Solutions.


The results of experiments carried out on embryo pigs by the technical procedures outlined in the previous section will be detailed here. The study was made on this animal because of the facility with which it could be obtained living and in good condition and also because it exhibits the characteristic meningeal anatomy of all mammals. The chick could not be used in this investigation on account of the dissimilarity between the avian and the mammalian menmges.


The chief problem concerned here was the actual physiological extent of the cerebro-spinal spaces. This apparently could be ascertained by the replacement of cerebro-spinal fluid by the ferrocyanide mass. But there was also to be considered the passage of fluid from the ventricles out into the periaxial* spaces, corresponding exactly to a similar passage in the adult.


If into the central canal of the spinal cord of a hving pig embryo of 9 mm., crown to rump measurement, an injection of the ferrocyanide solution be made under very mild syringe-pressure, the ventricles can be fairly well filled without rupture of any element. Incubation of this experimental embryo with its circulation continuing almost unabated for an hour should cause a further spread of the fluid throughout the normal canals. If at the end of this time the whole embryo is fixed in an acid medium the ferrocyanide will be precipitated in situ.


Such a specimen, subsequently cleared by the Spalteholz method, is represented in figure l.f In this drawing the spread of the injection solution is clearly shown. Running upward from the point of introduction, wholly within the central canal of the spinal cord, it reaches the bulbar region and extends outward into the large fourth ventricle, appearing as a dense collection of the prussian-blue. Cephalad from this region it spreads in diminishing intensity until it is finally lacking in the diencephalon.


The injected solution, then, in spite of the unavoidable increase in the normal intramedullary pressure, is contained only within the medullary-canal system (central canal of spinal cord and cerebral ventricles). There is no evidence of any spread outwards, either from the third or fourth ventricle.


In the next stage of meningeal development the replacement method can be used, as the embryo is no longer too small for its employment. In figure 2 is represented an embryo of 13 mm., in which the circulation continued for 90 minutes after the replacement. The same general picture shown in figure 1 results. The whole medullary-canal system is filled with the precipitated prussian-blue, which is densest in the region of the fourth ventricle. The roof of the ventricle, however, shows a striking difference from that of the ventricle in the embryo of 9 mm. Just posterior to the cerebellar lip is a regular oval, which is covered from within by a dense collection of prussian-blue granules, causing it to stand out in clear contrast to the thinner and more evenly distributed blue lining of the remainder of the roof. This oval area is comparatively large and comprises a portion of the superior or anterior half of the ventricular roof. This area, differentiated from the remainder of the rhombencephalic roof, is clearly shown in figure 2, a drawing of a cleared specimen of this stage.


  • Throughout this paper the term "periaxial " has been used in the seii90 of "around the central nervous system " or "around the ccrebro-spinal axis."
  • Throughout this work the reference "figure " 1, etc., refers to plate illustrations; the word "text-figure" refers to the illustrations inserted in the text.


With the exception of this strikingly dense area in the rhombic roof, the injection spread in an embryo of 13 mm., subjected to replacement of the cerebro-spinal fluid by the ferrocyanide, differs in no way from that in the embryo of 9 mm. Careful inspection of figure 2 is convincing that the spread still remains within the medullary canals, with no extension of the fluid into the spaces outside of the cerebrospinal axis. It seems justifiable, then, to speak of the cerebro-spinal spaces at this stage of development as being only mtramedullary in type, with no indication as yet of a meningeal fluid cushion (corresponding to the adult subaraclmoid space).


With the use of larger embryos, however, for the medullary replacement with ferrocyanide and citrate, the picture gradually changes. The first indication of a more advanced stage of development is obtained in embryos whose length exceeds 14 mm. Figure 3, of a pig embryo of 14.5 mm., is included here as representing this further extension of the injection fluid. The cerebro-spinal fluid of this specimen was replaced, by the compensating mechanism, by a solution of potassium ferrocyanide and iron-ammonium citrate. The embryo was then kept alive (as judged by the heart-beat) for a period of one hour. At the end of this time it was fixed in an acid medium and subsequently cleared in oil of wintergreen after careful dehydration.


The essential differences between an embryo of this stage and one of the stage represented in figure 2 concerns the spread of the injection fluid from the roof of the fourth ventricle. Both specimens show a complete filling of the intramedullary system (cerebral ventricles and central canal of the spmal cord) with the precipitated prussian-blue granules. The specimen of 13 mm. (fig. 2) is characterized bj' a dense oval collection of the prussian-blue on the upper and inner surface of the rhombic roof. In the specimen shown in figure 3, in contradistinction to this localized aggregation of granular matter, there is a deUcate extension of the injection fluid caudalwards from the roof of the fourth ventricle. This fusiform projection is here readily made out, lying beneath the skin over the ventricular roof and separated quite distinctly from the easily discernible line of the roof. This outward extension of the fluid has a fairly wide and deep origin from the upper portion of the roof, but tapers caudally to a sharp point with considerable rapidity.


At the stage of 14 mm. the roof of the fourth ventricle shows the small depression which marks the formation of the chorioid plexuses. With this depression occurrmg transversely the relation of the external surface of the embryo to the ventricular roof necessarily alters somewhat in this region. The chorioidal depression of the roof graduallj' becomes separated from the skin; and it is into this area between the skin and the ventricular ependyma that the first spread from the cerebral ventricles occurs. At this stage, illustrated in figure 3, the injection is intramedullary in type, with but sUght extension into the pericerebral tissues.


The pericerebral spread may be made out in nearly all replacements in embryos of 14 mm., but in a few cases the injection has remained intramedullary in type. In embryos of 16 mm. the spread into the pericerebral tissues is invariably found. Often, with this extension of the replacement solution outside the ventricles, the oval area noted in the stage of 13 mm. persists. (This phenomenon is especially well shown in a simple injection of silver nitrate, illustrated in figure 11.) The next stage of importance in the development of the cerebro-spinal spaces is represented in figure 4, a drawing of a pig embryo of 18 mm. in which a typical intramedullary replacement of the cerebro-spinal fluid with a solution of potassium ferrocyanide and iron-ammonium citrate had been made. Here, with the exception of the region of the roof of the fourth ventricle, the replaced fluid is contained solely within the central canal of the spinal cord and within the cerebral ventricles. The roof region, however, exhibits a new phenomenon, which distinguishes it from the stage shown in figure 3. The chorioid plexus invagination has become strongly developed, dividing the roof into two parts. These roof divisions have been termed superior and inferior, the former lying anteriorly and orally from the chorioid fold. The general surface outline is but little changed, due to the mesenchyme filling up the area between roof and skin. From two areas in the entire roof of the fourth ventricle the foreign fluid has escaped into the pericerebral tissue. These points of fluid passage he in the two divisions of the ventricular roof. The superior area of escape corresponds to the oval outlined by the prussian-blue in figure 2 and to the point of emergence of fluid shown in figure 3. The lower area of fluid escape is in the inferior half of the ventricular roof, where the ependymal lining and its supporting tissue are developing into a well-marked dorsal distension. This area corresponds to Blake's'3' caudal protrusion, though, as Heuser'23) has pointed out, the shape of the structure in the pig in no way resembles the "finger of a glove." The extraventricular spread of the injection fluid in this specimen is considerably greater than in the pig embryo of 14 mm. (fig. 3). On the whole, however, the distribution of the replaced fluid is not extensive as compared with the adult relationship, where the central nervous axis is entirely surrounded by its subarachnoid cushion of cerebro-spinal fluid. From the superior area of fluid passage the replaced solution (as shown by the resultant precipitation of the prussian-blue) has passed both superiorly and inferiorly. In the median line, and extending laterally but slightly, a projection of the blue may be seen occu])ying a large portion of the extraventricular area formed from the chorioidal invagination. This area of fluid passage occupies at this stage about one-third of the total transverse diameter of the ventricular roof. From it the blue tapers caudally, diminishing in all directions. Above, the precipitate may be made out extending superiorly over the cerebellar lip. Its extension into the pericerebellar tissue is not marked; here again it tapers from the area of fluid passage, its midline prolongation stretching farthest anteriorly. This relationship is easily made out in figure 4, a frank lateral view of such an experimental rei)lacement. The granules which result from the introduced ferrocyanide solution are found only in the central canal of the spinal cord and not in any perispinal arrangement.


In the pig embryo of 18 mm., shown in figure 4, the replaced solution has been carried somewhat farther than in the embryo of 14 mm. (fig. 3). The chief point of differentiation lies in the fact that in the latter stages two areas have apparently become permeable to the intraventricular fluid, so that a larger periaxial spread has resulted. Then, too, the extension of the ferrocyanide solution from the superior area is considerablj^ greater, overlapping the cerebellar Up and filling in some degree the pericerebral tissue in the chorioidal invagination.


With a definite periaxial spread established for the cerebro-spinal fluid in pig embryos of 14 to 18 mm., it seemed not unreasonable to expect a gradual increase in the extent of the future subarachnoid distribution in more advanced stages. The earliest extension of the fluid into the peribulbar tissues occurred with the inception of the infolding of the ventricular roof to form the chorioid plexuses of the fourth ventricle. Its further extension, particularly its passages through a second area, occurred with the greater development of the chorioidal invagination (i. e., 18 mm. stage). A still more extensive pericerebral flow of the ferrocyanide and citrate is illustrated in figure 5. Here the cerebro-spinal fluid in a hving pig embryo of 19 mm. was replaced by the ferrocyanide solution. The embryo was kept alive for about an hour after the replacement and was then fixed in toto in an acid fixing medium, which caused the precipitation of the prussian-blue. On clearing subsequently by the Spalteholz method the spread of the solution was found to be somewhat more extensive than in the stage of 18 mm. (cf. figs. 4 and 5). In figure 5 the whole periaxial area over the roof of the fourth ventricle is shown to be completely filled by a dense aggregation of the prussian-blue granules. The separation of the two areas of fluid passage can not be made out in such a specimen. This dense periaxial extension ahnost completely covers the cerebellar Up, not onlj^ in the medial region but laterally to the limit of the ventricular roof. The injection precipitate lies directly beneath the skin in this area, but more posteriorly its separation from the skin becomes more marked. Tracing this dense periaxial injection posteriorly, it is seen (fig. 5) to end somewhat abruptly in the region of the cephaUc flexure. The Une of termination of the denser mass, to the ventral surface of the medulla, tapers somewhat anteriorly. This extraventricular spread is medial to the otic vesicle, but extends peripheraUy along the caudal cerebral nerves, reaching outward as far as the peripheral gangUa. The periaxial spread also closely covers the ventral surface of the medulla and extends in this plane around the pontine flexure for a short distance upwards along the basilar surface of the mid-brain.


Examined from its dorsal aspects, the superior portion of the spinal cord is found to be covered (in a perispinal relation) by a fine deposit of the prussian-blue. This is shown in figure 5. Caudally from the higher cervical region there is no exidence indicating a further spread in the perispinal tissues. Such a spread from above downward is wholly at variance wdth Reford's^^' conception of a development of the spinal meningeal spaces before the cerebral. The complete filling here of the central canal of the spinal cord and of the cerebral ventricles with the replaced fluid, with no evidence of a periaxial spread except in the region of the fourth ventricle, indicates that in the pig cinbrj'o the adult human relationship between the cerebral ventricles and the subarachnoid spaces endures. There is apparently in this embryo no evidence of the foramina of Bichat and of Mierzejewsky, a findmg in accord with the observations of Dandy and Blackfan'i*.


In the slightly larger embryos the further extension of the embryonic extraventricular spaces progresses rapidly. Figure 6 represents such an extension in a pig embrj-o of 21 mm., in which the normal cerebro-spinal fluid was rei)laced by a dilute solution of potassium ferrocyanide and iron-ammonium citrate. In this specimen the central canal of the spinal cord and the cerebral ventricles are completely filled with the precipitated prussian-blue. But m addition there is almost a total filling of the periaxial spaces. Viewed laterally the densest aggregation of the blue granules is again in the region of the roof of the fourth ventricle. As in the embryo of 19 mm. (fig. 5), the whole extraventricular tissue posterior to this ventricular roof is filled with the granules precipitated from the foreign solution. The spread from this region is similar to that in the previous specimen, except in its far greater extent. The granules may be traced caudalwards in the perispinal spaces to the point of injection. The arrangement of the precipitated material, both withm the central canal of the spinal cord and surrounding it in the perispinal relationship, is well shown in figure 7, a frank dorsal view of the same specimen represented m figure 6. The greater density of the perispinal granules in the upper region of the cord, as contrasted with the granules in the thoracic region, is probably of importance in indicating the direction of the flow from above dowaiwards. The increased amount of the injection fluid in the region about the pomt of insertion of the spinal needle.is in all likelihood due to a local spread from the needle, such as frequently occurs in a very limited area. The phenomenon may, however, be due to an actual increase in the size of the potential perispinal space, though observations upon other embiyos of the same stage of development argue against this view. The segmental outlming of the caudal portion of the perispinal space is to be noted in this figure.


The cephalic regions in the specimen of 21 mm. show a quite extensive spread (fig. 6), and there is the same general distribution of the granules about the medulla, as in the specimen shown in figure 5. The rhombencephalon is completely surrounded by the blue, the ventral sheet inclosing it tightly. Laterally the prussianblue is shown in a dense mass, in intimate relation to the cranial nerves as they join the brain-stem. The cerebellum is practically completely covered by the precipitate; from the ventral portion of the pericerebellar granules the replaced solution (as evidenced by the granules of prussian-blue) spreads forward and surrounds a portion of the mid-brain. Only the ventral surface of the posterior half of the mid-brain is circumscribed by the granules; anteriorly it is wholly surrounded by the i)eriaxial injection; more anteriorly tlic extension is iimited to the mesial structures, leaving unsurrounded the cerebral hemispheres, althougli creejnng between the hemispheres and the mid-brain.


The pecuhar avoidance by the replacement fluid of the extreme dorsal half of the mid-brain is also to be made out in the dorsal view of the specimen (fig. 7).


The two lateral extensions from the ventral sheet of the injection granules approach on either side this mesencephalic eminence. The peculiar appearance of the injection spread caused by the chorioidal invagination of the roof of the fourth ventricle is also here illustrated.


In this specimen, then, of a pig embryo 21 mm. the periaxial spread is almost complete, the only areas not entirely surrounded being the aiiterior mesencephalon and the cerebral hemispheres. In an embryo but a few milhmeters larger this periaxial exten.sion of the solution is complete. The mesencephalon first becomes entireW covered by the jjrussian-blue precipitate, with later extension over the hemispheres. This complete periaxial injection occurs usually in replacements in embryos varying in length from 24 to 28 mm.


A specimen exhibiting a complete extension of the replaced solution around the central nervous system is shown in figure 8. This specimen was prepared by replacing the cerebro-spinal fluid in a Uving embryo of 26 mm. and then keeping the embryo alive for an hour. After fixation in an acid medium, dehydration, and clearing, the uijection was found to occupy the whole medullarj-canal system and also to surround completely the cerebro-spinal axis, as shown in the lateral view. The striking features of this stage are similar to those observed in the younger specimens — the dense accumulation of granular material in the region of the roof of the fourth ventricle, the surrounding of the central portion of the caudal cranial nerves, and the thin pericerebral covering by the replacement mass. In addition the specimen exhibits in the thoracic region an extension of the granular material laterally along each spinal nerve. An observation of this peculiarity reveals the prussian-blue extending outwards only as far as the ganglia on the posterior roots.


The relationships, then, observed in an embryo pig of 26 mm. are those which exist in the adult; the cerebro-spinal axis contains cerebro-spinal fluid within its cerebral ventricles and within the central canal of the spinal cord, while in turn it is cornpletely surrounded by cerebro-spinal fluid within the subarachnoid space. Communication between the ventricles or intra-medullary sj'stem and the perispinal spaces occurs only in the region of the fourth ventricle. Here again the adult human relationship holds. The evidence, therefore, from a study of the fluid spread in a replacement experiment with the use of true solutions, indicates that in pig embryos of about 26 ram. an adult distribution of cerebro-spmal fluid occurs.


The Results of Injections of True Solutions

In the preceding section there have been detailed the results of experiments on living pig embrj-os in which the cerebro-spinal fluid of both the central canal of the spmal cord and the cerebral ventricles has been replaced by a dilute solution of potassium ferrocyanide and iron-ammonium citrate. After the replacement, carried out so as to avoid any increase in the normal tension, the embryos were incubated for varj^ing periods of time so that the normal current of the fluid might cause an extension of the loreign solution. In the experiments which will be recorded in this section the same true solution was injected from an ordinary syringe and the salts immediately precipitated as prussian-blue. The purpose of these observations was solely to ascertain the effect of injections at pressures above the normal tension, so that the conclusions drawn from the replacement method might be more fully substantiated.


It was soon ascertained that the pressures caused by injections with a simple syringe could be fairly well controlled and that several degrees of tension might be employed. Thus it was found to be simple and serviceable to designate the injections as those made with mild, moderate, or strong syringe-pressure. Most of these injections were made into the central canal of the spinal cord, but occasionally into the perispinal spaces or cerebral ventricles. Injections under equivalent pressures in the central canal of the spinal cord or into the cerebral ventricles always gave corresponding results. It is necessary to record that the injections, even under strong pressure, were not carried to the point of macroscopic rupture.


The so-called mild syrmge-pressure, making use of solutions of potassium ferrocyanide and iron-ammonium citrate, resulted in extensions of the prussian-blue wholly similar to those obtained in the replacement experiments which were carried on for 30 minutes and over. This similarity indicates a complete filhng of the available cerebro-spinal system in the replacement method, for certainly (even in the mildest syringe injections) the intraventricular pressure must be excessively increased. Figure 1 shows a specimen under such conditions, with a marked thinning of the injection mass in the region of the fore-brain. This finding is customarily present in the injections under mild pressure, due to the pushing upwards of an existent ventricular fluid.


"When moderate pressures are employed with the syringe the picture gradually changes. The essential difference in the results obtained by moderate syringe injection and by the replacement method lies in the greater extension of the foreign solution in the smaller embryos. Thus in figure 9 the spread of the injection precipitate in a pig embryo 16 mm. is shown to be about as extensive as that obtained by the replacement method in an embryo of 19 mm. (fig. 5). The extra ventricular distribution of the injected solution around the medulla, the extension (even more marked here) along the central roots of the caudal cranial nerves, and the localized perispinal spread are easily made out in this specimen of 16 mm.


This general rule applies to all of the results obtained with the use of syringepressures above the mildest. Dependent upon the degree of syringe-tension, the spread extends in simple ratio. Thus, by the use of moderate pressures of injection into the central canal of the spinal cord, a complete intramedullary and periaxial spread was secured in a pig embryo of 22 mm. somewhat earlier than the equivalent stage was obtained by the use of the replacement method.


With the highest syringe-pressures (insufficient, however, to cause macroscopic rupture) the same general type of injoctiou spread was obtained, bringing the more complete stages down into smaller and smaller embryos. Most of these embryos, however, on microscopic section showed obvious rupture of some part of the central nervous system.


The most important feature of these findings in the embryo pig injected with true sohitions under moderate pressures from a syringe concerns the fact that the extension of the injection coincides, except as to the size of the embryo, in every instance with that obtained by the replacement method. Thus similar and analogous spaces are filled by injections under syringe-pressures in small embryos and by the solution under normal tension in larger embryos. It must be assumed, then, that the pressure of injection is sufficient to dilate potential cerebro-spinal spaces which normally would not be concerned in the pathway taken bj' the cerebro-spinal fluid. No evidence of new or abnormal pathways for the fluid is afforded by the observations made with the increased pressure; these phenomena indicate great potential strength in the tissues which limit the immature cerebro-spinal spaces.


Injections with a simple syringe may he made with such a degree of pressure that gross rupture of the tissues becomes apparent. In such an injection into the central canal of the spinal cord the infundibular region ordinarily ruptures in the smaller embryos (under 15 mm.), while in larger embryos rupture usually occurs into the subcutaneous tissues of the back of the neck over the fourth ventricle.


In discussing the effects of the introduction of solutions of ferrocyanide under pressures higher than normal into the central canal of the spinal cord, it may be appropriate to record observations made in the attempt to inject the cerebro-spinal spaces from the perispinal space. In embryos under 15 mm. in length it is quite difficult to make a perispinal injection. As the embryos exceed this measurement the injection becomes increasinglj' easy, but not until a length of 20 mm. is attained can it be made under the mild pressure advisable. These observations tend to substantiate the findings alreadj' recorded in both the intramedullary replacements and the injections under mild pressure.


Results Of Injections Of Nitrate Of Silver

In a number of experiments a dilute solution (0.5 per cent) of nitrate of silver was injected into the central canal of the spinal cord and the salt then reduced in the sunhght. This solution, although a true one, is wholh- unsuited for the replacement type of injection, on account of its great toxicity and its power to coagulate protein. It was employed here onh' for the simple type of injection.


The results obtained by this intraspinous injection of solutions of nitrate of silver were of but little value in the determination of a pathway for the cerebrospinal fluid, but they vividly present certain aspects of the problem. Thus, in figure 11, a drawing of a specimen (pig) of 16 mm., the area through which fluid passes in the superior portion of the roof of the fourth ventricle is clearly outlined by a denser deposition of the silver. This specimen was prepared by introducing the solution of nitrate of silver into the central canal of the spinal cord under the so-called moderate sj-ringe-pressure. The drawing shows a shght, cone-shaped extraventricular spread of the injection fluid. This spread takes place solely from the superior area of fluid passage, a result in accord with the finding that the solution of potassium ferrocyanide and iron-ammonium citrate passed first through the superior area. Of course it is realized that the precipitant action of the silver may have exerted a more potent action on the structures constituting the lower area of fluid passage.


Another interesting phenomenon of the injections of silver nitrate is shown in figure 12. The embryo of 13 mm. here represented was injected under strong syringe-pressure with a sohition of silver nitrate into the central canal of the spinal cord. On subsequent reduction and clearing it was found that the excessive pressure had resulted in a comjilete intramedullary injection with a localized pedunculate spread into the tissues from the roof of the fourth ventricle. This bulbous extravasation into the extravcntricular tissue has not been observed in any specimens except those into which the solution of silver nitrate was injected. Such a spread is probably to be accounted for by an immediate coagulation of the surrounding tissue.


The extensive use of solutions of sUver nitrate as a means of demonstrating vascular channels naturally suggests a careful comparison of the results obtained from its use and those obtained from the employment of other available true solutions, in regard to the evidence afforded by the two methods of intraspinous injections. The chief objection to the use of silver nitrate, as has aheady been mentioned, is its power to coagulate protein. This is illustrated by many features of the specimen shown in figure 11 — by the sharp outhning of the area of fluid passage, the markings on the caudal process of the fourth ventricle, and the delimitation of the cerebellar hp. But much more marked are the evidences of this coagulative power as shown in figure 12, the pedunculated extraventricular spread, the transverse corrugation of the cerebellar hp (amountmg to circumscribed indentations), and the peculiar outlining of the roof attachment to the bulb. These phenomena obtained by the intraspinous injection of solutions of silver nitrate must be classed as artifacts. The different degrees of this corrosive action of the silver probably result from the varj^ing rates of reduction of the salt to the metal, a factor which is not easily regulated. The findings, therefore, with this method are worthless unless controlled.


Many embryos of varying sizes were injected with the silver nitrate. In the main these observations followed the course of development of the cerebro-spinal spaces as evidenced by the replacement experiments with the ferroc3'^anide. The injections required moderate pressure in the syringe in order to secure more than a local extension from the roof of the fourth ventricle, and to secure the same extent of spread it was generally necessary to use embryos a few millimeters larger than those required in the replacement experiments; but this is to be expected, m view of the i^robability of a constant precipitation of the albuminous tissues by the injection fluid.


Specimens prepared by the intraspinous injection of silver nitrate, then, afford but little reliable evidence in this j)roblem except of a substantiative sort. The findings by this method indicate that the perispinal and pericerebral spaces, in pig embryos of 25 mm. and upward, could be filled by an injection of silver nitrate under moderate pressures into the central canal of the spinal cord. The point of passage of the fluid from the intramedullary to the periaxial system was in the region of the roof of the fourth ventricle.


The Injection Of India Ink

The objections to the use of any fluid of insoluble particles in suspension have already been discussed in considering the methods of injection which were possible for use in this study; but for comjiarisou Avith results obtained by more promising methods and to ascertain to what extent injections with India ink are reliable they will be further considered here. No granular substance other than India ink (carbon granules) was employed in this investigation. In every way this suspension possesses advantages over other possible masses — in its ease of preparation, in the small size of the granules, and the insolubihty in the reagents used for microscopic technique.


Suspensions of nidia ink (diluted from 4 to 10 times) were introduced first into the medullary-canal system of living pig embryos by the replacement method. In no case, however, even though the circulation of the embryo may have continued for 90 minutes, was there any evidence of an extension of the replaced mass outward into the periaxial spaces. The carbon granules remained wholly within the ventricles, a striking difference from the results obtained by the ferrocyanide replacements. It would appear, then, without the further evidence afforded by microscopic section of the specimens, that there is an existing mechanism which prevents the passage of the carbon granules from the fourth ventricle into the periaxial spaces. This finding was found to be constant ui all the living embryos subjected to the cerebro-spinal replacement.


Quite similar to these results by the replacement method are those from the injection of a suspension of india ink under mild syringe-pressure. In no instance, provided the pressure was maintained at a low enough degree, was there any passage of the granular material into the periaxial tissue. In embryos of over 30 mm., however, even with the lowest pressure, it becomes increasingly difficult to prevent a sudden spread into the periaxial spaces. The type of spread indicates a sudden release of some restraining agent and suggests a rupture of a membrane. This spread is usually local and takes place from the roof of the fourth ventricle.


With moderate and strong syringe-pressures, however, it is possible to secure a periaxial spread, but this is quite different from the distribution of the uijections by the use of ferrocyanide solutions. Figure 10 illustrates a specimen of a pig embryo of 21 mm. into whose central spinal canal india ink was injected under strong syringe-pressure. The resultant spread of the injection is easily discerned; the cerebral ventricles are quite filled with the carbon, while from the superior portion of the roof of the fourth ventricle a dense but localized periaxial spread is made out. This extraventricular extension of the ink is well defined; it stretches caudalwards for a slight distance, curving about the bulbous caudal portion of the ventricle and extending lateralwards but a short distance. The median portion of the cerebellar lip is covered by the granules. E\adeuces of the excessive pressure at which the injection was made are shown by the lines of mvasion of the spinal cord and mid-brain. A comparison of the spread of this injection mass with the extension of a ferrocyanide replacement in an embryo of the same size (21 mm.) is afforded by figures 10 and 6. With such a divergence in the results obtained by the two methods of approach it is not surprising that observations such as Reford's) fail to coincide with these findings. The unsuitabiHty of suspensions of granular material in the investigation of the cerebro-spinal sjiaces has been many times verified in this work.


In the further study of the course of the spread with injections of india mk it was found that, in pig embryos of approximately 22 mm. and over, a partial periaxial injection could be secured by plunging the syringe-needle into the perispinal spaces. The carbon granules could subsequently be seen filling the perispinal spaces and also mounting upwards in jjartial pericerebral relationships, particularly around the medulla. This result was obtained by the use of strong syringe-pressures. Apparently the resistance to the spread of the ink in injections or replacements in the medullary-canal system occurs in the passage of the fluid from the roof of the fourth ventricle into the periaxial spaces. So far as is known, Reford^"*^) did not control his injection pressures. These results with the injection of india mk under strong pressures coincide with the idea of his observations afforded by the abstracts given by 8abin(*3) and Cushing^^'. Suspensions of india ink. then, injected under mild syringe-pressure or by the replacement method, offer no evidence, in the pig embryo, of a passage of the cerebro-spinal fluid into the j^eriaxial spaces. Only by employing pressures much above the normal tension can such evidence be obtained.

V. Undescribed Structures in Roof of the Fourth Ventricle

The results of the replacement of the existing cerebro-spinal fluid by a true solution of potassium ferrocyanide and iron-ammonium citrate in a living pig embryo indicated, as detailed in the foregoing section, that the fluid passed from the ventricular system into the periaxial tissues in the region of the roof of the fourth ventricle. This important transit of the fluid, agreeing with the established conception of the relationship in the adult, was first observed in an embryo pig of 14 mm. (fig. 3). At this stage the exudation of the replaced fluid occurred in one defined area, seemingly corresponding to the dense oval in a smaller embryo shown in figure 2.


Such a passage of fluid from ventricle to i)eriaxial tissue is necessarily a ])hysiological phenomenon, and it was in the hope of finding an anatomic basis for this phenomenon that the roof of the fourth ventricle was studied histologically. It was reahzed that failure to demonstrate anatomically differentiated structures would not vitiate the physiological observations, but that a correspondence between function and structure was most desirable. Hence observations were undertaken to determine, if pn.ssible, an area of histological differentiation in the roof of the fourth ventricle which might be concerned in the primary passage of fluid from the cerebral ventricles into the periaxial tissues. The investigation concerned first the examination of this region in pig embryos of 14 to 15 mm., at which stages the fluid passes from a single area. Subsequently, similar studies were undertaken in regard to the second, more inferior area (shown in figure 4). The results of these studies will be given here.


AN UNDESCRIBED AREA IN THE SUPERIOR PORTION OF THE ROOF OF THE FOURTH VENTRICLE.


The Area Membranacea Superior In The Pig Embryo

Examination of the roof of the fourth ventricle in a pig embryo of 14 mm. revealed a peculiarly differentiated area in the superior portion. The general topography of this area is show7i in the rectangular area marked off in figure 32 — a median sagittal section from a pig embryo of this critical stage. In figure 33 this rectangular area is enlarged to show the morphologA' in greater detail.


In this figure the densely staining ependj'ma lining the fourth ventricle approaches from both sides. The superior portion of the ependyma ends abruptly, while the inferior Hne of the layer tapor.s more slowly. Between these two jjoints is an area having none of the characteristics of the ventricular lining at all other points. The comparatively smooth contour of the ependymal cells is replaced by an irregular cell-border. The pyknotic nuclei of the cells have been replaced by less densely staining, elongated, spindle-like nuclear bodies. The cell-layer lining the ventricle is here really only of a single cell in thickness, although blood-capillaries closely applied to it suggest a greater thickness. The mesenchj-nie lietween this layer and the peripheral epidermis is quite thin, but resembles in everj' way the mesenchyme in the immediate neighborhood.


There is, therefore, as pictured hi figiues 32 and 33, an area in the roof of the fourth ventricle which is morphologically dissimilar to the characteristic ependyma lining the cavity. Is this the result of some distortion in fixation or in the routine histological technique? Is it a constant finding and, if so, what is its historj'? Does it arise at a definite period and persist throughout intra-uterine life onlj' or through adult life also?

The question of the actual existence of this area, or of its being caused by technical manipulations, is one which must be answered. That this differentiated portion of the roof of the fourth ventricle is not an artifact is verified bj* the general history of its formation, by its invariable occurrence (not onh' in the pig but in other animals), and by its general histological appearance. Moreover, the physiological importance of this area undoubtedh' incUnes one completely from the possible explanation that it is due to an artifact. Xo single finding wholly excludes such a possibility; rather is one convinced, by many features, of its actual occurrence.


Considering the fact, then, that this differentiated structure in the roof of the fourth ventricle may be found in all embryo pigs at the stage of 14 mm., it becomes necessary to ascertain at what time in the development of the embryo it first appears and how it is formed. Obviously the most satisfactory' method is to trace the area through the lower stages and also through the older embryos. For the sake of greater clearness, however, a description of the area \s-ili be given from its first differentiation through its maximum transformation to its final disappearance — for the structure is only temporary.


In pig embryos of 8 mm. and less in crown-rump measurement, the roof of the fourth ventricle is fcfrmed of cells morphologically and tinctorially different from those lining other parts of the ventricular cavities. These cells are quite unlike the deeply staining ependjTiial cells, which can be so readily identified as the lining cells in older embryos. In this yoiuiger stage of 8 mm., the entire ventricular roof is composed of several layers of cells with round or somewhat oval nuclei and fairly abundant cytoplasm. The cell-boundaries are not well defined. The nuclei are not deeplj' tinged with hematoxylin. The chromatin material is sparse and irregularly distributed. Nucleoli are prominent. The cytoplasmic border hning the ventricular cavity is rough and ragged at times, often blending with the coaguUvted albumen of the cerebro-spinal fluid. Altogether, these lining cells bear a much greater resemblance to the epithelial cells than to the ependymal.


These characteristics of the lining cells of the roof of the fourth ventricle are shown in figures 24 and 25, from a pig embryo of 8 mm. The close association of the roof cells to the surface epithelium is easily made out in figure 25, as well as the general character of the lining cells.


At the stages of 8 mm. and under, in the pig embryo, the roof of the fourth ventricle is relatively quite large. In its whole extent it is formed of the peculiar lining cells described above. With the growth of the embryonic nervous sj^stem, the roof of the fourth ventricle is subjected to alterations in form and position; to some extent these changes influence the cells which line the cavity in the early stages.


In pig embryos between 8 and 12 mm. in length the roof of the fourth ventricle undergoes a change. The ependyma, which from comparison with later stages is regarded as typical, begins to encroach upon the epithelial-like cells which are so numerous in the 8 mm. stage (fig. 25). The area occupied by these cells diminishes, not only relatively but absolutely. It becomes smaller and the cells gradually change their character. These changes are shown in figures 26 and 27, from a pig embryo of 11 mm. Figure 26 gives the location, in a sagittal section near the midline of the area in figure 27, taken at a higher magnification.


In figure 27 the densely staining lips of ependymal and nerve cells are seen approaching each other. For a considerable space in the central portion of the photograph there is an area similar to that shown in figure 33. But considered in connection with figure 25 this area represents the epithelial-like cells of the roof of the fourth ventricle. This relationship is more clearly shown in figures 28 and 29, taken in a more lateral plane from the same embryo (11 mm.). Examination, however, of th(; area in figure 29 shows the epithelial-like cells again a])i)arent in the roof of the fourth ventricle.


