Paper - Primitive neurons in the embryonic human central nervous system

From Embryology
Embryology - 17 Jun 2019    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)

Humphrey T. Primitive neurons in the embryonic human central nervous system. (1944) J. Comp. Neurol. 81(1): 1-45.

Online Editor  
Mark Hill.jpg
This historic 1944 paper by Humphrey describes early neural development in human.

See also earlier paper: Humphrey T. The development of the olfactory and the accessory olfactory formations in human embryos and fetuses. (1940) J. Comp. Neurol. 431-468.
Our understanding of neural development has improved significantly over the last 100 years, some current research looks at early molecular patterning mechanisms as well as the development of connections between different regions.

See the links below for the current notes pages.

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 | Postnatal | Postnatal - Neural Examination | Histology | Historic Neural | Category:Neural
Historic Neural Embryology  
1883 Nervous System | 1893 Brain Structure | 1892 Nervous System Development | 1900 fourth ventricle | 1905 Brain Blood-Vessels | 1909 corpus ponto-bulbare | 1912 nuclei pontis - nucleus arcuatus | 1921 Neural Development | 1921 Anencephaly | 1921 Brain Weight | 1921 Brain Vascular System | 1921 Cerebellum | 1922 Brain Plan | 1923 Neural Folds | 1904 Brain and Mind | 1904 Brain Structure | 1909 Forebrain Vesicle | 1922 Hippocampal Fissure | 1923 Forebrain | 1927 Anencephaly | 1934 Anencephaly | 1937 Anencephaly | 1945 Spinal Cord | Santiago Ramón y Cajal | Ziegler Neural Models | Historic Embryology Papers | Historic Disclaimer
Historic Disclaimer - information about historic embryology pages 
Mark Hill.jpg
Pages where the terms "Historic Textbook" and "Historic Embryology" 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 and interpretations may not reflect our current scientific understanding.     (More? Embryology History | Historic Embryology Papers)

Primitive Neurons In The Embryonic Human Central Nervous System

Tryphena Humphrey

Department Of Anatomy, University Of Pittsburgh, Pennsylvania

Eighteen Figures


The possible presence of a transitory primitive nervous mechanism in mammals, comparable to that described for amphibians by Coghill (’13, ’14 and elsewhere) and others, has been the subject of conjecture and speculation, but, up to the present time, it has not been reported. In connection with the studies 011 the development of behavior in mammals, the apparent lack of any such primitive nervous mechanism has been suggested by some observers (\Vindle, ’34, pp. 501-502; ’40, p. 146; Hanson, ’43, p. 100) as the reason for a totally different pattern of behavioral development than that observed for amphibians by Coghill. Indeed, the absence of primitive neurons apparently has been regarded by these observers as a sufficient justification for considering a similar pattern of behavioral development to be impossible. Other workers (Hooker, ’43, p. 15) in the field of mammalian behavior have not considered that even the fact that a primitive nervous mechanism has not been noted necessarily excludes a similar pattern of development of activity.

During the course of other studies in progress on the spinalcord it has been of considerable interest to note, incidentally, over a period of 3 or 4 years, isolated unipolar cells, obviously of sensory ganglion cell type, in the spinal cords -of fetuses of 16 mm., 26 mm., 89.5 mm. and 145 mm. crown—1'ump length. When, more recently, a unipolar neuron of obviously primitive character was seen in the thoracic cord of a 5—mm. human embryo, it seemed worth While to examine the suitable younger specimens available and report the findings. The present paper is concerned, then, with the description of such cells as have been observed and a discussion of their possible significance.

Materials And Methods

This study is based on human embryonic material ranging from 5 mm. to 14;") mm. in crown—rump length (table 1). The younger specimens were serially sectioned in toto; the last two series listed are of the spinal cord after removal. All of the series were transversely sectioned and are only a part of the collection of human embr_volog.>;ic material at the Department of Anatomy of the University of Pittsburgh.


N0. 01‘ (MR ‘ FVGTH APPR-OX1MA’l‘E THICKNESS OF 1~‘F.TI.'SES IN I\.“‘1‘“ 3 MENSTRUAI; sEo'r1oi~:s IN svmrx (“0IA1.zEC’[‘I(7N' ‘ * ‘ ‘ AGE3 MICRA

N 5.0 5 wks. 10 Activated protargol “ 1 16.0 7 wks. 5 Erythrosin and ‘roluidin blue

22.5 8 wks. 10 Activated protargol 22 26.0 8.5 Wks. 10 Er_vthrosin and toluidin blue 19 26.3 8.5 wks. 10 Pyridine silver 68 89.5 14.5 wks. 10 Erythrosin and toluidin blue 3 145.0 18.5 wks. 15 l"yridine silver

‘The numbered fetuses in this table are from the series on which cinematographic records of fetal behavior have been made by Hooker (’36, et seq.). The material was prepared by Dr. Ira D. Hogg, Mr. Reinhardt Rosenberg and Miss Beryl Dimick.

" The measurements, with the exception of Homo N. were made before fixation.

3 The approximate menstrual age was computed by the use of Streeter Tables (Streeter, ’20).

‘Method of Bodian (’36 and ’37).

A careful examination of each section of the nervous system of Homo N (5 mm.) was made from the level of the trigeminal nerve to the caudal end of the spinal cord, with both the 4-mm. and the 2 mm (oil immersion) objectives, using a binocular research microscope. All levels of the spinal cord of Home O (22.5 mm.) were examined with the 4-mm. objective. All sections of the spinal cord of Home 3 (1.45 mm.) were examined with the 16-mm. objective. The various neurons observed in the remaining‘ series listed were found in the course of other studies (involving particularly Cl—7 inclusive) as were part of those observed in Home N and in Home 3.

  • The material studied is part of the human embryological collection which forms the basis of the physiological studies on human prenatal behavior being carried on by Hooker (’39a, ‘39b, and elsewhere), These physiological and morphological studies, of which this paper

is publication no. 12, have been aided by grants from the Penrose Fund of the American Philosophical Society, the Carnegie Corporation of New York and the University of Pittsburgh.

In order to give the approximate level of the various cells figured and described, the different segments of the spinal cord were determined as accurately as possible by checking the root fibers in all material sectioned in toto. In Homo N, however, since root fibers are not present as yet at caudal levels, the determination was made on the basis of ganglionic crest segmentation cephalically and, behind the point where a constriction indicates the ganglia, on the recognition of somites. The results obtained were checked with the descriptions of Ingalls ( ’07) and of such standard publications as Keibel and Mall (’10). The spinal cord levels for the two oldest fetuses, from which the central nervous system was removed, have been approximately determined primarily by observation of the ventral horn cell columns of the enlargements.

The photomicrographs are all at the magnification of approximately 950 in order that the sizes of the neurons illustrated may be compared more readily. None of the photomicrographs has been retouched in any way. In certain instances, however, a supplementary drawing has been added either in order to show features not illustrated by the photograph or in order to combine the structures in two photographs or in neighboring sections. In all instances a key figure has been inserted to show the location of the neuron described. The author is particularly grateful to Dr. John C. Donaldson without whose skill in

photography many of the photomicrographs would not have been possible.

The location of each neuron illustrated, that is, the series number, slide, et cetera, is given with the figure description (footnote, p. 6).

In cases in which there is no figure, this information is included with the account in the manuscript.


Since any attempt to discuss the significance of the neurons studied here, or to homologize them with the transitory neurons described for lower vertebrates, necessarily involves some review of the literature, the pertinent papers will be referred to in the discussion. However, no attempt will be made to cover the entire field, particularly that dealing with these neurons in fishes. Extensive reviews of the early work are to be found in the contributions of Dahlgren (1897), Sargent (1898a, 1898b), Harrison (’01) and van der Horst (table on p. 173 of Ariéns Kappers, Huber and Crosby, ’36), and reference should be made to these if further information is desired.


Altogether three morphologic types of neurons— unipolar, bipolar and multipolar cells, none of which has been described previously in mammalian embryos of comparable ages—have been observed in the material studied. Of these neurons, the multipolar cells are fewest in number. Both bipolar and unipolar sensory cells have been more often seen and the latter has been found at a wide variety of ages. it seems better to consider consecutively each type of neuron for all ages in which it has been noted rather than to take up all the neurons observed at any given age. Consequently, the bipolar neurons will receive attention first.

Bipolar neurons of sensory ganglionic cell type

Definitely bipolar neurons of a sensory type have been seen at four stages, 5 mm., 16 mm., 22.5 mm., and 26 mm. In the first embryo only a few such cells have been identified; in the 22.5-mm. embryo, for which the entire spinal cord was examined, a total of seventy—one were found.

5 mm embryo (5 weeks)

The best differentiated bipolar cell seen in this embryo is located at C6. just within the mantle layer in the region at which dorsal root fibers will enter the left side of the spinal cord (fig. 1, A and C). Due to the thickness of the section (10 ,u) it is not readily seen. The long axis of the neuron is oriented from dorsolateral to ventromedial, and the fusiform cell body measures 11 pt by 6.6 ,u and has a large nucleus with relatively clear karyoplasm containing one large nucleolus. The cytoplasm is densely granular. A fine process, almost clear of granules where it leaves the cell body and therefore considered the axon, extends ventromcdially from the deeper pole of the neuron. The other process extends cephalad and dorsolaterally from the superficial pole of the neuron. It becomes very fine as it passes cephalad through the external limiting membrane, but then increases in caliber and, in the third section cephalad, lies again almost in the plane of sectioning. There it can be followed ventrolmedialwarcl through the ganglion to a point just. dorsal to the ventral root, Where it turns lateralward parallel to the Ventral root fibers. After a short course in this direction, the process appears to end as a tiny swelling. \Vit-h the exception of two small intervals, each about 7 ,u long, the course through the ganglion and along the ventral root occurs in one section. Since no other fibers are present in the ganglion as yet, there is little difficulty in identifying the peripheral course of the fiber.


A, Al, axon P1, P2, P3, processes described (or their point 0, cell overlying neuron described of origin) D, D1, D2, D3, D4, D5, dendrites or their R, right

points of origin S, S1, synapsing fibers

DH: dorsal 1101'“ SC, satellite cells DHC, typical dorsal horn cell SG, sensory ganglion DOES: d°"sa1 SGC, L, largest sensory ganglion cells at age DR, dorsal root figured E’ epldemus SGC, T, typical sensory ganglion cells at age EP, ependyma figured G, Nissl granules SO’ somite

L, left

LF, lateral funieulus

LP, pigment mass of lipochrom type M, external limiting membrane (or its

SU, surface of spinal cord VENT, ventral VH, ventral horn

location) VHN, typical ventral horn motor neuron at ML, marginal layer 383 figured N, cell body of neuron described VR, Ventral I'00t N0, notocord X, Y, Z, primitive motor neurons in the

NT, neural tube 16-mm. embryo

In this embryo the ganglionic crest is not completely segmented to form the sensory ganglia even at the best differentiated levels such as the mid-cervical region. At the level at which the intramedullary bipolar neuron just described is found, no root fibers connect the sensory ganglia with either the neural tube or the ventral root. It is only rarely that a root fiber connects the ganglion with the spinal cord at other levels (C3, for example). For the most part, there is little differentiation of cells Within the ganglia (fig. 1B).

Fig. 1 Illustrations of neurons from a 5-mm. human embryo with a key drawing to show their location. Activated protargol preparation. A, Photomicrograph of a bipolar sensory cell in the alar plate in the region of C6 (N:6—2-9: L). X 950. B, Photomicrograph of the sensory ganglion in the region of 06 (N: 6—1—4_: R) to show the undifferentiated character of the cells at this age. X 950. 0, Composite drawing to illustrate the structure, location and processes of the neuron in A. The dendrite is drawn in from the three sections rostral to that which contains the cell body.

In the lumbar region (L4) bipolar neurons have also been identified and, Where differentiable, are situated at the border of the cellular layer. At. this level, however, the bipolar cells are less well developed than the one found in the cervical region. One of these neurons at this level (N:8—3—1:R)‘-’ is almost opposite the sulcus limitans. It is fusiform in shape with a dorsolatcral-ventro— medial orientation, granular cytoplasm, a light staining nucleus with one nucleolus, and diameters of 9.6;». by 5.9 ,u. A neuron of similar nature lies

Fig. 2 lntramcdullary bipolar sensory neuron, and spinal ganglion cells, from a lti—nnn. human embryo. Erythrosin and toluidin blue preparation. X 950. A, Bipolar sensory neuron from the region of the lower medulla oblongata (1: 7-1-2: R). B, Spinal ganglion cells from the region of C6 where they are well differentiated at this age (1: 17~:2—4: R).

just dorsal to this one in the same section, but is more rounded (8.1 pt by 6.2 fl.) in outline and has more darkly staining cytoplasm. Undoubtedly other cells of bipolar nature and simi1a.r significance are present elsewhere in the spinal cord at this age, but it is possible to identify them satisfactorily only if the plane of the section coincides with the long axis of the neuron and it is not obscured by overlying cells. At lumbar levels the ganglionic crest is small and without constrictions to indicate the different ganglia.

16 mm (7 weeks)

One neuron of bipolar type (fig. 2A) was found in the lower medulla at the junction of the marginal and mantle layers. It is approximately the same size (14 p by 9.6 fl.) as the unipolar neuron seen in this embryo (p. 11) and lies ventral to the vagal fibers. One process extends ventromedially, the other dorsolaterally toward the emerging fibers of the vagus nerve. The

“ The first number, or letter, indicates the series used. This is followed by the slide number, number of the row of sections and number of the section in the row, in the order given. The letter at the end, R or L, indicates whether the neuron described is on the right or the left side of the central nervous system of the embryo. If there are no figures to illustrate the text material this information is inc-luded with the description; in all other instances it is given with tho figure description.

eccentric nucleus is granular and has a single large acidophilic nucleolus. The cytoplasm contains fine blue staining granules scattered uniformly throughout it.

At this age, spinal ganglion cells in the cervical region measure not over 12 p. in their greatest diameter and are bipolar (fig. 2B). The Nissl granules are fine, compactly arranged and give the scanty cytoplasm an intense dark blue stain with toluidin blue. The nuclei are denselygranular and contain two or more small acidophilic nucleoli.

22.5 mm (8 weeks)

A survey of the entire spinal cord for bipolar sensory cells revealed a total of 71, 47 on the right side and 24 on the left. The distribution of these in the transverse plane is shown in figure 3A; the cephalocaudal distribution is indicated by figure 3B. From the first figure it will be noted that all of these neurons lie either at or above the level of the sulcus limitans in the sensory portion of the spinal cord. The majority of them are just ventral to the dorsal root fibers, a somewhat lesser number lie dorsal to the root fibers and only very few were found as far ventralward as the sulcus limitans itself. Almost without exception, the bipolar neurons are situated within the mantle layer, but usually nearer the marginal layer than the ependyma. For the most part these neurons are strictly bipolar (fig. 4A) in type, but some of them show a definite change toward a unipolar cell type (fig. 4B) and others are entirely unipolar in nature. The neurons are the most numerous in the lower lumbar and sacral regions and fewest in the cervical region (fig. 3B).

The bipolar cells which lie ventral to the dorsal root fibers send one process dorsolaterally into the dorsal root and the other ventromedially (fig. 4A). Those located dorsal to the sensory root fibers send one process ventrolaterally toward the dorsal root fibers but the other usually extends dorsomedially. These neurons average 11.9 ,u. by 8.6 [A in size and vary from 18.4 [L to 8.1 ,u. in greatest diameter to 12.5 p. to 6.6;; in their smallest diameter. The nucleus is large, centrally placed, and contains one large nucleolus and a distinct chromatin net. The cytoplasm is finely granular, although less so at the point of origin of the central fiber, the axon, which is fine in character (A in fig. 4, A and B). In a few cases this process has been seen to bifurcate in a T— or Y-shaped type of division. The peripheral process is coarser in nature and granules are present where it leaves the cell (D in fig. 4, A and B).

The vast majority of the spinal ganglion cells at this age (fig. 4C) are smaller than the bipolar cells inside the cord, being about 11 ,u. or even less in diameter. They also have little cytoplasm and a low degree of nuclear differentiation. However, certain of the best differentiated spinal ganglion cells (compare fig. 4A and 4C) at comparable levels equal and even exceed in size (being as much as 16 ,u. in greatest diameter) the bipolar cells inside of the spinal cord. Otherwise, except for the presence of two or more nucleoli and a more darkly staining karyoplasm, they differ little from the intramedullary bipolar neurons.

26 mm (8.5 weeks)

A bipolar neuron was found at this age on the right side of the spinal cord (T5) above the level of the sulcus limitans and about half way through the mantle layer (A, fig. 5B). One process is pointed dorsalward and slightly lateralward, the other ventrolateralward. The nucleus has a distinct chromatin net and one large nucleolus (fig. 5A). The cytoplasm contains fine, deeply staining granules, uniformly distributed. The neuron measures 14.7 ,u. by

Fig. 3 A, A diagram of a cross section of the spinal cord to show the distribution of the intramedullary sensory ganglion cells in the dorsal horn region of the 22.5-mm. embryo. B, A diagram to illustrate the longitudinal distribution of the eveuty-one neurons of sensory type found in the spinal cord of the same embryo. In both diagrams each dot represents

Fig. 4 Photomicrographs of sensory neurons from a 22.5-mm. human embryo and an orientation drawing to show the location of these cells. Activated protargol preparation. Phot0micrographs, X 950. A, Bipolar neuron of sensory ganglion cell type from the region of T7 (0: 29-4-2: R). B, Neuron of the same type from T3 (0: 25—3—<2:L). In this case, however, the cell is changing from a bi- to a unipolar shape. 0, Spinal ganglion cells at the same level (T7) as the bipolar neuron in part A. D, A drawing to show the location of the neurons figured in A and B. In each case that part of the course of the dendrite which is present in adjacent sections is indicated by a broken line.

8.1a, a size slightly greater than that of the average neuron of similar type in the 22.5-mm. embryo. This cell is not so definitely bipolar as that in figure 50 but shows less indication of a change toward the unipolar type than does that in figure 4B for the 22.5-mm. embryo. Another bipolar neuron is present in T4 on the left side dorsal to the entering sensory root fibers and near the surface of the gray matter (C, fig. 5B). This neuron shows no change toward unipolarity and has less cytoplasm in which only a few very fine granules are found (fig. 50). Other intramcdnllary bipolar sensory neurons are present at this age as well. The thickness of the sections, however, together with lack of much differential staining, makes recognition difficult so that the entire spinal cord has not been examined.

The intramedullary bipolar sensory cells found in the 22.5-nnn. and 26-mm.

stages are obviously better developed than those at comparable levels of the 5—mm. embryo. This fact is indicated by the increased size of the cell body, the

Fig. 5 Bipolar neurons from the human spinal cord at the age of 8.5 weeks (26 mn1. C.R.L.). Erythrosin and toluidin blue preparation. Photomierographs, X 950. A, Bipolar neuron, from the fifth thoracic segment, which is undergoing early changes toward a unipolar form (22: 70—3—2:R). B, Drawing to show the location of the cells illustrated in A and C. C, Bipolar neuron from the region of T4 (22: 68-2-3: L).

greater length of the processes of the neurons, the staining reaction of the nucleus and the additional amount of cytoplasm. A comparison of the bipolar neurons within the spinal cord in each case with the spinal ganglion cells in the same embryo shows that, in all instances, the intramedullary bipolar cells are better developed than are the sensory ganglion cells. These dilferences in degree of ditferentiation are greatest in the youngest embryo, however, and have decreased markedly in the older ones for which these cells are described (22.5 mm. and 26 mm.) so that, although the majority of the sensory ganglion cells are distinctly less well diflcrentiated than the bipolar cells within the cord, the ganglion cells which are best developed show much less difierence even at 22.5 mm. It should be mentioned, however, that a typically unipolar neuron has been found as early as 16 mm., presumably as the result of the transformation of one of the intramedullary bipolar cells just described (1). 1] and fig. 6A).

Unipolar neurons of sensory ganglion cell type

Neurons which are distinctly unipolar in type have been found in the spinal cord in 16-mm., 26—mm., 89.5-mm. and 145—mn1. stages. The structure of these unipolar neurons and the occurrence of such cells in varying stages of transition from a bi» to a unipolar shape (figs. 4, 5 and 6) indicate that they represent a further stage in the development of the intramedullary bipolar sensory neurons just described in the 5—mm., 16-mm., 22.5-mm., and 26-mm. embryos. Other neurons which occur in the 5-mm. embryo, and might be considered essentially unipolar in nature, are of a different type — more nearly comparable to the cells of Rohon—Beard (p. 13). Consequenty these nerve cells will be considered later.

16 mm. (7 weeks). A single neuron of unipolar type (fig. 6A and C) was seen in the medulla, near the junction with the spinal cord, of the 16-mm. embryo. It is located on the left side, slightly ventral to the level of the sulcus

Fig. 6 A, Photoxuicrograph of a unipolar sensory neuron from the lower medulla oblongata. of a l6—mn1. human embryo (1: 8-2-13: L). Erythrosin and toluidin blue preparation. X 950. B, Photomierograph of a unipolar sensory neuron in T5 of the spinal cord of a 26—mm. embryo (222: 70-2-7: L). Erythrosin and toluidin blue preparation. X 950. 0, Drawing of the left half of the spinal cord at T4 to show the location of the cells figured in A and B.

limitans, about midway in the mantle layer. The cell body is large (14 ,u by 10.3 M) and ovoid in shape with the smaller pole pointed dorsolaterally toward the periphery of the medulla. From this pole a rather heavy process extends about 3 ;,a. dorsolaterally and cephalad. The nucleus is large, also ovoid, granular and contains a single nucleolus which takes an aeidophilic stain with erythrosin and toluidin blue. The cytoplasm is granular, particularly near the nucleus, where blue staining granules are massed irregularly, and at the larger pole of the cell, where a few blue staining granules are present at the periphery. The remainder of the cytoplasm is acidophilic in reaction. This cell is larger and more highly differentiated than are the spinal ganglion cells at this age (fig. 2B).