The i^rocess of transformation, then, as shown in these photographs from an embryo i)ig of 11 mm., concerns a gradual encroachment upon the area of epithelial-like cells by the more densely staining and more closely packed ependymal cells. Gradually the epithelial-like cells in the central portion of the area lose their former character (fig. 27), while around the periphery, especially on the lateral sides, the epith('lial-lik(! ajjpearance persists (fig. 29).


On the lateral side of this area, just as the tyincal ependymal lining is al)out to become isolated (fig. 29), the epithelial-like lining cells form a several-celled layer.


The nuclei are poor in chromatin material and the cytoplasm somewhat small in amount. The inner cytoplasmic border hning the ventricle is in contrast, by its ragged outline, with adjacent smoother ependj^ma on both sides. At this stage of the pig embryo the characteristics of the epithehal-like cells are still to be made out, but a gradual transformation is becoming evident.


The metamorphosis becomes much more marked in the central portion of the area, as shown in figures 26 and 27. In these figures the whole central area seems to have lost some of its former character as an intact cell-laj'er. Closer examination, however, under higher power demonstrates that it still possesses an intact surface as a hning for the ventricle. Delicate cytoplasmic strands stretch in a continuous line across the whole area between the Ups of denser tj-pical ependjina. The nuclei in this difi'erentiated area are seeminglj' altered from their rounded form and have elongated almost into spindles. The inner cytoplasmic border is characteristically rough, with, small amounts of coagulated albumen adhering to the processes. The area, then, in its central portion, at the stage of 11 mm., has assumed the character of the stage of 14 mm. (fig. 32). On the periphery, however, the cells stiU resemble those of smaller stages (8 mm.).


From the pictures presented by the intermediate stages (figs. 27, 28, and 29) the differentiation goes on verj' rapidly, so that in the pig embryo of 13 mm. there is rarely any evidence of the epithelial-like cells. Figures 30 and 31 are photomicrographs of a sagittal section of an embryo pig of 13 mm.; here there are no evidences of the epitheUal-like cells. The whole area, pictured in figure 31 as sharply delimited from the tongues of tjijical ependyma above and below, has become well differentiated. The cell-character observed in figures 27 and 33 (elongated nuclei and scanty strands of protoplasm) has become very obvious. The ragged and roughened intraventricular border, the coagulated albumen, and the abrupt transition from the neighboring tj-pical ependyma are well shown in the photomicrographs of this specimen.


The differentiation of this area in the roof of the fourth ventricle of the pig embryo proceeds at a very rapid rate, so that within the growth of a few millimeters (from 8 to 13 or 14) a great histological change occurs. Figures 32 and 33, already described, show the extent of this metamorphosis in a pig embrj^o of 14 mm. The process, however, continues, modified possibh' bj- the changing of the roof of the fourth ventricle. For this roof structure is subjected to marked alteration in stages of 14 mm. and upwards, both by the lateral development of the chorioid ple.xuses and b}^ the readjustment of the cervical and pontine flexures. Its maximal differentiation may be said to appear at a stage of 18 mm.; this is maintained through several millimeters, until undergoing final retrogression.


This maximal change in the roof of the fourth ventricle is shown in figures 34, 35, 36, and 37. Several points of interest are brought out in these photomicrographs. Figure 35 represents an enlargement of the rectangular area in figure 34, taken from transverse sections of an embryo pig of 18 mm. The area is particularly well shown in this figure, in which, from the right, the typical ependyma, in a fairly smooth single-cell layer, approaches the differentiated cells in the central portion. On the left, too, similar typical ependyma is shown. In the central area, which has been repeatedly described, the elongated nuclei, the strands of protoplasm, and the ragged, iiTegular intraventricular surface are well presented. The photomicrograph has been reproduced to show the relation of this differentiated area to the various blood-channels in the supporting mesenchyme. Apparently the whole ventricular roof is, at this stage, a site for an extensive capillary plexus; from both sides, as shown in figure 35, vessels (one of great caliber) approach the central area of differentiation. Directly beneath this area smaller capillary channels can be made out, from which, apparently, a sUght extravasation of red blood-cells has occurred. Here, as in the greater part of the basilar pericerebral region, extravasation of the blood-cells is very frequent. This phenomenon has already been pointed out by INIaU'^"".


The large extent and the great differentiation of this pecuUar area in the roof of the fourth ventricle are well shown in figures 36 and 37, taken from a transverse section of a pig embryo of 18 mm. In the photomicrograph of higher magnification the two sharp tongues of typical ependyma are quite striking. Their abrupt termination in the wide, differentiated area has nowhere been more convincingly shown. The resemblance of these lining cells in the central area to the mesenchymal elements adjoining is here also seen. The most interesting of all the phenomena exhibited in this reproduction, however, is the attachment, apparently by precipitation, of the coagulated albumen of the cerebro-spinal fluid. This coagulation, in this specimen, deUmits the differentiated area in the roof of the fourth ventricle. The phenomenon is seemingly only an ampUfication of a similar attachment of small fragments of the albuminous precipitate shown in other figures.


Beyond the stage of 18 mm., which may be termed the maximal stage, the differentiated area in the roof of the fourth ventricle undergoes a regression. This is apparently due to the morphological alterations in this rhombic roof. The chorioid plexuses in embryos over 18 mm. long deeply invaginate the fourth ventricle, possibly drawing some of the true roof with them, but surely encroaching upon the mid-hne with their lateral tuftings. This growth tends to decrease the available extent of the differentiated area, but an even more potent factor is the rapid development of the cerebellum. The caudal growth of the cerebellar lip soon largely occupies or replaces the superior half of the roof. These two factors, the cerebellar growth and the enlargement of the chorioid plexuses, render the persistence of the differentiated area impossible, so that a regression or disappearance is to be expected.


With these considerations before us, the study of sectioned pig embryos of a greater length than 18 mm. becomes important. The process of disappearance, however, does not occur at once. Thus, in an embryo pig of 19 mm. (figs. 42 and 43) the differentiated area is as large and as characteristic as in the stage of 18 mm. This same aijpoarance and maintenance of size may be observed through the next several millimeters' growth, but in pig embryos of 23 mm. the chorioid plexus has usually developed to such an extent that a continuation of the former size becomes impossible. This is shown in figures 44 and 45. Figure 45, the enlarged squared area from figure 44, is a photomicrograph from a pig embryo of 23 mm. The differentiated area, duo to the factors favoring its regression, now appears in close proximit}^ to the chorioid plexus. It has more the appearance of a degenerating area at this stage than in any of the younger embryos, but it still shows a characteristic delimitation of both edges — on the one from the typical ventricular ependyma, and on the other from the differentiated ependj'ma of the chorioid plexus. The cytoplasmic strands of the area which forms the ventricular border do not show to advantage in the photomicrograph, but the .same ragged character with the covering of coagulum may be made out. The process of regression, mechanical as it perhaps is, has begun at this stage in the pig, and in the course of the next few millimeters' growth will become even more active.


With the encroaclmaent of the chorioid plexuses and the downward growth of the cerebellar lip, the superior portion of the ventricular roof soon disappears, and is practically non-existent in embryos of 30 mm. and more in length. The differentiated area thus encroached upon from the sides and above becomes a mere vestige of its former size. Thus in a pig embryo of 32 mm. (figs. 46 and 47) it appears as a very small break in the lining continuity of the ventricular ependyma. Without the intermediate stages such a picture would undoubtedly be considered as an artificial erosion of the ependjinal lining of the ventricle, but when studied in connection with figure 45 the true vestigial character of the area becomes established.


The final fate of this differentiated area in the roof of the fourth ventricle is a complete disappearance, with the occupation of the region by chorioidal epithelium and cerebellum. In this study it was impossible to find traces of the differentiated areas in pig embryos of over 33 mm. in length; vestiges maj' persist, but so small as to present difficulties of decision. The persistence of such a differentiated vestige in rare instances would not be surprising; the transitory character of the area and the method of disappearance make this seem not unlikely.


This transitory area of differentiation in the roof of the fourth ventricle of the pig has not, so far as can be determined, been noted or described bj- any p^e^•ious author. His'25\ in a retouched photomicrograph of a sagittal section of a human embryo of 17 mm., reproduced the area as differentiated from the roof, but he has made no comment upon it. I have called this differentiated area in the superior portion of the rhombic roof ventricle the "area membranacoa superior ventriculi quarti." This terminology is based on the anatomical character of the area as a continuous membrane, but chiefly on its physiological significance. For, as will be shown in the succeeding section of this paper, the transit of embryonic cerebrospinal fluid from ventricle to periaxial tissue occurs in this area, which functions apparenth- as a physiological membrane. With such a physiological conception of the area, the ttim "area membranacea" seems most suitable, inasmuch as it also meets the anatomical requirements.


The Area Membranacea Superior in the Human Embryo

The finding of the diflferentiated area in the superior portion of the roof of the fourth ventricle in the embryo pig suggested the value of a study of the same region in the human embryo in the further solution of the problems underlying its occurrence. Hence this region in the roof of the fourth ventricle has been examined in the sectioned human embryos of the Department of Embryology of the Carnegie Institution of Washington. It was found that a similar area occurred in the human embryo of approximately the same age.


The study of the roof of the fourth ventricle is usually more difficult in the human embrj^o than in the pig. This is due to the fact that the roof of the fourth ventricle quickly suffers from poor fixation and dehydration — collapse or inversion of the whole structure being commonly met with. It is rarely possible, in the younger embryos, to secure the most satisfactory fixation, whereas in the pig these factors may be controlled as desired. Furthermore, the undue pressures to which the human OAaim is frequently subjected in abortion may cause crushing of the more delicate parts of the nervous sj^stem.


It is probably best, in the human embryo as in the pig, to trace the formation of the area membranacea superior ventriculi quarti from its beginning, through the various differentiations.


In a human embryo of 4 mm. (No. 836 of the Collection of the Carnegie Institution of Washington) the entire roof of the fourth ventricle is composed of cells with round or slightly oval nuclei and palely staining cytoplasm. The nuclei of the cells are poor in chromatin material as contrasted with the pyknotic character of the typical ependymal cells. The lining tissue is of the thickness of several cells. The ventricular cytoplasmic border is fairly smooth at this stage. This characteristic ventricular lining is shown in figures 40 and 41, both taken from embryo No. 836. The whole picture is similar to that exhibited by the pig embryo of 8 mm. (figs. 24 and 25).


A similar accumulation of epithelial-like cells is found in a human embryo of 7 mm. (No. 617 of the Carnegie collection). This is pictured in figures 48 and 49. The photomicrograph of higher magnification shows these poorly staining cells heaped up in a rather localized part of the ventricle, fairly sharply dolimitod from the adjoining ventricular lining. This accumulation of cells in the roof of the ventricle invariably occurs, and it must not be considered as being due to the distortion of the ventricular roof. The reason for the asymmetry of the rhombic roof shown in these figures hes in the fact that in this embryo, as in practically all the embryos of similar stages in this collection, some degree of distortion of the roof of the fourth ventricle is present. Photomicrographs (figs. 50 and 51) taken more posteriorly (from embryo No. 617) give strong evidence of this distortion. They are reproduced not only to show the possible distortion, but also to give a further picture of the lining of the ventricle, with its epithelial-like cells in several layers (fig. 51).


Similar accumulations of these epithelial-like cells are to be found in human embryos of 9 mm. Reproductions of a much fragmented specimen of this size (No. 721) are given in figures 52 and 53. In the latter figure the complete occupation of the ventricular roof by these cells is well illustrated. Moreover, the specimen shows the many-layered stage to a degree but seldom found. It is unfortunate that such a degree of fragmentation and distortion is found throughout this specimen.


Thus far, in human embryos up to and including 9 mm. in length, the roof of the fourth ventricle has shown the same architecture as appears in the pig. As will be recalled, the first evidence of a further differentiation of these cells in the pig embryo was found at a stage of 11 mm. (figs. 26 and 27). In one human embryo of this stage (No. 544) a distinct break in the roof of the fourth ventricle can be made out. This is shown in two photomicrographs (figs. 54 and 55). The picture in this cas3 is somewhat obscured bj' the shrinkage and distortion of the ventricular roof, but a distinct differentiation of the lining epithelium can be made out. On the caudal side of figure 55 considerable nervous tissue is seen. Just superior to this (toward the left) the lining tissue is almost lacking, a few nuclei, only, preserving the contour of the ventricle. Above this area appears again the ventricular lining of many layers of cells. It has been quite difficult to interpret these findings. The area under discussion shows a rather typical adherence to the coagulated albumen ; there is evidence of its extension also into the adjacent mesenchyme, a finding observed in no other similar stage. The caudal position of the opening, the character of the tissue approximating the ventricular cavity, and the presence of the albumen in large amount in the adjacent mesenchyme — all indicate that in great measure the pictures presented in this specimen are largely artifacts. It seems most likely, though, that some differentiation of the tissue in this area has occurred.


In a human embryo of 14 mm., as in the pig of the same stage, the area membranacea superior has attained a great degree of differentiation. This is particularly well shown in figures 56 and 57, the latter being an enlargement of the squared area in the former. These photomicrographs are from embryo No. 144 of the collection of the Carnegie Institution of Washington. Figure 57 shows a characteristic which distinguishes the area membranacea from that of the pig, although in the later stages of the pig embryo (figs. 45 and 47) this feature is present. This concerns the marked decrease of cellular tissue in the membranous area. In figure 57 the deeply stainmg typical ependyma is shown approaching from below. These cells end abruptly at the border of the area membranacea; the ventricle in this area is lined by cells possessing small elongated nuclei and long cytoplasmic processes, which unite to form a ventricular lining. The oval nuclei along the ventricular border become more closely massed together in the superior portion of the area, but nowhere is there the same architecture as in the equivalent stage in the pig (fig. 33). A feature of the histological appearance of the membranous area in the pig embryo is also shown in figure 57; this is the marked adherence of the coagulated albumen of the cerebro-spinal fluid to the area membranacea superior.


The roof of the fourth ventricle in the human embryo is subjected to the same factors causing changes in the form and relationships which were commented upon in the pig; but these play little part until the chorioid plexuses become of sufficient size to divide the ventricle into a superior and inferior portion. In the human embryo, as in the pig, the superior half of the ventricular roof is sacrificed to the greater growth of the cerebellum.


  • Measured on the slide after mounting.

In human embryos of 17 mm., however, these factors have not begun to influence the membranous area. This is shown in figures 58 and 59, photomicrographs from embryo No. 57G. The section is somewhat to the side of the midline, but in the superior portion of the roof of the fourth ventricle the differentiated membranous area can be made out. The sharp delimitation of this area from the denser t3T5ical ependyma on both sides is quite apparent. The ragged character of the ventricular border, with its few elongated spindles, seems wholly in keeping with the transverse view of this area afforded by figure 37.


Embryo No. 576 exhibits one characteristic of the area membranacea superior very frequently seen in human embryos, but almost invariably absent in these stages in the pig. Along the lateral margins of the superior membranous area are dense borders of the many-layered epithelial-like cells which lined the ventricular roof in younger stages. This feature is well shown in figures 60 and 61, the latter figure being a higher magnification of the former. The cellular border of the superior area reaches transversely only through a few 15-micron sections, but it extends throughout the whole cephalo-caudal diameter of the area. It seems likely that this represents purely a survival of the epithelial-like cells in the younger embryos. In rarer instances the whole area membranacea superior may be surrounded bj^ such a border of many-layered cells, but even in these cases the superior and inferior margins are quite thin.


No apparent agencies favoring the disappearance of the superior membranous area m the roof of the fourth ventricle of the human embryo are apparent in stages up to the fetus. Thus, in human embryos of 18 mm. this differentiated area in the roof has reached its maximal differentiation. A section from an embryo of this size (embryo No. 409) is reproduced to show the distortion and its influence upon the topography of the area membranacea. The two photomicrographs (figs. 62 and 63) show the extreme collapse and distortion of the roof of the fourth ventricle. In the figure of higher power (No. 63) the membranous area appears facing posteriorly, due to the shrinkage; the proper leader runs to this area. It shows the differentiation from the adjoining tj'pical ei)endyma which is characteristic of the full}' developed area membranacea superior.


In a beautifully preserved and sectioned human embryo of 21 mm. (No. 460) in the collection of the Carnegie Institution of Washington the area membranacea superior appears as a sharply delimited area (figs. 64 and 65). These figures give a very good idea of the definiteness of this area when the fixation and dehydration approach the perfect. The tissue of this membranous area lining the ventricle here appears to be wholly lacking in an epithelial covering; the mesenchyme seems to serve as the ependymal lining. Study of this area, however, through different stages argues most strongly against such a view.


he process of regression of the area memhranacea sujjerior in the human embryo differs somewhat from that described in the pig. This alteration in the mode of disappearance is largely due to the fact that in the period of growth from 20 to 35 mm. the superior portion of the roof of the fourth ventricle in the human embryo is not sacrificed to the cerebellar lips; for in the human the cerebellum grows largely into the fourth ventricle, enlarging beneath the superior part of its roof. Thus, the attachment of this part of the roof is not greatly interfered with by the rapid development of the cerebellum. The total extent, then, of the superior portion of the roof is hardly altered in these stages in the human, while in the pig embryo the roof is shortened by its attachment to the inferior portion of the cerebellar hp, which retains its earUer characters. These differences in the relationship of the superior portion of the ventricular roof in human and pig embryos may be seen by comparison of figures 74 and 89.


Another factor which renders the mode of disappearance different in the two embryos concerns the greater tufting and development of the chorioid plexuses of the fourth ventricle in the pig. This greater size and complexity of the plexus causes an encroachment upon the roof structures which, in the pig embryo, seems of considerable importance in the t'nal closure.


In the human embryo, however, it has been found very difficult to explain the final disappearance of the superior membranous area on the same mechanical factors w^hich seemed so well to account for its transitory characters in the pig; but at approximately the same stage of growth the process of regression occurs in the human fetus. The area maintains a fair size in stages up to a length of 23 mm. Thus, in figures 89 and 90 (No. 453 of the Carnegie collection) a sagittal section from a human fetus of this size is illustrated. In the higher power (fig. 90) the superior membranous area is short, rather sharply delimited on its superior border by the typical, dense ventricular ependyma. Below, its edge is irregularly formed by the deeply staining ependyma over the invagination of the chorioid plexus. The cell-character of this area resembles that shown in the photomicrographs from the specimen of 21 mm. (figs. 64 and 65). There is left in the area no indication of the cellular architecture which characterized the original ventricular ependj-ma; the cells with their elongated cj'toplasmic processes here have the oval nuclei which are found almost invariabh' in this membranous area.


In the human fetus of 26 mm. (No. 1008 of the collection of the Carnegie Institution of Washington) there is but sHght evidence of a superior membranous area in the upper portion of the roof of the fourth ventricle. The evidence present in this specimen consists in a localized thickening of the lining cells of the ventricle in the situation of the area in other stages. This thickening is illustrated in figures 91 and 92; it consists of several layers of epithelial-hke cells, similar in all respects to the many-layered border shown in figure 83. The picture is somewhat obscured by the vascular pl-^xus directly beneath the ventricular lining.


There is difficulty in determining exactly when the last evidences of the superior membranous area in the roof of the fourth ventricle may be found. This is due to the likelihood of artifacts disturbing the character of the ventricular lining in human material, where the freshness and fixation of the specimen may not be ideal. In the larger specimens in the collection of the Carnegie Institution, which are well fixed and sectioned, the existence of the area membranacea superior could not be wholly verified. Thus, in specimen 405 (26 mm.) the presence of the area seemed probable though not definite. In another embiyo of this same size (So. 782) the existence of this area was still more questionable. In a larger embryo (30 mm. No. 75) the presence or absence of the area could not be assured; many indications suggested its existence, but the resemblance to an artificially separated ependyma was strong. In all specimens of human embryos of over 30 mm. examined, no evidence of the area membranacea superior could be found. It appears likely, then, that the final disappearance of this differentiated area in the roof of the fourth ventricle occurs at a slightly earlier stage in the human embryo than in the pig.


The final disappearance of the area membranacea superior in the human embryo is not accompanied by the same ingrowth of typical ependyma that characterizes the process in the pig. There is a great tendency, in the human, as indicated in figure 92, for a replacement of the area by the same type of epithelial-like cell which comprised the whole ventricular roof in the earlier stages (fig. 41) and later formed lateral borders for the superior membranous area (fig. 83). Thus, in a human embryo of 24 mm. (No. G32 of the Carnegie collection) there is evidence of a very small membranous area surrounded by a border of epithelial-like cells. In a slightly larger specimen (No. 840, 24.8 mm.) the whole membranous area is occupied by the epithelial-like cells. The frequent association of these cells with the area indicates that in disappearing the area membranacea is probably replaced first by these cells, which in turn disappear, so that the whole roof is finally composed of the typical, densely staining ependyma.


The Area Membranacea Superior in Other Animals

In order to ascertain whether the area membranacea superior existed in other animals examinations of serial sections of the rabbit, cat, sheep, and chick of suitable stages were made. All of these animals were found to possess a differentiated area in the roof of the fourth ventricle.


Opportunity was afforded for the study of serial sections of the head of a chick* of 121 hours' incubation. The head was carefully dehydrated and embedded by Dr. E. R. Clark, and was subsequently sectioned bj' Dr. C. R. Essick. The material was beautifully fixed and dehydrated, showing practically no evidence of shrinkage. Typical portions of the superior membranous area are reproduced in figures 66, 67, 68, and 69. Figure 67, taken near the crown of the embr3-o and representing the squared area in figure 66, shows the two dense masses of ependyma separated by the more lightly staining area membranacea. The cellular character of this differentiated zone resembles more the histological features of the similar afea in the pig than those of the human embryo. This resemblance is also to be seen in figure 69, taken more posteriorly than the two preceding figures. The dense ependynia approaching on both sides is sharply delimited at the edge of the hroad membranous area. This is composed of cells having elongated, chromarin-poor nuclei, and long cytoplasmic processes, which form the ventricular roof. The adlierence of the albuminous coagulum occurs here also.


  • The chick measured 14 mm. in 40 percent alcohol.


In the rabbit the occurrence of the superior membranous area was verified as in the other species studied. In a rabbit embryo of 13 mm. (series x in the embryological collection of this laboratory the area was well differentiated from the surrounding typical ependyma. The cells of the area resembled those of the adjacent mesenchyme. The ventricular surface was roughened by the projection of numerous protoplasmic processes. An albuminous coagulum was attached to the cells of the membranous zone.


One sheep embryo from the collection of this laboratory was also studied. The sections, although labeled as an embryo of 10 mm., resembled in every way a pig embryo of 18 mm. The area membranacea was easily identified in the roof of the fourth ventricle; it is similar in every respect to the same area in the pig and the human embrjo.


In a cat embryo of 10 mm. a smiJl but highly differentiated area membranacea superior was made out. The most striking feature in this specimen is the great adherence of the coagulated albumen to the cells of the area and the resemblance of these cells to the mesenchymal elements adjacent. The edges of this differentiated area are sharp and clear-cut.


No attempt was made to identify the area membranacea superior in other animals — as further suitable material was not immediateh- available. The chief study has been made on pig embryos and on human embryos. The occurrence of the area in the cat, sheep, and rabbit probably indicates its existence in aU mammals. The finding of such an area in the chick is also suggestive.


General Consideration of the Area Membranacea Superior

The occurrence of a definite area of differentiation in the superior portion of the roof of the fourth ventricle has been pointed out in preceding subdivisions of this paper. It has been described in detail in the pig embryo and in the human embryo; it has been identified also in cat, sheep, rabbit, and chick embryos. It remains here to discuss the general characteristics of this area.


No description of such an area of differentiation in the ventricular roof has been found in the literature. It may be that the distortion of this structure in the course of the usual embryological technique has rendered its discovery Iess likely. His'-=, m his description of the ventricular roof, has not commented upon the occurrence of this membranous area, even though in a retouched photomicrograph of his fetus C-1 (a human specimen, of the beginnmg of the third month) the area membranacea superior can be made out. Likewise in his description of the pUca chorioidea he faJs to mention any differentiated areas in the roof, although plate I, in his "Die Entwickelung des menschUchen Rautenhims, von Ende des ersten bis zum Beginn des dritten Monats," shows a slight irregularity in the roof. Practically all of the contributions to the anatomy of the roof of the fourth ventricle deal with the lower half of the structure, with particular reference to the occurrence of the foramen of Magendie.


The general biological process involved in the formation of the area membranacea superior concerns a dilTerentiation of the epidermal elements which Une the ventricular cavity. This differentiation, both in human and in pig embryos, first begins with the occurrence in the ventricular roof of an area of epithelial-hke cells. These, in the course of enlargement of the roof, become more or less isolated in the superior portion of the structure, and then undergo a metamorphosis into the typical cells of the membranous area. They are characterized by oval or elongated nuclei (rather poor in chromatin as compared with the nuclei of the typical ependjonal elements) and by cytoplasmic strands (in which the cell-boundaries are very poorly marked) which compose the ventricular border. The ventricular surface in the area membranacea is more ragged and u-regular than where lined by typical ependjana. In many instances, as in figure 57, from a human embryo of 14 mm., this transformation has proceeded to such an extent that the epithehal character of the lining cells is almost wholly lost, and the ventricle seems, in this area, to be lined by mesenchyme. Study of the membranous area in many stages convinces one that such an hypothesis is untenable; in every case the ventricle must be considered as being lined by epidermal elements, no matter to what extent the process of differentiation has proceeded. There is no real evidence to support the view that the ependymal lining of the ventricle has been replaced by mesenchymal elements to form the area membranacea superior.


In general the area membranacea superior is a rounded oval. Its measurement is quite difficult except when fixation and dehydration have been excellent, because of the highly abnormal distortion of the ventricular roof which frequently occurs in the technically poor specimens. Measurements have been made in a considerable number of favorable specimens, both of human and pig embryos. With the history of this area in mind, it will be realized that the size of the structure necessarily varies with the length of the embryo, attaining its greatest dimensions at about the length of 18 or 20 mm. Herewith is a short table of the measurements taken.


Dimensions of area membranacea superior.

Species.

No. of specimen.

Length of embryo.

Width of area.

Length of area.

Species.

No. of specimen.

Length of embryo

Width of area.

Length of area.

Pig

98 107 144 119 106

mm.


12 13 14* 14 14


mm. 0.37 0.95 1.25 0.45 0.65


7iim. 0.5 0.4 1.1 0.6 0.85


Pig


1 mm.


121 1 16 576 1 17 108 18 (?) 45 18 84 1 22


mm. 0.6 1.5 0.8 0.9 0.8


mm.


0.48 0.9 0.8 0.4 0.7


Rabbit Human Pig


Human Sheep Pig


Chick


Pig


In a rough way, then, we may consider the area membranacea as an oval; in some cases the longitudinal diameter exceeds the lateral, and in others the reverse holds. The measurements given above were taken from mounted sections and are probably somewhat disturbed by the histological tcchnifjue which was followed.


  • MuisurecI on alidc uftpr iiiouDting.

The borders of this oval area membranacea are usually fairly regular and smooth, but in some instances they are irregular, due to the fact that small extensions of the area run into the bordering ependyma. These extensions are more commonly met with at the stage when the area has reached its maximum size, as in figures 38 and 39, photomicrographs from an embryo pig of 19 mm. The higher power of these two photographs shows two areas in the smoother ependymal wall. These are extensions of the area membranacea, and within a section or two directly connect with the differentiated area. Both of these small spots on the circumference resemble technical errors; their ragged appearance, the relative excavation of their surface, and the intact ependymal borders w'ould seem to encourage such a view; but when considered in connection with the character of the whole area membranacea they assume a definite relationship in this regard. Other similar areas, rather rare in occurrence, are found separated entirely from the main area membranacea. These isolated areas are of the same size as those shown in figure 39. In significance and character they are probably identical with the larger area membranacea superior.


Most of the general features of the area membranacea superior have been commented upon in descriptions of the various stages of differentiation in both pig and human embryos. The characteristics most commonly observed concern the differentiated character of the cells of the area, the sharp borders of the typical ependyma, the ragged ventricular surface throughout the whole extent, and the peculiar adhesion of the albuminous coagulum from the embryonic cerebro-spinal fluid to the lining cells. The area membranacea superior should be considered, then, as a transitory focus of differentiation of the typical ependymal hning of the roof of the fourth ventricle.


An Undescribed Area In The Inferior Portion Of The Roof Of The Fourth Ventricle

With success attending the effort to find in the superior portion of the rhombic roof an anatomically differentiated area which would furnish a morphological basis for the jihysiological phenomenon of the extraventricular passage of the cerebrospinal fluid, attention was necessarily directed to the inferior portion of this roof (considering the whole roof structure to be divided by the chorioid plexuses). The spread of the replaced injection fluid (fig. 4) into the periaxial tissues through two points in the roof of the ventricle suggested a study of this stage (pig embryo of 18 mm.) as the basis of the investigation. As a histologically differentiated area in this inferior portion of the roof is easily made out, the complete history of the area will be given chronologically. It has been termed the "area membranacea inferior ventricuU quarti," the terminology being based on the same physiological and anatomical features which led to its adoption in the case of the analogous area in the upper portion of the roof.


The Area Membranacea Inferior In The Pig Embryo

The inferior portion of the fourth ventricle shows no evidence of a differentiation from the typical lining ependyma until the length of 15 mm. is reached. In this development consideration must be given to the factors concerned in the process. It will be recalled that in the younger embryos, both pig and human, up to and including a length of 9 mm. the whole roof of the ventricle is occupied by the epithehal-like cells. With rapid growth of the medulla and corresponding enlargement of the fourth ventricle the roof becomes elongated and widened. This process results in the isolation of the area composed originally of the ejiithelial-like cells and the subsequent formation of the superior membranous area. The epithehal-like cells remain in the superior portion of the enlarged ventricular roof, while the whole inferior half is composed of the densely staining, typical ependyma. The division of the roof by the laterally developing chorioid plexuses becomes evident in pig embryos of 14 mm. At this stage the whole inferior portion shows a ventricular lining composed of the typical ependyma.


The first indication of a differentiation in this inferior half of the roof was found in a pig embryo of 15 mm. This is illustrated in figures 70 and 71. The sagittal section from which these photomicrographs were taken is near the mid-line of the embryo, as is indicated by the partial section of the central canal of the spinal cord (fig. 70). The division of the ventricular roof into two parts is also indicated in figure 70 by the invagination of the chorioid plexus. The squared area in the lower half is reproduced in figure 71 under higher magnification; here the first evidence of an ependjonal differentiation is observed. The dense Une of the typical ependjona appears from both sides, but in the center of this ventricular lining a small area of differentiation is seen. This area, isolated by the abruptly terminating pyknotic ependymal elements, is composed of two or three layers of less deeply staining cells. The nuclei are round, rather larger than those of the adjacent mesenchyme, and contain httle chromatin. The cytoplasm stains fairly well with eosin and is not scanty in amount. The cells resemble those epithelial-hke elements which so largely make up the ventricular roof in the earlier stages. No albumen is found near this point of differentiation, although the whole ventricular cavity is filled with the normal amount. In figure 70 the marked zone of the area membranacea superior may easily be seen.


After this initial indication of a differentiation in pig embryos, the further differentiation of the tissue proceeds but slowly until the length of 18 mm. is attained. Thus, in a similar specimen from an embryo pig of 18 mm. the area of differentiation is not greatly hicreased in size. This is shown in figures 72 and 73. In the higherpower figure (fig. 73) both the superior and inferior membranous areas can be made out by the attachment to these areas of the protein coagulum of the ventricular cerebro-spinal fluid.


In the higher-power figure (fig. 73) of the squared area from figure 72, the area membranacea inferior shows the same character as exhibited by the specimen of 15 mm. (fig. 71). The opening maintains the same approximation to the lateral lip of the medulla, but the area is larger and the histological character more nearly approaches the permanent feature of the tissue. The nuclei in this zone are paler than those of the adjoining ependymal elements and contain less chromatin. The cytoplasm is not scanty, nor is it very abundant in amount. The area is also characterized by the occurrence of the cells in a layer, two or three cells in thickness.


In view of the very slow differentiation of the area membranacea inferior in the growth of the embryo from 15 to 18 mm., the enormous enlargement of the region within the next few millimeters growth is very astonishing. This period, as has been jjointed out, is a critical one in the extension of the embryonic cerebro-spinal fluid from a ventricular to a periaxial relationship. Apjjarently, in the course of the embryo's growth during these next few millimeters the whole inferior roof of the ventricle undergoes a transformation and enlargement, so that the differentiated area membranacea comes to occupy practically the whole inferior half of the roof. This portion of the roof, persisting, enlarging, and suffering no extension of nervous t'ssue upon it, becomes the tela chorioidea inferior.


The rapid differentiation of the whole iiiferior half of the roof of the fourth ventricle is a very interesting process. Apparently the typical ependymal elements, visible on both sides of the membranous area in figure 73, undergo a very rapid alteration, so that in the course of a few millimeters' growth the cubical lining of the ventricle is replaced by a low- type cell, with round or oval nuclei, staining much less densely than do the ependymal elements. The whole area membranacea rapidly becomes a membrane in the true sense of the word; it is a continuous, intact laj'^er of cells, generally only one cell in thickness, closing in the fourth ventricle from the chorioid plexus above and the bulbar lips on the sides.


The general characteristics of this transformation are seen in figures 74 and 75. These photomicrographs are taken from a sagittal section of a pig embryo of 23 mm. On one side of the sharply deUmited membrane shown in figure 75 is a tongue of nervous tissue of the medulla; on the other is the differentiated ependjina of the chorioid plexus; between these two structures stretches uninterruptedly the area membranacea inferior. The flattened cells of the membrane, with their oval nuclei and almost continuous cytoplasm, effectually close the whole ventricle. The photomicrograph also shows an interesting characteristic of this membranous area which is universally present in the larger forms; this is the relatively unsupported character of the membrane. The highly vascular mesenchyme posterior to the area has gradually developed, during growth, larger and larger interstices between the cytoplasmic processes. The phenomenon is not due to shrinkage, but is intimately connected with the formation of the future cisterna cerebello-medullaris. This phase of the mesenchymal differentiation will be more fully considered in an appropriate section of this paper. It will suffice here merely to record the lack of support of the membrane.


Another phenomenon of unportauce in the cerebro-spinal fluid relationships of this stage is shown in figure 75. In the mesenchjanal spaces directly beneath the membranous area there is a large amount of albuminous coagulum. This phenomenon does not occur to any appreciable extent in earlier stages or in other parts of the mesenchyme, except about the nervous system. The c1o.se association of the coagulum from the ventricular cerebro-spinal fluid with the inner border of the area membranacea (shown in figure 75 as a slight roughening of the border) is of very great significance in this connection. In one i)oint in the membranous area (fig. 75) the albumen can be traced almost without interruption from the ventricle into the wide spaces of the mesenchyme (cf. fig. 8). This observation strongly suggests that the embryonic cerebro-spinal fluid, which is rich in protein material, is passing, in this stage of embryonic growth, from the ventricle into the periaxial mesenchyme; and such an interpretation becomes established by the comparative findings in the embryo of the same stage in which a replacement of the cerebro-spinal fluid by the ferrocyanide solution had been effected. These comparable findings are surelj^ of the utmost importance for the final solution of the problems centering about the embryonic cerebro-spinal fluid.


In the later stages of development of the area membranacea inferior in the pig embryo the same structural relationships persist that are shown in figure 75. Figures 76 and 77 are photomicrographs taken from a sagittal section of a specimen of 32 mm. In the enlargement of the squared area, from the first of these figures, the continuity and completeness of the membrane are well established. The photograph shows well the flattened character of the cells comprising the membrane and its sharp differentiation from the nervous tissue and ependyma below and from the ependj'ma and chorioid plexus above. Most important in this case is the distribution of the albuminous coagulum. Within the ventricular cavity this appears in considerable amount, and in several places it is in close adhesion to the lining area membranacea. This albuminous precipitate may likewise be traced in some places apparently through the cellular membrane into the periaxial spaces. For here, as indicated in figure 75, the clotted albumen from the cerebro-spinal fluid apjjarently exists in large amounts in the space just posterior to the membrane — the future cisterna cerebello-medullaris. Delicate strands of mesenchyme are still observed running through the wide space, but in general the whole tissue has returned to the line of the future arachnoid. The relative lack of substantial support of the membrane is well brought out in figure 77. A characteristic feature of this membrane, which Blake'3^ has championed, and which is indicated in figures 76 and 77, is the posterior bulging of the roof — "the caudal process like the finger of a glove."


Another section from the same pig embryo, taken more laterally, is represented in figures 78 and 79. In the photomicrograph of higher power the flattened character of the lining cells, the intactness of the membrane in isolating the ventricular cavity, the unsupported freedom of the membrane, and the relation to the albumen coagulum on both sides are of |)articular interest.


The ultimate fate of the area membranacea inferior will not be more fully entered into until the early history of the similar area in the human embryo has been detailed. For in this connection the occurrence of the foramen of Magendie requires discussion, and it seems best to delay the further consideration of the present topic until the whole question can be reviewed.


THE AREA MEMBRANACEA INFERIOR IN THE HUHMN EMBRYO.


The same process in the formation of an area of differentiation in the inferior portion of the roof of the fourth ventricle may also be followed in the human embryo. Unfortunately, however, human omliryological material can rarely be subjected to the immediate fixation and preservation which j'ield excellent histological results in the more plentiful specimens. It does not seem strange, therefore, that the determination of the exact stage at which an area of differentiation can be made out in the ventricular roof should be practicall}' impossible; for, in poor technical procedures, the roof of the fourth ventricle suffers almost more than does any other portion of the specimen.


In a human embryo of 13 mm. (No. 695 in the collection of the Carnegie Institution of Washington) there is slight evidence of a differentiation in the lower portion of the rhombic roof. The changing character of cells in this specimen is not marked, but as the central portion of this inferior roof is reached the ependymal cells seem to assume gradually a more cubical morphology. Associated with this change in shape, there is also a slight loss of the deeply staining character of their nuclei. The whole differentiation, however, is slight and would be commented upon only from the conception of this area in the pig embryo.


The first definite evidence of differentiation in the inferior portion of the ventricular roof was found (specimen 390 in the Carnegie collection) in a human embryo of 15.5 mm. This initial differentiation occurs, then, in the human embryo of approximately the same length as in the pig. The specimen showed the same change in character of the hning ependj-ma as was found in the pig. The deeply staining ependymal elements are replaced in a limited central area in the inferior portion of the roof by cells with more elongated nuclei, poorer in chromatin, and resembling somewhat the epithehal-like cells which early filled the ventricular roof. These cells tend to compose a layer of more than one cell in thickness — a feature particularly noticeable in the peripheral portions.


The size of the area membranacea inferior observed in specimen 390 suggested that the earUest evidence was probabh' to be observed in somewhat smaller specimens. This could not, with the material at my disposal, be verified, but it is probablj^ safe to assume that the first signs of an ei)end\Tnal differentiation will be found in human embryos of about 15 mm. This time of appearance of the area in the human would coincide with its time of primarj* differentiation in the pig embryo. In this limitation of the first appearance of the area membranacea inferior, the standard has been an unmistakable differentiation of ependyma and not an isolated change of a lining-cell or two which might have been the result of the technical procedure. Such a criterion was necessitated by the verj' marked changes in the ventricular borders observed in specimens in which distortion of the chorioidal roof had occurred.


The area membranacea inferior very rapidly increases in extent after the onset of the process of ependymal differentiation. This was hkewise observed in the pig embryo, although perhaps more stages could be made out. In a human embryo of 16 mm. (No. 406 of the collection of the Carnegie Institution) the area nierabranacea inferior is quite extensive, as is shown in figures 80 and 81. In the photomicrograph under higher power (fig. 81) the denselj' stained ependyma approaches the membranous area (ami) as tongue-hke processes from above and below. These tips gradually lose their dense character and are prolonged as a delicate membrane, lining, in this localized area, the ventricular cavity. The nuclei of the cells here are not heavily laden with chromatin; they are oval and somewhat larger than the more densely packed nuclei of the typical ependj-mal element. Unfortunately, the middle portions of the membranous area in this specimen are surrounded by extravasated red blood-cells obscuring somewhat the structure (fig. 81). The process, though, of the differentiation of these ependymal elements into paler and larger epithelial-like ceUs is quite apparent.


As in the pig, the tendency of the differentiated ependymal cells forming the area membranacea inferior to lose in some degree their distinctive appearance and to approach in character the undifferentiated mesenchymal element is apparent in the human embryo very shortly after the original steps in the process of differentiation have occurred. Photomicrographs from two human embryos of 17 mm. have been included to show this phenomenon. Thus, in figure 88, an enlargement of the blocked area from figure 58, the area membranacea inferior (ami) is well defined. The sagittal section from which this photomicrograph was taken is from embryo No. 576, in the Carnegie collection. Above and below the dense fine of ependj-ma may be made out; this tapers quite abrupt!}', to be succeeded bj' the cells of the area membranacea inferior. These cells, products of ependymal differentiation, have lost much of their epithelial-like appearance; they now show rather small, oval or rounded nuclei, poor in chromatin. The cytoplasm of the cells is small in amount, but not disproportionate for the size of the nucleus. The ventricular border of these cells (fig. 88) exhibits a rather characteristic phenomenon, the adherence of a shght albuminous coaguluni. The fine processes of this coagulum fuse^ith the cytoplasmic borders of the cells and render these borders vague and indefinite. Beneath the cells of this inferior area small vascular channels may be made out. These tend to make the membrane appear denser than its cellular character warrants.


In another section from this same embryo (No. 576) the inferior membranous area is shown in relation to the tufted chorioid plexuses (figs. 82 and 83). In the reproduction under higher magnification (fig. 83) the ependymal lining may be traced caudalwards to a gradual fusion into the area membranacea inferior. From the rather high cubical cells in the immediate proximity to the plexuses the ependymal elements become reduced in size and in height, and then rather abruptly the pyknotic character of the ventricular lining is lost. This loss of the deeply staining character coincides with the superior border of the area membranacea inferior {ami). The membrane of this area shows the same cell-character as already desorilied for this embryo. On the superior side of the plexuses (fig. 83) the lateral border of the area membranacea superior (ams) is shown composed of epithelial-like cells.


The apparent tendency of the cells composing the inferior membranous area to lose the epithelial-like character, as shown in the figures from embryo Xo. 576, is not an invariable phenomenon. Rather is an aggregation of epithelial-like cells met with in human embryos very commonly in this area, not onlj- in embryos of small size, but also in small fetuses. This phenomenon is illustrated in figures 84 and 85, reproductions of photomicrographs from a human embryo of 18 mm. (No. 409 in the collection of the Carnegie Institution). In figure 85 the total transverse extent of the area membranacea inferior (ami) is illustrated, with the villous chorioid plexuses appearing to the left. Although this membranous portion of the embryo has been distorted somewhat by the technical procedures to which the specimen was subjected, the cellular character of the membranous area is well indicated. The most striking feature, apart from the characteristic tinctorial differentiation from the typical ependymal elements, consists in the marked clumping of the cells in certain parts of the membrane. On one lateral extent the membrane is thickened into a bulbous swelling several cells in thickness. These cells have palely staining nuclei, poor in chromatin, with an oval or round form. In other places in the membrane smaller but no less characteristic clumps of similar cells maj' be made out. Between these cellular aggregations the membrane stretches in a continuous line with but few nuclei.


Analogous clumps of cells, with pale, rounded or oval nuclei, may be made out in figures 86 and 87, taken from a human embryo of 19 mm., No. 431 in the collection of the Carnegie Institution. Only a small portion of the membrane is reproduced in the figure under higher magnification, but a characteristic clump of epithelial-like cells (epc) is shown. These cells of the differentiated ependyma here again have oval and rounded nuclei, poor in chromatin, similar to those which have been pointed out manj^ times in the foregoing pages. A second broadened area in the inferior membrane is also showm in figure 87.


The further development of the area membranacea inferior proceeds in the human embryo in a manner very similar to that described for the pig. In the stages but shghtly above those already described the differentiation goes on slowly, but withui a few millimeters the cellular pictures resemble those given for the embryo of 17 mm. (figs. 82, S3, and 88). The cellular clumps which appeared quite frequently in the embryos under 20 mm. have not been found in the larger forms. Thus, in an embryo of 23 mm. (No. 453 in the collection of the Carnegie Institution) the inferior membranous area {ami) appears as an extensive membrane comprising almost wholly the inferior portion of the chorioidal roof. The membrane is here of a single cell in thickness; these cells are rather small, with oval nuclei, simulating in some measure those of the surrounding mesenchj-me. The most mteresting phase of the membranous area at this stage of 23 mm. concerns its completed cellular differentiation and its rather slow increase in size.


Wholly similar pictures of the inferior membranous area of the roof of the fourth ventricle are afforded by a human fetus of 26 mm. (figs. 91 and 92). These photomicrographs were taken from embryo No. 1008 in the collection of the Carnegie Institution. In this specimen (fig. 92) the fourth ventricle seems almost to lack a lining of oj^endymal (epidermal) elements in the area membranacea inferior (ami). The cells of this area are small, inconspicuous ua their distinctions from the underlying mesenchjme. The whole character resembles that of the superior area membranacea shown in figure 57.


The appearances exhibited by the inferior membranous area in the stages above 26 mm. are modified in great part by the development of the great cisterna cerebellomedullaris. As in the pig, the breaking-down of mesenchyme to form this cistern results finally in the almost total isolation of the inferior membranous area. The cistern is fairly rapidly formed when once the process begins, and so in an embryo of 35 mm. (No. 199 in the Carnegie collection) the isolated character of the area membranacea inferior (a?ni) may be easily made out. This is shown in figure 94, an enlargement of the blocked area in figure 93. The general architecture of the membrane, particularly its intact character, appears in this photomicrograph, but its finer structure is obscured by the albuminous coagula which adhere on both surfaces. The cell structure of the area membranacea resembles closely that described in the embryos already pictured.


Discussion of the final disposition of the area membranacea inferior will be undertaken in the following subdivision of this paper, in order that the findings in the pig and in the human embryo may be correlated.

General Consideration of the Area Membranacea Inferior

The ependymal lining of the caudal portion of the roof of the fourth ventricle undergoes a process of differentiation which results in the formation of the area membranacea inferior. This transformation has been observed in pig and human embryo


s; in both, the first definite evidence of the cellular change has been observed in specimens of 15 mm. The essential phases of the process are identical in the two embryos. The tendency of the deeply staining typical ependymal elements is to lose their highly pyknotic character; the nuclei become poorer in chromatin and the cytoplasm somewhat more abundant. In the first stages of the metamorphosis the lining cells come to assume epithelial-like appearances, but in the final change the nuclei become small oval bodies, poor in chromatin, resembling to some degree the nuclei of the adjoining undiff'erentiated mesenchyme. In the human embryo, a tendency for the epitheUal-like characters to persist in isolated cellular aggregations is apparent.


After the initial process of differentiation has begun, the area membranacea inferior increases rapidly in extent and the differentiated cells which characterize it come to occupy the greater portion of the caudal part of the chorioidal roof. In the somewhat later stages the area membranacea is almost wholly unsupported by other tissues, due to the development of the cisterna cerebello-medullaris. As soon as the cistern forms, the area membranacea serves as practically the sole dividing membrane between the ventricular system and the future subarachnoid spaces.


The ultimate fate of this area membranacea inferior is necessarily involved in the distribution of the tela chorioidea inferior. Likewise it necessitates a discussion of the possible formation of the so-called foramen of Magendic and its mode of origin from the "caudal process" of Blake. It is proposed to discuss briefly some of these questions in the hope that some phases of the problem may be brought forth.


It must be clearly understood that the questions of the ultimate fate of this area membranacea inferior probably differ considerably in the different species of mammals. In the horse and in the pig the absence of the medial foramen (]\Iagendie) is fairly well established, but in man its existence .seems to rest on equally firm grounds. While, ))rimarily, this investigation has not been concerned with the possible existence of the foramen of IMagendie, the question has been presented manj' times in regard to the pig and human embryos examined.


As far as can be determined, no descriptive study of the development and differentiation of the inferior portion of the rhombic roof has been published. Heuser's'23' studies on the form of the cerebral ventricles of the pig have afforded a very good conception of the gradually changing relationships in this region. Hess(22j has devoted attention to the histological appearances of the inferior roof in the embryo. One of his interesting obseivations concerns the caudal portion of the rhombic roof in a fetal cat of 10 cm., where he noticed a very sudden interruption in the epithelial lining of the ventricle, with a complete closing by a fibrous net. This description by Hess is the only comment upon the histological appearance of the ventricular roof that has been found. His^^s^ pictures, without comment, in a retouched jihotomicrograj^h, a differentiated area in the proper situation in his fetus C-1 (beginning of the third month) .


The many writers in embryology have commented upon the roof of the fourth ventricle. Minot, in 1892, stated regarding it:

"Several writers have thought that the membrane was broken through at several points, but it probably is really continuous throughout life. The fourth ventricle is to be regarded, then, as an expansion of the central canal permanently bounded by the original medullary walls."

Kollman (32) on the other hand, advances the view that during the third month the rhombic roof is broken down to form the foramen of Magendie and the two foramina of Luschka. Streeter^S'*), in his chapter on the development of the nervous system in the Keibel-i\Iall Handbook of Embryology, advances a similar view. The majority of investigators to-daj' incline to the beUef that the roof of the fourth ventricle in man is perforated to form the median foramen of ]Magendie.


jjpgg (22) has advanced a conception of the foramen of Magendie that is supported by numerous observations. To test KolUker's statement that the fourth ventricle remained closed during human embryonic life. Hess sectioned the region in human fetuses, new-born infants, and in adults. The lengths of the fetuses cut were as follows: 7, 12.5, 15, 16, and 17 cm. In the 47 cases the roof showed a medial opening (IMagendie), except in one case, in which it was closed by a "thin pial membrane." Hess's conception of the process of formation of this membrane was that in earh' embn'ological life the rhombic roof was bordered by a regular, meshed tissue. Later the small meshes in this tissue fused to form the larger foramen of Magendie.


Blake's (2) hypothesis of the formation of the medial foramen has been quite extensively quoted in the more recent publications on this subject. In a study of the chorioidal roof Blake found a caudal bulging of the inferior velum; this outpouching became more and more extensive in the older embryos.. In man this pouch became sheared off at its neck, leaving the foramen of IMagendie.


In addition to the few studies referred to above, there have been in the past 25 years a great mmiber of articles (notablj' those of Wilder^^s) and Cannieu^^^) offering evidence that this median foramen of the fourth ventricle is an existent, functional opening. Into this literature it is not proposed to go in the present communication; it may be stated that in the larger part the views presented have been in favor of the consideration of the true occurrence of the foramen of IMagendie.


The material on which this study is based has been purely embryological in type, so that no relialile data regarding the foramen of IMagendie could be obtained. But even in the largest fetuses examined, there was no evidence which indicated a breaking-down or a shearing-off of the inferior roof of the fourth ventricle. In the largest human fetus at my disposal, in which the histological material was good enough to permit an accurate examination of the chorioidal roof (embryo No. 448, 52 mm. in the Carnegie collection) the area membranacea inferior appeared as an intact membrane supported only be a few pial cells. In the pig the material at hand has been such that accurate study of the roof could be made in specimens up to 20 cm. ; in all of these later fetal pigs the roof has been wholly without foramina. If, however, in these larger stages the histological procedures have not been of the best, ruptures and other artificial separations are very frequently found.


The area membranacea inferior, then, may be regarded as a region of ependymal differentiation. Whether it persists as an intact membrane or undergoes, in certain animals, a perforation to form a foramen of Magendie can not be here answered; this study has been concerned solely with the embrj^ology of the cerebro-spinal spaces, and it affords no evidence in favor of or against the existence of such a foramen. Nor has any study been made of the two foramina of Luschka, the two openings from the lateral recesses of the fourth ventricle into the subarachnoid spaces. It can Ije stated, however, that these foramina arc not in existence at the time of estal)lishment of the circulation of the ccrebro-si)inal fluid. This phenomenon, as recorded in the previous section, occurs in pig embryos of 26 mm.; at this time the lateral reces.ses are anatomically and physiologically closed.

VI. Passage Of Fluid Through Roof Of The Fourth Ventricle

On pages 20 to 'SO is a description of the passage of a true solution, substituted without increase in pressure for the embryonic cerebro-spinal fluid, through the roof of the fourth ventricle into the extravcntricular or periaxial spaces. This extension of fluid occurred in two localized areas, one in the superior half and the other in the inferior half of the rhombic roof. Histological study of these regions revealed a localized differentiation of the ependyraa, both in the upper and lower halves of the ventricular roof. It becomes necessary, then, to correlate, if possible, the areas of this fluid-passage to the anatomical differentiations pointed out.


The Accumulation of Injection-Masses in the Superior Membranous Area

It has already been recorded that the first evidence of a change in the reaction to a replacement injection occurred in an embryo about 13 mm. long (fig. 2). This stage was characterized by a dense collection of the precipitated granules in a definite area in the roof of the fourth ventricle. At this stage also the area membranacea superior is well differentiated (fig. 31). That the site of the granular accumulation is this membranous area is easily proved by an inspection of figure 117, which represents an enlargement of the squared area in figure 116. In the low power photomicrograph the prussian-blue granules are not represented, but are found scattered through the ventricles, with a definite collection in the posterior region of the fourth ventricle. Under a higher magnification (fig. 117) the blue can be traced in but small quantity along the normal ependj'mal lining (shown to the left in the figure), but as soon as the differentiated area (area membranacea superior) is reached the granular material is heaped up in a dense mass, which extends as a thickened pad into the ventricle.


The same phenomenon of the accumulation of the injection fluid in the superior membranous area is shown in figures 112 and 113, the second photomicrograph representing the area outlined in the first, but reproduced under much higher magnification. In this specimen (an embryo pig) a dilute solution of silver nitrate was injected into the central canal of the spinal cord. On histological examination the accumulation of the silver also shown in figure 11 was found. Thus, in figure 113 the ventricular epithelium can be made out in the upper right-hand corner, while below (in the area membranacea superior) the silver is densely accumulated.


The explanation of this phenomenon of accumulation in the superior membranous area is not whoUj' clear. It occurs only in stages in which the histological differentiation of the ventricular roof has proceeded to some degree and in stages where the fluid-passage into the periaxial tissues is not wholly unobstructed. This aggregation of the precipitated granules of prussian-blue and of the reduced silver in a locaUzed area certainly suggests a phj'sical exi^lanation, as in these cases the physical laws of precipitation and reduction must hold. The many figures of the superior membranous area of the ventricular roof show that in the stage under consideration the cell-outlines projecting into the ventricles are rough and ragged as contrasted with the smoother and more regular surface of the adjoining ependyma.


Could not these roughened, irregular cell-surfaces become the site of the first and most extreme precipitation of the prussian-bluc and of the reduction of the silver? Certainly they would serve much more efficiently as the foreign substances about which precipitation would occur in greatest amount. This phj^sical explanation finds man}' arguments for its suj^ijort in these studies.


Another explanation of the phenomenon concerns the normal flow of the fluid and the relation of the direction of this flow to the roof of the fourth ventricle. As has already been emphasized, it is difficult to assume that there is any marked production of cerebro-spinal fluid before the periaxial spread occurs. Such an assumption would argue against the development of any special current toward the roof of the fourth ventricle in any stage smaller than that represented in figure 3, and would vitiate the explanation of the occurrence of the granular accumulation shown in figure 2 (a pig embryo of 13 mm.). In the later stages (16 mm., cj. fig. 11) this explanation would probably suffice for the phenomenon exhibited.


The Sites of Fluid Passage Through the Roof of the Fourth Ventricle

With consideration of the evidence presented as to the accumulation of the precipitates of the injected fluid about the area membranacea superior during certain stages in the development of the cerebro-spinal spaces, it would seem that the same area must be concerned in the passage of fluid from the ventricular cavities into the periaxial tissues. This view receives support from the reproduction of a cleared specimen (fig. 11) in which an injection of silver nitrate had been made into the central canal of the spinal cord. The pressure employed was great enough to force the fluid into the periaxial spaces, but the resultant picture clearly showed the oval outline of the area membranacea superior.


The study of the passage of fluid from the ventricular to the extraventricular spaces can best be made by simple histological serial sections. In these observations pig embryos in which the cerebro-spinal fluid had been replaced by the compensating device, supplying a true solution of potassium ferrocyanide and iron-ammonium citrate, were sectioned and examined with reference to the sites of fluid passage. The results of these studies are given here in order that the whole question of the connection of the cerebral ventricles with the subarachnoid spaces may be discussed.


In the stage represented by figure 3 (in which fluid passes from one area in the roof of the fourth ventricle into the extraventricular tissues) histological sections show that the point of fluid ])assage is localized and concerns solely the area membranacea superior. The replaced fluid (as demonstrated by the subsequent precipitation of the prussian-blue) passes through this entire membranous area into the adjoining mesenchyme. The process is wholly confined to this area; the adjoining ependyma is entirely imj)ervious to the ferrocyanide. This phenomenon of passage of the replaced fluid through the sujjerior membranous urea is well shown in figures 14, 18, and 23.


The distriltution of the minute granules of ])russian-l)lue in the cells of the superior membranous area is of iini)ortance in any discussion of the passage of fluid through a membrane; for this area (in the superior portion of the roof of the embryonic fourth ventricle) must be considered as a memljrane permeable in certain degrees to the fluids bathing it. That the area mcml:)ranacea is intact and does not contain stomata or other minute foramina has been demonstrated histologically. Further evidence of the entire lack of intercellular stomata is afforded by the distribution of the prussian-blue granules precipitated in situ after the replacement of the cerebro-spinal fluid by the ferrocyanide solution.


Figure 14 is a reproduction of the superior area from a transverse section of a pig embrj^o in which the routine replacement had been made. The position of the area is shown by the squared outline in figure 13. On both sides the impermeable ependyma is seen, with granules of the blue adhering to the ventricular border of the cells, but not penetrating them at all. To the left of the drawing the few ependymal cells possess, beneath their central border, a chain of the granules which have entered from the abrupt edge of the area membranacea. In the cellular border between the two hps of the ependyma, the area membranacea superior, the passage of the replaced fluid is easily made out by the resultant blue granules. The area is roughly deUmited by a ventricular collection of the blue granules. Examination of these cells shows that the prussian-blue is present within the cytoplasm, avoiding the nuclei with perfect precision. Some of the cells are rounded and almost free from the granules; others, particularly those whose cj'^toplasm is elongated, are completely filled with the granules, the nuclei standing out in a blue granular cytoplasm.


The question of the passage of the fluid between the cells must also be answered by the histological evidence. In the same drawing (fig. 14) in one or two places there are indications of a slight stream of granules between the cells of the area membranacea superior. This apparent transit of the fluid through intercellular passages is particularly clear in the small areas where the cellular cytoplasm is relatively free from the granular deposits. But upon careful examination of these areas under oil immersion it is always apparent that the adjoining cytoplasm is also mvolved in the granular precipitation, indicating that the cells, although almost free from the deposit, are also engaged in the process of the fluid passage. Compared to the whole area of fluid transit, the points indicative of a passage through possible intercellular stigmata are almost negUgible. It seems not unlikely that the outlining of canals between cells may be a physical phenomenon, as in most cases no cellular borders (as demonstrated bj' the precipitated granules) can be made out. These pecuUarities of fluid passage may be seen in figures 14, 18, and 23.


Consideration of all the evidence afforded by histological examinations of the essential character of the area membranacea superior and of the passage of fluid through it incUnes one inevitably to the belief that this area functionates as a cellular membrane. The fluid passes through it as through any permeable liv-ing membrane. Histologically the passage is for the most part through the cj-toplasm of the cells, but occasionally an intercellular course is suggested. Both processes are wholly compatible with the accepted view of a cellular membrane de\dsed for the passage of fluid through it.


The same phenomenon of the passage of fluid from the fourth ventricle into the periaxial spaces is beautifullj' illustrated in figure 23. This drawing is from a transverse section of a pig embryo (23 mm. in length) in a stage when the superior membranous area is rapidly being encroached upon by the developing cerebellum and by the caudal chorioid plexuses. Between the deeplj' staining epondj-mal cells on cither side the membranous area is densely outlined by the deposition of the granules of prussian-blue in the cytoplasm of the cells of the area membranacea superior. The avoidance of the nuclei of these cells by the ferrocyanide is well demonstrated in this reproduction, as is also the impenetrability of the ependymal cells. In a specimen of this nature the question of the passage of the injection fluid through possible intercellular foramina loses its significance; for the drawing shows clearly the importance of considering the entire area membranacea as a functioning whole — a permeable, Uvmg, cellular membrane.


It has been shown in a foregoing section of this memoir that histologically the area membranacea superior decreases to an almost neghgible remains in specimens of embryo pigs over 30 mm. long. This same rule apparently holds for its functional importance, as determined by the relative and absolute amount of prussian-blue granules deposited in the cells of the superior membrane. This decrease in the functional imi;)ortance may be inferred from figure 47, a photomicrograph from a pig embryo of 32 mm. Apparently the size of the membrane determines in large measure the amount of the replaced fluid which passes through it.


Thus far we have been concerned solely with the passage of fluid through the area membranacea superior. In the earUer stages of from 14 to 23 mm. the importance of the superior membrane functionally is great, but in the later stages the inferior membrane assumes far greater significance. This is demonstrated not only by the structural history of the two areas, but by the functional index afforded by the replacement of the cerebro-spinal fluid by a foreign solution.


In the foregoing section the first evidence of any histological differentiation in the inferior portion of the roof of the fourth ventricle was shown to occur in pig embryos of 15 mm. in length. From this stage upwards (figs. 4, 5, etc.) a portion of the inferior roof allows fluid to pass through it. The exact point of fluid iiassage is the localized ependymal differentiation forming the area membranacea inferior. This relationship is easily verified by reference to figure 18. In this drawing of a median sagittal section of a pig embryo the two localized pomts of fluid passage into the periaxial tissue are readily identified; they are quite Umited in comparison to the extent of the periaxial spread.


Figure IG represents the inferior membranous area of the roof of the fourth ventricle from a i)ig embryo of .similar size (18 mm.). The histological character of the inferior area is well shown in this drawing. It will be seen that, except in small areas, the histological differentiation of the ependyma has not proceeded to any great extent; the fluid from the ventricular cavity (as traced by the precipitated granules) closely follows the points of greatest cellular differentiation. There is no possibility of an intfri)r('tation of the findings concerned with the existence of intercellular stomata; the passage of fluid is here again to be looked upon as a transit through a cellular membrane.


The same general i)heiioinena of the pa.ssagc of fluid through a localized area (the area membranacea inferior, hi the caudal portion of the roof of the fourth ventricle) that have been observed in the superior portion of the roof are shown in figure 18. Cliief among these phenomena is the careful avoidance bj- the precipitated granules of the ependymal lining of the ventricles and the adherence of the granules to the Uning walls at the points of fluid passage. The ependymal lining, except in the two areas of differentiation, is everywhere impenetrable to the solution of the ferrocyanide.


As the size of the embryo increases the functional importance of this more caudal area becomes much greater (c/. figs. 3, 4, 5, and 76). The whole caudal half of the fourth ventricle becomes an area of ependj-mal differentiation and of fluid passage. It serves everywhere as a complete diff'using membrane, un))roken by the occurrence of stomata. Through this whole membrane the replaced solutions of potassium ferrocyanide and iron-ammonium citrate pass with apparent ease, as demonstrated by the precipitated granules of prussian-blue (fig. 18). From stages of 24 mm. and over, the lower membranous area is the onlj' one of significance in the total fluid passage.


The areas, therefore, through which the replaced solution of potassium ferrocyanide and iron-ammonium citrate jiassed, in the experimental pig embryos, are the two areas of histological differentiation in the roof of the fourth ventricle — the areae membranacese superior et inferior. There is no evidence whatsoever of any other point of escape of the fluid from the ventricular system into the periaxial spaces. The precipitated prussian-blue does not penetrate any of the lining cells of the ventricle except in the two areas under consideration. Nor is any evidence afforded by histological study of the escape of ventricular fluid through the described foramina of Bichat and of ^Mierzejewsky.


Factors Concerned in the Experimental Fluid Passage

It becomes necessary to discuss the question of the passage of the replaced fluid through the two cellular membranes in order to ascertain to what extent the results obtamed by the method may be relied upon. Naturally in such questions the factors concerned in the normal transit of body-fluid through such structures must be considered.


Probably the most essential element in obtaining rehable results in any injection is the control of the pressure at which the foreign fluid or mass is introduced. This matter has been fully discussed in the resume of the methods employed; it is suflRcient to reaffirm here that, in these observations, the normal cerebro-spinal tension has not been disturbed because of the use of a compensatory replacement. Other experiments, carried out under increasing pressures of injection, have been made, in order to compare the results with those furnished by the replacementmethod.


Consideration must next be given to the factors of diffusion, filtration, and osmosis in the passage of fluid through the roof of the fourth ventricle. The third factor, however, may be largely excluded, owing to the fact that the solutions of potassium ferrocyanide and iron-ammonium citrate employed were for the most part practically isotonic with the body-fluids. Furthermore, the use of hypertonic solutions apparently gave no difi"crent results (except in the increased density of the resultant precipitate) from those obtained by the employment of the isotonic solutions. Finally, it was found to be of service to use hypotonic replacement solutions in order to obtain very sUght precipitates; in these experiments also the spread of the replaced ferrocyanide solution was similar to the standard result afforded by the isotonic solution. These observations with varying concentrations of the foreign solutions replacing the cerebro-spinal fluid serve to indicate that osmosis plays but little part in the passage of fluid through the roof structures of the fourth ventricle. Undoubtedly the factor of osmosis can not be ignored in any consideration of the passage of fluid through a cellular membrane, but it seems unhkely that with solutions of practically the same salt-content it should be of great importance.


The influence of diffusion in this passage of the solution of the ferrocyanide and citrate from the cerebral ventricles into the extraventricular space is probably great. The whole plan of the experiment concerns the introduction of salts foreign to the body-fluids, even though in analogous concentrations. It seems not unhkely that as soon as the replacement of the existent cerebro-spinal fluid is effected the ferrocyanide and citrate must immediately begin to diffuse out into the periaxial tissues and the normal salts return to the ventricles. Probably, however, this same phenomenon plays a normal role in the human body. Jacobson's^^?) extensive and important studies on the chemistry of cerebro-spinal fluid have shown that the ventricular cerebro-spinal fluid is not identical with the subarachnoid fluid. The differences in the two fluids are probably to be accounted for by the fact that the ventricular fluid represents the pure elaboration of the chorioid plexuses, whereas the lumbar subarachnoid fluid is composed not only of the products of the chorioid plexuses but also of the fluids from the perivascular system. In this transference of the ventricular fluid to the subarachnoid space diffusion may play some part, the relative importance of which can hardly be estimated.


But will diffusion alone accoimt for the passage of the experimental fluid in the ventricle through two well-defined areas into the periaxial tissues? Will diffusion account for the varying extent of the injection in different stages of embryonic develojjment? There are several arguments against according diffusion a maximal imjjortance in the process. In the first i)lace, an injection of the solution of the ferrocyanide under mild syringe-pressure will give a spread similar in every re.spect to thos(; obtained by the replacement method. This indicates that the course taken by the two solutions is not necessarily the result of diffusion, but rather of the capabilities of the tissues for fluid-spread; and similarly the jiassage of this true solution through the roof areas need not be solely a diffusion process, but may be accounted for by the true flow of the fluid in this direction. Again, in the stages represented in figure 2 one would expect as extensive a spread of the replacing solution into the periaxial tissue were diffusion the active force in the movement of the fluid. Instead of such a periaxial spread the injection fluid remains wholly within the ventricular system, indicating that other forces than that of diffusion play an active role, in the more advanced stages, in the movement of the fluid. Finally, if diffusion is to be considered the sole agent in the distribution of the replacing fluid, why does not the ferrocyanide penetrate all the cellular structures lining the ventricular cavity? Surely it would he expected that diffusion between the body-fluids and the ferrocyanide solution would occur in each ependymal cell — a phenomenon observed only in the cells comprising the ventricular surfaces of the membranous areas of the rhombic roof.