26 mm. (8.5 weeks)

A unipolar neuron has been identified in the fifth thoracic segment of this series. It is at the level of the sulcus limitans on the left. side, about one-third of the distance from the surface of the mantle layer to the ependyma (fig. 6, B and C). The cell body is large (14.7 ,u by 11 ,u.), ovoid in shape and the single process extends dorsolaterally and slightly cephalad from the smaller pole of the cell for about 17 ,u. in the plane of the section. The large oval nucleus has one acidophilic nucleolus and a fine chromatin net. The cytoplasm contains a. few blue staining granules which are not uniformly distributed but situated at the periphery.

The largest spinal ganglion cells at this level equal the intramedullary sensory neurons in size and also have an acidophilic nucleolus. Some are undergoing initial changes from bi- to unipolar shape. In all of them the Nissl granules are very fine and uniformly closely packed together in the scanty cytoplasm. The majority of the spinal ganglion cells are smaller than the unipolar cells within the spinal cord at this age and are strictly bipolar in type.

89.5 mm. (14.5 weeks)

One neuron of sensory ganglion cell type has been noted in the lower part of Cl at 14.5 weeks. This cell (fig. 7) is located iii the middle of the neck of the dorsal horn where it can be observed in two adjacent sections. The cytoplasm is filled with deeply staining, fine Nissl granules uniformly closely arranged. The large, centrally placed nucleus contains a single large nucleolus which has a basophilic staining reaction. The ovoid cell body measures 17.2 ,1; by 14. ,u., as compared with an average of 17.9 p. by 14.3 p. for the largest ganglion cells in the lumbar area (the only region in which a ganglion remains), and is obviously of sensory ganglion cell type although no process can be seen with the Nissl stain. This neuron is the only one of such character observed in this specimen. but no special search for them has been made.

Fig. 7 A, Photomicrograph of an intramedullary sensory ganglion cell in the first cervical segment of the spinal cord of a 14.5-week fetus. Erythrosin and toluidin blue preparation. X 950. B, Outline drawing of a transverse section of the spinal cord at the level of C1. The location of the neuron in A is shown by the dot in the right dorsal horn.

145 mm (18.5 weeks)

Only two sensory neurons of unipolar type were found upon examination of the entire cord of this fetus; both are on the left side. One of these neurons is in the region of T6, about midway in the gray matter of the neck of the dorsal horn (figs. 8 and 9D). It has a large ovoid cell body which measures 26.5 p. by 19.1 p. and can be identified in two sections. An average size for ten spinal ganglion cells at this level is 22.6 M by 18 ,u and the largest measures 28.7 pl. by 27.9 pi. l\’eurofibrils are plainly visible in the cytoplasm of this intramedullary unipolar sensory neuron and extend out into a process which is directed dorsolaterally and cephalad for 25.7 p. in the plane of the section before being broken. It cannot be recognized in the adjacent sections or seen to branch. Satellite cells are clearly differentiated in the spinal ganglion at this level (fig. 9C), but no comparable cells are present around the unipolar sensory cell in the spinal cord.

The second neuron of this type (fig. 9, LA and B) is much farther caudad, at about the level of S2. It is located just dorsal to the retrodorsolateral cell column, immediately ventral to the level of the central canal, but near the surface of the gray matter (fig. 9D). Although larger in size than the neuron just described (33.1,u. by 22.8 ;u.), this cell is not strictly unipolar in type. Instead, two processes come off near the same point. One of them, the axon, becomes very fine soon after leaving the cell body but can be followed for about 90,; dorsomedially toward the dorsal horn in the plane of the section. The other process, the dendrite, is coarse, but gradually narrows in caliber. It extends almost directly ventralward, but a little laterally, for about 90 }L among the ventral horn motor neurons. The greater part of its course is in the section just rostral to the cell body, but due to the much greater diameter of the process it is readily distinguished. Neurofibrils can be identified extending into both processes. The structural characteristics of this neuron classify it unmistakably as of sensory ganglion cell type (fig. 9C). Again, any cells com~ parable to the satellite cells of the sensory ganglia are lacking.

Fig. 8 Photomicrograph of a unipolar neuron of senory ganglion cell type found in the dorsal horn of an 18.5-week fetus at the level of T6 (3: 475: 8~—8: L). The location of this cell is shown in figure 9D at 8A. Pyridine silver preparation. X 950. B, Drawing of the same neuron to show the origin of the process. X 950.

Neurons of Rohrm-Beard cell type

5 mm (5 weeks)

Most intramedullary sensory neurons seen in the 5-mm. embryo differ from those just described in various ways, particularly by their location in the marginal layer rather than in the gray matter. They appear to be a more primitive type of sensory cell and not comparable to the neurons just described but rather to the giant cells of Rohon-Beard. The question of homologies is taken up in the discussion (1). 32).

Fig. 9 Photomicrograplis of neurons from a fetus of 18.5»week of menstrual age, together with a sketch to show the location of the neurons. Pyridine silver preparation. Photo micrographs at a magiiiflcatioii of 950. A, An intramedullary sensory ganglion cell which has not become completely unipolar, but shows separate points of origin for axon and dendrite (3: 475:—-3—4: L). B, Photograph of the same neuron to show the origin of the axon. (J. Photomicrograph of a spinal ganglion cell from the dorsal root, near the spinal cord, at flulevel of T6 (3: 475—3—4: L). D, Diagram of :1 transverse section of the spinal cord to show the location of the neuron illustrated in A (and B) of this figure and that shown in A (and B) of figure 8.

Fig. 10 A, Photomicrograph of a unipolar type of sensory neuron found in the 5—mm. human embryo and considered homologous to the Rohon~Beard cells of lower vertebrates. This neuron is the best differentiated of those of this type found in the 5-mm. embryo (T6; N: 8-6-3: L). Activated protargol preparation. )< 950. B, Drawing of the neuron shown in A at the same magnification. With the exception of the bit of fiber indicated by the dotted line, all of the processes drawn may be seen in the section in which the cell body is located. 0, Sketch to show the location of the neuron figured in A and B.

The first unipolar neuron (fig. 10) to be noted in this embryo, and the most highly differentiated one, is located at about T6, on the left side of the spinal cord and near the middle of the basal plate region. The cell body is situated just within the external limiting membrane, in the marginal layer, and is ovoid in shape with diameters of 7 .4 M by 5.9 M. The nucleus is similar to the cell body in outline and only slightly smaller than the cell itself (4.4 ,u. by 3.7 It). It is located toward the larger, more deeply situated pole of the cell, opposite the single process. The karyoplasm stains lightly, appears finely granular, and contains one large, deeply staining nucleolus. The cytoplasm is heavily stained with silver and finely granular; neurofibrils are not identifiable. It is impossible to determine the nature of these cytoplasmic granules, but Nissl granules do not stain in this manner with activated protargol and the cytoplasm remains relatively clear (Hogg, ’44). It may be assumed, therefore, that these granules are of some other type, probably yolk material (p. 32).

A single process leaves the neuron to penetrate the external limiting membrane immediately, then branches just outside of the membrane. Before it bifurcates, this process is coarser than the axons of the motor neurons developing at the level. The more dorsally situated branch (P1) does not divide again but ends in a tiny terminal swelling, 22.1 .11. from the external limiting mem— brane. The ventral branch redivides almost at once. Of its two subdivisions the more ventral (P2) can be followed only 16.2 n from the external limiting membrane but the dorsal branch (P3) extends out into the mesenchyme for 36.8 pt, not quite halt’ (43.5%) of the total distance (84.5 fl.) from this neuron to the myotome. Neither of these subdivisions can be followed to its termination, for, after branching, the peripheral processes are fine and difficult to follo-W. Since no other fibers can be identified leaving the spinal cord in this section, however, those described can be traced with a fair degree of certainty at the level, although they cannot be recognized in adjacent sections. There is no evidence of a fiber within the spinal cord in the section in which the cell body is located. In the section immediately caudal to the cell, however, there is a short segment of heavy fiber which appears to belong to this neuron since it lies just inside of the external limiting membrane at the same dorsoventral level. This fiber (represented by the dotted line in figure 10B) passes dorsally for a distance of 18.4 ,u., just within the membrane and parallel to it, but cannot be identified in any other sections. Other neurons at this and nearby levels—obviously developing motor neurons—do not show difierentiation of nucleus and cytoplasm as does this one, are piriform in shape, and have a single unbranched process, with the characteristics of an axon, which has penetrated the external limiting membrane either for a shorter distance (7.4//. to 22.1,u.) as a rule or not at all.

Two other less well developed neurons of a similar type are found at about L2. The more clearly delineated of these (fig. 11, A and B) is located in the basal plate region of the right side of the spinal cord (fig. 11(3). In this case, too, the cell body is ovoid in shape but not quite so large (6.6 ,u. by 5.9a in diameter) and again lies with the smaller end against the external limiting membrane, the single process passing through the membrane as it leaves the cell. There is less differentiation between nucleus and cytoplasm than in the unipolar neuron just described. The nucleus is more densely stained, is finely granular in nature and contains a single nucleolus. The cytoplasm stains more lightly than does the karyoplasm but contains darker granules of varying size which resemble in appearance the yolk granules described by Coghill (’14). The process divides just outside the cord into three branches. The shortest one (P) —that located most ventrally——terminates in a small swelling 7 .4,u. from the external limiting membrane. The other two can be followed 15.4,; (P1) and 25.7 p. (P2) into the mesenchyme but neither one can be traced to its termination in this section or identified in adjacent sections. There is no evidence of a process within the cord. For the most part, at this level, the motor neuroblasts have not yet developed axons.

A very similar neuron (fig. 11D) is present slightly cephalic (170,; rostral) to the preceding one but in the same cord segment and also on the right side of the basal plate (fig. 110). In this case the process is broken from the cell body at the external limiting membrane and a portion of the process is present in the section cephalad to the cell body and another part. in the section caudal to it. The continuity between these branches and the cell body is lost; the probable course in such regions is indicated in the figure by broken lines. The nucleus is clear, rather than granular, and has a single large nucleolus. The cytoplasm is finely but densely granular. The greater diameter of the cell body is 7.411; the lesser diameter is 6.611. The cell body itself is somewhat ovoid in shape with the smaller pole against the external li111iti11g membrane. It appears to have a single process which divides into two at the surface of the cord. One branch (P) is traceable almost directly lateralward into the mesenchyme for 14.1 _p. witho11t further division, although not to its termination. The other branch (P1) extends dorsalward more or less parallel with the surface of the external limiting membrane and ultimately divides into three branches. Two of these branches have terminal swellings, one of them close to the cord (P2), the other out in the mesenchyme (P3). The third branch cannot be followed to its ending (P4).