While acknowledging that diffusion and osmosis may play important parts in the process of the passage of fluid from the fourth ventricle into the periaxial tissues, it seems apparent that some other factor or factors must be the determining agent or agents. It is not unlikely that the formation of cerebro-spinal fluid by the cells of the chorioid j)lexus may cause, in the replacement experiments, further passage of fluid into the extraventricular regions. Such an elaboration of fluid, with the ventricles filled with the experimental solution, would result in an increase in the normal ventricular tension. If this be the real explanation, the passage of the fluid into the extraventricular spaces would result in part from the increase in pressure on one (the ventricular) side of the membrane. The process, then, would be one of filtration through the membrane from the point of higher to that of lower pressure. This explanation best seems to cover the results obtained by the replacement method, and is supported by the histological examination of the developing chorioid plexuses and by many other features which are dealt with in other sections of the paper. This view is also strongly supported by the results of injections under mild syringe-pressure .


On the basis that the passage of fluid from the fourth ventricle into the periaxial tissues is in large measure a process of membrane filtration, the phenomenon of the fluid transit of the replaced solutions may be taken as a real index of the circulation and distribution of the cerebro-spinal fluid. It may be assumed, therefore, that the resulting distribution of the prussian-blue granules represents the course and extent of the fluid channels of the embryonic cerebro-spinal fluid.


The discussion of the fluid passage outward from the cerebral ventricles into the subarachnoid spaces has thus far been concerned with the processes involved for the transit of the true solutions of the salts. There is, however, an undoubted passage outward, as has already been indicated in a foregoing section, of the protein content of the normal cerebro-spinal fluid. This occurs in specimens in which a trul}- definitive membrane, intact throughout, can be seen inclosing the chorioidal roof. The explanations which suffice for the passage outward of the true solutions will not serve for this phenomenon.


The cells of the body probably are equipped to handle colloidal solutions in several ways, but two methods seem possible as explanatory of the problem at hand.


In the first place, it is conceivable that the cells in the differentiated areae membranaceae could phagocyte the colloidal albuminous particles of the ventricular fluid and excrete them into the subarachnoid spaces on the other side of the membrane; but it does not seem probable that this explanation is correct. Much more likely is it that the colloidal masses may follow the same laws of fluid-passage as the true solutions. But in such a passage through a cellular membrane the rate of passage will be much slower with the colloid.


These two theories regarding the passage of the albumen of the ventricular cerebro-spinal fluid into the subarachnoid spaces are not based on any findings presented in this article, but are ventured as being in keeping with current physiological explanations of such phenomena. On the basis of the second hypothesis, the failure of granular material to pass through the cellular membrane of the chorioidal roof must be explained as being due to the inability of the cells to handle the foreign material except in sizes which could be absorbed. The fact that the original unit was not phagocyted or passed through the membrane probably depended on the size of the molecule and the specific character of the lining-cells.


THE PASSAGE OF SILVER NITRATE AND INDIA INK THROUGH THE MEMBRANOUS AREAS IN THE ROOF OF THE FOURTH VENTRICLE.


Thus far in the discussion of the passage of the experimental fluids through the ventricular roof, true solutions of potassium ferrocyanide and iron-ammonium citrate only have been considered. This solution, as has been pointed out in this and in a previous article^^^)^ jg non-to.xic and is not taken up by the cells. With the dilute solutions (0.25 to 0.5 per cent) of silver nitrate, a far different problem is presented. Replacement experiments with this salt are rendered impossible bj- its intraspinous toxicity and by its precipitating action upon protein; but beautiful preparations may be made by this method by the simple injection with a syringe into the central canal of the spinal cord.


With mild syringe-pressure the result of such an injection with silver nitrate is in all cases a simple ventricular spread, with no extension into the ])eriaxial tissues. This general rule holds in all stages in which the central canal can be definitely entered without causing a spread into the j^erispinal tissues. This failure of the spread to extend into the periaxial tissues under mild pressure is undoubtedly due to the coagulating effect of the silver, which renders further passage of the fluid impo.ssible. The reduced silver collects about the superior membranous area in the roof of the fourth ventricle, outlining it distinctly. This phenomenon is illustrated in figure 115 (a transverse section of a pig emljryo of 19 mm.). At this stage the rejjlacement of cerebro-spinal fluid by a ferrocyanide solution results in a quite extensive spread {cf. fig. 5).


With increa.sed pressures of injection the silver may be pushed into the periaxial tissue through the roof structures of the fourth ventricle. The transit of the injection-mjuss occurs in the area meml^ranacea sujjcrior in practically all cases (c/. fig. 12). The inferior membranous ar(>a, in the earlier stages, is almost invariably impermeable to the silver (unless the injection-jiressure i.s extreme). When the superior area is examined after such an injection under high pressure the silver is found deposited throughout the cells of the area, extending only a short distance into the adjacent tissue. This feature of the injection is pictured in figure 113. In these injections the high pressure undoubtedlj' suffices to force the silver through the coagulated area membranacea. Its coagulating effect on the ependjina is almost equally marked, but the point of least resistance is apparently in the membranous area, allowing the fluid to pass through it.


Replacements of the cerebro-spinal fluid with diluted solutions of mdia ink within the medullary-canal system of small pig embryos never result in any extension of the granules into the periaxial tissues, for under the normal tension in the ventricles of the pig the arese membranaceaj are impermeable to the passage of granular material. After such a replacement the carbon masses maj' be found everywhere throughout the ventricles, but not in the periaxial tissues. However, india ink may be forced into the periaxial tissues by the use of high pressures of injection, as shown in figure 10. In this s])ecimen of a pig embryo (21 mm. m length) the periaxial spread occurred solely from the superior membranous area. This is analogous to the results obtained with silver nitrate, shown in figure 12. Without doubt in the earlier stages the superior area is much more permeable than the inferior. Histological examination of these specimens after an injection of india ink under high pressure reveals that the carbon granules gain the extraventricular space onh' through the area membranacea superior; some cells in this area are crowded with the granules, but for the most part extensive intercellular stomata have been made. The whole process must be viewed as a result of the excessive pressure of injection.


In the more advanced stages of the pig embryo (30 mm. and upwards) the pressure necessary to occasion an extraventricular spread of the india ink after intraspinous injection decreases somewhat, so that \\ith mild syringe-pressure a local periaxial sjn-ead from the fourth ventricle may be obtained from an injection into the central canal of the spinal cord. This is in accordance with the observation of MalU'^), who found that the injection flowed "through the medial opening of the fourth ventricle." The opening in these cases is in the area membranacea inferior, and in many instances subsequent examination showed rupture of the membrane with escape of the ink, even though the injection-pressure was moderate.


Taken as a whole, then, the findings are against the passage of solutions of silver nitrate or suspensions of india ink from the ventricles into the periaxial tissues, except when injected under pressures far above the normal intraventricular tension.


RELATION OF THE EPENDYMAL DIFFERENTIATION TO THE PASSAGE OF FLUID.


Under this heading it is jjroposed to discuss the relationship, if any, existing between the stages of differentiation of the ependjona of the roof of the fourth ventricle and the prssage of fluid through the two membranous areas. The discussion must necessarily be of a temporal character, with an attempt to consider possible factors in the process.


The most important question in this connection is whether the ependymal differentiation is necessary for the passage of fluid through it. In tlie \ng embryo of 13 mm. the area membranacea superior has reached a stage of marked differentiation (fig. 31), but at this same stage (fig. 2) there is no evidence of any pas.sage of fluid through the roof of the fourth ventricle into the periaxial tissue, only an outlining of the oval membranous area. Here, then, the histological differentiation has definitely preceded the assumption of function on the part of the area membranacea superior. The passage of fluid through the lower area occurs at a relatively earlier stage than it does through the superior opening. The first evidence of differentiation of the inferior roof of the fourth ventricle was observed in pig embryos of 15 mm. in length. At 18 mm., even though the process of differentiation was far from complete, some of the replaced fluid was able to pass through the lower area (figs. 4, 16, and 18).


A consideration of these observations leads to the assumption that some histological differentiation of the ependj^ma is necessary for the extraventricular passage of the replaced fluid. In the case of the superior area the differentiation occurs at a considerable developmental interval before fluid passes through it; in regard to the inferior area the assumption of function occurs at a somewhat earlier period in its differentiation. This slight difference between the two areas may possiblj' be explained on the basis that as soon as the stage of 14 mm. is attained (by the pig embryo) a greater amount of cerebro-spinal fluid is produced than can be cared for by the more slowly enlarging ventricular cavities. As soon as this disproportion occurs the excess of fluid is poured into the periaxial tissues through the already differentiated area membranacea superior; therefore, when the inferior area first shows evidence of formation there is still this excess of fluid in the ventricles. The fluid apparently avails itself almost at once of the new opening and its functional existence becomes immediate. It is apparent, moreover, that the capacity of the membranous areas for the passage of fluid is considerably in excess of the demands made upon them, and furthermore, that the provision for the passage of increasing amounts of fluid is completed before the demand arises.


In the passage of fluid from the ventricles into the mesenchyme, there is one other factor which has not as yet been considered. This concerns the iKitentiality of the adjacent mesenchyme to afford channels for the fluid poured into it. Were resistance offered to the flow of solutions through the mesenchymal tissue spaces, fluid could escape from the ventricles in only very small amounts, if at all ; as soon, however, as easily traversed fluid channels became established, the cerebro-spinal fluid could readily escape through the two membranous areas. The question as to what part the embr\'onic cerebro-spinal fluid i)lays in the further development of the meningeal si)aees also arises in this connection. It is at i)resent impossible to assign to any one of these factors a specific role in the passage of fluid from the fourth ventricle into the jieriaxial spaces, but it is important to consider them as possible determining agents. The evidence all indicates that the rate of production of the embryonic cerebro-spinal fluid is the most important factor, by far, in the extraventricular escape of the fluid.


VII. General Histological Differentiation Of The Cerebro-Spinal Spaces

The general problems concerned in the formation of the meninges and of the spaces inclosed within them deal with the gradual adaptation of a primitive undifferentiated mesenchyme to the anatomical and physiological requirements of the adult. Originally the meninges were held to be derived from the same epidermal infolding which gave origin to the central nervous system; then, with increasing knowledge of the structure, the dura alone was said to be a product of the middle germ-laj'er; and finally, by the researches of His^^s) and of K6lliker,'3i) the mesenchymal origin of the three meninges was established. The general process of the differentiation and the stages in this transformation have not been reported in great detail; here, too, the investigations must have an outlook for physiological anatomy as well as for pure morphology.


It may be well to comment briefly on the relationships of the three meninges found in adult mammals. The dura is well estabUshed as the fibrous-tissue envelope of the leptomeninges and the central nervous system. But there is a tendency to regard the arachnoid and pia mater as constituting one structure — the leptomeninges or "pia-arachnoid," in the terminology of Middlemass and Robertson'^"). This difference of opinion in regard to the two inner meninges is due to their structural and intimate relationships. The arachnoid may well be a.ssumed to be a single membrane, worthy of being regarded as a single structure if one considers only its outer continuous membrane as the essential structure. But the inner surface of this membrane sends processes inward to fuse with the pia mater, which is so closely applied to the ner\-ous tissue. These processes divide the subarachnoid space (mcluded between arachnoid and pia) into the well-known meshes in which the cerebro-spinal fluid circulates. From the standpoint of these channels (the subarachnoid spaces) the arachnoid constitutes the parietal and the pia the visceral layer. Thus the intimate structural unitj' of the two membranes seems, in the opinion of many investigators, to warrant their designation as a single membrane. This view, however, has been strongly opposed by Poirier and Charpy^*^', who considered the distinction of three meninges very essential. Hence, in considering the transformation of tissues in the embryo, regard must be had for the dura as a well-differentiated structure, and for the leptomeninges as units, but certainly to be regarded from the standpoint of the subarachnoid spaces. In tliis connection Sterzi's'^^') observations on the comparative anatomy of the meninges are of interest. It will be recalled that the dura in lower forms becomes well estabhshed before the leptomeninges emerge from a primitive mesenchyme.


THE PERIAXIAL MESENCHYME.


Surrounding the central nervous system in young embryos is a rather thick cushion of undifferentiated mesenchyme, similar m all respects to the undifferentiated tissue in other parts of the embiyo. But verj' soon in the course of development the nuclei in this mesenchyme increase along the clear marginal zone of the spinal cord ami ha.silar structures, forming the initial indication of the pia mater. This phenomenon is indicated somewhat in figure 40, a photomicrograph taken from a human embryo (Xo. 836) of 4 mm., the earliest stage here illustrated.


The next essential change in the great differentiation of the meninges concerns a blastemal condensation of this same mesenchymal tissue to form ultimately the bony covering of the central nervous sj'stem and a portion of the dura; but between these two zones of differentiation the mesenchyme remains for a time almost unaltered. A portion of this tissue will go to form the arachnoid membrane and the trabeculaD which mark off the subarachnoid spaces. This process in the formation of the arachnoid will be discussed here; the formation of the pia mater and dura will be detailed in succeeding divisions of the paper. The differentiation will be discussed as a general process, in regard to both human and pig embryos, for in no respect has any essential difference between the two been observed.


The general character of the periaxial mesenchyme may be commented upon here. The tissue is of a verj' loose and typical structure, forming a syncytial network of rather small mesh, but fragile. The nuclei of the cells are oval, with a definite chromatin content; the cytoj^lasm is largely devoted to the maintenance of long processes which connect with adjacent cells. Adhering to the cj^toplasmic processes are very tiny albuminous coagula, of such small amount as to be hardly noticeable; also in the meshes of the mesenclwme very small quantities of this albumen may be identified. These albuminous coagula undoubtedly represent the protein of the tissue fluids in the undifferentiated stages.


THE FORMATION OF THE ARACHNOIDEA.


A general consideration of the problems here involved will surely shed light on some of the various factors concerned. It must be noted that in its development this membrane proceeds from an undifferentiated but small-meshed mesenchyme into the adult structure which contains the relatively large cerebro-spinal channels. Then, too, the enlargement of the tissue meshes in certain places — as the future cistcrnae — must be enormous. Besides this necessary dilatation of the spaces in the periaxial m(>senchymc, the outer portion of the tissue must separate from the future dura and form the outer surface of the arachnoid membrane. Here the process must be one of tissue condensation and proliferation. A similar agencj' is involved in the growth of the mesothelial cells which cover the outer surface of the arachnoid and also the inner suliarachnoid spaces.


The g(>n<'ral process, then, in the formation of the arachnoid membrane concerns a thiiuiing and readjustment of the primitive mesenchyme in certain areas, while in others the process is reversed, the membrane reaching the adult form through proliferative and condi-nsing i)henonieiia. Such alterative processes must naturally result from the apj)lication of certain mechanical or vital agents in the growth of th(? embryo. Is the mere growth of the central nervous system sufficient to furnish these alterative agents, or must we likewise trace the corres])onding development of the bony coverings of the brain and spinal cord? X{>ith(>r factor seems relatively of great importance when comi)ared to the possible influence of the presence and circulation of cerchro-spinal fluid on this periaxial tissue. This seems to be the most important factor, an internally-modifying influence to which the periaxial mesenchyme is subjected in the formation of an arachnoid and its subarachnoid spaces. It will therefore be from this standpoint that the development of the spaces will be discussed; for, as has already been pointed out, the periaxial mesenchyme becomes a functionally active tissue for the circulation of the cerebrospinal fluid at a stage when difi"erentiation has not begim. On this basis, the lack of differentiation shown in the |)eriaxial mesenchyme in the stages before the ventricular cerebro-spinal fluid is poured into the mesenchyme in the neighborhood of the roof of the fourth ventricle is not surprising. The character of the periaxial mesenchyme in the early stages is reproduced in numerous photomicrographs (figs. 25, 49, 51 , and 53). The mesenchj'me is here characterized by a rather dense meshwork of cytoplasmic processes, interspersed b^' a considerable number of oval nuclei. The content of the interstices in albumen, as judged by the persisting coagula, is very small. This picture of the periaxial mesenchyme persists untU cerebro-spinal fluid is poured from the ventricle through the area membranacea superior.


As will be seen in figure 3, the first indication of an extraventricular spread of the replaced fluid in the ventricles occurred in a pig embryo of 14 mm. At this stage the membranous area in the superior portion of the roof of the fourth ventricle has already' become well differentiated. The fluid from the ventricles, however, does not reach any considerable spread until after a length of 18 mm. is attained; the periaxial spread during this period of growth is wholh^ confined to the peribulbar tissues. It is quite important in this connection that the first obvious differentiation of the mesenchyme for the formation of the arachnoid .should appear during this period and should involve the peribulbar tissues.


The first change to be noted in the transformation of primitive mesenchjone into the future arachnoid is an obvious thinning of the structure with a decrease in the number of nuclei per unit-volume. This is made out in a photomicrograph (fig. 57) of a section from a human embryo 14 mm.* long, when contrasted with a similar mesenchymal area posterior to the ventricular roof (fig. 53). In the pig embryo this thinning of the mesenchyme is as obvious at this early stage.


The process of dilatation of the mesenchymal spaces at this stage hardly seems to concern a direct disruption of the syncj'tial strands, but resembles more the spreading of the cell-bodies by the introduction of more fluid into the tissue spaces. This process would certainly result in an ap])earance similar in everj' way to that represented by figures 35 and 57. It probably also concerns other factors, as, possibly, the growth of the whole embryo without a corresponding degree of mesenchymal proliferation.


In a human embryo of 17 mm. (Xo. 576) evidences are apparent of such a thinning of the mesenchyme about the medulla. Thus, in figures 58 and 59, from this specimen, the cellular decrease can be made out both in the region of the roof of the fourth ventricle and around the basilar surface of the medulla. It wdll be noted that the differentiation (i. e., the thinning) about the roof has proceeded more rapidly than along the anterior bulbar surface. This is perhaps to be expected in view of the initial pouring-out of the cerebro-spinal fluid into the mesenchyme just posterior to the roof.

  • This embryo measured 14 mm. on the slide.


In this mesenchymal differentiation a slightly increased amount of albuminous coagulum may be noticed. The truth of this is made obvious by an examination of figure 61, a photomicrograph from a human embryo of 17 mm. The almost entire freedom of the mesenchyme from albuminous detritus is most noticeable at earlier stages.


As was pointed out in the description of the results of replacing the cerebrospinal fluid, a marked change in the rate of development of the cerebro-spinal spaces in the pig-embryo ensues just after attaining the length of 18 mm. Within the growth of 2 mm. the injection spreads completely down the spinal cord and about the basilar structures of the cerebral cavity. This rapid extension finds its analogous process in the equally rapid changes which may be traced in the periaxial mesenclnTTie. Thus, in figure 72, a photomicrograph from a sagittal section of a pig embryo of 18 mm., the whole nervous tissue appears surrounded by a very thin, Ughtly staining tissue; this is the periaxial mesenchyme, which is undergoing its rapid metamorphosis. It will be noticed in this figure that the posterior structures (rhombencephalon) are surrounded by a much less dense mesenchyme than are the anterior (mesencephalon). This relative differentiation between the bulbar tissue and that around the mid-brain is only of temporal character; the mesenchyme about the medulla, as has already been pointed out, begins to differentiate first, the differentiation of the mesenchyme about the other nervous structures following somewhat later.


Figure 73 is a photomicrograph of higher power, taken from the squared area in figure 72. It shows to what a surprising degree the mesenchymal differentiation has proceeded during a few millimeters' growth. Two striking features of the process are brought out in this reproduction. In the first place, many of the mesenchymal trabecular have apparently been broken down, sacrificed to a few larger remaining strands. The cells connected with the destroyed trabeculie appear to recede until one of the heavier surviving strands is met with, when they adhere and apparently aid in the future development of a permanent arachnoid trabecula. The second feature of importance in figure 73 concerns the large amount of allnimen seen in the periaxial space. There is here a much greater amount of albumen than is ever found in the periaxial mesenchyme before the differentia ting process which results in the future subarachnoid space has become definite. The occurrence of this large amount of albuminous coagulum is apparently related directly to the outflow of the cnibrj'onic cerebro-spinal fluid, for the embryonic fluid is very rich in protein material, a.s can be readily seen by the partial filling of the embryonic cerebral ventricles with the clotted albumen.


This process of the l)reaking-down of the mesenchymal spaces to form fewer and hirger spaces goes on very rapidly in pig embryos as the\' exceed the length of 18 mm. Thus, figure 75 (from a pig embryo of 2.3 mm.) shows a marked decrease in the mesenchymal elements al)out the medulla; the strands are becoming fewer in number, and the albumen-filled spaces are increasing rapidlj' in size, but decreasing in number. About the mcsencej)halon, however, the process has only just begun (also shown by fig. 74). In this photomicrograjih (fig. 75) the mesenchj^mal elements have broken down somewhat; the spaces are Ijecoming enlarged, and a fine albuminous coagulum fills the interstices between the mesenchymal processes. The whole picture conveys an excellent idea of the forces which convert the many-spaced mesenchyme into the much fewer cerebro-spinal channels.


This general plan of the formation of the larger subarachnoid canals reaches its maximum in the formation of the various cisternae for cerebro-spinal fluid. The process is probably best illustrated in the case of the cisterna magna, which persists in the posterior cerebello-bulbar angle. Figures 74 and 75, taken from an embryo pig 23 mm. long, give an idea of the initial formation of the cisterna cerebellomedullaris. The mesenchymal strands, as shown in figure 75, are already broken down in part, and are profusely covered with albuminous coagula. The process has not proceeded to any extent in this specimen of 23 mm., but in the course of the next 10 millimeters' growth extensive changes occur, as arc shown in figures 76 and 77, photomicrographs from an embryo of 32 mm. In the space outside the inferior membranous area the mesenchymal trabecula? have almost disappeared; the space — or cistern, as it should now properly be called — is almost completely filled with the clotted albumen. The mesenchjTne is seen running through this embryonic cistern as a few isolated strands, but most of the tissue appears now as a fairly definite membrane on the outer side of the space. This membrane will go to form the inner surface of the dura and the continuous outer layer of the arachnoidea, as it furnishes a visceral layer for the subdural space.


More laterally in this same specimen the formation of the cistern has progressed to an even greater extent. In figures 78 and 79 the total freedom of the lower portion of the cistern from trabecular strands is seen; above, the mesenchjTne still sweeps down as a supporting structure for the chorioid plexus. A definite differential hne of mesenchymal condensation indicates the future outer border of the arachnoid as it incloses the cisterna cerebello-medullaris. This general process of mesench^Tnal breaking-down, altering the original small spaces into the larger arachnoid channels, holds as the embryo develops into larger forms.


In addition to this formation of the subarachnoid spaces in the adult through the enlargement of the embrA-onic mesenchjTnal spaces, the perimedullary mesench\Tne undergoes in these same localities condensations which result ultimately in the formation of the arachnoid membrane and the trabeculse diAnding up the cavum subarachnoidealo. Mention has already been made of the adhesion of the cellbodies of the disrupted mesenchymal elements to the persistmg strands — the initial step apparenth- in the ultimate differentiation of the mesothehal cells which line these spaces. Gradually ^vith the increasing growth of the embryo these cells seemingly become arranged in definite columns covering the persisting arachnoidal trabeculae. At the same time a differentiation of these primitive mesenchj-mal elements occurs, the cells ultimately being transformed into the very low cuboidal mesothelium of the subarachnoid spaces. This differentiation begins first in the basilar portions of the cranium and spreads upward, in a way similar to the course of development of the cranium and of the enlargement of the pericerebral spaces.


While such a general process as outUned accounts for the formation of the arachnoidal trabeculae and the subarachnoid spaces, it has but little bearing on the development of the outer intact membrane of the arachnoidea. This portion of the arachnoidea (which might be termed the arachnoid membrane as distinguished from the arachnoid trabecule) first appears as a distinct line of mesenchymal condensation separating the mesenchyme into the primitive arachnoid and dura mater, as in figures 76 and 77, dmc. This rather thin zone of cellular density in reality represents not only the outer surface of the arachnoidea, but also the inner surface of the dura mater. At first these develop in close fusion with a later separation of the two membranes. With this cleavage of the two surfaces, the arachnoid membrane rapidly differentiates, forming an intact layer over the subarachnoid spaces. The cells covering the surface membrane seem to change gradually into the low cuboidal type, similar to those covering the arachnoidal trabeculae. The details of these processes may be most easily studied in the region of the cerebral hemispheres; in this situation the transformation of the tissues occurs at a later period than in the basilar regions, for the differentiation of this mesenchyme follows the general plan of development of the cartilaginous and bony cranium.


The greatest problem in connection with the development of an external arachnoid membrane naturally concerns the separation of this leptomeningeal tissue from the pachymeninx. In the solution of this particular problem gross dissections have been found of benefit. For this purpose, pig embryos of larger size were used, and attempts were made to ascertain at what stage of development a true anatomical separation of the two membranes occurred. It was found that in embryo pigs of about 40 mm. the dura over the calvarium could be well separated from the arachnoid, but areas of unseparated tissue still persisted at this stage. This was also found to be true in pig embryos of 50 mm. ; on the inner surface of the dura at this stage a mesothelial cell pattern could be demonstrated, although areas of attachment to the arachnoid existed. However, the differentiation of the periaxial mesenchjTne into the adult arachnoid does not occur coincidently with the possibility of a forceful separation of the dura from the surface of the brain; ])ut before this separation of the pachymeninx can l)e made the mesechyme which will go to form the arachnoid must undergo some differentiation. This process invohcs a condensation or accumulation of mesenchymal elements directly in the secondary dural thickening; the cells, with oval nuclei, soon form a continuous membrane of two or three cells in thickness. Apparently soon after the cellular accumulation has been accom])lished, a separation of the dnni from the arachnoid may be made. In certain areas, varying greatly in size, there is still an intimate connection between dura and arachnoid. These connections are particularly prominent over the developing cerebral hemispheres, and it is with this differentiation in the formation of the arachnoid spaces that we will now deal.


In a human fetus of 76 mm. (No. 1134) the arachnoid was found to constitute, in the region about the great sagittal sinus, a cellular layer which adhered quite closely to the dura, even though a Une of difTerentiation between the two meninges could be made out. This adhesion could undoubtedly be separated, even bj' gross dissection, although the tendency to adhesion was stronger than the attachment of the pia to the cortex. From its cell-character and general histology' the arachnoid at this stage must be considered as a formed membrane, but in a primitive state.


A somewhat similar but more advanced stage in the formation of the arachnoid membrane is seen in a human fetus of 100 mm. (No. 928-E) and in a fetal pig of 114 mm. In both the arachnoid membrane is verj' cellular, adhering to the dura only along the superior longitudinal sinus and in certain isolated areas. The cells comprising the arachnoidea possess oval, rather large nuclei which stain palelj- with hematoxyhn. No typical arachnoidal trabeculae could be made out in specimens in this cortical region.


The cellular character of the arachnoid persists in the larger embryos and fetuses as a layer, several cells in thickness, constituting the outer arachnoid membrane. In a fetal pig 190 mm. in length the membrane was practically differentiated, its outer wall being covered by me.sotheUal cells with large nuclei lying about a small fibrous-tissue base. The arachnoid trabeculae were developed only in the larger sulci, where they appeared as typical cellular cords about a core of fibrous tissue. At this stage, too, the vessels traversing the arachnoid spaces were found covered with similar cells. These may now be justly termed the mesotheUal cells.


Quite similar stages of arachnoidal differentiation occur in human fetuses of 200 (No. 870) and of 240 mm. (No. 1131). The arachnoid has everj-where practically become adult in character, except for a further decrea.se in the number of the peripheral layers of mesothelial cells. The fibrous tissue underlying this covering membrane possesses, as in the adult, almost a minimum of support.


In certain areas, however, the differentiation of the mesenchA-me into the adult arachnoidea does not keep pace with the general process. In the present study this phenomenon of unequal develoj^ment was especially well shown in fetal pigs of 150 mm. and upwards. It concerns the development of arachnoid trabeculae in the cerebral sulci. As is well known, the arachnoid membrane bridges the cerebral fissures, wliile the pia follows the cerebral contour. In the fetal pigs of the stages specified above, certain furrows showed a typical adult relationship with the covering arachnoid membrane and lining pia, the intervening space being traversed bj' definite arachnoid trabeculae. Other of the sulci were filled with an almost emb^^•onic tj-pe of mesench}Tne — a loose meshwork of cytoplasmic processes containing rather small oval nuclei. The explanation of this embryonic type of tissue seems to be that it occurs in the newly developing sulci and that some time must elapse in this formation before the tissue fully differentiates into the adult arachnoid membrane. Strangely enough, a similar collection of an embryonic type of tissue is sometimes met with, in these stages, between the two hemispheres.


The general process, then, of the formation of the arachnoidea involves both a breaking-down (or thinning-out) of the mesenchymal spaces and a condensation of the cells. The first of these processes results in the transformation of the interstices of the periaxial mesenchyme into the larger subarachnoid spaces, divided off by arachnoid trabeculae; the second finds its final accomplishment in the development of the outer arachnoid membrane which, covered with mesothelial cells, forms the inner surface of the subdural space. The transformation begins in the basilar regions of the cranium and spreads upward over the hemispheres.


THE CIRCULATION OF FLUID THROUGH THE SUBARACHNOID SPACES.


In view of the processes of differentiation involved in the formation of the arachnoidea and the subarachnoid spaces, the circulation of fluid through this pecuhar membrane must be considered. It seems important to ascertain, if jjossible, the relationships between the beginning of the passage of the cerebro-spinal fluid and the onset of the histological changes.


The conceptions of the development of the circulation of the cerebro-spinal fluid which are presented in this conmiunication are dependent, in large measure, upon the results of the replacement of the fluid, in living embryos, by the ferrocyanide solution. Additional evidence was obtained from the identification of albuminous coagula in the periaxial tissues. The correlation of these findings with the development of the chorioid plexuses and with the results of injections under low I^ressures, from a syringe and so forth, gave evidence of their correctness.


The differentiation of the mesenchyme into arachnoid membrane may be said to keep pace with the establishment of the periaxial channels for the cerebro-spinal fluid. In the main, the passage of this fluid into the undifferentiated mesenchyme about the nervous system precedes the process of histological change. This phenomenon is shown in figure 14, from a pig embryo of 18 mm. The replaced fluid is seen passing out into the mesenchyni(^ through the two membranous areas in the roof of the fourth ventricle. The mesenchyme at this stage has already differentiated somewhat, but hardly in proportion to the length of time during which the fluid has been passing into the space.


There are several features of interest in the course of the fluid through the periaxial spaces. In sections of embryos in which the cerebro-spinal fluid has been replaced by a foreign solution the granules of the precipitated salts may be identified in the periaxial mesenchyme in situations corresponding exactly to the extent of th(! spread shown in the cleared specimens (figs. 1 to 9). The exact location of the prussian-blue granules is of importance in this connection, as the exact form and distribution of the periaxial spaces and their relation to the adult subarachnoid spaces may thus be determined.


Kxamiiiation of serial sections from an embryo in which the embryonic ventricular fluid has been replaced by the ferrocyanide will reveal, if the embryo exceeded 14 mm. in length, granules of prussian-blue in the peribulbar mesenchj'me (fig. 14). The granule.s are not found in any cell-bodies in this tissue; they are made out, in large measure, adhering to the mesenchymal cell-processes or lying free in the mesenchymal interstices. The granules do not penetrate the pia mater or the dura mater, a finding which will be discussed more fully in the sections dealing with these membranes. Everywhere the transit of fluid into the nervous tissue seems to be prohibited by the pia; in some areas, however, the outer condensation of mesenchyme to form the dura-periosteum has not j'et occurred. This is shown particularly well in the region of the roof of the fourth ventricle (fig. 18), where the epidermis offers the only barrier to the passage of fluid from the pericerebral spaces.


In the earlier stages in which the phenomenon of fluid passage about the central nervous system may be observed, the outer layer of the arachnoid is not at all differentiated. Here the barrier to the fluid is the blastemal condensation of mesenchyme (fig. 16). In the later stages, when the outer layer of the arachnoid is beginning to appear as a mesenchymal thickening, the fluid (as indicated by the precipitated prussian-blue) is confined strictly within the immature arachnoid membrane.


The course, then, of the fluid which has replaced the cerebro-spinal fluid in the embryo follows that of the aduh cerebro-spinal fluid (as shown by the resultant blue granules). It is everywhere contained within spaces which topographicaUj' and embryologically correspond to the subarachnoid spaces in the adult. The spread of the replaced solution from the embryonic ventricle into the peribulbar tissue is analogous in every way to the passage of cerebro-spinal fluid from the fourth ventricle of the adult into subarachnoid spaces.


VIII. A Consideration Of The Embryonic Pia Mater

Our present conceptions of the embryolog}' of the pia mater are largely due to the work of His^^s) and of Kolliker^^D. who first firmly established the idea that this inner leptomeninx was mesodermal in origin. While generally accepted (Farrar'i^*). this view has not been widely referred to in the literature; but the absence from all embryologies of any information concerning the development of the meninges is quite striking and it does not seem strange, therefore, that our information regarding the pia mater has not advanced in keeping wnth our knowledge of the embryology of other structures of the body. In the present section of this conmiunication it is purposed to present merely a general consideration of the process by which the pia mater is formed and to point out some of its functional characteristics, especially in regard to the fluid channels.


The term pia mater is accepted throughout this article as designating solely the cellular membrane which adheres closely to the outer surface of the nervous sj-stem, but it is in direct connection with the arachnoidal trabecular which traverse the subarachnoid space Whether the two membranes should be considered together as the pia-arachnoid or as the leptomeninx is a question in regard to which there is some disagreement ; it will suffice to consider the pia as a separate membrane.


THE GENERAL HISTOLOGY OF THE PIA MATER.