Fig. 11 A, 1-’hoto1nicr0graph of another unipolar sensory neuron of the Rohon—Beard type in the 5mm. embryo. This neuron, located farther cautlalvvard, is less well differentiated (L2; N : 8—5——8: R). Activated protargol preparation. X 950. B, Drawing of the neuron in A to indicate cellular detail and illustrate the other branches of the process. X 950. C, Sketch to show the location of the neurons illustrated in A, B and D. D, Drawing of another neuron of the type shown in A and B. This cell is found in L2 of the same embryo (N: 8-4-1: R). The drawing includes parts of the processes from adjacent sections, both rostral and caudal to the cell body. X 950. The series at this level is cut from caudal to rostral.

Other cells, either less well ditferentiated or possibly undergoing degeneration, are also present in the l11mbar area. In the region of L2 to 41, five neurons, all of which send their rather long processes dorsally, are identified on the left side of the basal plate. One of these (fig. 12B), in L2, has a piriform cell body, in which no structure can be discerned, and a single dorsally directed process with three branches. Two of these parallel the surface of the mantle layer and terminate within the cord in tiny swellings 14.7 ,u and 27.2 ,u. from the cell body. The third branch extends out to the external lim.iting membrane b11t cannot be followed farther. In another instance a cell of piriform type (fig. 12A), in which structural det.ail cannot be determined, has one branch of its process extending dorsalward within the external limiting membrane, a second branch penetrating the membrane for 30 [L in a dorsocaudal direction and a third one terminating at the surface of the spinal cord. A third neuron, at L3 (N: 8—3~10: L), has a piriform cell body situa.ted at the border of the mantle layer and sends a single unbranched process dorsolaterally through the external limiting membrane, ending in a tiny enlargement 22.8,u from the cell body. The cytoplasm stains more darkly than does the finely granular nucleus, which contains a single nucleolus. ln L2 (N:8—5—8:L) a similar neuron, likewise having a light staining nucleus with one nucleolus, sends a single process out clorsolaterally through the external limiting membrane to terminate in the mesenchyme in a tiny swelling 22.1 ,1 from the cell body.

Two other neurons of this unipolar type have processes which extend dorsally only within the cord. One of these (N: 8—3—1 : L), in L4, has a process which extends dorsolaterally along the mantle layer to terminate 18.4,; from the cell body. Nuclear detail can be distinguished within the piriform cell body only with difficulty. The other cell of this type (N:8—5—2:L) is in L2. No structure can be noted and the cell body is more ovoid in shape. The single process extends for 19.1 ,u. almost directly dorsally.

Certain other cells, from which such long processes cannot be traced, appear to be of the same unipolar type. Three of these are on the left side of the cord in L2. The process (7.4 ,u in length) of one of them (N : 8-5-5: L) does not even penetrate the external limiting membrane although the cell body in each case is difierentiated to show a nucleus containing a single nucleolus and dark staining cytoplasmic granules of varying size. The other, in the same section, has a process which turns sharply dorsalward outside of the cord. In still another instance a globular cell (fig. 12C) just within the external limiting membrane of the left basal plate sends a single process directly inward toward the mantle layer.

Fig. 12 Photomici-ographs and drawings to illustrate tl1e less well diflerentiatcd neurons of the Rohon«Beard type iii the 5-mm. embryo. Activated protargol preparation. Photo» micrographs, X 950. A, Neuron with process having three branches, one of which penetrates the external limiting membrane (L2; N: 8—5—10: L). B, Neuron similar to that in A, but with two branches of its process extending dorsalward within the neural tube (L2; N: 8——5—4: L). C, Photomicrograph of a neuron which has the process directed rnedialward (L2; N 2 8-1-10: L). D, Sketch of the spinal cord at L2 to show the location of the neurons in A, B and C. The neurons are labeled with the appropriate figure letter. E, Neuron from the region of the lower medulla (N: 4-1-7: R). F, Cell from 06 which shows initial degcnera— tive changes (N: 6~—2—9: L). G, Photomicrograph of a neuron in the alar plate in C6. The single process extends ventralward within the cord (N: 6—2—9: L). H, Outline drawing of a transverse section of the spinal cord at C6 to show the location of the neurons in E, F and G. These cells are labeled with the appropriate figure letter. Note that right and left are reversed as compared with the sketch for the lumbar region, where the sections are cut from caudal to rostral.

ln lower thoracic regions of the spinal cord a few motor neurons have axons which either approach or penetrate the external limiting membrane. No cellular detail is visible in such cells, however, and the single process is not seen to branch. Although cells of this type grade over gradually into the sensory nerve cells which have just been described in the lumbar region, the motor neurons differ in the lack of cytoplasmic and nuclear differentiation and in the general appearance of the cell process.

Two cells, showing difl'erent degrees of dit"ferentiation, are present in the alar plate on the left side of the same section .in C6. The smaller of these (fig. 12G) is a unipolar cell of ovoid outline measuring 4.4,; by 6.6 ,u.. It situated at the middle of the alar plate near the border of the mantle layer. The cell body is oriented dorsoventrally and the single process, which arises from the smaller pole of the cell, passes almost directly ventrally (but a little medially) for 22 [L before being lost. Althougli both nucleus a11d cytoplasm are densely granular, so that the nucleus is scarcely distinguishable, two nucleoli may be seen.

The second unipolar neuron (fig. 12F) found in this section lies just beneath the external limting membrane immediately above the level of the sulcus limitans. This cell is larger, 8.8 ,u. by 5.9 [.L, and is more rounded in shape. The nucleus is almost clear and contains two nucleoli. The cytoplasm very dense, granular and small in amount. Fragments of the heavy process may be seen in the three sections located cephalad. It branches, with one division extending ventrally within the spinal cord and the other penetrating the external limiting membrane, but. neither can be followed for more than a few micra. The cell itself appears to be undergoing degenerative changes similar to those occurring in neurons found in the lower medulla (fig. 12E). Tliese unipolar cells are in the same section as is the bipolar neuron described on p. -1.

Two neurons located in the right alar plate in the lower medulla oblongata should also be mentioned. Both are situated dorsal to the roots of the vagus and the single process of each passes ventrolaterally toward the nerve roots. The larger cell (fig. 1213) has a diameter of 11 ,u., a large, lightly staining but finely granular nucleus with one large nucleolus and a very narrow irregular rim of homogeneous darkly staining cytoplasm around the nucleus. Only a small part of the process can be observed. The smaller cell (N:4—1—5:R) measures 8.1 p. in diameter and its single unbranched process can be followed 49.2 p. ventrally and a. little lateralward——al.most to the external limiting membrane near the point of emergence of vagal fibers. The nucleus stains more darkly than does that of the preceding cell and contains two nucleoli, but it is finely granular in character. The cytoplasm forms a. dark, irregular, solidlooking rim about the nucleus. This neuron is located only 20 [L cephalad to the preceding one.

The least differentiated of all of the neurons of this type observed is found in S4 in the right alar plate. At this level the somite is un.ditl"erentia.t.ed and lies against the neural tube. The neuroblast in question (fig. 13) is oval in shape and oriented in a. dorsolateral-ventromedial direction with the peripheral pole against the external limiting membrane. From this pole a branching protoplasmic process extends out through the external limiting membrane for about by to reach the cells constituting the somite. Another protoplasmic process arises from the cell body at the same point, turns directly ventralvvard along the external limiting membrane for 5.2 pt, diminishes sharply in size and apparently does not divide again. Both processes contain fine granules such as those seen in the cytoplasm. The cell body is comparable i11 size (8.1 ,u. by 5.9 ,4») to that of the best differentiated cell of this type (p. 15) but is of a highly undifl"erentiated nature, in that the cytoplasm and the nucleus both contain vacuoles and discrete granules of varying size. There is little distinction between cytoplasm and karyoplasm. No other cells in such an early stage of development have been seen in this embryo. Obviously they could be recognized only if the plane of the section passes through the process where it leaves the cell body. Consequently more of them, no doubt, may be present.

Fig. 13 A, Photomicrograph of a sensory neuroblast found in the sacral region of the spinal cord of a 5-mm. human embryo (S4; N : 7—3—~l: R). Note that the branching protoplasmic process ends on cells of the adjacent somite. Activated protargol preparation. X 950. B, Sketch to show the location of the cell illustrated in A. In the sacral region the sections are cut from caudal to rostral so that the right and left sides of the neural tube are reversed as compared with the drawings at cephalic levels.

The seventeen cells of unipolar type in the 5 mm. embryo, for which a descriptive statement has just been given, do not represent all of those either present or even observed. A few others, which are less representative, might be included and there are still more for which unfavorable plane of sectioning makes identification ditlicult. A study of the distribution of those enumerated, however, suggests that more are present in the lumbar region (11 of the 17), that more are found on the left side of the neural tube (12 of the 17), that at caudal levels (thoracic and below) they are located more often in the basal plate. and that in more cephalic regions (medulla oblongata and cervical cord) the cells are situated in the alar plate. A survey of the structure of these neurons indicates that development occurs in a cephalocaudal direction.

Motor neurons

Two types of primitive motor neurons have been found in the spinal cord of the human embryos used in this study. One type is present in the 5—mm. embryo and appears to be homologous to the primary motor neurons described by Coghill (’13) for amphibians. The other type of primitive motor neuron has been seen in 16—mm., 22.5-mm., and 26.5-mm. embryos. All of the primitive motor neurons observed will be described, beginning with that found in the youngest embryo.