The findings in this mvestigation are wholly in accord with the conclusions of His(25), of Kolliker('i), and of Farrar^^^), that the pia mater is derived from the middle germ-layer. In the earliest stages the mesenchymal elements may be made out adhering to the outer i)ortion of the primitive nervous system. In the course of growth these cells are grouped about the mantle zone of the spinal cord in a rather dense laj'er, two or possibly three cells in thickness, with the tyjjical oval nuclei of the mesench5Tnal elements. Certain stages of this process may be made out in the figures in this paper. Thus, in a human embryo of 4 mm. (No. 836 of the Carnegie collection) the mesenchymal elements form a definite layer around the neural axis (fig. 41). The nuclei are oval in shape, possessing a moderate amount of chromatin, and are found in a layer two cells in thickness. This membrane, with its fairly scant cytoplasm, is sharply differentiated by its existence between two layers, in one of which nuclei are wanting, and in the other somewhat widely separated — the mantle zone of the spinal cord and the periaxial mesenchyme.


This typical arrangement of the mesenchj'mal elements about the cerebro-spinal axis holds in almost unchanged form throughout the whole embryonic growth. Thus, about the nervous tissue in figures 48 and 52 (from human embryos of 7 and 9 mm., respectively) the same condensation of the mesenchymal elements to form the pia mater are made out. This ajipearance is so familiar that further description in the later stages seems needless, but certain characters of this embryonic arrangement seem to require comment.


The general appearance of the pial layer is greatly altered by the early formation of the capillary blood plexus about the nervous sj'stem. This plexus tends to render the pial tissue more cellular, on first microscopic examination, as the endothehal channels branch greatly outside of the nervous tissue in this mesenchymal pia. The general character of the pial layer, however, as a membrane with prominent nuclei and scanty protoplasm, is not altered at all by the vascular plexuses.


The ultimate fate of these undifferentiated mesenchjinal elements forming this initial \nti\ condensation is a gradual transformation of the cells into ver}' low cuboidal mesothelial elements constituting the adult pia. The transformation concerns not only the differentiation of the cells but also a rearrangement so that the original layer of two or more cells in thickness becomes finally of but a single cell in thickness. The jjrocess, in a way similar to the development of the subarachnoid spaces, begins in the basilar portions and spreads upward; the process, hence, may often be studied in a single suitable sjiecimen.


More imjjortant, for our consideration, is the i)eculiar relationship of the pia mater to the roof of the foiutli ventricle, and in particular to the two area^ membranaceje. In this situation, in place of the slight mesenchymal condensation which characterizes the jjia, and which ^Minot*"^ pictures in his figure 114, the mesenchyme seems altered. The condensation to form the pia, which takes place in other situations about the true nervous tissue, has not here occurred. This absence of the typical pial arnuigenit-nt may be noted even in very small embryos — those in which the roof of the fourth ventricle is composed of the many-layered, epitheUal-like cells. This is well shown in a photomicrograjih (fi}^. 53) from an injected human embryo of 9 mm. (Xo. 721) of the Carnegie collection. Likewi.sc, in this region in a [)ig embryo of 8 mm. (fig. 25), the same absence of a real pial condensation may be made out. But this peculiarity of the pia is most striking at the period of maximal differentiation of the superior membranous area in the rhombic roof. In figures 37 and 43, photomicrographs from pig embryos of this stage, the mesenchymal condensation, augmented by some vascular endothelium, is shown in adhesion to the ependyma on both sides of the membranous area; but directly behind the differentiated cells of the area membranacea evidence of a condensation of mesenchyme is whollj' lacking, even though both specimens show vascular channels in close appro.ximation. Similarly, in a human embryo of 14 mm. (No. 144, Carnegie collection) a total lack of the true pial thickening is to be observed (fig. 57).


Quite similar is the failure of a pial thickening about the inferior membranous area. This can be made out in figures 83 and 87, from human embryos in which the process of differentiation of the area is proceeding. In later stages of the formation of the area membranacea inferior, the marked absence of a true pial condensation in the mesenchyme in this region is noted in figure 75 (a specimen from a fetal pig of 23 mm.) But this apparent failure to form the typical mesenchymal condensation of the pia mater in certain areas in the roof of the fourth ventricle must not be construed as indicating an absence of pia mater. Such does not seem to be the case here, for in the later stages of the formation of the cisterna cerebellomedullaris the area membranacea inferior is found entirely unsupported, except for a layer of mesenchymal cells. This is shown in figures 77 and 79, both taken from fetal pigs of 32 mm. This mesenchymal laj'er must be considered as pia mater apparently modified for a specific purpose.


The general process, then, of formation of the pia mater concerns a condensation of mesenchymal elements to form an embryonic membrane about the central nervous system. From its earliest beginning very slight modification is needed to reduce it finally to the histological character of the adult membrane. The general process holds, except in the regions of the area? membranaceic in the roof of the fourth ventricle; here, apparently, a modification of the pia for a specific purpose, involving an absence of the primary pial condensation, takes place.


THE RELATION OF THE PIA MATER TO THE FLUID CHANNELS.


The cerebro-spinal fluid in its normal pathways comes everywhere into contact with the pia mater, which serves as the inner retainer for the subarachnoid space; therefore the functional relation of this membrane to the fluid which bathes it becomes of interest. To some degree the results of the experiments recorded in the earlier portions of this paper throw light upon the relation of the pia mater to the circulating fluid. The most important question in this connection is naturally that dealing with the possible penetration of the normal fluid through this embryonic membrane. In this regard the findings in replacement experiments with ferrocyanide solution serve to elucidate the problem. These observations give no evidence of any penetration of the pia mater by the fluid. This is well brought out in figures 14 and 18. In every respect (as demonstrated by numerous experiments of this type in pig embryos of varying lengths) the pia mater is wholly impenetrable to true solutions of foreign salts when injected so that the normal tension is not altered. The whole subarachnoid sjiace may, in such an experiment, be filled with the prussianblue, but none of these granules are found within the cells of the pia mater or in any layer between these cells and the nervous system. Evidence that the fluid has bathed the outer pial cells is afforded by the adhesion of granules of prussian-blue to the outer cytoplasmic borders of the cells.


Likewise the cells comprising the embryonic pia have been found to be impenetrable to true solutions (ferrocj'anide) when injected under varying pressures from a syringe. In these cases, rupture of the roof of the fourth ventricle or of the infundibulum may be produced by great pressure, without causing any of the fluid to penetrate the intact layer of the pia mater. The same result is obtained when india ink is substituted for the true solution.


The pia mater, then, even in its embryonic form, serves as an efficient fluidbarrier. This is demonstrated, in regard to the adult pia mater, in the report^^^) of the observations made on adult cats, dogs, and monkej's. But the barrier which the pia offers to the entrance of fluid from without exists also for fluid coming in the reverse direction. This is shown by the well-known phenomenon of the so-called subpial extravasation, which occurs in blood vascular injections when the injections are continued for too long a time at too high a pressure. The perforating vessels in such cases rupture as they enter the nervous system, and the injection mass spreads extensivelj' beneath the pia, stripping it away from the nervous tissue. Of interest in this discussion is the fact that the injection mass in these extravasations does not rupture the pia, which seemingly is an equallj' efficient fluid barrier to pressure exerted on it from within. Similar subpial spreads of the injection fluids have been observed in the course of this work. These extravasations resulted from the rupture of the whole nervous tissue from within, particularly in the region of the infundibulum, when the inj(>ction was made into the ventricular system under excessive pressure. In this respect, too, the i)ia seems to be wholly efficient as a retainer for true solutions or for granular suspensions. It is realized that the embryonic pia mater will not resist the passage of fluids through it under the highest pressures afforded by the syringe, but the membrane serves as an efficient barrier for all pressures such as are employed in careful anatomic injections.


With this conception of the impenetrabihty of the pia mater to fluids under ordinary pressures, it does not seem strange that there is a variation in the process of formation of the pia mater in the region of the roof of the fourth ventricle. It has been shown in the foregoing paragraj^hs that the phenomenon of mesenchymal condensation which results in the formation of pia elsewhere does not occur in the region of the two area? membranacefe. In view of the passage of cerebro-spinal fluid through these two membranous areas, the pia mater must necessarily be altered in these places. For were it not adapted to the jjurpose of affording fluid passage the cerebro-spinal fluid would, in its course from the ventricle to the subarachnoid space, form a subpial extravasation. It would seem that this modification of the pia is designed to meet the particular need and function of this region.


The Adhesion of the Pia Mater to the Cerebral Tissue

It is a well-known fact in embryology that tlie pia mater and the peria.\ial mesenchyme in poorly dehydrated specimens split away from nervous tissue, but in adult preparations, if the meninges and brain are dehydrated in a block, the separation of the tissues occurs between the dura and the arachnoid, or (in more exceptional instances) the dura and arachnoid come away, leaving the pial layer closely applied to the cortical tissue. It is quite difficult in any adult mammal to separate the pia from the brain tissue. Realization of these peculiarities in the degree of adhesion of the pia led to an attempt to ascertain what structures were involved in the attachment of this mesodermal layer to the epidermal nervous system. The results of this attempt add nothing to the ultimate solution of the problem, but are perhaps of sufficient interest to justify brief presentation.


Two theories in explanation of this adhesion of the pia immediately suggested themselves. One of these concernod a possible growth into the pia of neuroglial elements, causing an intimate association between the pia and the cerebral cortex. Our findings in reference to the neuroglial outgrowth in fetal pigs gave no reUable basis for the assumption. The second theory dealt with a diminution in the elasticity of the walls of the perforating blood-vessels which supply the nervous system. The early embryonic vessels, with walls comjjosed solelj- of endothelium, when subjected to the distortions of poor dehydration, might possibly offer less resistance to the separation, so that the pia would come awaj' from the nervous tissue. In the later stages, however, the thicker-walled perforating vessels w^ould naturally oppose such a cleavage, so that the pia would remain adhering to the cortical tissue. This theory is also purely an hj^Dothesis, although it does not seem unlikely, especially if one takes into account a possible connection of the pia with the perivascular system. In examining blocks of the meninges and brain tissue taken together it was found that the pia mater separated cleanly from the nervous tissue in fetal pigs 15 cm. in length. Beyond this stage the arachnoid might remain in adhesion to the dura, but in such cases there was always found a layer of cells on the outer side of the cortical tissue, constituting a true pia mater.


IX. The Development Of The Cranial Dura Mater

The dura mater, like the two other meninges, is derived from the mesenchj-me about the central nervous system. The researches of Sterzi'"' on the comparative anatomj" of the meninges furnish additional evidence for this conception in the higher mammals. The origin of the pachymeninx from the middle germ-layer is now well established. But there is lacking in the literature a comprehensive account of the formation of this fibrous envelope. The gross generaUties of the process are given in pari, liut there is an almost total absence of the more intimate details of the transformation. One of the most essential points in the process concerns the relationship of the dura to the bony coverings of the cerebro-spinal axis. Does the adult dura serve as the periosteum of the bony skull? In the standard text-books of anatomy the adult human dura is described as being composed of two layers. In the skull these layers split, to comprise the walls of the great venous smuses. The outer layer of the dura serves as the periosteum for the bony skull, but below the foramen magnum the two layers separate to inclose the epidural space. The outer dural layer in this spinal region adheres to the inner surface of the bony vertebral column, where it functions as the periosteum; the inner layer here becomes the spinal dura.


In this account of the adult dura mater there is indicated a very suggestive periosteal relationship which implies an embryological basis for the disposition of the two laj-ers of the membrane. It must be granted, however, that this division of the cranial pachymeninx into two layers is quite arbitrary; there is nothing in the general histology of the fibrous covering to suggest such a halving except its division about the sinuses and its spinal relationships.


THE GENERAL PROCESS OF THE FORMATION OF CRANIAL DURA.


The first evidence of the development of the pachymeninx is found in the basilar region of the skull, where the mesenchyme thickens, to form eventually the bony covering of the brain. This thickening of the mesenchymal elements results not only in the formation of the chondro-cranium, but also in the final formation of the bony skull and possibly its internal periosteum and dura. In the process of differentiation the condensation of mesenchyme in the early stages gives no index of the varied character of the resultant tissues, so that, in the first place, the study of the process was necessarily related to the more adult stages. In this paper, however, the whole history of the dura will be detailed chronologically, beginning with the earliest stages.


Bardeen(2) has given data on the first appearance of the mesenchymal condensations which go to form the blastemal phenomena in both the cranial and spinal regions. The blastemal vertebra? become fairly well differentiated in human embryos during the first month of intra-uterine growth. At the end of the first month, in the occijiital region, three fairly well-marked occii)ital myotomes may be made out; these afterwards disappear. "During the early part of the second month the membranous anlage of the skull becomes extensively developed. The roof of the cranial cavity is formed by a dense membranous layer, which fu'st becomes marked at the side of the head in embryos from 9 to 11 millimeters in length" (Bardeen).


These evidences of a primary mesenchymal condensation about the central nervous system are concerned in the problem of the differentiation of the dura only in so far as they indicate the onset of the process which will give rise to the bone and possibly the periosteum — a part of the dura about the cerebro-sjjinal axis. Gaupp^'^) has already pointed out that this cranial l)lastemal condensation gives rise to these adjacent Init wholly different structures. These cranial mesenchymal condensations persist in simple form until after the cerebro-spinal fluid begins to fill its extraventricular bed; then, within a short time, the tissue becomes transformed by the development within it of cartilage, so that in the human embrj^o the caudal half of the chondro-cranium forms a ring of cartilage about the posterior portion of the brain. On the inner side of this ring of cartilage the mesenchyme later shows a marked condensation in the midst of the rarefied perimedullary tissues. In this layer the nuclei soon become fewer in number and the cytoplasmic structures fibrillar, the whole resulting ulthnately in the formation of the fibrous adult dura. The mesenchymal condensations in the regions of the skull, where membranous bone formation holds, go directly into a membrane of fibrous tissue, in the outer portions of which bone is laid do\\Ti. The details of these processes will now be taken up.


In figures 30 and 32, photomicrographs from pig embryos of 13 and 14 mm., respectively, the well-established vertebral differentiations and the now poorly differentiated base of the skull are shown. From this stage upward the mesenchymal condensation in the head region proceeds rapidly. Thus at a stage of 17 mm. in the human embryo (fig. 60) the ventral portion of the vertebral canal has become cartilaginous, while the base of the skull has also undergone the chondrogenous transformation in its mere posterior portions. But of especial interest in our problem is the line of mesenchjTnal condensation, which may now be traced whollj^ around the brain-stem and hemispheres (fig. 60). The nature of this condensation is well shown in figure 61, an enlargement of the squared area of figure 60. The mesenchymal nuclei have become closely packed and rather sharply differentiated from the looser mesenchj^me which in part goes to form the arachnoidea. Figure 59 similarlj' shows this condensation proceeding upward to the vault.


Examined in another plane, the process of mesenchymal condensation seems to proceed much more rapidly in the posterior than in the anterior region. This is brought out in a transverse section of a human embryo of 18 mm. (fig. 62). Here the condensation is much more extreme about the medulla and roof of the fourth ventricle than in the more anterior parts of the mesencephalon. The same general appearance, typical of this stage, may be made out in figures 56 and 57 from a human embryo of 14 mm.* (No. 144, Carnegie collection). In the slightly larger stages the process of mesenchj^mal condensation about the nervous system becomes rapidly more marked. This increase in the number of cells comprising the denser membrane is shown in figures 64 and 65, photomicrographs of embryo No. 460 (21 mm.).


The degree of condensation of the mesenchyme in the various stages of the human embryo is followed quite closely in the pig embryo. The comparative degree of differentiation coincides within a millimeter or two. Thus, in a section from a pig embryo of 19 mm. (fig. 38), the degree of condensation about the roof of the fourth ventricle is practically similar to that in human embryos of the same length.


The phenomena just commented upon represent the stages concerned in the formation merely of a cranial blastema and are related to the formation of the dura only so far as it is out of this mesenchymal condensation that the periosteal portion of the pachymeninx may be derived. The degree of condensation referred to in the figure.s has been solely of the blastemal type, but in some of the specimens this simple condensation is seen only in the more cephalic portions of the cranium. Thus, in the figures (64 and 60) taken from embryo 460, the mesenchymal condensation is still of the simjile undifferentiated type, whereas in this same embryo the more caudal sections show a chondro-cranium which is well develojied. The i)rocess of formation of the cranial dura, then, is one which begins in the basilar j^ortions of the cranium and proceeds from these points into the region of the calvarium. In general, all of the phases of this transformation into dura may be found in one specimen of sufficient and suitable size, the basilar differentiation re]:)resenting the advanced stages, while the steps in the differentiation are found in the areas nearer the vertex.


  • Measured on slide after sectioning.


It is quite difficult to decide exactly what importance the primary condensation of mesenchyme maintains in the formation of the dura, because, coincident with the chondrification of the blastema, there occurs another condensation which forms the line of division between the inner surface of the dura mater and the outer arachnoid membrane. The first evidence of this secondary perimedullary condensation is found in pig embryos of about 17 mm. In these specimens, in the narrow space formed by the mesencephalic flexure, mesenchymal cells collect together in the form of a fairly definite membrane. After its primary beginning in this area, the narrow line of its thickening may be traced to the basis cranii in embryos a little larger. In slightly older stages this secondary fine of condensation is found to be fairly extensive throughout the area between the middle and posterior cranial chambers.


At a stage of 20 to 21 mm. the whole basilar portion of the cranium shows evidence of this secondary line of condensation lying between the pia mater and the cartilaginous skull. The condensation occurs in the outer portions of the loose tissue which, as shown in a foregoing section (No. vii) becomes the subarachnoid trabeculae. The line of condensation is not broad on section; it comprises a cell-layer from three to six cells in thickness. Between this cellular border and the cartilaginous skull the tissue rapidly differentiates (a process seemingly sj^nchronous with the develojiment of this membrane). This tissue, which maintains dural relationships, is far more cellular and compact than the original perimedullary mesenchyme. Even without the rather dense line of division in the mesenchymal ti.ssue, the dural structure can be easily outlined by its characteristic apjiearanco.


The original dural condensation between the two wings of nervous tissue which unite in the me8encei)halic flexure can be traced in slightly later stages around into the tentorium cerebelli. This structure develops as a wholly similar mesenchymal thickening in the midst of the jK-rimeduliary mesenchyme. The tentorium consists in these embryos of 20 to 2") nun. of two thin lateral plates which widen at their cranial attaclunents into prismatic areas. These areas, which finally lodge, in the two layers f)f dura, the sinus transversus, arc characterized by the same dense tyjie of mesenchyme. The jx-ripheral edges of the prismatic jwrtion of the tentorium sjjreads caudalwards as a definite line of condensation. In the earlier stages this line becomes indefinite as it extends from its teiitori:il attachment, but finnllv a similar line of condensation about the whole posterior chamber may be made out. This lies within the area of the cartilaginous skull and bounds the subarachnoid spaces.


This same process of formation of dura holds for the formation of the basilar dura in the more anterior portions of the cranium. The appearance of the .secondary zone, narrow and rather dense, may be made out inclosing the more cellular mesenchyme w^hich extends to the cartilaginous skull. The same process also endures for the formation of the dura of the calvarium, but here the addition of tissue from the undifferentiated mesenchyme is undoubtedly very small in amount. This will be discussed in a later paragraph. The various stages in the formation of this secondary condensation which goes to form the major portion of the dura may be fairly well studied in any one embryo of suitable stage, because the process, as pointed out above, begins in the basilar portion of the cranium and extends upward. Likewise, the condensations directly beneath the region of the dorsal membrane are delayed as compared to those of the lateral regions.


Some of the phenomena shown in the formation of the dura mater are illustrated in figures 46, 76, 77, 78, 79, and 94. Throughout these figures the letters dmc refer to the Une of the secondary mesenchymal condensation, which borders internallj^ the dura and which goes to form the outer membrane of the arachnoidea.


In figure 46, a photomicrograph of a pig embryo of 32 mm., the dura mater (dmc) is shown as a somewhat condensed tissue separated a sUght distance from the chondro-cranium. On the basilar surface, the inner line of dural tissue is quite remote from the inner surface of the basioccipital. Tracing this line of condensation forward, it is soon seen to merge more closely with the perichondrium* of the basioccipital. More anteriorly it again leaves the occij^ital plate and after a brief interval it fuses with the temporal perichondrium. Continuing slightlj^ more anteriorly the dural process toward the mesencephalic angle maj' be made out; this appears as a doubled membrane at its basal attachment. In its further prolongations the dural surface is at times a distinct structure; at other times it is completely fused with the perichondrium.


Posteriorly, in this figure 46, the line of dural condensation (incorporated also with the outer arachnoidea) may be traced upward around the cisterna cerebeUomedulla


ris. This hne of condensation is seen to lose its definitive character as it curves inward toward the chorioid plexus of the fourth ventricle — a phenomenon shown particularly well in figures 77 and 79, taken from the same pig embryo of 32 mm. The dura in this termination maj' be said to be in its formative stage; but dorsally, over the mesencephalon, the inner surface of the dura again becomes a definite membrane, as shown in figures 76 and 78. In the latter figure it is shown inclosing a wide mesh of dural vessels, between the arachnoidal surface and the membranous skull. Anteriorly, again, it seems to lose its definite hne of condensation.


•The term "perichondi.L n" is used tliroiighout this paper to designate only the ver>' dense cellular line delimiting the edge of the cartilngo. This dense zone is composed of the nuclei of the cartilage, crowded together, and represents probably some phenomenon of the growth or resorption of the cartilage. In a much broader sense, the whole dural tissue, lying between the line of secondarj- condensation and the cartilaginous border, could be termed "perichondritmi," as it probably represents the sole internal membrane which could be stripped from the cartilage.


Quite similar pictures are obtained regarding the dura mater in the human embryo. The relationships of the dura to the cisterna cerebello-medulla


ris are shown in figure 94, a photomicrograph of a human fetus of 35 mm. (No. 199 in the collection of the Carnegie Institution). In this rejiroduction the line of secondary mesenchymal condensation (representing the outer membrane of the arachnoidea and the inner surface of the dura) becomes widely separated from the occipitale superius in its superior portion.


In a fetal pig of 8 cm. the same general arrangements of the dura mater could be made out. The inner surface of the dura was in places still fused with, the outer arachnoid membrane, but in other places the areas of attachment were lacking, so that a true separation of arachnoid from dura had taken place. Along the ])eripheral points of the tentorium the dura and arachnoid were still closelj' a{)plied to each other. The dura itself was of the same cellular, rather loose tissue, with a dense inner surface. In i)laces, as described in the younger stages, the dural tissue was incorporated with the definitive perichondrium over certain cartilages or even over parts of the same structure. In other places a definitive perichondrium may be wholly lacking; in these areas the indefinite cartilaginous border gradually merges into the dura. In still other situations an intermediate arrangement of dura and perichondrium exists, where the cartilage is bounded bj' a somewhat condensed but not fully developed perichondrium which is continuous with the dura. Everywhere in the membranous sutures between the cranial cartilages or bones, the dura bridges the gap as a loose, cellular tissue. Over the calvarium the dura appears solely as a dense, rather fibrous membrane which is incorporated with and serves as the inner periosteum. This dura over the hemispheres is continuous with the fibrous sutures of the cranial vault.


The findings in a fetal pig of 98 mm. were not dissimilar to those just recorded. The dura was everywhere quite well developed, a rather loose cellular tissue except over the hemispheres, where it showed a more fibrous character. In the region of the occipito-atlantoid ligament the dura was fused with the ligamentous tissue, while above (over the occipitale superius) the dura became a distinct, thick cellular layer. The structure of the tentorium was wholly similar to the occijiital dura. In the basis cranii there are areas in which the dura is wholly fused with the periosteum or jierichondrium; in other areas it bridges the sutures or exists as a definite membrane on the inner surface of a definite perichondrium.


The dura mater in a fetal pig of 15 cm. did not vary greatly from those larger stages already tlescribed. The tissue, however, had become somewhat more fibroiis. The prismatic attachment of the tentorium was no longer as large proportionately, but the dura lining the occipitale superius remained a thick bulbous swelling on the dorsal surface. Hut most striking of all the features in the specimen was the very dense fusion of the dura of the calvarium with the fibrous sutures of the cranium. No line of demarcation betwecm dura and fibrous suture could be made out; the two fibrous layers are anatomically one structure.


The falx cerebri forms in the pig and human embrj-o by a process similar to that of the inner portion of the dura mater. In the sulcus between the two cerebral hemispheres the mesenchyme remains undifferentiated until quite late; then there appears in the posterior portion a narrow zone of condensation which soon presents two lateral surfaces separated by a layer of rather loose cellular tissue, similar in all regards to the dural tissue already described. This zone of condensation spreads forward to comprise the whole falx. The double surfaces of this membrane finally separate into two parts, forming the outer surface of the arachnoidea and the inner surface of the falx. At the cranial attachment of the fabc, the loose tissue forms a prismatic base, containing the suj)erior sagittal sinus and spreading laterally over the denser dura of the calvarium. The whole appearance of this region, which will again be referred to, is that the falx has been added onto the dura of the vertex. Its time of initial appearance is later than that of the rest of the cranial dura and there is apparently no additional acceleration of development. Hence the dural tissue in the fabc cerebri presents, in appropriate stages, a more immature type of differentiation than does the adjoining dura.


The process of the formation of the dura is not wholly a simple one due to the relation of the adult dura to, or its function as, the inner periosteum of the skull. In the figures already referred to, the almost complete fusion in some areas of the inner Une of dural condensation with the perichondrium has been commented upon. In other situations definite separations of the inner dural surface from the perichondrium occurred; in still other regions no perichondrium could be made out as a definite membrane. These differences in relationships of the dural tissue to the line of the perichondrium can not at present be wholly explained, but some indication of the meaning of the process can be given.


Out of the original cranial blastema, as described by Gaupp'^i^\ there develops the cartilaginous and bony skull, the periosteum, and the dura. But the observ^ations recorded above indicate that by far the major portion of the dura is formed by a secondarj^ mesenchymal condensation, which was indicated bj' a thin zone of more condensed cells on its inner border. This inner zone ultimately separated to form the inner surface of the dura and the outer membrane of the arachnoidea. The tissue included between this inner line of condensation and the cranial wall gradually differentiated into a more condensed but still a loose cellular tissue and finally became a fibrous dura.


In all cases the dural tissue extends from the inner line of condensation to the cranial blastema, to the perichondrium, or to the cartilage of the skull. The presence of a definitive perichondrium can not at present be explained, but apparently the perichondrium is manufactured by the cells of the original cranial blastema and not by the dural tissue which lies in approximation to it. When a definite perichondrium is found, it seems quite uninfluenced by the dura; at other times a fusion of an indefinite perichondrium with the dura seems to occur. The fusion of the perichondrium with the dural tissue derived from the secondary mesenchjinal condensation may occur, so that the small outer portion of the dura may be derived from this laj^er. The findings, however, in this investigation, are against any addition of perichondrium to the dural tissue; histologically, a definitive perichondrium is a membrane entirely apart from the dural condensation.


Over the cerebral hemispheres the dura of the cranial vault offers more difficulties of study than does that of the basilar regions. With the formation of a blastcmal condensation over the whole vertex — an extension of the dorsal membrane to form the membranous skull — there occurs very quickly a condensation to form the dura. This condensation may be first detected as a continuation anteriorl)'of the leaflet of the tentorium cerebelli, which stretches forward from the prismatic zone of the tentorial attachment. This zone of condensation is wholly similar to the narrow line of the mesenchymal thickening which was found in the more basilar regions of the skull. This zone of condensation occurs just wathin the cranial blastema and may be traced upward over the mesencephalon and laterally around the rapidly enlarging hemispheres. As the distance from the cerebellar attachment increases, the zone tends to approach the blastema, except in those regions in which the precursors of the dural veins occur. In such a situation this inner dural zone swings inward to encompass the vessels. Between this inner line of the dura (representing also the outer surface of the arachnoid) the same rather loose cellular tissue exists.


From the fabc cerebri a zone of dural condensation in the mesenchyme spreads laterally also; this gradually may be traced anteriorly and laterally until fusion with the similar lines of condensation from the basis cranii and the prismatic zone of the tentorium are reached. The condensation connected with the falx cerebri, however, is not an extensive process, the greater part of the hemispheres being covered by the development from the basis cranii and from the tentorium. It must be understood, however, that there is no active migration of this line of condensation, for the whole process is a development in situ. The appearance of an active extension is derived solely from the study of various stages and the increased area of condensation appears as an increment which has developed at the terminal points of the previous condensation.


The amount of dural tissue delimited in the mesenchyme by the secondary zone of condensation is not great in the region of the vertex. It is a thin layer which fuses to the inner surface of the cranial blastema. At the stage of this fusion the blastema has become somewhat fibrous and it constitutes the membranous skull. In this fibrous tissue (the union of the blastema and the dura) bone is deposited, but only in the outer layers. The phenomenon is easily studied in any suitable stage, for the sutures between the flat cranial bones remain incorporated with the inner memliranf!— the dura which includes the periosteum. Hence, over the cranial vault, the dura and periosteum become incorporated as a single membrane; this serves as the membranous .skull, into the outer layer of which bone is deposited.


In the basis cranii, as soon as ossification of the cartilaginous skull takes place, tin; dura becomes inc()r|)orated as the periosteum in a manner similar to that which takes i)lace in the cranial vault. While no definite relationship of dura to the peri-chondrium could bo made out in the earlier stages, the later function of the dura as the inner cranial periosteum is Cjuite obvious. Thus the adult relationships of the dura are obtained. But it is quite difficult to decide to what extent the dura (or internal cranial periosteum) is derived from the primary cranial blastema. It seems probable that this blastemal condensation, in its final resolution into bone, may contribute, in the form of a periosteal element, somewhat to the formation of the dura. Such an addition is verj' difficult of verification; certainly the greater part of the dura is derived by the secondary condensation from the perimedullary mesenchyme.


Before giving details of the fibrosis of the dura, it may perhaps be interesting to point out a peculiarity of the primarj' cranial blastema, which does not seem to be connected directly with the formation of the dura. This concerns the tendency of the membranous skull to form more than one layer in its original zone of condensation. In certain areas, as in figure 64, from a human embrj^o of 21 mm., the dorsal membrane is .shown split into two layers. Somewhat similar to this is the occurrence of two zones in the cranial blastema of a pig embryo of 23 mm. (figs. 22 and 101). Inthis latter figure a less cellular outer layer and a more cellular inner layer are seen. Neither of these have particular significance in the formation of the meninges, although the inner layer in early stages actively functions as a fluid retainer.


The question of the development of fibrous tissue in the dura mater in the course of its development requires consideration here. This phase of the problem concerning the formation of the pachymeninx has been followed, in this stud}', in the dura of the vertex about the sinus sagittalis superior. The tissue was removed in blocks, including the meninges and cortex cerebri, and was then sectioned in the coronal plane. For the most part the deposition of fibrous tissue was studied in sections stained with hematoxyhn and eosin; the findings were controlled by treating other sections from the same blocks with IMallory's connective-tissue stain. In this way the general histogenesis of the dural tissue could be well investigated.


Sections from such a block from a human fetus of 76 mm. (Xo. 1134, Carnegie collection) showed the dura to be composed of fibrous tissue everywhere except in the region of the great sagittal sinus. About this sinus an immature, almost embryonic, tj'pe of loose mj-xomatous tissue was observed. The fibrous tissue comprising the dura elsewhere is of a quite cellular, somewhat immature type of white connective tissue, with a considerable number of true fibrils. A wholly similar picture is found in a section, stained by Mallory's method, of a block from a fetal pig of SO mm. (fig. 104). Unfortunately the cellular character of the fibrous dura is not brought out, but the photomicrograph shows well the avoidance of the lateral walls of the sinus by the process of fibrosis. The more embryonic type of tissue in the region between the hemispheres is also well presented.


The dura mater of a human fetus of 100 mm. (No. 928-E, Carnegie collection) possesses fewer nuclei in a given area than does the dura from the specimen of 76 mm. (No. 1134). The tissue is fibrous, except in the immediate region of the sinus sagittaUs superior; but interspersed among the connective-tissue fibrils are many stellate or spindle-like nuclei, greatly exceeding in number the nuclei found in the dense dura of the adult. Bone is being laid down in the outer portion of this dura where it merges \\dth the membranous skull. The lateral walls of the great sinus are still free from fibrillar depositions. A somewhat analogous picture is afforded b}' a photomicrograph of a specimen stained after IMallory's method, from a fetal pig of the same length (fig. 105). In this specimen the outer portion of the dura, incorporated into a part of the membranous skull, is quite dense with the fibrous tissue; about the superior sinus, however, the decrease in the amount of fibrous tissue is very striking. The falx is beginning to exhibit a fair degree of fibrillar structure; it forms a definite division between the two hemispheres.


In the larger fetuses, above 100 mm. in length, the process of formation in the dura of denser and denser connective tissue proceeds rather slowly. It is realized, however, that this fibrous transformation in fetuses of 10 cm. is veiy extensive, the region about the sinuses alone remaining comparatively free from the development of the fibrils. The chief difference between the dura of this stage and the dura of the adult is a greater number of cell-nuclei in the fetal membrane. It is well, then, to consider the cellular character of the fibrous membrane and the region about the sinuses in the larger stages.


In a human fetus of 125 mm. (No. 900-H) the dura is quite fibrous, but still contains an increased number of the stellate and spindle forms of nuclei; likewise, about the superior sinus the tissue is an immature m30iomatous structure, fairly free from connective-tissue fibrils. This increased number of nuclei in the dural tissue holds also for human fetuses of 165 mm. (as in No. 745), but seems slightly decreased as compared with the smaller specimens. The lateral wall of the great sagittal sinus in this stage possesses distinct bands of white fibrils, but the tissue is much looser and more cellular than the fibrous dura over the hemispheres. These phenomena may be made out in similar stages of the fetal pig, as shown in figure 106, a photomicrograph from a specimen of 17 cm. In this specimen, treated by Mallory's stain, the superior longitudinal sinus is shown surrounded b}' a clear zone in which the deeply staining fibrils are comparatively few in number. On each lateral wall of the venous channels distinct fibrous bands may be made out, lying in the looser, more immature tissue. The lower portion of the falx has assumed quite an adult character.


Gradually the conversion of the tissue about the cerebral sinuses into the adult structure progresses. Thus, in both human and pig fetuses of 20 cm. length, the dura mater has acquired practically all of its adult features. Everywhere over the cereljral cortex the dura is characterized ])y dense layers of interlacing strands of white fibrous tissue, but the number of nuclei in the.se bundles may still be slightly greater than in the adult structure. In the more posterior regions, at this stage of 20 cm., the lateral walls of the sinus sagittalis superior are found to be completely occupied by the white fibrous tissue; in the anterior portion of the sinus much thinner tissue, resembling myxomatous structure, appears, as shown in figure 107. But in this specimen the invasion of the area about the great venous channel by fibrils has begun; isolated bundles may be made out everywhere in the lateral walls of the sinus. This freedom from connective-tissue formation does not persist, however, and the area is gradually invaded by the continued growth of the fibrils. The avoidance of the region about the sinuses by the connective-tissue resolution will be further commented on in the following subdivision of this paper.


The dura, then, develops probably first in connection with the mesenchjinal condensation which ultimately forms the bony skull and a portion of the dura (the cranial periosteum). It first becomes apparent, as a structural unit, as a more cellular layer differentiated, by a secondary condensation, out of the peria.xial mesenchyme. As the chondrogenous stage is approached it becomes differentiated as a distinct layer, maintaining varying relationships with the inner perichondrium of certain of the cranial bones. At a stage of 40 mm. m the fetal pig, the dura of the vertex may be dissected out as a distinct, somewhat fibrous laj'er. The process of fibrous-tissue transformation, however, is slow; the dura until late in fetal life shows an increased number of nuclei, as does any young connective tissue. The invasion of the region about the superior longitudinal sinus by connective-tissue fibrils is much more tardj' than is the transformation over the hemispheres.


THE SUBDURAL SPACE AND THE MESOTHELIAL LINING OF THE DURA.


The subdural space (cavum subdurale) has been the subject of controversy in regard to its role in the pathway of the cerebro-spinal fluid. Before the work of Key and Retzius^^s) ^\^q yj^^^y ^.f^g ]^q.](\ ^)jat the cerebro-spinal fluid occupied the subarachnoid space in the spinal cord, but that in the cranium the subdural space afforded an analogous pathway. This conception was largely due to the fact that, in dissection on fresh material, the dura and arachnoid in the spinal region are found to be in approximation; in the cranium the greater adhesion, bj' trabeculse, of the arachnoidea to the pia renders the freeing of the dura from the leptomeninges the simplest line of cleavage. This view was entirely disproved by the beautiful injections of Key and Retzius, who demonstrated the anatomical and physiological continuity of the subarachnoid spaces.


With the introduction of this latter view by Key and Retzius the conception of the subdural space naturally changed. These Swedish investigators demonstrated a typical mesothehal cell-lining on the inner surface of the dura, as shown by the method of silver reductions. Without an intimate connection with the true cerebro-spinal fluid, the subdural space has come to be looked upon as somewhat analogous to the serous cavities of the body. Quincke^-*^', after a subdural injection of cinnabar granules, ascertained that communications existed between the subdural and subarachnoid spaces, but only in the direction from subdural to subarachnoid. Leonard HilK^^), from the results of physiological experiments, assumed that fluid passed from the subdural to the subarachnoid space, and in the reverse direction, with great ease. The more recent investigations, however, lend evidence to the view that in the normal animal with undisturbed intracranial pressure relations the two spaces are physiologically as well as anatomically separate. The current impression that the subdural space is in manj' respects a serous cavity will probably finally have greatest support; intimate connections with the lymphatic system are, however, entirely lacking in the dura.


The development of the subdural space must necessarily follow the develoj)ment of the dura. It has been mentioned that in fetal pigs of 50 mm. the dura can be freed from the arachnoid by gross dissection, but that at this stage many areas of adhesion between the two membranes exist. Such an observation has considerable bearing on the subdural space. For in the development of this space two processes must proceed far enough to permit the separation of the dura and arachnoidea by the capillary layer of fluid. The first of these processes, in order of probable importance, concerns the condensation of mesenchj'mal cells to form the outer membrane of the arachnoidea. This thickening and resolution into a true membrane takes place in close approximation to the inner surface of the dura. The second factor concerns the covering of this inner surface of the dura with mesothelial cells.


The lining of the subdural space by mesothelial cells can be readily demonstrated on the inner surface of the dura by silver reductions, but the outer membrane of the arachnoid does not permit of a similar technique. This technical failure in regard to the outer arachnoid surface is probably to be accounted for by the dissimilarity in cell-structure in the two situations. Similar difficulties have been encountered by other observers.


In order, then, to ascertain, if possible, at what stage a really adult subdural space could be demonstrated, the inner surface of the dura from fetal pigs of various lengths was subjected to treatment with silver nitrate. After the reduction had taken place to a sufficient degree, the whole dura was washed with distilled water, stained with hematoxylin, and cleared in glycerin. The i)ictures afforded by this method were quite satisfactory, and the technical procedure was so simple and reliable that considerable faith could be placed in the absence of the intercellular reduction lines.


The smallest fetal pig in which a typical mesothelial cell-j)attern could lie demonstrated on the inner surface of the cranial dura was one of 50 mm. In this specimen the inner surface of the dura was not uniformly covered with the mesothelial cells; certain ragged areas seemed to represent the points of adhesion of the arachnoid to the dura. Figure 108 is a reproduction of a drawing made from one of the areas in this specimen where a mesothelial cell pattern could bo seen. The drawing shows many of the characteristics of mesothelial ci'll-i)ut terns of other parts of the body. The irregularities in the cell-borders, the frefiuent accumulations of the reduced silver in the cellular angles, and thegeneral cellular pattern are quite typical ; but the variation in the size of the cells, as shown in figure 108, is also somewhat different from the usual finding in the adult, where there is considerable constancy in the size of the cells. About half the cells in this fetal pig of 50 mm. are diminutive in size; the smallest are hardly a third the size of the largest. Transitions between the smallest and largest cells are al.so shown in this figure. It i.s difficult to a.scertain whether these smaller cells represent young elements which have not yet reached their maximal growth; no evidence of cellular division, as evidenced by mitotic figures, has been observed, although in this connection it must be granted that the cleared si^ecimens are hardly the most favorable. Und()ul)tedly this explanation of the smaller cells would seem to be the true one, but there is little proof for the view, except their absence from the adult dura and their disappearance in larger specimens.


This disappearance of the smaller mesothelial cells is not rapid, but is seemingly delayed over into the larger fetuses; thus, in figure 109, a similar preparation from a fetal pig of 75 mm., corresponding smaller cells arc outhned. On account of the absence from the field of the drawing of the larger elements, these cells do not appear relatively as great in number as in the preceding figure. Likewise, in figure 110 every gradation in cell-size is shown, in a specimen made in the same manner from a dura of a fetal pig of 90 mm.


Very slowly in the course of growth of the fetus the cells Uning the inner surface of the dura reach their standard size and compose the mesothehal surface, with very little variation in size. The process, however, is apparently very tard}', even though the fetus at 16 cm. shows an inner surface to the dura which is largely composed of standard cells (fig. HI); but even in this figure, from a relatively large fetus, the standard size of the cells has not been attained, for a few cells of small size appear in the drawing. In other respects the whole pattern, in general appearance, resembles closely the adult.


It seems most fair to assume that the occurrence of a true mesothelial cellpattern on the inner surface of the dura represents the initial estabUshment of a subdural space. On this basis the subdural space may be said to occur in fetal pigs 50 mm. in length; in the present investigation it has been found impossible to demonstrate the existence of the mesothehal cell-pattern in fetuses smaller than 50 mm. The separation of the dura, possible bj'^ gross dissection in pig fetuses of 40 mm., suggests that the space may be found at a sUghtly earUer stage than that in which the mesothehal cells have been demonstrated.


Anatomically the subdural space in pig fetuses resembles m everj' particular the adult space in cats and dogs; this was described in a paper^^S) pubhshed in 1914. In the large pig fetuses injections of solutions of potassium ferrocyanide and ironammonium citrate were made into the spinal subarachnoid space. After precipitating the foreign salts as prussian-blue, the injection is found to be wholly within the subarachnoid spaces, both in the spinal and cranial regions; the subdural space is absolutely free from any evidence of connection with the subarachnoid space. These findings wholly accord with the opinion concerning the adult subdural space which has been repeatedly expressed.


THE COMPETENCY OF THE EARLY DURA AS A CELLULAR MEMBRANE.


During the stage when the condensation of mesenchyme to form the cranial blastema is prono'inced the spread of the cerebro-spinal fluid becomes more and more extensive. In these stages, when the pig embryo measures from 16 to 25 m m. approximately, the outer membrane of the arachnoid is not 3'et formed, the arachnoid spaces extending from pial to blastemal condensation. WTien in these embryos the cerebro-spinal fluid is replaced by the ferrocyanide solution and the embryo kept alive for some time, the course of the injection may be traced to varying extents throughout the periaxial tissue. To this spread of the injection fluid (a true solution, during the progress of the experiment), however, the blastemal condensation of mesenchjme opposes an absolute barrier. This pecuUarity of the early condensation may be readily seen in figures 16 and 18. At this stage in development the blastemal thickening may be said to play the role of the outer membrane of the arachnoidea or of the inner surface of the dura.


This feature of the blastema as an impenetrable membrane — an absolute barrier to the passage of fluid — is found also to endure during injections of the ferrocyanide solution under pressures sufficient to rupture other parts of the central nervous system. Similarly, the early blastemal condensation resists the inflow of the other injections used (india ink and silver nitrate) under similar pressure conditions. In later stages the injection solutions, from ventricular or subarachnoid mtroduction, do not reach the dura. This is due to the development of an outer membrane of the arachnoidea and the formation of the subdural space. The arachnoid membrane when formed does not permit fluid to pass outward into the subdural space; but the competency of the early blastemal condensation in the mesenchyme affords a very good example of the perfect function of a tissue as a fluid barrier.


An interesting phase of the competency of the secondary mesenchymal condensation (forming dura and outer membrane of arachnoid) may be seen in the region of the cisterna cerebello-medullaris. Here, as shown in figure 77, the zone of secondary condensation, while complete below, does not remain definitive above as the mesenchyme sweeps inward to the chorioid plexuses. At such a stage of 32 mm. in the pig, a replacement experiment would show no penetration of this secondary dural condensation by the foreign solution, where the condensation made a definitive membrane; above, however, in the region of the plexuses, a limited penetration by the introduced fluid could be made out.


X. THE RETURN OF CEREBRO SPINAL FLUID TO THE VENOUS SYSTEM.

The question of the exact mode of return of the cerebro-spinal fluid to the general circulation has interested many investigators. It has occasioned a large amount of work, with the presentation of several hypotheses. Key and Iletzius<2»), from the results of injections of colored gelatin into the spinal subarachnoid space, held that the cerebro-spinal fluid returned through Paccliionian granulations into the great dural sinuses. Other workers, following Key and Retzius, were dissatisfied with this theory, because of the apparent lack of these granulations in infants and in the lower animals. Cathclin(6), with but little evidence, hypothecated an absorption of the fluid by way of the perineural sheaths into the lymphatic system, although the physiological findings of Ziegler(57j^ Reiner and Schnitzler(<«), Leonard 11111*2^), and others made it necessary to consider a direct absorption into the blood system. Cushing(9) premised the drainage of fluid into the great sinuses through a valve-Hke mechanism. Dandy and Blackfan'"' .-suggested its absorption l)y the capillaries of the pia-arachnoid — an untenable hypothesis in view of the work of Kadyi'^Sj, Shroeder van der Kolk(5i), Ekker(i*>, Adamkiewicz^*), and others. Still another conception of the process has been advanced by Mott('**), namely, that the absorption of cerebro-spinal fluid isone of the functionsof the cerebral capillaries. Ina previous investigation'^), making use of a method similar to the one here employed in the rej)lacement experiments, evidence was presented indicating the drainage of cerebro-spinal directly into the great dural sinuses through arachnoid villi. These structures represent an invasion of arachnoid tissue through the lateral wall of the sinuses.


In view of the findings in adult laboratory animals, interest naturally turned, during the course of this work, to the process of drainage of the embryonic cerebrospmal fluid. The evidence afforded by the replacement experiments with the ferrocyanide solution indicated that in pig embryos of over 20 mm. cerebro-spinal fluid circulated throughout most of the periaxial tissue, and that in embryos of about 26 mm. the periaxial distribution was complete, the relations of the fluid at this stage becoming adult. With this evidence before us, the question of the drainage of the fluid became important. as the absorption process similar to the normal adult procedure, or was it entirely lacking, the production of the fluid being balanced by the growth of the nervous system and its meningeal spaces? The question of the absorption of cerebro-spinal fluid was approached in the embryo in a similar manner to that employed in the adult animal. The problems incurred by the use of abnormal intracranial pressureswere eliminated by the method of replacing, without disturbing the normal tension, the embryonic cerebro-spinal fluid with the ferrocyanide solution. The embryo was then kept aUve and was finally fixed in a preservative which would precipitate the replaced fluid as prussianblue. This procedure was carried out in many embryos of varying lengths and the specimens were subsequently stained in serial sections.


The smallest embryo in which any evidence of absorption of the fluid from the periaxial tissue was obtained was a pig embryo, 23 mm. in length. In this specimen granules of prussian-blue could be traced through the mesenchymal spaces (arachnoidal) to the inner wall of the sinus transversus. The sinus is well differentiated at this stage in the human embryo of 21 mm., as demonstrated by Streeter^**). The wall of the sinus in this pig embryo was quite thin, the mesenchvTne lending the endothelium but Uttle support. The prussian-blue granules could be traced directly through the endothehnl wall of the sinus, and a few were identified lying free in the lumen. The conditions of the observations, permitting a flow of venous blood through the sinus, undoubtedly accounted for the fact that but few of the granules were found Ijing free in the sinus. This passage of the replaced fluid into the lateral sinus is portrayed in figure 21, taken from the pig embrj-o of 27 mm.


The same process of drainage of cerebro-spinal fluid may be obser^-ed in pig embryos more than 23 mm. in length. In all but one particular it corresponds exactly to the process observed in adult laboratory animals. There is the same lack of absorption on the part of the cerebral veins and embryonic capillary plexuses. In the adult, however, the process is not diffuse, but is confined to the arachnoidal villi, while in the embryo a considerable extent of the inner wall of the sinus lying in the mesenchymal tissue, which is breaking down to form the arachnoidal spaces, serves as a site for the fluid passage. In these earlier stages the sinus transversus functions as the chief sinus of absorption. This is probably to be explained by the primary basilar spread of the replaced cerebro-spinal fluid and also by the fact that the true sinus sagittalis superior is a much later addendum. In the human embryo, according to Streeterf^^), it is found in stages of over 50 mm.


The absorption of cerebro-spinal fluid in the embryo seems to follow the directing agencies which operate in the adult. Increase in the pressure employed in the injection of true solutions results in the drainage of more of the fluid, as determined by subsequent microscopical examination. This suggests that the process is determined by factors other than that of difTusion; it seems most likely that here, too, the process is one of filtration, with a possible distension of the cellular membrane, so that intercellular spaces are opened. The histological picture of the sinus waU, however, undoubtedlj^ gives the impression that the fluid has passed almost solely through the cytoplasm of the endoth(>lial cells and likewise through the layer of supporting mesenchyme. These findings are in accord with observations on the adult.


With dilute suspensions of india ink as the injection mass, the results are quite different in regard to the passage of the material into the sinus. Replacement experiments making use of this suspension of particulate matter yield no evidence, as the carbon granules do not leave the ventricular system. Likewise, simple injections of the suspension into either the central canal of the spinal cord or into the perispinal spaces furnish no information unless the syringe-pressure be high. In this case the carbon granules may be traced into the sinus transversus, which is apparently the point of least resistance. Because of the obscuring of the picture by the carbon it can not be determined histologically whether the granules pass into the sinus in the same manner as does a true solution, or whether the passage is effected by numerous small ruptures of the tissue. The impression gained from our study would incline one toward the latter view.


If the injection of india ink be made under very great pressure from a syringe, the segmental veins may be filled with the carbon. This filling is always subsequent to its flow into the sinus transversus. But in no case was an evidence of a flow into lymphatic channels observed.


The process of drainage of the cerebro-s])inal fluid into the venous system of fetuses will not l)e (U^tailed here. This undoubtcdl}' concerns a study of the formation of arachnoidal villi and of the differentiation of the lateral walls of the superior sagittal sinus, the best site for this study. The material at hand is not suited for this investigation, so that postponement is necessary.


XI. The Chorioid Plexuses and the Elaboration of Cerebro-Spinal Fluid

With the realization that at a definite period in embryonic Ufe, cerebro-spinal fluid passes from the cerebral ventricles into the periaxial spaces, it seemed desirable to ascertain what relationship existed between the developing chorioid jilexuses and the elaboration of the fluid; for with the extension of the fluid into the periaxial tissue it becomes obvious that the balance between the development of the intraventricular fluid and the volume of the ventricles is destroyed and that more fluid is being elaborated than can be contained within the medullary-canal sjstem. This relationship between the ventricular volume and the production of cerebro-spinal fluid has been described at some length in a preceding section of this communication.


The determination, then, of the exact role plaj-ed by the chorioid plexuses in the further extension of the fluid into the periaxial tissue appeared to be of importance, for it could be conceived that the embryonic ependymal cells might be capable of elaborating the excess of fluid. With this purpose in mind the chorioid plexuses were investigated from morphological and cytological standpoints, in the hope that some index might be afforded as to the assumption of function on the part of the developing chorioid plexuses. These methods of study were apphed solely to the chorioid plexuses of pig embryos, for it is from them alone that evidence of the period of extraventricular extension of the cerebro-spinal fluid has been obtained.


THE DEVELOPMENT OF THE CHORIOID PLEXUSES.


The development of the chorioid plexuses is so well understood that only a ven,brief outline will be given here. The general scheme of origin of these glandular structures concerns a gradual histological differentiation in certain localities of the ventricular ependyma. The ependyma of the roof of the fourth ventricle thickens along the transverse invagination (phca chorioidea) and then gradually becomes tufted in villous projections into the ventricle, following the ingrowth of a capillar^' plexus and supporting mesenchyme. This general process of differentiation occurs at first along the lateral portions of the phca; the central portion of the ependj-ma remains unaffected by the villi even when the tufts have become quite well differentiated (fig. 23).


Quite similar to this process of development of the plexus chorioideus of the fourth ventricle is the differentiation of the other plexuses. The plexus of the third ventricle develops as an infolding of the tela chorioidea of the roof. In every case the process holds of ependymal invagination and subsequent vascularization and suspension by mesenchymal ingrowth.


The histological differentiation of the ependjTnal cells into the glandular tj-pe of plexus, as first determined by Luschka^^ and Faivre^^^)^ jg hardlj- satisfactory as an index of the production of fluid, as the secretory phenomena of the adult cells have not as yet bten completely established. The researches of Pettit and Girard(^\ dealing with the correlation of histological changes in the chorioidal cells and their functional state, first furnished reliable evidence that these cells give rise to cerebrospinal fluid. Since the publication of their investigations in 1900 many workers — Meek(37), Findlay("), Pellizzi(2^, Mott^^D, Hworostuchin(26), Engel(>2), and othershave been concerned with this problem and have established on fairly definite bases the relationship of the plexuses to the production of the fluid. The histological appearances of the secretory cells, however, does not rest on incontrovertible ground, as has been stated in a previous paper^^^).


The process of diff'erentiation of the ependymal cells which form the glandular elements of the chorioid plexuses occurs with the invagmation and tufting of these structures. The various stages of transformation from the low type of cubical epithelium constituting the ependymal layer are shown in various figures in this paper. The nuclei of these cells assume basilar positions and the outer zones of the C3-toplasm become granular with their greater height. The process is rather a slow one, as might be expected from the fact that the whole villus is gradually enlarging and becoming more and more tufted.


The histological differentiation of the plexuses need hardly concern us here, except as an index of the assumption of function. The final completion of this change into the adult morphology occurs at a much later stage of development than our evidence indicates for the establishment of a cerebro-spinal circulation. It becomes obvious, then, that the final liistological changes are not necessary for the process of elaboration of the fluid. This assumption seems warranted also.b} the fact that the embryonic fluid contains much more albuminous material than does the adult fluid.


The time of appearance of the chorioid plexuses in relation to the extraventricular spread of the fluid would surely seem to offer undoubted evidence in regard to the first elaboration of the fluid by the plexuses. It has been shown that in pig embryos over 14 mm. in length the replaced solution in the cerebro-spinal system spreads from the roof of the fourth ventricle into the periaxial tissues. This extraventricular extension occurs practically simultaneously with the first indications, in the pig embryo, of the formation of the chorioid plexuses of the fourth ventricle. Thus, in a pig embryo of 14 mm., the primitive thickening and tufting of the ependyma of the roof of the fourth ventricle may be observed (fig. 32). In earlier stages no definite evidence of this developmental process is found.


From the first indication of a developing chorioid plexus in a pig embryo of 14 mm., the growth of the tufts progresses rapidly, so that at 18 nmi. the process is well advanced. In embryos of 20 mm. and over the tufts of the plexuses in the fourth ventricle are quite marked, as shown in figures 22, 44, 40, and 92.


The chorioid plexuses of the third and lateral ventricles api)ear at a somewhat later stage than do those in the more caudal v(>ntricle. Thus the first indication of their ai)i)earance in pig eml)ryos is found in si)ecimens measuring 19 mm. in length. This coincides well with the further extension of the replaced fluid in specimens of 19 mm. and over. The definite differentiation of these plexuses, however, does not actually take place until the embryo reaches a length of 23 mm.— a fact suggestive of some relationshif) to the complete periaxial spread found in embryos of this measurement.


Considered, then, as a whole, there seems to be a very definite relationship between the developing chorioid plexuses and the periaxial spread of the embryonic cerebro-spinal fluid; for immediately after the first appearance of chorioidal tufting in the roof of the fourth ventricle (at 14 mm.) the replaced injection sjjread appears in the periaxial tissue (fig. 3). This extraventricular spread does not become marked until a length of 19 mm. is attained (fig. 5) — a factor in accord with the elaboration of the villi in the chorioid plexus of the fourth ventricle. The periaxial spread remains localized in the rhombencephalic region until the 20 mm. stage is attained, when it rapidly becomes pericerebral and perispinal (figs. 6 and 7). This coincides with the first indications of the chorioid plexuses in the more cephalic ventricles. But the further spread is here delayed (as in the stages between 14 and 19 mm.) until a length of at least 24 mm. is reached — which is perhaps of importance in the further development of the cerebral plexuses and the greater elaboration of the cerebro-spinal fluid. Thus it seems possible to conclude that coincident with the first appearance of the chorioid plexuses a more rapid production of cerebro-spinal fluid occurs, necessitating the passage of the fluid into the periaxial tissues.


THE GLYCOGEN CONTENT OF THE CHORIOID PLEXUSES.


In the hope that some cytological method might afford direct and incontrovertible evidence of the time of the assumption of function by the chorioid plexuses, stains demonstrating the intracellular presence of glycogen were applied to these structures. The quantitj' of the starch in the chorioid plexuses of rat and mouse embryos, as shown bj' Goldmann, suggested that this substance might be associated with the early elaboration of the cerebro-spinal fluid. Furthermore, the presence in the adult fluid of a definite reducing bodj', demonstrated by XawTatschi to be dextrose, added some weight to the hope that a definite conclusion might thus be afforded.


Several important studies concerning the presence of glycogen in the cells of the embryonic and fetal chorioid plexuses have been made. Creighton W found that the glycogen of the chorioid plexus was verj^ abundant about the middle of embryonic life, while von Loeper concluded that the great content in the cells of the fetal plexus was characteristic. Goldmann^^o) found large quantities of gh'cogen in the plexus in rats and mice, not only in embryonic life but also in animals from two to three weeks old. In the adult plexuses the cells contained no trace of glycogen.


The observations here included were made after fixing the chorioid plexuses of various pig embryos in absolute alcohol and staining the sections (cut either from celloidin or paraffin blocks) by Best's carmine method. This technique is similar to that employed by Goldmann. The staining reaction is such that a very striking differentiation of the glycogen occurs, but the shrinkage of the embryonic tissue in the fixation in absolute alcohol is a disadvantage. In these obsers'^ations the plexuses from the fourth and lateral ventricles were used.


As shown in the table on page 94, glycogen could be identified in the cells of the chorioid plexuses in pig embryos varying in length from 28 to 155 mm.


Below the first measurement no glycogen was demonstrated bj' the method employed; above the higher limit in only one instance (series No. 41) was glyocgen found. This finding of a limited period in the embryonic hfe of a pig during which glycogen occurs in the cells of the chorioid plexuses does not coincide with Goldmann's observations on the rat and mouse. Furthermore, it was found here that in stages up to 100 mm. the glycogen was practically generally distributed throughout all the cells of the chorioid jilexus, occurring with gn^it regularity in every villus and cell. This general distribution was not found in the plexuses of embiyos over 110 mm. in length; in these more advanced stages the cells containing starch occurred in clumps, giving a localized distribution. In the stages under 100 mm. the glycogen was present in very large amount, as estimated histologically. As the stages advanced the quantity of glycogen decreased rapidly. This great amount of starch was present in the same stages in wliich the general distribution of the cells occurred.


Occurrence of glycogen in the chorioid plexuses of embryo pigs.

C.P.


series, No.

C.R.


naeasure, mm.


Glycogen.


Globular forms of glycogen.


Plaques of glycogen.


Amount of glycogen.


Distribution of glycogen throughout plexus.


Intracellular distribution of glycogen.


16 3 13 12 14 6 9 8 4 1 42 17 15 20 10 18 27 39 25 32 40 41 24 19 23 11 21 22 26 5


18 23 28 33 36 39 40 55 66 80 90 100 105 118 132 155 155 155 158 160 163 170 173 185 195 209 213 223 244 260




Present Present Present Present Present Present Present Present Present Present Present . .


Present Present Present Present Present Absent


Present Present Present Absent Present Present Present Present Present Present Present Present Present Present Present Present


Present Present Present Present Present Present Absent Present Present Absent Present Absent Absent Absent Absent Absent


Great Great Great Great Great Great Great Great Great Great Small Small SmaU SmaU Small SmaU


General Localized General General General General General General General Genearl General Localized General LocaUzed Localized Localized


Basilar. Basilar. Basilar. Basilar. Basilar. Basilar. General. General. General. General. Basilar. General. General. General. General. General.




Absent



Present Absent


Present


Absent


Small


Localized


General.


Absent


Absent



Absent


Absent



Absent




Cioldmann'20) pictures the glycogen as occurring throughout the cells of the chorioid i)lexus in the form of globules of larger or smaller size. Some of these globules may be .seen even in the s\n'rounding cerebro-sjiinaj fluiil. This general intracellular disposition was observed in this series in specimens measuring 6G mm. and over (fig. 95). Below this measurement the glycogen occurred practically entirely in the basilar portion of the cell, central to the nucleus. Furthermore, in the stages between 30 and 00 mm. the glj'cogen globules were present in but small numbers and the glycogen was found in crescentic plaques (fig. 96). This formation of definite plaques is ai)parently to be ascribed to the fusion of the globules when the amount of glycogen becomes extreme. As far as is known tliis plaque formation with glycogen has not previously been noted; in one of Goldmann's figures the fusion of some of the globules has apparently taken place.


The table on page 94 records the findings in these observations.


The occurrence of glycogen in the cells of the chorioid plexus only during a certain portion of embryonic life is, as shown l)y the foregoing table, a fairly definite phenomenon, but there is surelj' no indication that this temporary presence of the animal starch bears any relation to the assumption of function on the part of the chorioid plexuses. The evidence afforded by the extraventricular flow of the replaced fluid, with the apparent relationship of the developing chorioid plexuses to the periaxial extension of the fluid, argues strongly against such an assumption.


XII. PERIVASCULAR SPACES IN THE EMBRYO.


In 1865 His' using a puncture injection, found that each nerve-cell existed in a so-called space. These pericellular spaces connected, as demonstrated by the flow of the injection mass, with an extensive perivascular network, more complex in its gray matter than in the white. In all of His's cases continuation of the injection led to a peripheral spread toward the pia, both in the spinal medulla and in the brain.


]Mott('*^\ working on the brains of animals in which an experimental cerebral anemia had been produced by ligation of the head arteries, found the perivascular spaces enormously dilated and the perineuronal spaces Likewise verj' evident. Direct connections between the perivascular and perineuronal spaces are pictured in Mott's communication.


The deduction which ^Nlott made from his findings, regarding the possible absorption of cerebro-spinal fluid by the cerebral capillary bed from this perivascular and perineuronal system, was discussed by the present author in a paper two years ago(55). It was there shown that, with the use of true solutions as the injection (potassium ferrocyanide and iron-ammonium citrate), the whole perivascular sj'stem could be filled. This injection of the spaces, however, occurred only when the pressure conditions within the cranial cavity were such that the subarachnoid pressure exceeded the vascular tension. This reversion of the pressure relations was accomplished by maintaining at normal the subarachnoid pressure with the injection fluid, and occasioning a simultaneous and complete vascular anemia. Under the routine conditions of injection (with undisturbed pressure relations) no injection of the perivascular system from the subarachnoid space resulted. It was found impossible to mject the perivascular system, using granular suspensions as the injection mass, without employing pressures far above the normal.


From these results here recorded briefly, the belief was expressed in tliis former paper that each nerve-cell was surrounded by a capillary space which drained along the perivascular channels into the subarachnoid spaces. Probably this sj'^stem represents a mechanism for accessory tissue drainage comparable ph3^siologically to the lymphatic channels of the other parts of the body.


In view of these findings in the adult mammal it seemed desirable to ascertain at what period of intra-uterine life such function was acquired. It also seemed not unhkely that information of interest might be acquired from the embryonic intramedullary circulation which would amplify our knowledge of this sj^stem in the adult. It was thought that there might be a correlation between the production of the perivascular fluid and the enlargement of the subarachnoid channels, similar to the evident connection between the chorioidal invagination and the extraventricular spread of the fluid.


Experiments to demonstrate possible perivascular and perineiu-onal spaces were first attempted on rather large fetuses (pig), as follows: The spinal meninges were exposed in a fetus in which the heart was still beating vigorously. Into the spinal subarachnoid space was introduced a needle connected with a small reservoir, containing the injection solution (potassium ferrocyanide, 0.5 gm.; iron-ammonium citrate, 0.5 gm. ; water, 100 c.c). The reservoir was then adjusted so that a pressure of 160 mm. of water was maintained in the subarachnoid sjiace. The arteries and veins in the neck of the fetus were then severed, and the subarachnoid pressure maintained at its former level. At the end of 20 minutes the head was placed in a fixative containing 1 per cent hj^drochloric acid.


This procedure, as outlined above, in the adult laboratory mammal, usually resulted in a complete injection of the perivascular system. In the embryo, however, the procedure was uniformly unsuccessful. The injection solution, as shown subsequently by the precipitated prussian-blue, rarely ascended over a centimeter above the point of injection. This indicated that the existent cerebro-spinal fluid was not replaced by the injection solution, and that the failure to demonstrate the perivascular system was to be explained on this basis, if the system were functional at this stage. Attempts were then made to replace the subarachnoid fluid with the injection solution before the cerebral anemia occurred. These attempts likewise m^t with failure, because of the impossibility of keeping the heart beating for any length of time in the larger pig fetuses. Other attem]its were also made to demonstrate these channels, in larger pig embryos, by means of a procedure which in the adult gave at times good injections of these intracortical canals. This method differed from the method first employed only in the maintenance of a high pressure (100 mm. Hg) in the spinal subarachnoid si)aces. It likewise met with failure, due ai)|)arently to the same causes which occasioned its failure in the adult: the high sul)arachnoid pressure o})erated chiefly to compress the cerebral and spinal tissues, rendering the injection of the i)erivascular spaces impossible.


The same procedures were attempted in smaller pig embryos (15 to GO mm.). The method usually successful in demonstrating the spaces (subarachnoid pressure slightly above normal, with subsequent cerel)ral anemia) failed, ajjparently because the cranial cavity at these stages is in no sense a rigid closed box. as in the adult.


Any method of service in the adult — which must have in consideration the physical character of the skull as a closed box — was here necessarih' doomed to failure.


Together with these technical failures to demonstrate a perivascular system, it must be borne in mind that these are merely failures to demonstrate the existence of the perivascular system in the pig embryo. The system wall probably be demonstrated as soon as a suitable technique is devised. The spaces are very likely present soon after the capillary plexus invades the nervous system, but the observation in many histological preparations of the spaces around the cerebral vessels must not be considered as offering proof of their existence, because of the likehhood of shrinkage influencing the picture. It is interesting, how^ever, to note that elasticity of the cerebral tissues seems greatest along the course of the blood-vessels, for here the phenomenon of shrinkage is most frequently observed. The existence of the perivascular and perineuronal spaces, probably of only capillary thickness, must remain — in the embryo as in the adult — a subject of physiological demonstration; histological e\adence, except with proper physiological regard, is of no value.


The early development and function of such a system as the perivascular and perineuronal canals afford seems most likely from the standpoint of pure speculation. It is not improbable that fluid is poured from this system into the embryonic subarachnoid space at a period soon after the capillary plexus invades the cerebrum. There is no evidence, however, from the observations recorded in foregoing paragraphs, that adequate subarachnoid channels are afforded until the pig embryo reaches a length of about 25 mm. The hypothesis of Essick^^^) regarding the damming of the perivascular fluid as the cause of the two cava corporis striati is of extreme interest in this connection. It remains, however, for future work to afford real evidence in regard to the embryonic perivascular system.


XIII. THE PERINEURAL SPACES IN THE PIG EMBRYO.


The question of the existence of potential or actively functional spaces around the peripheral nerves is of great interest, partly because of the possible relation of these spaces to the developing lymphatic system, and also on account of the anatomical evidence of the possible existence of such spaces.