5 mm (5 weeks)

A very interesting neuron of motor type—-one showing a high degree of differ-e11tiation— is located at about 05, near the middle of the left basal plate at the border of the mantle layer (fig. 14D). This cell (fig. 14, A to C) has a large, oval nucleus oriented dorsoventrally, as are the neurons giving rise to the fibers of the spinal accessory nerve at this level, rather than mediolaterally, as are those of the other motor neurons. The nucleus is granular in character, has three darkly staining and is eccentrically placed with no cytoplasm visible ventral to it. Dorsally the cytoplasm, filled with relatively coarse neurofibrils, is continued into a large process extending almost directly dorsalward (fig. 14A) for a distance of 43.4 ,u.; its further course in this or adjacent sections cannot be determined definitely due to the presence of other fibers, such as those of the spinal accessory nerve, which have a similar course and position, but it appears to be present farther dorsalward ( D1 in fig. 140). Three branches come off from the cytoplasm before it narrows down into this process; one extends medially and two laterally. The entire medial process (D2 in fig. 14, A and 0) appears to be present, since, at its end, 9.6 p. from the nucleus of the cell, there is a tiny rounded swelling, representing the terminal cone of growth. This medial branch grows out from the superior surface of the cytoplasm near the nucleus. The longer (A in fig. 14, B and C) of the two lateral processes also leaves the superior surface near the nucleus, then extends dorsally for a distance of 16.2 p. on the surface of the main process and curves ventrolaterally toward the external limiting membrane for 4.4,u. before losing continuity. In its short course ventrolaterally it accompanies the ventral root fibers but, although it is heavier than these fibers, it cannot be followed through the surface membrane in this or in adjacent sections. The uniform caliber and the differences in staining reaction of this process as compared with that of the other cell processes of this neuron, as well as its course with the ventral root fibers, indicate that it is the axon. In line with this fiber in the ventral root outside of the external limiting membrane but in the same section is a heavy fiber (A1 in fig. 14B) which might be a part of it. However, the two are separated by a distance of 23.5}; and the surrounding root fibers make recognition questionable. In any event, the course of the more proximal part along with the ventral roots to within 12.5p. of the external limiting membrane strongly suggests that the fiber passes out of the cord with them. In the section just caudal to this a long (15.4 ,i) isolated segment of coarse nerve fiber, present in the niesenchyme dorsal to the ventral root fibers, is undoubtedly a. part of this axon. The second lateral branch (D3 in fig. 14, A and C) comes ofl’ 6.6 p. from the nucleus and near the axon. It is lost almost at once and measures only 1.5 [L in length. It appears to extend caudalward, but cannot be identified in adjacent sections.

The neuron just described has more cytoplasm than do other motor neurons at this level although the greatest dimension of the nucleus is about the same (5.9 ,M as compared with 6.3 a for the largest of the motor cells present). In this neuron, too, the neurofibrils are relatively coarse and readily seen. lt is only rarely that fine neurofibrils can be distinguished at all in the other neurons.

16 mm (7 weeks)

Two very large multipolar neurons have been observed on the left side of the ventral horn just medial to the dorsolateral cell column in the seventh cervical segment. The more cephalic one (figs. 15A and 16) measures 31.6 M. by 14.7 pt in size and is triangular in outline with one angle pointing ventrolaterally, another ventromedially, and a third almost directly dorsally. The nucleus is relatively small compared with the size of the cell body, but only slightly smaller than the nuclei of the developing motor neurons

Fig. 14: Illustrations of a primitive type of motor neuron found in the cervical region of the 5-mm. human embryo (C5; N:6—1—-2: L). This neuron is considered homologous t.o the primary motor neurons in amphibians (Coghill, ’13). Activated protargol preparation. A, Photomicrograph of the primary motor neuron with the microscope focused on the deeper (more caudal) portion of the section to show the various dendrites. X 950. B, Photomicrograph of the same neuron with the microscope focused on the superior (more rostral) part of the section to show the axon leaving the cell. X 950. 0, Drawing combining the features shown in A and B and giving cytological detail. X 950. D, Sketch of a transverse section of the spinal cord to give the location of the neuron in A, B and C.

in the same sections (VHN, fig. 15A). However, it differs from them in struc tural character in that it has the chromatin net of an adult type of neuron and has a single nucleolus. The cytoplasm is uniformly finely granular but stains only faintly bluish with toluidin blue as compared with the dense blue Nissl granules in the developing motor neurons nearby. Naturally, with this stain, the processes cannot be followed or the ‘possible presence of neurofibrils derm-mined. Nevertheless, the general contour of the cell indicates that. unquestionably three large dendrites are given off—one dorsal.l_V ( D1), a second ventromedially »(D2),tand a third ventrolaterally (D3). Likewise, from the contour of the neuron it appears that at least two more processes arise in this plane, one on the lateral side of the cell (at D4) and another on the medial side.

Fig. 15 Photomicrographs of primitive motor neurons found in the 16—nm1. human embryo. These neurons, and the ones illustrated in figures 17 and 18, are of the same type. Although of primitive nature, they are not comparable to the primary motor neuron illustrated in figure 14. The location of these cells is shown in figure 16. Erythrosin and toluidin blue preparation. X950. A, Primitive motor neuron situated medial to the dorsolateral cell column in C7 (1: 18-2-6: L). B. Motor neuron found in the lower medulla medial to the somatic motor cell column (1: 8-‘2-14: L).

The second cell (X in fig. 16; 1:19—‘2—7:L) is situated 285;; caudal to the one just described. It has the same location, orientation and structural characteristics, but is less favorably sectioned and slightly smaller (29.4 pi by 10.3 '11.). The two neurons are evidently of the same type. They are not comparable to any of the ventral horn motor neurons (VHN in fig. 15A), although the latter, where best developed, do have a triangular outline. In C5, in which the Ventral horn cells are the largest, they average only 13.4 [L in their greatest diameter and are less well differentiated.

Other smaller cells of this type have been seen in the lower medulla of the 16-mm. embryo at approximately the level at which the unipolar sensory cell described on p. 11 is located. The plane of section of one of these (Y in fig. 16: 1:: 8-3—4:L) is such that only the dorsal dendrite and a large ventromedial dendrite can be seen, but the latter ca.n be followed for some distiance and observed to branch. Before this dendrite bifurcates the axon is given off of its lateral surface to extend in a ventrolateral direction. The contour of the cell body indicates that dorsomedial, ventrolateral and dorsolateral dendrites also are present. This motor neuron is located deep in the mantle layer near the region in which the special visceral efferent cell column develops. It is not so large (18.4 [L by 11 ft) as the two neurons just described but contains a relatively larger and less highly differentiated nucleus and proportionately less cytoplasm in which fine, blue staining granules are present, particularly medial and lateral to the nucleus.

Two other neurons which are about the same size are present in nearby sections in this region of the medulla. One of them (figs. 15B and 16) is located medial to the somatic motor cell column, the other (1 : 8—3—14:L) lateral to it (Z in fig. 16). The neuron situated medially (fig. 15B) lies at the junction of the marginal and mantle layers; the cell lateral to the somatic motor cell column is deep in the mantle layer. Both have Well differentiated nuclei with a single nucleolus and a proportionately large amount of cytoplasm containing a few fine, blue staining granules. The smaller cell (14.7 ,1 by 8.8 [.I.), located medial to the somatic motor column (fig. 15B), has a large medial dendrite (D), a smaller ventrolateral dendrite (D1) and a large dorsolateral dendrite (D2) from which the axon arises. The other neuron (14 ,1 by 9.6 M) has a large ventro— medial dendrite, a smaller dorsal dendrite and a. very large, branching \'entro— lateral dendrite from which the axon arises.

Fig. 16 Diagram of a trasverse section through the spinal cord of the 16—mm. embryo to show the location of the primitive motor neurons found in the 16-mm., 22.5—mm. and 26.5-mm. stages studied. The neurons illustrated in figures 15, 17 and 18 are labeled with the appropriate figure numbers. The letters designating the other cells are referred to in the text description of these motor neurons in the 16-mm. embryo.

22.5 mm (8 weeks)

In the lower part of C8 a large multipolar neuron of primitive type (fig. 17) is located just dorsal to the developing retrodorsolateral cell column near the surface of the gray matter (17 in fig. 16). The periphery of the cytoplasm is not well impregnated with the silver and the dendrites are difiicult to follow, but it is possible to recognize a large dendrite (D) directed dorsalward toward the dorsal horn, a smaller process extending ventromedially and branching (D1), with one dendrite passing among the ventral horn motor neurons and the other almost directly medialward and then somewhat ventralward. The third large dendrite (D2) passes latera.lly into the lateral funiculus and branches in the section just caudal to the cell body. The axon (A, fig. 17) arises from the dorsal branch, is fine as it passes through the surface of the cord, and increases in caliber outside. Then it can be followed, in its course toward the ventral root, through fourteen sections caudalward (140 ,u), running more or less parallel to the outer surface of the cord. The cell body is large, 14.7 ,u by 11 it, as compared with the ventral horn motor neurons in this region (11.9 ‘U. in greatest diameter) and shows evidence of a higher degree of differentia.tion. For example, the nucleus "stains less heavily, contains one nucleolus and has a distinct but fine chromatin net. Neurofibrils are present in the cell body and can be traced into the dendrites.

In the main mass of the cytoplasm, ventrolatera.l to the nucleus and somewhat overlying it, but not in the cell processes, there is a deeply staining, finely granular mass (LP, 17) of material which has characteristics suggestive of the pigment known as lipochrome (Penfield, ’32; Strong and Elwyn, ’43). This pigment is regarded by certain observers as a metabolic product and is present in motor neurons in the adult and increases in amount in older individuals (Strong and Elwyn, 43).

It is of particular interest that two fibers synapse with this primitive motor neuron in the plane of the section. The synapsing fiber more readily demonstrated is labeled as S in figure 17B and C. This axon, which contacts the ventral dendrite, may be followed some distance toward the cpendyma, but the source of the impulses carried by it cannot, of course, be determined. The other axon synapses (S1) with the dorsal dendrite of this motor neuron near the nucleus. Only a small segment is present, but this synapsing fiber appears to come from a more dorsal portion of the cord.

26.5 mm (8.5 weeks)

In the lower part of T1, on the right side of the cord, a large (18.4/A by 11.8 ,u) multipolar neuron similar to that described in the 22.5—n1m embryo, is located at the surface of the gray matter just dorsal to the retrodorsolateral cell column (fig. 18 and 18 in fig. 16). The large nucleus with its single nucleolus fills much of the cell. Neurofibrils are clearly visible in the cytoplasm, in the axon and in the larger dendrites. A large mass of granular, dark staining material (LP in fig. 18) occupies the main part of the cytoplasm and obscures the lateral side of the nucleus. This is evidently the same type of pigment, lipochrome, described for the multipolar neuron found in the 22.5—mm. embryo.

In the transverse plane of the section there are altogether five dendrites (fig. 18, A and B). Of these, two extend almost directly lateralward into the fiber tracts of the lateral funiculus. The more dorsal ( D1) of these two lateral den~ drites is heavy and can be followed almost to the surface of the cord. The ventral one (D2) is much finer and cannot be traced so far, probably because it passes out of the plane of the section. A third dendrite (D3), obviously cut near its origin, extends straight dorsalwarrl toward the dorsal horn area. From its size it is to be expected that it could reach the dorsal horn cells. A fourth dendrite (D4) passes medially into the gray matter but. can be followed for only a short distance. The fifth dendrite, also large, comes off of the ventral surface of the cell and bifurcates (D5). One branch runs ventromedially into

the retrodorsolateral portion of the motor area; the other branch passes ventrolateralward, also into the retrodorsolateral area, but obviously does not extend so far as does the more medial branch. The axon (A) arises near the region of origin of this ventral dendritic process, before it bifurcates, and passes ventrolaterally and caudally to emerge from the spinal cord between the dorsal and ventral roots three sections (30 p.) farther caudally. It can then be followed ventrally toward the ventral root, parallel to the outer surface of the cord, for two sections more (20 ,1) before becoming lost. It is evident that it becomes a ventral root fiber.

Fig. 17 Photomicrographs of gt primitive motor neuron located in C8 of the 22.5-mm. embryo (0: 22-—2—2: R3.’ The location of this neuron in transverse section is indicated by the figure number (17) on figure 16. Activated protargol preparation. X 950. A, Photograph to show the nuclear and cytoplasmic structure as well as some of the dendrites. B, In this photograph of the same neuron the cell body is not well focused, but at S and S1 axons of other nerve cells are in synapse with this motor neuron. C, Drawing to combine the features illustrated in A and B and to give the course of the axon as seen in adjacent sections.

Fig. 18 A, Photomlcrograph of a primitive motor neuron from Tl of a 26.5-mm. embryo (19: 28-1-3211). The location of this neuron is indicated by the figure number (18) on figure 16. Pyridine silver preparation. X950. E, Reconstruction of the neuron shown in A to illustrate the cellular details observed at other levels, the dendrites, and the general course of the axon in caudal sections. The ventral horn cells are drawn at the same magnification.

The distribution of this type of primitive motor neuron, as seen in transverse sections, is shown by a diagram (fig. 16) in which the approximate position of each such cell described is represented. The longitudinal distribution of these neurons observed is of no significance, since for only one of the series (Homo O), in which this type of motor neuron was found, has every section of the spinal cord been examined.


The occurrence, in the embryonic human spinal cord, of well difl’cren— tiated sensory and motor neurons of a type not comparable to cells found elsewhere in the embryonic or adult human central nervous system suggests the possibility that these cells may represent a mammalian homologue of the primitive neurons occurring in lower vertebrates. The primitive neurons to which reference is made are the giant ganglion cells of Rohon-Beard, the primary motor neurons described for amphibians (C-oghill, ’13) and the primitive type of motor neuron seen in adult fishes. In order that comparisons may be made, and similarities and differences pointed out, a brief review of the literature and a. summary of the characteristics of these cell types in lower vertebrates will be given.

The first sensory cells to develop in amphibians, and the only ones present when motility first occurs, are, according to Coghill (’14, ’24, ’29) the giant ganglion cells or cells of Rohon-Beard. These neurons are also known as intramedullary ganglion cells (Ariens Kappers, Huber and Crosby, ’36) since they are located within the central nervous system. Such ganglion cells are present in larval Amblystoma, but not in the adult. Theyiare also transient structures in plagiostomes (Beard, 1889, 1892, 1896a, 1896b; von~ Lenhossék, 1892; a11d van der Horst, ’36) and in reptiles (Ariéns Kappers, Huber and Crosby, ’36, p. 196). In cyclostomes, however, (Kolmer, ’05; Becarri, ’09; Tretjakoff, ’09; Ariéns Kappers, Huber and Crosby, ’36) and in some ganoids and some teleosts (Studniéka, 1896; Dahlgren, 1897, 1898; Sargent, 1898a, 1898b; Tagliani, 1899; Johnston, ’O0; van der Horst, ’18; Burr, ’28; Ariéns Kappers, Huber and Crosby, ’36, p. 175 and p. 280), intramedullary ganglion cells or cells of Rohon-Beard remain as permanent structures. Ganglion cells of this type have not been described, in so far as the author is aware, for birds or mammals.

According to Coghill ('14 and ’16) and Herrick and Coghill (’15) the essential features of the “transitory dorsal giant cells of RohonBeard” are as follows. The cell is ovoid in shape with two processes coming off together at the smaller pole of the cell which lies near the external limiting membrane. One process passes to the periphery and may send a branch both to muscle and to skin. The other, or central branch, the axon of the cell, ascends in the “dorso—lateral sensory tract” located superficially along the lateral aspect of the spinal cord. A dendritic process which extends caudalward in the cord is also described by Coghill (’14, p. 169). The peripheral fiber, according to this author (’14, p. 203) in some cases has been seen to arise from the descending process and also may arise from the cell body (Coghill, ’26b, p. 96). Yolk granules are present in the cytoplasm and, in the figures in which nucleoli are shown (’14, figs. 16 and 27a), a single large nucleolus is illustrated. All of Coghill’s figures show the RohonBeard cells situated peripherally, just deep to the external limiting membrane. As estimated from Goghill’s figures, their greatest diameter is approximately 32 p in the early swimming stage.

The Rohon—Beard cells in larval Amblystoma (Coghill, ’14) are found throughout the spinal cord and comparable neurons occur as far cephalad as the trigeminal nerve, but are “least numerous in the rostral portion of the spinal cord” (Coghill, ’16, figs. 51, 53—54; "24, p. 105). These cells are located in the dorsal part of the cord at cephalic levels, but occupy a more ventral position at both the caudal levels and in the brain—stem from the “fourth postauditory myotome rostrad” (Coghill, ’14l-, p. 167). Caudally they may be in the basal plate. Coghill considers these giant ganglion cells to be unipolar in character (’14, p. 168), but states that at caudal levels the cells are bipolar in type and oriented longitudinally (’14, p. 175).

As estimated from Coghill’s figures of the non-motile stage of Amblystoma the distance traversed by the peripheral fiber of the Rohon— Beard cell from the external limiting membrane to the muscle is approximately 17 p; to reach the skin, approximately 23;; (Coghill, ’14, figs. 3 and 5). In the early flexure stage these distances have increased to approximately 30;: and 6611 respectively (Coghill, ’14; fig. 23). At later ages, of course, the distances are much greater. Spinal ganglia develop in Amblystoma while the Rohon~Beard cells are still present; at a later date—~the exact time, according to Det— wiler (’37), is unknown——the Rohon-Beard cells disappear; at least they are absent in the adult. Earlier accounts of these cells in amphibians, such as the descriptions of Burckhardt (1889) and van Gehuchten (1898), give no additional information on the structure of these cells. Youngstrom ( ’38, p. 368 and fig. 6), however, has observed neurofibrils in R-ohon-Beard cells of Rana pipiens “by the time the animal responds to light touch on the body”.

One of the most informative descriptions of the Rohon—Beard cells in fishes is that given by Sargent (1898a) for a teleost, Ctenolabrus coeruleus. Here, as in other fishes, these cells are located on either side of the midline dorsally, close to the dorsal median septum. Only rarely is a cell deeply buried in the cord. The cell bodies are spherical to piriform in shape, with a process which enters a fiber bundle located just dorsolateral to the central canal. Some processes pass cephalad, others caudad and still others, after bifurcation, in both directions. In the adult teleost, the cells measure 40 u to 70 u in diameter, in the 3—cm. fish only 7 u to 8 p. The nucleus is large, eccentric, of the same shape as the cell body and usually contains a single eccentrically placed nucleolus, rarely two nucleoli. According to Kolster (1898), the cytoplasm of the Rohon-Beard cells contains very fine granules which stain deeply with Nissl stains. The number of cells varies in different fishes and, in one plagiostome, at least, according to Beard (1896a), these neurons are completely lacking. According to Tagliani (1899), upper levels of the spinal cord of the teleost, Solea impar, contain the greatest number of cells and they extend as far cephalad as the level of the vagus nerve. In an elasmobranch, however, Beard (1892) found the maximum number of cells in the eleventh to twenty sixth segments with a decrease both cephalad and caudad.

References to the Rohon-Beard cells in reptiles are few. Apparently these neurons are the “cellules dorsales medianes” of van Gehucten (1897) for Tropodonotus. According to van Gehuchten the commissural cells described and figured by Retzius (1894, 1898a, 1896b) in the embryonic spinal cord of reptiles are of this primitive type, but examination of Retzius’ figures indicates that his cells are probably not Rohon-Beard cells. The" neurons which have been regarded as giant ganglion cells of the cord by van Grehuchten a11d which are apparently represented by commissural types in Retzius’ material (1894, 1898a, 1898b) are probably longitudinally conducting neurons, partly of commissural and partly of non-commissural type. The neurons described by Terni (’22) appear to be comparable to those of van Gehuchten and Retzius. The accounts given by these authors do not figure or mention a peripheral process. It is questionable as to whether these neurons represent the exact homologue of the Rohon-Beard cells of other forms, but at least they are the only neurons thus far described for reptiles which could be considered comparable. Details of the structure of these cells in reptiles are not given by the authors quoted.

A comparison of the well differentiated cells found in the sixth thoracic and second lumbar spinal cord segments of the 5 mm. human embryo (pp. 15 and 16) with the Rohon-Beard cells described by Coghill reveals certain striking similarities, many of which are also seen upon comparing the figures. In each case the cells are ovoid in shape, are situated just beneath the external limiting membrane, and have an eccentrically placed nucleus at the opposite pole from the point of origin of the fiber. In each case a single nucleolus is present. It is not possible to determine whether or not the cytoplasmic granules seen in the human embryo are yolk granules as they are in the Rohon-Beard cells in amphibians, but at least they are not N issl granules since they stain deeply with the activated protargol and the cytoplasm of neurons which contain Nissl granules remains relatively clear with this stain (see also Hogg, ’44). The fact that the three well differentiated neurons just mentioned and most of the other cells of this type in the 5—mm. human embryo are situated in the basal plate region rather than in tlie alar plate does not exclude them from being considered as homologous to the Rohon—Beard cells of amphibians, for in amphibians the more caudal Rohon—Beard cells particularly may be basally located. In the human embryo, moreover, neurons of this type in the lower medulla and cervical cord are situated in the alar plate (p. 20). Thus in man, as in Amblystoma, cells of this type found cephalad may lie dorsally even though the available evidence suggests that the majority of them have a ventral location.

There are some similarities, but also slight differences, between the cell processes of the Rohon—Beard cells in amphibians and those of the neurons of this type observed in the human embryo. Both have the appearance of unipolar cells and in both a single process penetrates the external limiting membrane and then branches immediately. Only with one of the three well developed neurons of the 5 mm. human embryo, however, is there any evidence of a central process comparable to that which forms the “dorso—lateral sensory tract” described by Coghill. In the less well developed cells a central process, directed toward the region where this tract should be located, occurs frequently (fig. ]2A, B and G). In fact two processes of this type may be seen in some cases (fig. 12B) and in others the only process to be observed at all is of this nature (fig. 12G).