It is realized that before much dependence can be placed on any theory regarding these potential spaces around the cerebro-spinal nerves, the possibility of their being purely artifacts must be dealt with. The methods of demonstration, in the adult, in the hands of the earliest workers were such as to favor the production of artifacts. As far as can be ascertained, Cotugno('), dealing with the ner\-us ischiadicus, was the first to conceive of these possible spaces. His method of demonstration consisted in filling the spinal subarachnoid space with mercurj' (in a cadaver placed in the erect posture) . Globules of the mercurj^ were subsequently found about the sciatic nerve in what then became the perineural spaces.


Modern anatomical interest in these spaces was aroused by the remarkable injections of Key and Retzius^^s). These investigators, by means of gelatin injections into the spinal subarachnoid space, were able to demonstrate perineural spaces around the cranial nerves, especially around the optic pair. Their results, however, are open to criticism, because of the excessive pressures employed ("not over 60 millimeters of mercury") and because the injections were made in fresh cadavers kept warm for periods of 10 or more hours.


Some of the difficulties concerned in the problems of the perineural spaces were cleared up in a studj'^^^) of the cerebro-spinal circulation published in 1914. In this work injections of true solutions (similar to those used in the present study) were introduced into the spinal subarachnoid space in living cats and dogs, under pressures but sUghtly exceeding the normal intraspinal tension. These injections were continued for several hours, and the course of the injection fluid was then estabhshed bj precipitating the solution in situ. By means of this procedure, which it was beUeved approached the physiological, the perineural spaces around the cranial nerves could be demonstrated. In these adult laboratory mammals the cerebral nerves without exception showed prussian-blue granules in a perineural relation, extending outward along the nerves bej^ond the termination of the dural cuff. This extension of the injection mass outward was more striking around the first two cranial nerves than about any of the others. Thus, the olfactorj'^ nerves uniformlj'^ showed perineural deposits beyond the cribriform plate, extending downwards into the nasal epitheUura, while the optic nerves were surrounded by the granules in the inf ravaginal sheath, which spreads out over the posterior surface of the eyeball. The caudal cranial nerves were likewise characterized by extensive perineural injections.


These findings were interpreted as e\'idencing a true perineural space, probably of only capillary thickness, which could be injected by filling the cerebro-spinal spaces with a demonstrable true solution. As far as could be made out under the microscope, they had no appreciable existence except when fiUed with the precipitated true solution. These spaces were not filled in the early moments of the injections under low pressures, and could be demonstrated only when the injection had been continued for several hours.


The perineural spaces are quite different from the spaces surrounding the spinal ganglia and the gangUa of the cranial nerves. These ganglia he in the true subarachnoid space, wath the dura investing the arachnoid membrane. Distal to the ganglion the dura ends upon each nerve. In the injection under low pressure with the ferrocyanide the cranial and spinal ganglia were all surrounded bj^ the precipitated salts; the cranial nerves .showed extensive perineural injections, whereas the spinal nerves rarely showed a true perineural injection, and then only of hmited extent.


The existence of perineural spaces in the embryo, however, has been under dispute. The larger nerv^es in sectioned embryos almost invariably show spaces about them, either a complete separation of the surrounding mesenchyme or a partial dilatation of the mesenchymal interstices. Sabint'*^), in 1902, noted that in pcrispinal injections with inflia ink the spinal nerves could be outlined by the carbon granules, but in no case did such an injection run into true lymphatic channels. No evidence was afforded by her work of any lymphatic channels arising from these apparent perineural channels.


In the course of this investigation of the cerebro-spinal spaces interest naturally turned to the perineural spaces. In the typical experiments (a replacement of the embryonic cerebro-spinal fluid with a demonstrable true solution in the living embryo), there was evidence of a spread of the replaced solution around the cranial nerves. Because of the procedure u.sed (merely a filling of the ventricles and central canal of the spinal cord) no evidence of a perineural spread occurred until the foreign solution passed into the periaxial tissues. Here the spread chiefly involved the caudal cranial nerves curving around the lateral surface of the medulla in fanshaped processes (figs. 5, 6, 8, and 9). The spread, however, was not extensive. In figure 8 a similar slight spread along the spinal nerves is to be made out. Closer study of these cleared specimens, and examination of the same and of similarly injected embryos after serial sectioning, convinces one that the apparent perineural spread in these cases extends around the sensory ganglia and not further toward the periphery. In no case, cither in the caudal portion of the cranial or in the spinal region, has the replaced injection fluid passed the blastemal condensation of mesenchyme. This finding is well shown by the distribution of the injection fluid in figures 9, 16, and 18.


The optic nerves, however, jjossessing gangha in the retina, usually show, in the typical replacements in the living embryo, a partial or complete surrounding of the nerves by the prccij)itated pru.ssian-bluc. An incomplete example of this — more typical, according to these observations, than a total circumvention — is given in figures 19 and 20. The higher-power reproduction of this field is very interesting. It shows in the central portion the fiber bundles comprising the optic nerve, surrounded by mesenchyme and the developing ocular muscles. In the region between the ners'e and the muscles is an undiflferentiated mesenchyme which is characterized by a crescent of the precipitated granules of prussian-blue. The non-penetration of the surrounding tissue by the ferrocyanide is verj- well brought out in this drawing. The prussian-blue has reached its position about the nerve by extension from the pericerebral spaces; actually it has still the same distribution as noted in figure 8 above. The adult dura will completelj- surround the optic nerve in its whole extent; the subarachnoid space will likewise extend unbroken to the posterior surface of the eyeball. Hence it must be assumed that in this case the perineural space does not extend beyond the peripheral ganglion. With regard to the olfactory nerves, no evidence of a perineural spread was obtained in specimens of pig embryos up to 45 mm. in length.


It seems obvious, then, that in the embryo pig true solutions, when substituted for the cerebro-spinal fluid, do not extend peripherally along the nerves any further than does the dura in the adult. The replaced fluid (if, as appears most likely, it indicates the true circulation of the cerebro-spinal fluid) extends only through the future subarachnoid space. Such a conclusion is best supported bj' the observations. The only discreparcv between the findings in the pig embryo and those in the adult with the same method lies in the fact that in the adult the cranial nerves showed a much more extensive permeural injection. This seeming discrepancy may be accounted for in two ways. In the first place, the experimental replacement in the embryo pigs lasted at most one hour (due to the fact that the embryo's heart frequently ceased beating at the end of this time), while in the adult cat or dog they were continued for several hours; and it was only in the long-continued experiments in the adult that the extensive perineural injections were obtained. On this basis it seems more than likely that the communications between peripheral perineural spaces and the subarachnoid space are very small and that diffusion must account for the slow filhng of the peripheral system. The second explanation seems undoubtedly to concern the time of development of these perineural spaces in the embryo. It may be that the spaces are morphologically non-existent until late in fetal life; in that case, of course, it is not strange that they have not been filled wdth the injection fluid.


From the observations recorded above it is quite apparent that in the typical experiment in which the normal cerebro-spinal tension is not increased no evidence of the perineural space, as injected by Miss Sabin, has been adduced. However, the possibility of mjecting these spinal spaces as was done by Miss Sabin is easily demonstrated. The injections may be made with ease, either with . granular suspensions or with true solutions. Success invariably attends such an injection into the perispinal tissues. The injection solutions easily run out around each nerve, more readily, apparently, in the younger embryo than in the older. It is not clear whether this difference is due to the fact that in younger embryos the resistance is greater to the perispinal flow and less peripherally, or merely to the fact that a greater amount of fluid must be introduced in order to attain the same result. Careful repetition of these observations has led to the conclusion that such a demonstration of the spinal perineural spaces results from excessive pressures of injection. WTienever the pressure exerted by the injection is but slightly above the normal, or does not exceed the normal (as in replacements), the perineural spaces are not injected around the spinal nerves. Miss Sabm's conclusions from her results, that no connection exists between the spaces and the lymphatic system, seem to be wholly substantiated by these observations.


The apparent perineural spaces around the embryonic nerv^es must be looked upon as artifacts. In tissue carefully fixed, dehydrated, and embedded, there is no real evidence of these spaces. Theh- size apparently varies with the care observed in the histological technique.


XIV. General Summary

In the foregoing sections of this communication some of the problems concerned with the embryology of the cerebro-spinal spaces have been discussed and observ^ations have been presented in the hope that a better conception of the processes might obtain. It is purposed to present here briefly the results of these obser^^ations and to attempt to correlate the findings so far as is possible; and in this, as in the detailed reports in the preceding pages, the relationship of the jihysiological processes concerned will be referred to the morphological changes in the developing embryo.


As a means of studying the physiological extent of the embryonic cerebro-spinal spaces, a method of replacing the medullary fluid with a foreign solution was devised. The procedure consisted in substituting, in the hving embryo, a solution of potassium ferrocyanide and iron-ammonium citrate for the cerebro-spinal fluid. The embryos were then kept alive, for periods of about an hour, by placing them with the attached placenta; in an incubator at 38°. At the end of this time, which varied in the many experiments, the whole embryo was fixed in a medium containing hydrochloric acid, thereby precipitating an insoluble prussian-blue. Specimens prepared in this manner were studied after sectioning or after clearing by the Spalteholz method.


Pig embryos, subjected to such experimental replacements, exhibited only an intraventricular retention of the foreign solution until after a stage of 14 mm. was attained. In the earliest specimens, embryos of about 9 mm., there was no characteristic distribution of the foreign solution, except that it remained within the medullary-canal system. In stages of about 13 mm. the replaced fluid also was retained within the cerebral ventricles, but in these specimens a dense accumulation of the precipitated prussian-blue may be made out in a distinct oval in the superior portion of the rhombic roof. This granular aggregation occurs against a histological differentiated area in the roof of the fourth ventricle — an area which represents apparently the more epithelial-Uke elements of the earher roof-plate. This area must be considered solely as a differentiation of the epidermal lining of the medullary-canal system.


In hving pig embryos of 14 mm. and over, the result of the routine replacement of the ventricular cerebro-spinal fluid was a sUght extraventricular spread into the tissues posterior to the rhombic roof. The passage of this foreign solution outward occurred through the same area of ependjTnal differentiation, outlined bj' the collection of granules against its inner surface in the previous stage. The extraventricular spread remains definitely locahzed to a ver}' small conical area which does not rapidly increase in size.


The factors which cause this initial flow into the pericerebral spaces are of interest. It follows that in the growth of the embryo the production of the intraventricular and intraspinal cerebro-spinal fluid must necessarily keep pace with the increasing size of the cerebral ventricles. It is also necessarj' for the occurrence of an extraventricular spread of the fluid that the production of the fluid within the ventricles must exceed the amount required to keep the medullary-canal sj'stem filled. From our knowledge of the elaboration of the adult cerebro-spinal fluid, it is impossible to conceive of the production of a true cerebro-spinal fluid in the perimcdullary mesenchyme. Such a view would be a reversion to the old hypothesis of Haller, who regarded the leptomeninges as the elaborators of the fluid. Likewise, the passage of the replaced foreign solution into the extraventricular spaces would render such a hjiDothesis untenable.


Hence, it becomes incumbent to regard such an extraventricular spread of the experimental solution as an mdication that the production of the cerebro-spinal fluid within the cerebral ventricles exceeds the capacity of the ventricles to care for the fluid. This argues strongly that the process of elaboration of the fluid in these pig embryos of 14 mm. is no longer sluggish, but that an active production, sufficient to cause a sUght extraventricular flow during the observation, is now taking place. This acceleration of the flow is not great, but it represents a marked change in the relationship of the process of fluid elaboration to the increasing volume of the ventricles.


It seemed desirable to endeavor to correlate this extraventricular spread of the experimental fluid with the morphology of some intraventricular structure at this critical stage of 14 mm. m the pig embryo. The first evidences of villous tufting in the chorioid plexus of the fourth ventricles were found to occur at this stage in the pig. Other studies of this plexus, particularly those which concerned the occurrence of glycogen in these glandular cells, were found to offer no additional evidence of value in regard to the onset of function in these structures. The correspondence between the initial tufting of the ependyma to form the rhombic chorioid plexuses and the initial extraventricular spread must be regarded as of the utmost importance. It would appear most Ukely that as soon as the chorioid tufts occurred an increased production of cerebro-spinal fluid took place, necessitating an extraventricular expulsion of the excess of fluid. Such a view receives the utmost support from these recorded observations; it is in keeping with the best conceptions of the processes of production of cerebro-spinal fluid in adult mammals.


With the initial pericerebral extension of the experimental fluid occurrmg in pig embryos of about 14 nmti., the further extension of this spread did not occur until after a length of 18 mm. was attained. At this stage the replaced foreign solution passed from the fourth ventricle through two areas in the roof-plate. The chorioid plexuses now have divided the roof into two portions; from each, fluid escaped. The superior area of fluid passage is the same which was concerned in the mitial outpouring of the ventricular fluid. The inferior area, like the superior, is an area of ependymal differentiation, of which the first evidence may be made out in pig and human embryos of 15 mm. This differentiation consists in the transformation of the densely staining ependymal elements into cells with larger nuclei, poor in chromatin, and with more abundant cytoplasm.


After the functional employment of the two membranous areas is established at about 18 mm. in the pig, the further pericerebral spri.'ad of the replaced solution occurs very rapidly. The peribulbar tissues are filled with the fluid and from this region extensions occur downward into perispinal spaces and upward into the more basilar pericerebral spaces. Thus, the spinal spaces must be considered as develop>ing physiologically from above, and not from below upward, as Reford found. The complete filling of these perispinal spaces is found in pig embryos of 21 mm. At this stage the pericerebral spaces are filled, with the exception of those around the superior portion of the midbrain and about the cerebral hemispheres.


The final filling of all the periaxial spaces occurred in pig embryos of about 26 mm. This phenomenon may be taken to indicate the estabUshment of the true cerebro-spinal relationships of the adult, for in this case there is an intraventricular production of the fluid and an extraventricular spread. Likewise, the fluid returns to the venous system in embryos of over 23 mm., and this escape of the fluid from its periaxial bed is, as in the adult, directly into the venous sinuses of the dura mater.


The rapidity of the further extension of the replaced solution after the stage of 18 mm. is passed is apparently due to a second marked acceleration in the rate of production of the ventricular cerebro-spinal fluid. As in the first instance, this increased elaboration seems connected intimately with the formation of the chorioid plexuses of the third and lateral ventricles. As soon as these tufts develop, the cerebro-spinal fluid is produced in amounts which far exceed the quantities for which the more slowlj' enlarging ventricles can provide.


The histories of the two areae membranaceae of the fourth ventricle are dissimilar. Both are areas apparently differentiated from the normal lining ependyma for a specific functional purpose — the passage of fluid from the ventricles into the future subarachnoid spaces. The superior membranous area reaches its maximum functional importance in the stages of 18 to 20 mm. in the pig and also in the human embryo and from these stages on it slowly regresses. The final obUteration of the area, if it do not persist as an occasional small remnant, is due to the increasing growth of the cerebellum and the enlargement of the chorioid plexuses of the fourth ventricle. On the other hand, the inferior membranous area continues to increase both in size and functional importance after its initial differentiation from the ependyma; it finally occupies the greater portion of the velum chorioidea inferior. These observations can not solve the interesting question of a perforation of the inferior velum to form the foramen of Magendie.


Of the factors which influence the passage of fluid outward into the periaxial spaces, it must be reahzed that probably there is difference in this regard between the true solutions of the salts and the colloidal suspensions. For the true solutions (as in the experimental replacements) diffusion probably plays some role; but that this is not the sole factor is shown by the failure of the fluid to pass through the membrane in the stages under 14 mm. The findings of the granules of prussian-blue within the cytoplasm of the cells of this membrane mdicates that the fluid passage is similar in every way to that through a true membrane. There is also a possible site of fluid passage between the cells of this membrane. But, surely, the most important factor in this process is one of filtration of the fluid from the point of higher pressure to one of lower. This is mdicated by all of the findings : that the mcreased production of the fluid or the increased mtra ventricular pressure (whether due to normal or experimental agencies) causes a marked extraventricular spread seems firmly established. For the colloidal suspensions (particularly the protein of the normal ventricular fluid) a slower process of diffusion and filtration seems the probable agencj^ for passing the ventricular colloids into the subarachnoid spaces.


That the results obtained by the method of replacement were not solely due to diffusion, but represent a fiUing of the physiological extent of the cerebro-spinal spaces, has been shown in many ways, but probably the chief argument against such a view is that whollj^ similar extensions of the foreign solution may be obtained by injections under mild pressures from a syrmge; with increasing pressures these injections show the same type of spread, but always in a smaller embryo than the replacement method demonstrates as the standard for a given stage of the extension. The results recorded in the foregoing pages indicate also that suspensions (India ink) and true solutions (when powerful precipitants) are valuable only for affording comparisons in problems concerning the normal processes of absorption.


Of interest in any discussion of the results of injections into the perispinal spaces or into the spinal central canal are the findings in regard to the perineural spaces. It is possible to inject such spaces around each of the segmental nerves, but only when the pressures of injections are extreme. In no case, however, were such injections found to enter the lymphatic system — a finding in accord with the observations of Reford and Sabm. The physiological importance of these spaces in the adult is probably great, but the same methods of demonstration (with carefully controlled pressures) which suffice in the adult are unavailing in the embryo.


The origin of the three meninges from the perimedullary mesenchjTue is well established. His, Kolliker, Sterzi, Farrar, and others have placed this conception on a very firm basis. Most of the investigators have been concerned with the differentiation of the spmal meninges, while the observations here reported have been concerned solely with the cranial portion of these membranes. In general, the same phenomena in the transformation of the primitive periaxial mesenchyme as recorded by these earlier workers may be found in the cranium. The division of the primitive mesenchyme by a secondary condensation, a view advanced chiefly by Salvi, seems well supported. The findings in the cranium are in accord with this concej^tion; the outer portion of this primitive meninx becomes the dura mater, the inner forms both the pia and arachnoid. The processes in the formation of the arachnoid are, however, quite diversified and concern both the formation of the subarachnoid spaces and the outer membrane of the arachnoid.


Out of the rather loose-meshed periaxial mesenchyme, the subarachnoid spaces develop. The process concerns the transformation of the small " tissue spaces " of this mesenchyme into the larger subarachnoid channels, which are interrupted by the well-known arachnoid trabecule. Well-marked stages in this metamorphosis, which begins in the basis cranii, can be made out. The first appearance of a differentiation is seen in the gradual increase in the size of the mesenchj'mal mesh. This is closely as.sociated with an increased amount of an albuminous coagulum which in a measure fills the larger interstices. Following this initial dilatation of the spaces occurs a breaking-down of some of the syncytial strands; these ruptured mesenchymal processes then retract and adhere to the persisting trabeculue. The process continues with the formation of larger channels in this mesodermal tissue, with also the formation of the permanent arachnoidal trabeculae. Throughout these larger spaces, in the smaller fetuses, the coagula of protein material are everyT\-here found, the remains apparently of the albuminous portion of the circumambient fluid.


In the formation of the various cisternse, particularly the great cistema cerebellomedullaris, the process of the dilatation and confluence of the original mesenchymal spaces reaches its maximum. Here the breaking-down of the original sjTicytial strands proceeds to such an extent that very few of the strands remain to persist through life.


Such a process of the enlargement of mesenchj^ual spaces to form the larger subarachnoid spaces, as described in some measure by His for the spinal meninges, is apparently intimately connected with the circulation through these spaces of the embryonic cerebro-spinal fluid. The fluid flows everj- where through the spaces, as evidenced by the replacement experiments and by the increased content in albumen, before the process of enlargement of the mesenchymal spaces begins. It seems most likely that this circulation of the fluid acts as the causative agent in initiating and probably also in completing the enlargement of the "tissue spaces." The great content of albumen in the embryonic cerebro-spinal fluid has greatly facihtated the investigation, as the presence of the coagula from this protein has permitted the absolute exclusion of artifacts in the process of the tissue-dilatation.


This mechanism of enlargement of the tissue spaces finds its analogue in the formation of the anterior chamber of the eye and in the perilymphatic spaces of the ear (Streeter). In both these situations, as in the meningeal spaces, there are special body-fluids, more or less characteristic in their physical and chemical characters, obviouslj' subserving specialized functions. In both the eye and cranium, the absorption of the fluids is by way of special organs, directly into venous sinuses; in both, the origin of the specialized fluid is from epidermal organs; this fluid is at fijst poured into epidermal spaces and then subsequently into mesodermal spaces (subarachnoid space and anterior chamber of the eye). Thus, m these situations, the characteristic fluids have certain definite channels through rather larger spaces, connected finally with the venous system, and only indirectly with the Ij-mphatic system.


In no sense must the cerebro-spinal circulation be taken as a portion of the lymphatic system. Increasing knowledge of the cerebro-spinal fluid, of its physiology and chemistry, and of its pathway, have separated it permanently from any connection with the lymph of the lymphatic system, variable though that be. No longer may the meningeal spaces be compared to serous cavities, except possibly in the case of the subdural space, and this space is really a space apart from the true cerebro-spinal or subarachnoid spaces. Quite s i milarly, in place of the many varjdng channels in the dura and to a lesser extent in the leptomeninges, which older writers considered lyonphatic in nature, our increasing knowledge has caused the introduction of specialized arachnoidal cell-chains running throughout the pachymeninx. Unquestionably, the cerebro-spinal fluid i)ossesses its own peculiar and characteristic pathway, analogous in no way to the lymphatic vessels of other tissues.


The outer continuous membrane of the arachnoidea forms as a mesenchymal condensation, at first in common with the inner surface of the dura mater, but soon separated from it by the subdural space. The very low cubical mesothelium which covers the arachnoid membrane on both surfaces and also invests the arachnoid trabeculae differentiates apparently from the original mesenchymal elements in the periaxial tissues.


One of the most interesting features of this study has been the relation of the various mesenchymal condensations to the foreign true solution which was introduced into the medullary-canal system. This fluid circulated throughout the periaxial spaces which enlarge to form the subarachnoid channels, but it never penetrated the primary blastema which served as a primitive dura, nor did it ever invade the pial cells which so closelj^ adhere to the nervous tissue; hkewise, as soon as the secondary mesenchymal condensation dividing the dura from the arachnoid spaces appeared, this condensation was impervious to the true solution. No evidence of any penetration, as might be expected as due to diffusion, could be made out.


This summary has been included in order that some correlation between the topics discussed separately in the foregoing sections might be made. No attempt has been made here to present the findings in abstract form; these have been summarized at the end of each division of this communication.


XV. Conclusions

Based on the observations recorded in the foregoing sections, the following conclusions seem warranted:

  1. During the earlj' part of the growth of the pig embryo there is no extraventricular spread of the cerebro-spinal fluid. The first extension of the ventricular fluid into the periaxial tissues occurs in pig embryos of 14 mm.; the adult relationship of the ventricular and meningeal cerebro-spinal fluid is established in pig embryos of about 26 mm.
  2. The ventricular cerebro-spinal fluid escapes into the periaxial tissues through two areas of ependjmal differentiation in the roof of the fourth ventricle. Both of these areas differentiate at a shghtly earlier period than that at which they function actively. The area membranacea superior undergoes a gradual regression and obliteration due to the changing form of the rhombic roof; the area membranacea inferior gradually occupies the major portion of the velum chorioidea inferior.
  3. The embryonic cerebro-spinal fluid, as evidenced by the replacement with a true solution, spreads from the ventricles into the mesenchymal tissue about the central nervous system. It docs not penetrate the cranial or vertebral blastemal condensations, nor does it invade the pial cellular layer.
  4. The subarachnoid spaces arise by a process of breaking-down of the perimedullary mesenchj'mal sj'ncytium and a dilatation of the existent mesench3Tnal spaces. This phenomenon of the enlargement of the mesenchymal spaces is associated with the presence in the spaces of an increased amount of albumen. The process occurs at a period shghtly later than that at which the initial flow of the cerebro-spinal fluid into the spaces is recorded.
  5. The dura mater, arachnoid, and pia mater develop out of the perimedullary mesenchj'me. The arachnoid trabeculae are left by the breaking-down of the original mesenchymal strands, while the outer arachnoid membrane is formed, together with the inner surface of the dura, by a separate mesenchymal condensation. The dura develops between this secondary Une of condensation and the embryonic skull.
  6. There is indicated a very close relationship between the tufting of the chorioid plexuses of the fourth ventricle and the first extraventricular spread of the cerebro-spinal fluid.
  7. By means of the method of replacement it is possible to demonstrate perineural spaces as far out along the nerve trunks as the peripheral gangUa. The extensive injections of the perineural spaces along the segmental nerves are not obtained by the method of replacement.


The work, of which this paper forms the report, was done in the Anatomical Laboratory- of the Johns Hopkins Medical School. It was largely due to aid received from the Department of embryology of the Carnegie Institution of Washington that the completion and scope of this paper were possible. The wTiter gladly acknowledges his indebtedness to the Carnegie Institution. January, 1916.


Bibliography

Ad AMKiEwicz, Albert. Die Blutgef &S8e des menschlichen Rilckenmarkes. I Theil: Die Gefassc der RQckenmarks-Substanz. II Theil: Die Gefassc der Rlickcnmarksobcrflache. Sitjungshcrichte der k. Akad. d. Wiss. in Wien. Math.-naturw. CI., 1H81, Lxxxrv, pt. 3. 469; 1882, Lxxxv, pt. 3, 101.


Bardeen CR. XI. Development of the Skeleton and of the Connective Tissues in Keibel F. and Mall FP. Manual of Human Embryology I. (1910) J. B. Lippincott Company, Philadelphia.

Bardeen, C. R. The development of the skeleton and of the connective tissues. In Kcibel and Mall, Manual Fluman Embryology, Philadelphia and Ix>ndon, 1910, I, 292.


BuAKE, J. A. The roof and lateral recesses of the fourth ventricle considered morphologically and embryologically. Jour. Comp. Neurol., Granville, Ohio, 1900, X, 79.


Canniec, a. Contribution & I'dtude de la voOve du quatrifeme ventriculc chez Ics mammifires; le trou de Magendie. Jour, de Mfed. de Bordeaux, 1897, xxvii, 547.


Cappelletti, L. L'^coulemcnt de liquide cerebrospinal par la fistule c^phalo-rachidienne en conditions normales ct sous I'influence de quelques m6dicaments. Arch. Ital. de Biol., Turin, 1900, xxxv, pt. 2. 463.


.L'efflusso del liquido cerebro-spinale dalla fistola cefalo-rachidiana in condizioni normali e sotto I'influenza di alcuni farmaei. .Vtti dell' / ccademia d. Scienze Med. e Nat. di Fcrrara, 1900, ^xxiv, 85.


Cathelin, F. La circulation du liquide ctphalorachidien. Paris, J. B. Bailli^re, 1912.


CoTL'GNO [CoTUNNius], DoMENico. De ischiade nervosa commentarius. Viennae, R. GriifTcr, 1770.


Creighton, C. Microscopic researches on the formative properties of glycogen. London, A. & C. Black, 1896-99.


CusniNO, H. Physiologische und anatomische Beobachtungen iiber den Einfluss von Hirnkompres-sion auf den intracraniellen Kreislauf und ilber einigc hiermit verwandtc Erscheinungcn. Mittcilungen a. d. Grenzgeb. d. Med. und Chir., Jena, 1902, ix, 773.


. Some experimental and clinical observations concerning states of increased intracranial tension. Am. Jour. Med. Sciences, Philadelphia and New York, 1902, cxxi\-, 375.


. Studies on cerebro-spinal fluid: No. 1. Introductory. Jour. Med. Research, Boston, 1914, xxxi (n. s. XXVI), 1.


Dandy, W. E., and K. D. Buckfan. An experimental and clinical study of internal hydrocephalus. Jour. Am. Mod. Assoc, Chicago, 1913, Lxi, 2216.


. Internal hydrocephalus: An experimental, clinical and pathological study. Aiii. Joiir. Dis. Children, Chicago, 1914, viii, 406.


Ekker, H. De cerebri et meduUte spinalis systemate vasorura capillari. Trajecti ad Rhenum, 1853.


Kngel, E. A. Uelx^r die Secretionscrscheinungen in den Zellen der plexus chorioidei des Mcnschen. Arch. f. Zellforschung, Leipzig, 1909, ii. 191.


Essick, C. R. Transitory cavities in the corpus striatum of the human embryo. Contributions to Embryology, 1915, II, No. 6, 95. (Carnegie Institution of Washington, Pub. No. 222.)


14. Evans, H. McL. The macrophages of mammals, Am. Jour. Physiol., Baltimore, 1915, xxxvii, 1.

15. Faivre, J.-J.-A.-E. Des granulationes miningiennes. Thfee de Paris, 1853. (No. 142, vol. 540.)

16. Farrar, C. B. The embryonic pia. .\m. Jour. Insanity, Baltimore, 190& 7 LXiii, 295.

17. FiNDLAT, J. W. The choroid plexuses of the lateral ventricles of the brain — their histolog}-, normal and pathological. Brain, London and New York, 1899, xxu, 161.

18. Frazier, C. H. The cerebro-spinal fluid in health and disease. Jour. Am. Med. Assoc, Chicago, 1915, LXIV, 1119.

19. Gaupp, E. Die Entwickelung des Kopfskelettes. In Hertwig's Lehrbuch der Entwickelungsgeschichte d. Menschcn u. d. W'irbcltierc. Jena, 1906, K od., 692.

20. Goldmann, E. E. V'italfarbung am Zentralnervensystcm. Berlin, G. Reimcr, 1913.

21. Haller, a. Elementa physiologix corp^jris humanis. Lausanne, 1757-78.

22. Hess, Carl. Das Foramen Magendii und die Oeffnungen an den Reces.iu.s laterales des \*ierten Ventrikels. Morphol. Jahrbuch, Leipzig, 1885, x, 578.

23. Hecser, C. H. The development of the cerebral ventricles in the pig. Am. Jour. Anat., Philadelphia, 1913-14, XV, 215.


24. Hill, L. Physiology- and pathology of the cerebral circulation. London, J. & A. Churchill, 1896.


25. His, W. Ueber ein perivasculiires canalsystem in den nervoseu Ccntralorganen. Leipzig, Engelmann, 1865. . Die Hfiute und Hdhlen des Korpers; ein academisches Programm. Basel, Schweighauser, 1865.

. Die Entwickelung des mcnschlichcn Rautenhirns, von Ende des ersten bis zum Beginn des dritten Monats. Leipzig, Hirzel, 1890.

. Die Hiiute und Hohlen des Korpers. Arch. f. Anatomie u. Physiol., Anat. .\bth., 1903, 368.

. Die Entwickelung des menschlichen Gesims. Leipzig, Hirzel, 1904.


26. HwoROSTUCHiN, W. Zur Frage uber den Bau des Plexus chorioideus. Arch. f. mikr. Anat., Bonn, 1911, Lxxvii, 232.


27. Jacobson, C. Unpublished.


28. Kadvi, Heinrich. Ueber die Blutgefisse des menschlichen RQckenmarkes. Lemberg, Gubrj-nowicz 4 Schmidt, 1889.


29. Key, E. A. H., and G. Retzius. Anatomie des Nervensystems und des Bindegewebcs. Stockholm, Samson & Wallin, 1875-76.


30. Knower, H. McE. a new and sensitive method of injecting the vessels of small embrjos, etc., under the microscope. Anat. Record, Philadelphia, 1908, ii, 207.


31. KoLLlKER, A. Entwiekelungsgeschichte des Mcnschen und der hoheren Thiere. 2ed. Leipzig, Engelmann, 1879.

. Gnmdriss der Entwiekelungsgeschichte des Mcnschen. I^eipzig, Engelmann, 18S0.


32. KoLLMANN, J. Lehrbuch der Entwiekelungsgeschichte des Menschen. Jena, Fischer, 1898.


33. Lewandowskt, M. Zur Lehre von der CerebrospinalflQssigkcit. Zeitachr. f. klin. Med., Berlin, 1900, XL, 480.


34. vov LrscHKA, H. Die .\dergeflechte des menschlichen Gehimes. Berlin, G. Reimer, 1855.


35. Magen-die, Francois. Recherches sur le liquide csfephalorachidien. Paris, 1825.


. Recherches philosophiqucs et cUniques sur le liquide c^phalo-rachidien ou c6r6bro-spinal. Paris, 1842.

Mall FP. Hosp. Bull., XIV, 1903. Catalogue of the collection of human embryos in the Anatomical Laboratory of the Johns Hopkins University. Johns Hopkins Hopkins University. Baltimore, 1904.

36. Mall, F. P. On the development of the blood-vessels of the brain in the human embryo. Am. Jour. Anat., Baltimore, 1904, rv, 1.


37. Meek, Walter J. A study of the choroid plexus. Jour. Comp. Neurol, and Psychol., 1907, xvii. No. 3, 286.


38. Mestrez.\t, W. Le liquide c^phalo-rachidien. Paris, Maloine, 1912. Thise de Montpellier, 1911, No. 17. .


39. MiDDLEM.tss, James, and W. F. Robertson. Pathologyof the nervous system in relation to mental diseases. Part III. Morbid conditions of the dura mater. Edinburgh Med. Jour. 1895, xl, pt. 2, 704.


MiNOT, C. S. Human Embryology. New York, W.

Wood A Co.. 1892. MoTT. F. W. The Oliver-Sharpey lectures on the cerebro-spinal fluid. Liincct. London. 1910, ii, 1 ; 79. Peluzzi. B. Experinientelle histologische Untersuchungcn ilbcr die Plexus chorioidei (Adcrgeflechte).


Foli.i neuro-biologica, Haarlem, 1911. v, No. 4, 305. Pf.ttit. a., and J. Giiuhd. Sur la fonetion .s4or6toire ct la morphologie des plexus chorioldes des ventrilulcs latfraux du Syst^nie ncrveux central. Arch.


d'anat. micr., Paris, 1902, v. 214. PuLT, F., O. Kehm, and H Schott.mOller. Leitfadcn zur Untcrsuchung der ZerebrospinalflOssigkeit.


Jena, Fischer, 1913. PoiBiER, P., and A. Charpy. Traits d'antomie huniaine.