In none of the well developed cells described can the peripheral process be traced either to myotome or to dermatone although one of the processes can be followed without question for 44 it into the mesenchyme. It might be pointed out that this is over twice the distance necessary to reach the skin in non-motile Amblystoma (23 u, p. 30) and far more than that needed to reach the muscle in the early flexure stage of this amphibian (30 u, p. 30). Since only three well differentiated cells are available for study it is not surprising that the processes can be traced no farther. In the least dilferentiated cell of this type observed, however, that seen in the sacral region (fig. 13), a branching protoplasmic process, which is apparently all present in the one section, does terminate in the adjacent cells of the ‘undifferentiated somite in a manner much like that shown by C-oghill for non—motile Amblystoma in figure 12 of his 1914 paper.

It should be mentioned that in no case are these neurons so large as the Rohon—Bea rd cells in Amblystoma, in which the greatest diameter (as estimated from fig. 27a of Coghill’s 1924 paper) is approximately 32 p. In the 5—mm. human embryo the range in greatestdiameter is from 6.6 u to 11 p, However, this is about the size of the Rohon-Beard cells found-by Sargent (1898a) for the 3-cm. teleost, Ctenolabrus (7 u to 8 p), although in the adult teleost the cells are from 40 u to 70 u in diameter and may be as much as 200 u in another teleost, Solea impar (Tagliani, 1899). Obviously, then, there is a wide variation in size even within the same order of fishes and from embryo to adult, so that differences in size may be of little significance.

Some mention should be made of the less diflerentiated cells found in the lumbar region and those degenerating cells of this type in the lower medulla. of the 5-mm. human embryo. In the two cells in the medulla the unduly small amount of cytoplasm for the size of the neuron and its arrangement as a very dense irregular rim about the nucleus suggest degenerative changes, indicated by complete filling of the cytoplasm with granules (Cowdry, ’24). The less well ditferentiated cells described in the lumbar region of the cord have certain structural features which suggest that in some instances they represent earlier stages in the development, in other cases perhaps an abortive type of development or even an arrest of differentiation.

A second type of primitive cell is also described for Amblystoma by Coghill. (’13, ’26a, ’26b) and by Herrick and Coghill (’15). This is the “primitive motor neuron”, which precedes the typical motor neuron in development. According to these authors, the earliest root fibers which develop arise as collaterals of the descending processes (the ventrolatcral tract of Coghill) of neurons forming the somatic motor column. The peripheral fibers of these cells form the “primary root fibers”, consisting of never more than four fibers to a segment and most frequently only one. In the non-motile stage of Amblystoma,

these fibers are found only cephalad; by the early swimming stage they extend to the twenty-fifth segment. Such a primary root fiber varies both in thickness and in point of origin. It may arise from “the short, thick protoplasmic cell body” or it may be formed by the bifurcation of the descending process at some distance from its origin. The cell body itself is near the periphery of the cord, has a large spherical nucleus and an indistinct perikaryon containing yolk granules, some of which indent the nucleus (Coghill, ’13, p. 123, non—moti1e stage). According to Coghill, the cell nuclei of the motor column are smaller than those of the giant ganglion cells (’13, p. 129). There is “no satisfactory evidence” (Coghill, ’26b, 13. 109), that the primary motor cells “have dendrites in the transverse plane”. The number of root fibers is small compared with the “relatively large number of primary motor cells” (Coghill, ’26b, p. 108) which constitute the primary motor cell column.

Youngstrom (’40) confirms the presence and structural features in Amblystoma of the primary motor neurons described by Coghill (’13, ’26b, ’34). This author also states (’40, p. 140) that i-n cross sections of the spinal cord in larval stages the “principal axis of these cells is distinctly and regularly in a dorsovcntral plane” and that such a cell has both dorsal and ventral dendrites. He states further, as did Coghill (’34), that just before giving rise to the axon the large ventral dendrite is insynaptic relation with Mauthner’s fiber. According to Youngstrom (’40, p. 142), at the rostral end of the spinal cord some of the primary motor root fibers emerge almost directly opposite to their cells of origin. The axons of the primary motor neurons are coarse in nature as compared with the fine axons of the motor neurons to develop later ‘which are referred to by Youngstrom (’40) as “sec— ondary motor neurons”. This author states that even in adult Ambly— stoma (p. 145) the primary motor fibers are the most prominent constituent of the ventral roots, although after metamorphosis they are “conspicuously absent” in the frog. Youngstrom (’38, p. 366) found “considerable neurofibrillation” in these neurons in embryos of various Anura “by the time they responded to light touch with a fine human hair”.

The motor neuron in C5 (fig. 14), described on p. 22 of the present account, in the autl1or’s opinion, represents, Coghill’s primary motor neuron, although there are certain morphologic differences, some of which are probably more apparent than real. The principal axis of this primary motor neuron of the 5 mm. human embryo is the same as that of the primary motor neurons in Amblystoma — that is, dorsoventral—and the largest dendrite is representative of the dorsal dendrite described in amphibians. The ventral dendrite of Amblystoma is either lacking or is represented by the small segment (D3) which appears to turn sharply caudad in the human embryo. Since the ventral dendrite in amphibians is the process coming into synaptic relation with Mauthner’s fiber, ‘it is not surprising that this process should be either lacking or atypically represented in the human embryo in which Mauthner’s fiber is presumably absent. In amphibians and in the human embryo the axon of ‘the primary motor neuron is coarser in type than that of the permanent motor nerve cells, but in man the axon occupies the more dorsal part of the ventral root rather than the ventral part, as it does in Amblystoma (Youngstrom, ’40, figs. 7 and 8). The axon of the primary motor neuron in the human embryo arises from the rostral surface of the cytoplasm; in Amblystoma it arises either from the cytoplasm or, more usually, from the ventral dendrite at a level caudal to the cell body. However, it may emerge from the amphibian spinal cord almost directly lateral to its cell of origin, particularly at rostral levels. Although in the human embryo the axon arises from the rostral part of the cytoplasm itdistributes, as does its amphibian homologue, caudal to its point of origin since it courses a little caudally as it passes toward the surface of the cord and can be identified in the mesenchyme of the next most caudal section. The neurofibrils present in the human primitive motor neuron are comparable in degree of differentiation with those seen in Anura by Youngstrom (’38) at the age when these embryos “responded to light touch with a fine human hair” rather than to those in the nonmotile stage of Amblystoma for which Coghill (’13, p. 123) described yolk granules. Thus it seems that the diiferences between the primitive motor neuron of the 5-mm. human embryo and .the primary motor neurons of amphibians actually are not only minor but also those which would be expected.

The 5 mm human embryo not only has primitive sensory neurons comparable to the Rohon-Beard cells in amphibians, but a second type of intramedullary sensory cell of primitive character is also beginning to appear, that is, a bipolar sensory neuron (p. 4). Only a ‘very limited number of these cells has been identified definitely at this age; others are evidently present, but positive rcognition is diflicult unless the plane of the section is very favorable. At 5 mm. these bipolar sensory neurons are far in advance of the sensory ganglion cells in differentiation, although morphologically very similar to the sensory ganglion cells at a later age. No functional peripheral sensory connections of any type, however, have been established.

These bipolar sensory neurons are represented in older embryos as well, for one has been found in the 16—mm. embryo (p. 6), and a considerable number of them in the 22.5—mm. stage (p. 7) and a few at 26 mm. (p. 7). In the 22.5-mm. embryo there is every structural indication that these intramedullary sensory neurons have reached a functional stage of development and have established functional connections both centrally and peripherally. Such is undoubtedly the case as well for the best differentiated of the spinal ganglion cells, although the vast majority of these neurons are still at a much lower level of development. According to Windle and Fitzgerald (’37, p. 505) it is during the eighth week of human development that the first spinal reflex arcs are completed. It is at a slightly earlier age that both Hooker (’43) and Fitzgerald and Windle (’42, p. 166) have first observed movement in the human fetus. Windle and Fitzgerald (’37) evidently regard these reflex arcs as involving the spinal ganglion cells for they do not report the presence of any type of intramedullary sensory neurons either at this age or in the younger embryos studied by them (7 mm. to 24: mm. in C.R.L). It would seem that the afferent side of such a reflex arc might be provided for, in our material, in some instances by the more primitive but nevertheless more highly and earlier differentiated intramedullary sensory neurons as well as by the spinal ganglion cells.

Some of the intramedullary sensory neurons in the 22.5—mm. and 26-mm. embryos, such as those shown in figures 4B and 5A, are undergoing changes toward unipolarity and in’ both the 16—mm. and the 26-mm. stages unipolar neurons of sensory type have ‘been identified within the spinal cord (fig. 6). The author believes that these unipolar neurons represent a later stage in the differentiation of the intramedullary bipolar sensory cells found in the 5—mm., 16—mm., 22.5—mm. and 26-mm. embryos— a transition comparable to that undergone by extramedullary sensory ganglion neurons during development—and that the sensory cells of unipolar type found in the spinal cord at 14.5 weeks and 18.5 weeks (figs. 7 to 9) are undoubtedly still later stages in the differentiation of the intramedullary bipolar sensory cells present in younger embryos. Indeed, one of these cells at 18.5 weeks (fig. 9A) has not yet completely changed into a unipolar type of neuron. Structurally, then, the two intramedullary sensory cells present at 18.5 weeks are directly comparable with the sensory ganglion cells (fig. 9C) and appear equally capable of carrying impulses. There seems to be no legitimate reason to’ believe that a position of the cell body of a sensory neuron inside instead of outside of the central nervous system should prevent its taking part in a functioning system.

Whether or not the intramedullary bipolar sensory neurons of the human embryo are comparable to the bipolar cells of Amblystoma, found by Coghill (’14, p. 175) in the caudal part of the column of giant ganglion cells, it is impossible to say without further knowledge of structural detail. The neuron described by Coghill may be forerunners of the bipolar sensory neurons- in the human spinal cord, although they do differ in having a longitudinal rather than a transverse plane of orientation. It is also possible that the intramedullary bipolar sensory neurons reported in cyclostomes by Beccari (’09) are comparable to those described in this account for the human embryo.

In birds no intramedullary sensory neurons of any type have been described, but the neurons figured by Ramon y Cajal (’09) and Bok (’28) as motor nerve cells with axons passing out with the dorsal root fibers are comparable in position and, in part at least, in structure to the intramedulla.ry bipolar sensory neurons present in human embryonic development." In the absence of more specific information concerning the structure of these cells in birds it is impossible to make a definite comparison. It should be stated, however, that the peripheral process of the bipolar cells described for the human embryo has definite characteristics of adendrite and that the bipolar cells themselves are not structurally comparable to the motor neurons in the same embryo. N or. are the bipolar neurons -described for the human embryo a part of the spinal parasympathetic system of Kuré and others (see Ariéns Kappers, Huber and Crosby, ’36),'since they are much larger and of a different character than are the visceral efferent neurons at the same, age.