Pari,'*, BattaiUe, 1892-1902. Qci-vcKE, H. Zur Physiologie der Cerebrospinalflussigkeit. Arch, f . Anat., Physiol, u. wissensch. Med. (Du Bois-Reymond), Leipzig, 1872, 153. Refobd, L. L. Unpublished. Work referred to by Sabin (49) and Gushing (9). Reiner, M., and J. Schnitzler. Ueber die Abflu.sswege des Liquor cerebrospinalis. Centralblatt f. Physiol., Leipzig and Wien, 1894, viii, 6H4. . Zur Lehre vom Hirndruck. Wiener klin. Wochenschr.. 1895, viii, 371.

Sabin, F. R. Development of lymphatic systcm. In Keibel and Mall's Manual Human Embryology, London 4 Philadelphia, 1912, ii, 709. Salvi, G. Histogenise et structure des meninges.


Thise de Pari.", 1898. . Quoted in Poiricr et Charpy. Traitt d'Anatomie humaine. Paris, 1901, tome iii, 113. ScHROEDER VAN DER KoLK, J. L. C. Bau und Functionen der Medulla spinalis und oblongata. Aus dem Holliindischen iihertragen von Dr. Fr. W.


Theile. Braunschweig, Vieweg & Sohn, 1859. Spin.i, a. Expcrimenteller Beitrag zur Kenntniss der Hyperamic des Gehirns. Wiener med. Blatter, 1898, XXI, 17; 247. . Experimentelle Untcrsuchungen iiber die Bildung des Liquor cerebrospinalis. Arch. f. d. gesamte Physiol., Bonn, 1899, Lxxvi, 204.


52. Spina. A. Uebcr den Einfluss des hohen Blutdrucks auf die Neubildung des CerebrospinalflQssigkeit. Arch, f. d. gcsamtc Physiol., Bonn, 1900, i.xxx, 370.


. Untersuchungen uber die Resorption des Liquor bei normulcm und crhohtem intracraniellem Drucke. .Vrch. f. d. gesamte Physiol., Bonn, 1900-1901, Lxxxiii, 120; 415.


53. .'^TERZi, G. Recherches sur I'anatomie comparfee et sur rontogent-se des m6ninges. Arch. Ital. de Biol., 1902, XXXVII, 257.


. Ricerche inforno all' anatomia comparata ed air ontogenesi dcllc mcningi, e considerazioni suUa filogenesi. Atti del R. Instituto Veneto di scienze, letterc ed .\rti, 1900-1901, LX, parte ii, 1101.


54. Streeter, G. L. Development of the nervous system. In Keibel and Mall, Manual Human Embryology, London & Philadelphia, 1912, ii, 1.

. The development of the venous sinuses of the dura mater in the human embryo. Am. Jour, of Anat., Philadelphia, 1915, xviii, 145.


55. Weed. L. H. Studies on cerebro-spinal fluid. No. II: The theories of drainage of cerebro-spinal fluid with an analysis of the methods of investigation. Jour.Med. Research, Boston, 1914, xxxi (n. s. xxvi), 21. . Studies on cerebro-spinal fluid. No. Ill: The pathways of escape from the subarachnoid spaces, with particular reference to the arachnoid Wlli. Jour. Med. Research, Boston, 1914, xxxi (n. a. XX VI), 51. . Studies on cerebro-spinal fluid. No. IV: The dual source of cerebro-spinal fluid. Jour. Med. Research. Boston, 1914. xxxi (n. s. xxvi). 93.


56. Wilder, B. G. Notes on the Foramina of Mageudie in man and the eat. Jour. Nerv. and Ment. Dis., New York, 1886, xiii, 206. . The metapore (foramen of Magendie) in man and an orang. Med. News, Philadelphia, 1893, LXiii 439. . Meninges. Ref. Handbuch Med. Sciences, New York, 1893, ix (suppl.), 606.


57. ZiEOLER, P. Ueber die Mechanik des normalen und pathologischen Hirndruckes. Arch. f. klin. Chirurgie, Berlin, 1896, Liii, 75.

Explanation of Plates

KEY FOR FIGURE-LEGENDS.


ami, area membranacea inferior. dmc, dura mater cerebri (inner surface, in pme, pia mater cerebri.


amt, area membranacea .superior. approximation with arachnoid). j»p6, precipitated prussian-blue.


cbl, cranial blastema. epe, epithelial-like cells lining ventricle. pun, reduced silver nitrate.


cent, cisterna cerebello-mcduUari.i. epe, ependyma. s<u, subarachnoid spaces.


chp, plexus chorioiduus. 4"^, vcntriculus quarlus. tir, sinus trans\ersu3.


Plate I.

Weed1917 plate01.jpg

Fig. 1. Drawing of a pig embryo of 9 mm, into the spinal central canal of which an injection of 0.5 per cent solution of potassium ferrocyanide and iron-ammonium citrate was made under very mild syringe-pressure. The embryo was fixed in Camoy's fluid to which 1 per cent hydrochloric acid had been added. The specimen was carefully dehydrated and cleared by the Spalteholz method. The resultant precipitate of prussian-blue is found w holly within the central canal of the spinal cord and within the cerebral ventricles. Enlargement, 11 diameters.


Fig. 2. Drawing of a pig embryo of 13 mm, in which the cerebro-spinal fluid was replacc<l by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. The embryo was kept alive for 90 minutes after this replacement and was then fixed in 10 per cent formol containing 1 per cent hydrochloric acid, .\fter dehydration the specimen was cleared by the Spalteholz method. The occiurence of a definite oval, outlined by the denser mass of the granules, in the roof of the fourth ventricle, is characteristic of this stage. Enlargement, 9 diameters.

Fig. 3. Drawing of a pig embryo of 14.5 mm, in which the cerebro-spinal fluid was likewise replaced by the ferrocyanide solution, After the replacement, the embryo was kept alive for 60 minutes; it was fixed in Camoy's fluid (with 1 per cent hydrochloric acid added) and after dehydration it was cleared by the SpaltehoU method. The earliest indications of a peria.idal spread of the replaced fluid from the roof of the fourth ventricle is here shown. Enlargement, 8 diameters.

Plate II.

File:Weed1917 plate02.jpg

Fig. 4. Drawing of a pig embryo of 18 mm., in which a typical replacement of the spinal fluid had been made. The animal was kept alive for 45 minute." and was then fixed, dehydrated, and cleared in the usual manner. The extra ventricular spread of the replaced fluid from two are;is in the roof of the fourth ventricle is well illustrated. Enlargement, 9 diameters.


Fig. 5. Drawing of a pig embryo of 19 mm., in which likewise a typical replacement of the cerebro-spinal fluid by the ferrocyanide solution had been made. After this procedure, the embryo was kept aUve for 55 minutes and was then carried through the routine technique for the Spalteholz method. The ftirther pericerebral spread of the replaced fluid is recorded. Enlargement, 8 diameters.


Plate III.

File:Weed1917 plate03.jpg

Fig. 6. A frank lateral drawing of a pig embryo of 21 mm. The typical replacement of the embryonic cerebro-spinal fluid by the ferrocyanide solution was effected in this embryo and it was then kept alive for 45 minutes. At the end of this time the embryo was fi.xed in an acid fluid, dehydrated, ana cleared. The almost complete periaxial spread of the replaced fluid is indicated by the precipitated granules. Enlargement, 7.6 diameters.


Fig. 7. A dorsal view of the embryo illustrated in fig. 6. The perispinal spread of the replaced fluid is well shown. Enlargement, 7.8 diameters.


Plate IV.

File:Weed1917 plate04.jpg

Fig. 8. Drawing of a pig embryo of 26 mm. in which the typical replacement of the cerebro-spinal fluid has been made. After the introduction of the ferrocyanide solution the embryo was kept alive for one hour; at the end of this time it was fixed in an acid solution, subsequently dehydrated, and cleared in oil of wintergreen. The specimen shows a complete periaxial spread of the replaced fluid, as evidenced by the precipitated granules, in addition to a total filling of the intramedullary system. Enlargement, 6.5 diameters.


Fig. 9. Drawing of a pig embryo of 16 m.m., in which the central canal of the spinal cord was injected with the ferrocyanide so.u'ion under moderate syringe-pressure. After fixation in an acid mediiun the embn,-owaa dehydrated and cleared by the Spalteholz method. The extraventricular snread in the peribulbar region Ls easily made out. Enlargement, 9 diameters.


Plate V.

File:Weed1917 plate05.jpg

Fig. 10. Drawing of a pig embryo of 21 mm., in which an injection of diluted india ink was made into the central canal of the spinal cord. The preasure employed was the highest obtainable from the syringe, yet below the tension causing rupture. The specimen, after injection, was fixed, dehydrated, and cleared. The slight extent of the periaxial spread of the carbon granules can be easily seen. Enlargement, 7 diameters.


Fig. 11. Drawing of a pig embryo of 16 mm., in which an injection (under moderate syringe-pressure) of 0.5 per cent solution of silver nitrate was made into the central canal of the spinal cord. The silver was reduced in the sunlight, the embryo then fixed. After dehydration, the embryo was cleared in benzol and oil of wintergreen. Enlargement, 7. .5 diameters.


Fig. 12. Drawing of a pig embryo of 13 mm.; into the central canal of the spinal cord a dilute solution of nitrate of silver was injected under strong syringe-pressure. Reduction of the silver was accomplished by exposure to sunlight; the embryo was then fixed, dehydrated, and cleared. Enlargement, diameters.


Plate VI.

File:Weed1917 plate06.jpg

Fig. 13. Photomicrograph of transverse section of a pig embryo of 15 mm. Specimen obtained from an embryo in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and ironammonium citrate. After this replacement the embryo was kept aUve for 65 minutes. The resultant priL'Ssian-blue precipitate is not included in this photomicrograph. Enlargement, 13 diameters.


Fig. 14. Drawing of blocked area in fig. 13, under higher magnification and including the resultant precipitate of prussian-blue. The typical ependymal cells (epc) lining the fourth ventricle are shown on either side; between them occurs the area membranacea superior {ams). The transit of the replacement fluid through the membranous area and the spread through the adjacent mesenchyme are illustrated. Enlargement, 245 diameters.


Fig. 15 Photomicrograph of transverse section from embryo pig illustrated in fig. 13. Section taken from more caudal plane than that given in the former figure. The prussian-blue spread is not illustrated. Enlargement, 10 diameters.


Fig. 16. Drawing, under higher magnification, of the rectangular area in fig. 15. The passage of the replaced solution, as shown by the resultant precipitate of prussian-blue, through tlie area membranacea inferior (ami) is here illustrated. The extension of the replaced fluid through the adjacent mesenchyme and the nonpenetration of the solution into the condensed mesenchyme are shown. Enlargement, 140 diameters.


Flg. 17. Photomicrograph of sagittal section of a pig embryo of 18 mm. Specimen obtained from an embrj-o in which the cerebro-spinal fluid was replaced by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate. After this replacement the animal was kept alive for 45 minutes. Fixed for 5 minutes in 10 per cent formol containing 1 per cent hydrochloric acid; then over night in modified Boiiin's solution (saturated aqueous solution of picric acid 75, formaldi-hyde 10, glacial acetic arid 10). Dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraffin. Serial sections, stained by hematoxylin and eosin. The resultant precipitate of prussian-blue has not been reproduced in the photomicrograph. Enlargement, 8 diameters.


Fig. 18. Drawing of blocked area in fig. 17 under higher magnification. The granules of prussian-blue are here represented by the blue stenciling. The transit of the fluid, as shown by the granules, into the periaxial mesenchyme through the two membranous areas {arm and ami) in the roof of the fourth ventricle are well shown. Enlargement, 35 diameters.


Plate VII.

Weed1917 plate01.jpg

Fig. 19. Photomicrograph from a sagittal section of a fetal pig of 27 mm. The cerebro-spinal fluid in this specimen was replaced by a 1 per cent solution of potassium ferrocyanide and iron-ammonium citrate; the fetus was kept alive for 40 minutes; fixed in 10 per cent formol containing 1 per cent hydrochloric acid for 15 minutes; then over night in modified Benin's solution; dehydrated by 2 and 4 per cent grades of alcohol; embedded in xylol-paraffin. The prussian-blue granules are not represented in this photomicrograph. Enlargement, 8 diameters.


Fig. 20. Drawing of squared area in fig. 19. The center of the field is occupied by the optic nerrve; around it the developing extrinsic optic muscles are shown. The precipitate of prussian-blue occurs in the perineural mesenchyme. Enlargement, 190 diameters.


Fig. 21. Photomicrograph of rectangular area in fig. 19. The passage of the ferrocyanide solution into the sinus transvcrsus {sir) is represented by the precipitated blue granules. Enlargement, 133 diameters.


Fig. 22. Photomicrograph of a transverse section of a pig enibrj-o of 23 mm. The cerebro-spinal fluid was replaced in this embr>'o with a 1 per cent solution of potassium fenocyanide and iron-ammonium citrate. The embryo was kept alive for .50 minutes and was then fixed over night in 10 per cent formol containing 1 [ler cent hydrochloric acid. The granules of pnissian-blue are not shown in this reproduction. Enlargement, 13 diameters.


Fig. 23. Drawing of squared area in fig. 22. The area membranacea superior (rima) is shown, sunounded on either side by tufts of the chorioid plexus (rhp) .ind the typical ventricular ependyma. The transit of the solution is shown, as represented by the resultant granules, through the areji, with the subsequent spreail into the periaxial mesenchyme. Enlargement, 125 diameters.


Plate VIII.

File:Weed1917 plate08.jpg

Fig. 24. Photomicrograph of a transverse section of a pig embryo of 8 mm. Fixed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent grades of alcohol, embedded in xylol-parafiin. Enlargement, 30 diameters.


Fig. 25. Photomicrograph, retouched, of the blocked area in fig. 24. The character of the cells (epc) composing the roof of the fourth ventricle Uve) is showTi in this reproduction. Enlargement, 16.5 diameters.


Fig. 26. Photomicrograph of a sagittal section from a pig embryo of 11 mm. Fi.xed in modified Bouin's solution over night, dehydrated by 2 and 4 per cent gracles of alcohol, embedded in xylol-parafhn. Enlargement, 11 diameters.


Fig. 27. Photomicrograph of the blocked area in fig. 26. The area membranacea superior (ams) in the roof of the fourth ventricle is shown sharply delimited from the two processes of typical ependyma {epe). Enlargement, 67 diameters.


Fig. 28. Photomicrograph of a more lateral section of the pig embryo of 11 mm. given in fig. 26. Enlargement, 11 diameters.


Fig. 29. Photomicrograph, under higher magnification, of the blocked area in fig. 28. The lateral border of the area membranacea superior {ams) of the roof of the fourth ventricle is given. Enlargement, 50 diameters.


Fig. 30. Photomicrograph of a sagittal section from a pig embryo of 13 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 8 diameters.


Fig. 31. Photomicrograph, under higher magnification, of the squared area in fig. 30. The reproduction comprises a sagittal section of the area membranacea superior (ams) of the roof of the fourth ventricle. Enlargement, 67 diameters.


Fig. 32. Photomicrograph of a sagittal section of a pig embryo of 14 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 11 diameters.


Fig. 33. Photomicrograph of the blocked area in fig. 32 under higher magnification. The area membranacea superior (ami) in the roof of the fourth ventricle is reproduced. Enlargement, 75 diameters.


Plate IX.

File:Weed1917 plate09.jpg

Fig. 34. Photomicrograph of a transverse section of a pig embryo of 18 mm. Fixed in Camoy's fluid (6:3:1), deh3'drated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffin. Enlargement, 13 diameters.


Fig. 35. Photomicrograph, under higher magnification, of the blocked area in fig. 34. The area membranacea superior (atns) is here given, flanked on either side by typical ependyma (ept). Enlargement, 170 diameters.


Fig. 36. Photomicrograph of a transverse section of a pig embrj-o of 18 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 13 diameters.


Fig. 37. Photomicrograph of rectangular area outlined in fig. 36. The extent of the area membranacea superior (ams), with its adherent coagulum of albuminous material, is well differentiated from the adjacent typical ventricular ependyma (epe). Enlargement, 100 di.imeters.


Fig. 38. Photomicrograph of a transverse section of a pig embryo of 19 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades, and embedded in xylol-paraffin. Enlargement, 13 diameters.


Fig. 39. Photomicrograph, under higher power, of the rectangular area in fig. 38. A small break in the integrity of the lining ependyma of the roof of the fourth ventricle, representing the irregular boundary of the area membranacea superior (ams), is given. Enlargement, 290 diameters.


Fig. 40. Photomicrograph of a transverse section of a human embryo of 4 mm. (No. 836 of collection of Carnegie Institution of Washington). Enlargement, 33 diameters.


Fig. 41. Photomicrograph, retouched, of the blocked area in fig. 40. The epithelial-like cells (epe) composing the roof of the fourth ventricle (4^) are here shown separated from the denser nervous tissue. Enlargement , 100 diameters.

Plate X.

File:Weed1917 plate10.jpg

Fig. 42. Photomicrograph of transverse section of pig embryo of 19 mm. Fixed over night in modified Bouin's solution, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in x>-lol-paraffin. Enlargement, 13 diameters.


Fig. 43. Photomicrograph of squared area in figure 42, under higher magnification. The area membranacea superior (ams) with the attached coagulum of albumen is reproduced. Enlargement, 115 diameters.


Fig. 44. Photomicrograph of sagittal section of pig embryo of 23 mm. Fuced in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Enlargement, 5 diameters.


Fig. 45. Photomicrograph, under higher magnification, of squared area in fig. 44. The area membranacea superior (atns) is here shown, delimited by the cells of the chorioid plexus (chp) on one side and by the further ependymal prolongation (epc) of the cerebellar lip. Enlargement, 88 diameters.


Fig. 46. Photomicrograph of sagittal section of pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent changes, and embedded in xylol-paraffin. Certain portions of the dura mater (dmc) are indicated. Enlargement, 5 diameters.


Fig. 47. Photomicrograph of blocked area in fig. 46, under higher magnification. The small remaining area membranacea superior (ams) is quite surrounded by encroaching ependyma in the chorioidal folds. Enlargement, 88 diametere.


Fig. 48. Photomicrograph of transverse section of human embryo of 7 nun. (No. 617 of the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.

Fig. 49. Photomicrograph of squared area in fig. 48, under higher magnification. The epithelial-like cells (epc) composing the roof of the fourth ventricle at this stage are well shown. Enlargement, 100 diameters.


Fig. 50. Photomicrograph of transverse section of human embryo of 7 mm. (No. 617 in the Caniegie Institution of Washington). Enlargement, 10 diameters. Fig.


Fig. 51. Photomicrograph of blocked area in fig. 50. The marked invagination of the roof of the fourth ventricle (4if ) with the lining of epithelial-like cells (epc) is given. Enlargement, 33 diameters.

Fig. 52. Photomicrograph of transverse section of human embryo of 9 mm. (No. 721 in the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.


Fig. 53. Photomicrograph of squared area outlined in fig. 52. The pale, large cells {epc) comprising the roof of the fourth ventricle characterize the reproduction. Enlargement, 50 diameters.


Fig. 54. Photomicrograph of sagittal section of human embryo of 11 mm. (No. 544 in the collection of the Caniegie Institution of Washington). Enlargement, 6 diameters.


Fig. 55. Photomicrograph of blocked area in fig. 54. The apparent break in the continuity of the roof of the fourth ventricle with exudation of the ventricular albumen into the mesenchyme is brought out. Enlargement, .50 diameters.


Plate XI.

File:Weed1917 plate11.jpg

Fig. 56. Photomicrograph of sagittal section of human embryo of 14 mm. measured on the slide (No. 144 of the collection of the Carnegie Institution of Washington). Enlargement, 8 diameters.


Fig. 57. Photomicrograph, under higher magnification, of blocked area in fig. 50. The greater part of the ventricular wall shown is composed of the area membranacea superior (ams), bounded below by typical ventricular ependyma {epe). Enlargement, 67 diameters.

Fig. 58. Photomicrograph of sagittal section of human embryo of 17 ram. (No. 576 of the collection of the Carnegie Institution of Washington). Enlargement, 10 diameters.

Fig. 59. Photomicrograph of rectangular area in fig. 58, showing the area membranacea superior {ams) of the roof of the fourth ventricle. Enlargement, 50 diametere.


Fig. 60. Photomicrograph of sagittal section of human embryo of 17 mm. (No. 576 of the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.

Fig. 61. Photomicrograph of tlie blocked area in fig. 60 under higher magnification. The aggregation of epithehal-like cells (epc) on the lateral border of the area membranacea superior is here portrayed. Enlargement, 67 diameters.


Fig. 62. Photomicrograph of transverse section of human embryo of 18 mm. (No. 409 of the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.


Fig. 63. Photomicrograph, under liigher power, of squared field in fig. 62. The peculiar inversion of the roof of the fourth ventricle (/fVe) indicated in fig. 62, has resulted in a marked dislocation of the area membranacea superior (nm«), shown in this figure. Enlargement, 75 diameters.


Plate XII.

File:Weed1917 plate12.jpg

Fig. 64. Photomicrograph, retouched, of a transverse section of a human embryo of 21 mm. (No. 460 of the collection of the Carnegie Institution of Washington). The field taken consists of a portion of the fourth ventricle with the lining of typical ependyma (cjk) on either side. The area membranacea superior {ams) is shown between the two lips of ependyma. Enlargement, 33 diameters.


Fig. 65. Photomicrograph, retouched, of a similar section to that given in fig. 64, but taken from a more anterior plane from the same embryo. The field shown is analogous in everj' way to that in the preceding figure.


Fig. 66. Photomicrograph of a transverse section of an embryo chick of 121 hours' incubation. Fixed in Bouin's solution. Enlargement, 15 diameters.


Fig. 67. Itetouched photomicrograph, under higher magnification, of the blocked area in fig. 66. The area memhranacoH superior (ams) is here given, delimited sharply from the lips of ependyma {cpc) which line the rtmf of the fourth ventricle. Enlargement, 133 diameters.


Fig. 68. Photomicrograph of a more caudal section from the same embryo as portrayed in fig. 66. Enlargement, 15 diameters.


Fig. 69. Retouched photomicrograph, under higher magnification, of the blocked area in fig. 68. The area membranacea superior {am») is shown at the point of its greatest transverse diameter. Enlargement, 88 diameters.


Fig. 70. Photomicrograph of a migittul section of a pig embryo of 15 mm. Fixed in modlfied Bouin's solution, dehydratiil by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement , S diameters.


Fig. 71. Phutomicrograph, under higher magnification, of blocked area in fig. 70. The earliest evidence of the area membranacea inferior (ami) in the roof of the fourth ventricle is here shown. Enlargement, 12.'i diametere.


Fig. 72. Photomicrograph of sagittal section of a pig embryo of 18 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 11 diameters.


Fig. 73. Photomicrograph, under higher power, of the rectangular area outlined in fig. 72. The enlarging area membranucea inferior (ami) is shown in the midst of the typical lining ependyma of the roof. Enlargement, 100 diameters.


Fig. 74. Photomicrograph of sagittal section of a pig embryo of 23 mm. Fixed in modified Bouin's solution, dehydrated by 2 :\iid 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 6 diameters.


Fig. 75. Photomicrograph of blocked area in fig. 74. The area membranacea inferior {ami) is, at this stage, quite extensive, as shown in the reproduction; the early stages in the development of the cistema cerebellomedullaria may also be seen. Enlargement, 75 diameters.


Fig. 76. Photomicrograph of sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. Enlargement, 7 diameters.


Fig. 77. Photomicrograph, under higher magnification, of blocked area in fig. 76. The unsupported character of the area menibranarca inferior and the formation of the cistema cerebello-medullaris is here reproduced. Enlargement, 67 diameters.


Plate XIV.

File:Weed1917 plate14.jpg

Fig. 78. Photomicrograph of a sagittal section of a pig embryo of 32 mm. Fixed in modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xj'lol-parafiin. Enlargement, 7 diametera.


Fig. 79. Photomicrograph of the blocked area in fig. 78, under higher magnification. The intact area membranacea inferior (ami), unsupported by any mass of tissue, is shown separating the ventricular cavity from the developing cisterna ccrebello-medullaris. Enlargement, 67 diameters.


Fig. 80. Photomicrograph of a sagittal section of a human embryo of 16 mm. (No. 406 in the collection of the Carnegie Institution of Washington). Enlargement, 7 diameters.


Fig. 81. Photomicrograph of the area outlined in fig. 80, but under liigher magnification. An early stage in the differentiation of the area membranacea inferior (ami) is given. Enlargement, 50 diameters.


Fig. 82. Photomicrograph of a sagittal section of a human embryo of 17 mm. (No. 576 in the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.


Fig. 83. Photomicrograph, under higher power, of the area blocked in fig. 82. The chorioid plexuses of the fourth ventricle Ue in the central portion of the field; above is the thick cell-layer on the lateral side of the area membranacea suf)erior (ams), while below the upper limit of the area membranacea inferior (ami) appears. Enlargement, 67 diameters.


Fig. 84. Photomicrograph of a transverse section of a human embryo of 18 mm. (No. 409 in the collection of the Carnegie Institution of Washington). Enlargement, 5 diameters.


Fig. 85. Photomicrograph of the blocked area in fig. 84. The cellular character, and especially the clumping of cells, of the area membranacea inferior (ami) is shown. Enlargement, 25 diameters.


Fig. 86. Photomicrograph of a sagittal section of a human embryo of 19 mm. (No. 431 in the collection of the Camegie Institution of Washington). Enlargement, 5 diameters.


Fig. 87. Photomicrograph of the blocked area outUned in fig. 86. The area membranacea inferior (ami) appears separating the fourth ventricle from the developing cistema cerebello-medullaris. Enlargement, 25 diameters.


Plate XV.

File:Weed1917 plate15.jpg

Fig. 88. Photomicrograph from a sagittal section of a human embryo of 17 mm. (No. 576 of the collection of the Carnegie Institution of Washington), representmg an enlargement of the second blocked area in fig. 58. The area membranacea inferior (ami) appears sharply delimited from the adjoining tj^pical ependjina. Enlargement, 67 diameters.


Fig. 89. Photomicrograph of a sagittal section of a human embryo of 23 mm. (No. 453 of the collection of the Carnegie Institution of Washington). Enlargement, 6 diameters.


Fig. 90. Photomicrograph of the blocked area in fig. 89. The area membranacea superior (ams) appears in the stage of closure, while the area membranacea inferior (ami) is becoming well differentiated from the typical ependyma lining the other portions of the fourth ventricle. Enlargement, 26 diameters.


Fig. 91. Photomicrograph of a sagittal section of a human embryo of 26 mm. (No. 1008 of the collection of the Carnegie Institution of Washington). Enlargement, 4.5 diameters.


Fig. 92. Photomicrograph, under higher magnification, of the blocked area in fig. 91 . The area membranacea superior has been almost completely closed by the dense ependyma of the superior half of the roof of the fourth ventricle, while the inferior area (ami) has become a membrane lacking wholly the character of ependyma. Enlargement, 23 diameters.


Fig. 93. Photomicrograph of a sagittal section of a human embryo of 35 mm. (No. 199 of the collection of the Carnegie Institution of Washington). Enlargement, 3 diameters.


Fig. 94. Photomicrograph, under higher powers, of the blocked areas in fig. 93. The formation of the cistema cerebello-medullaris is shown in relation to the ventricular roof. Enlargement, 23 diameters.


Fig. 95. Drawing of cells of the chorioid plexus from the lateral ventricles of a fetal pig of 132 mm. The specimen waa fixed in absolute alcohol, and stained by Best's carmine stain for glycogen. The glycogen occurs in the form of globules within the epithelial cells. Enlargement, 950 diameters.


Fig. 96. Drawings of the cells of the chorioid plexus from the lateral ventricles of a fetal pig of 36 mm. The specimen was fixed in absolute alcohol and stained by Best's carmine method. The glycogen appears in the epitheUal eclls in the form of basilar plaques. Enlargement, 950 diameters.


Plate XVI.

File:Weed1917 plate16.jpg

Fig. 97. Photomicrograph of a transverse section of a pig embryo of 10 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent f;radcs of alcohol, and embedded in xylol-paraffin. Enlargement, 10 diameters.


Fig. 98. Photomicrograph, under higher magnification, of the blocked area in fig. 97. The double condensations of mesenchyme to form pia mater (pmc) and cerebral blastema (cbl) appear separated by a region of mesenchyme which is breaking down. This central area of mesenchyme, with the marked albumen-content, is to become the arachnoid spaces. Enlargement, 133 diameters.


Fig. 99. Photomicrograph of a transverse section of a pig embryo of 20 mm. Fixed in modified Bouin's fluid, dehydrated by 2 and 4 per cent changes of alcohol, and embedded in xylol-paraffm. Enlargement, 10 diameters.


Fig. 100. Photomicrograph, under higlior powers, of the blocked areas in fig. 99. The relations of the pial condpn.sation ipmc) of mesenchyme to tlie nervous system, as well as the infiltration of tlie arachnoid mesenchyme (sas) with albumen, is reproduced. Enlargement, 133 diameters.


Fig. 101. Photomicrograph, under higher magnification, of the blocked area in fig. 22. The reproduction is included here to show the double condensjition {cht) of mesenchyme which goes to form ultimately bone and possibly a portion of the dura. Enlargement, 132 diameters.


Fig. 102. Photomicrograph of a transverse section of a pig embryo of 18 mm. The embryo was one in which the cerebro-spinal fluid was replaced by the ferrocyanide solution. Subsequently the embryo was fixed in 10 per cent formol containing 1 per cent hydrochloric acid for a few minutes to precipitate the prussianblue. It was then transferred to modified Bouin's solution, dehydrated by 2 and 4 per cent grades of alcohol, and embedded in xylol-paraffin. The granules of prussian-blue are not represented in this figure. Enlargement, 10 diameters.


Fig. 103. Photomicrograph of the squared area in fig. 102. The relation of the thinning mesenchyme in the arachnoid areas to the caudal cranial nerves is shown. The granules of prussian-blue, scattered through the area of thin mesenchyme (sas), are not reproduced. Enlargement, 40 diameters.


Fig. 104. Photomicrograph of a coronal section of a tissue block which includes the meninges and cerebral cortex in the region of the sinus sagittalis superior. The block was obtained from a fetal pig of 80 mm., fixed in Zenker's fluid, and stained, after embedding in celloidin, by Mallory's technique for coimective tissue. Enlargement, 27 diameters.


Plate XVII.

File:Weed1917 plate17.jpg

Fig. 105. Photomicrograph of a coronal section of a tissue block including cerebral cortex and meninges in the region of the sinus sagittalis superior. The block was obtained from a fetal pig of 10 cm., fixed in Zenker's fluid, and stained by Mallory's technique for connective tissue. Enlargement, 13 diameters.

Fig. 106. Photomicrograph of a coronal section, similar to that in figs. 104 and 105, except in that it was obtained from a fetal pig of 17 cm. The same technical procedures employed in the other specimens were used in this. Enlargement, 27 diameters.

Fig. 107. Photomicrograph of a similar section to those of the foregoing figures. The specimen was obtained from a fetal pig of 20 cm. and was treated in the manner outlined above. Enlargement, 20 diameters.


Fig. 108. Drawing of the cell pattern from the inner surface of the dura mater of a fetal pig of .5 cm. The specimen was prepared by the reduction of a dilute solution of silver nitrate in sunlight. The preparation was subsequently stained by hematoxylin. Enlargement, 190 diameters.


Fig. 109. Drawing of a preparation, similar to that of fig. 108, but obtained from the inner surface of the dura mater of a fetal pig of 7.5 mm. Enlargement, 28.5 diameters.


Fig. 110. Drawing of a preparation, similar to those of figs. 108 and 109, obtained from the inner surface of the dura mater of a fetal pig of 90 mm. Enlargement, 28.5 diameters.


Fig. 111. Drawing of a preparation from the inner surface of the dura mater of a fetal pig of 16 cm. The specimen was made in the same manner as outlined in fig. 108. Enlargement, 285 diameters.


Fig. 112. Photomicrograph of a sagittal section of a pig embryo of 17 mm. An injection of an 0.5 per cent solution of nitrate of silver was made into the central canal of the spinal cord; the silver was reduced in sunlight and the embrjo fixed in formalin. Enlargement, 13 diameters.

Fig. 113. Photomicrograph, under higher powers, of the blocked areas in fig. 112. The accumulation of the reduced silver (p»n) again.st the area membranacea superior is represented in black. Enlargement, 117 diameters.


Fig. 114. Photomicrograph of a transverse section of a pig embryo of 19 mm. An injection of 0.5 per cent solution of silver nitrate was made into the central canal of this embryo and the silver immediately reduced in sunlight. The embryo waa fixed in formalin, carefully dehydrated, and embedded in xylol-peiraflBn. Enlargement, 10 diameters.


Fig. 115. Photomicrograph, under higher magnification, of the blocked area in fig. 114. The collection of reduced silver {psn) against the cells at the inferior end of the area membranacea superior is illustrated. Enlargement, 100 diameters.


Fig. 116. Photomicrograph of a tranverse section of a pig embryo of 16 mm. The central canal of the spinal cord of this embryo was injected with a 1 per cent ferrocyanide and citrate solution under mild syringe-pressure; the embryo was then fixed in 10 per cent formol containing 1 i>er cent hydrochloric acid. Enlargement, 10 diameters.


Fig. 117. Photoniicrogrnph of the blocked area in fig. 116, under higher magnification. The accumulation of the precipitated injection fluid against the area membranacea superior is represented in black. A slight extravcntriculai Bprciul of the fluid, which is found in this as in all embryos of this stage, can not be miule out in the reproduction. Enlargement, 07 diamuters.


Cite this page: Hill, M.A. (2024, April 24) Embryology Book - Contributions to Embryology Carnegie Institution No.14. Retrieved from https://embryology.med.unsw.edu.au/embryology/index.php/Book_-_Contributions_to_Embryology_Carnegie_Institution_No.14

What Links Here?
© Dr Mark Hill 2024, UNSW Embryology ISBN: 978 0 7334 2609 4 - UNSW CRICOS Provider Code No. 00098G
Retrieved from ‘https://embryology.med.unsw.edu.au/embryology/index.php?title=Book_-_Contributions_to_Embryology_Carnegie_Institution_No.14&oldid=391999’
Powered by MediaWiki