There is no indication, of course, as to the extent of the peripheral connections of these intramedullary bipolar sensory neurons. Since they are a primitive type of neuron, however, and obviously not re— tained to any extent in later fetal life, it seems likely that the peripheral relations are general in nature rather than specific, that is, one neuron provides sensory innervation for a large area and not for a more restricted territory as do many of the sensory ganglion cells. Moreover, the fact that all of these intramedullary neurons are not only relatively large but also essentially alike in size and structure, rather than having a wide rangein diameter and considerable variation in morphologic characteristics, suggests further that they are capable of carrying a strong stimulus of uniform type from a wide area rather than varying specific stimuli from localized regions.

In the 16 mm embryo, and again in the 22.5 mm. and 26.5 mm. embryos, motor neurons of a primitive type are present. These neurons, however, are not structurally comparable with the primitive motor neuron which is present in the 5-mm. embryo and is considered homologous to the primary motor neurons described by Coghill (’13) and Youngstrom (’40) for amphibians. Instead they resemble structurally the type of motor neuron found in fishes. In plagiostomes (Ariéns Kappers, Huber and Crosby, ’36, p. 160), for example, the motor neurons have three types of dendrites. Those which extend ventramedially into the ventral funiculus of the opposite side constitute the smallest group. Those which extend into the dorsal horn toward the entering dorsal root fibers are more numerous, but the largest number of dendrites extends out into the lateral funiculus and forms a marginal dendritic plexus at the periphery of the cord. The axon of such a neuron arises from one of the dendrites. A similar arrangement of dendrites and manner of origin of the axon are shown by Ramon y Cajal ('09, fig. 223 and 229) for the frog and the lizard during development.

The large motor neuron found in the 26.5-mm. embryo has all three of these types of dendrites shown in the transverse plane of sectioning —- dorsal, lateral and ventral-— and the axon arises from the ventral dendrite to pass caudalward as well as laterally toward the ventral root fibers. The lateral dendrites can even be traced well toward the periphery of the cord and certainly simulate, at least, the marginal dendritic net found in lower vertebrates. The extension of the dorsal dendrite into the dorsal part of the cord to .establish there the sensorymotor synapse is a characteristic of primitive spinal cords _(Ariéns Kappers, Huber and Crosby, ’36). Only in higher forms is the main sensory-motor synapse in the ventral horn region. The primitive motor neuron described for the 22.5-mm. embryo has a similar distribution of dendrites, but the axon arises from the lateral rather than from the ventral dendrite. The structure of the large motor neurons found in the 16-mm. embryo indicates a similar distribution of dendrites. Obviously such motor neurons would receive impulses from all tracts in the lateral funiculus as well as from the incoming dorsal root fibers.

The large size of these motor neurons and the wide spread of the dendritic net are not only indicative of their primitive nature, but also suggest that they probably discharge to a relatively wide area rather than to a small group of muscle fibers as is the case with the permanent motor neurons. From the appearance of the neuron it should be capable of functioning. Indeed, in one instance, there is evidence of synapsing fibers (p. 26). The fact that only a very limited number of such cells has been observed in silver stained material (pp. 26 and 29), in View of the well known selectivity of silver staining technics, is not significant, for others may well be present without being impregnated by the silver. It should be remembered, too, that primitive neurons, whether motor or- sensory, are always relatively few in number as compared with the permanent nerve cells of the same type. No motor neurons of this type are mentioned by Windle and Fitzgerald (’37) in their study of the spinal reflex mechanism in human embryos of comparable ages.

It will be evident from the foregoing discussion, then, that in the course of human embryonic development of the spinal cord not merely one but two types of primitive nervous reflex mechanisms appear. The sensory side of the system is represented first by neurons of unipolar type which are homologous to the Rohon-Beard cells of amphibians and other lower vertebrates and the motor side by a type of motor neuron comparable to the primary motor neuron described by Coghill for Amblystoma. These neurons occur at _a very early age (5 weeks) long before functional activity begins. Apparently the sensory cells, at least, disappear (p. 33), but nothing less has been observed to indicate the possible fate of the primitive motor neurons.

Appearing at the same age (5 weeks), but reaching a well developed stage at the time at which functional activity begins (7 to 8 weeks), is a second type of primitive sensory cell - an intramedullary bipolar sensory neuron. Cells of this type, too, are relatively few in number, but structurally are more nearly comparable to the sensory ganglion cells than are the Rohon-Beard neurons, and, like the sensory ganglion cells, are transformed into unipolar sensory neurons. At comparable ages (7 to 8.5 weeks) a second type of primitive motor neuron has been found, one which is structurally like the motor neurons present "in adult fishes rather than the primary motor neurons described by C-oghill. At this age, likewise, at least some of the permanent motor neurons and sensory ganglion cells also appear to have reached a functional stage in development.

It should be stated that functionally the primitive nervous mechanism, with its few large neurons which supply a wide territory, is suited to a response of a. very generalized nature—the type of reaction designated by Coghill as a total pattern response. But when adaptability and selectivity of response is the desirable thing, then large numbers of neurons, which may become specialized for different reactions, meet the requirements of the organism more satisfactorily. Such is particularly the case with terrestrial animals.

On the sensory side of the reflex arc, this specificity in reaction is brought about in the nervous system by an increase in number of the sensory neurons, the development of different types of nerve endings and the corresponding changes in size of cell bodies and degree of myelination of nerve fibers. On the motor side of the arc, the same result is obtained again through increased numbers of cells but here through the differentiation of these neurons into cell columns supplying impulses to related groups of muscles and even further differentiaation into subgroups of neurons for still more specialized reactions. Thus it becomes possible to register a specific type of stimulus and bring about a localized muscular response. With the primitive nervous mechanism, where there are no specialized nerve endings and a single neuron supplies a large area, only differences in intensity of stimulation or in ar.ea stimulated (i.e., deep or superficial) are possible.

From the anatomical facts presented here it would seem that when human fetal activity first begins, the types of response which occur may vary, depending somewhat upon the nature of the stimulus. For instance, light stimulation may result in a generalized response through the primitive sensory and motor neurons; or a deep type of stimulus of considerable strength may produce a more localized response through the permanent sensory and motor neurons. It is even conceivable that both responses may succeed each other in the same embryo, or that sometimes one, sometimes the other may apparently occur exclusively in embryos of the same age, neither occurring first consistently. Considering the marked telescoping of developmental structures on the anatomical side of the picture, a similar telescoping in functional activity might be expected. Nor would it be surprising to find, particularly in a highly specialized animal in which early use of some specific part of the body is of basic importance to survival, that further structural and functional telescoping might result in localized activity of that part alone as the first response to stimulation. In any event, it should be emphasized that a generalized response, for which primitive neurons are morphologically adapted, represents the more fundamental type of reaction.

At first, it may seem surprising that primitive neurons which exist only as transitory structures in amphibians should occur at all in the human embryo, particularly since they have not been noted in any other mammal. It need be no more surprising, however, than that a non-nervous structure, the pronephros (which begins its development later in embryonic life than does the nervous system) appears in human embryos even though it never functions and disappears by the time the embryo has attained a length of 4.9 mm., the same period at which the earliest type of primitive motor and sensory neurons are present. Indeed, it would be more startling if none of the primitive nerve cells of lower vertebrates ever made an appearance in any of the mammals. That these nerve cells should be relatively very few in number in the human embryo, exist for a short period of time, in some cases be. nonfunctional and perhaps even fail to occur at all, particularly in forms which are highly specialized in type, is to be expected. In fact, it was suggested to the author, in 1938, in a personal communication from Dr. Elizabeth Crosby, that there should be some representation in the developing human spinal cord of the primitive motor and sensory neurons present in lower vertebrates. As indicated by Prof. H. H. Donaldson (’37), a type of nervous structure which has developed early in phylogeny is not readily abandoned, but is carried on to higher forms and modified. It seems that the same principle holds true for the nervous system in embryonic development.


The findings reported in this paper are based on the serially sectioned central nervous system of human embryos and fetuses of 5 mm., 16 mm., 22.5 mm., 26 mm., 26.5 mm., 89.5 mm., and 145 mm. in crownrump length. Three types of staining have been used-—activated protargol, pyridine silver and erythrosin and toluidin blue.

Two types of primitive sensory neurons are described in the embryonic human central nervous system. The earlier of these is present at 5 mm., but has not been observed in the later stages studied. This type of sensory neuron is considered to be homologous with the RohonBeard cells of such lower vertebrates as amphibians. At this age these sensory neurons vary in their degree of development at different levels of the nervous system and definitely show differentiation in a cephalocaudal direction.

The second type of primitive sensory neuron is an intramedullary bipolar neuron which first appears when the Rohon-Beard cells are still present. Such cells, however, can be recognized in larger numbers at the age at which activity begins. In the material studied these bipolar sensory cells are best represented at 8 weeks (22.5 mm.) of menstrual age, at which time some of them are changing to a unipolar shape. Such intramedullary sensory cells develop earlier than do the sensory neurons in the spinal ganglia, but the disparity in degree of differentiation is much greater at 5 mm. than later, so that at 22.5 mm. there is little difierence between these neurons and the best differentiated spinal ganglion cells. The isolated intramedullary unipolar sensory neurons found at later ages (14.5 and 18.5 weeks) are considered to be cells of this bipolar sensory type which have developed into unipolar neurons and have been retained.

Two types of primitive motor neurons are also described in the embryonic human spinal cord. The earlier to develop appears at the same age (5 mm.) as do the primitive sensory neurons of Rohon-Beard, and is considered homologous to the primary motor neurons described by Coghill for Amblystoma. It-has not been observed at later ages.

The second type of primitive motor neuron is present only at later ages (16 mm., 22.5 mm., and 26.5 mm.), but, ‘like the second primitive sensory neuron, it is present at the time when activity first appears. These motor neurons are not comparable structurally to the primary motor neurons described by Coghill for Amblystoma, but resemble the motor neurons of adult fishes in the dendritic pattern and in the dendritic origin of the axon.

Both the intramedullary bipolar sensory neurons and the second primitive motor nerve cells which develop are present at the age when fetal activity first occurs. Morphologica.lly there is every indication that these primitive cells, both motor and sensory, are just as capable of functioning as are the permanent motor neurons and the sensory ganglion cells at comparable ages. The limited number of such neurons reported, particularly of the motor type, does not exclude the possibility of functional activity, for the number of cells observed is no true indication of the number which may be present, partly because of the selectivity of silver stains and partly because identification with the cellular stains used is only possible with a very favorable plane of sectioning.

The greater number of the primitive neurons of the Rohon-Beard type found in the earliest embryo studied (5 mm.) are on the left rather than the right side of the central nervous system. At least one other portion of the central nervous system which begins to develop but later disappears, the accessory olfactory bulb, remains best differentiated on the left in the human fetus (Humphrey ,’40) and is represented on the left and not on the right in the young macaque (Crosby and Humphrey, ’39).

Morphologically the primitive neurons are adapted to a generalized type of activity rather than to specific, localized reactions. This is evidenced in part by the large cell body and wide dendritic spread of the motor neurons (p. 40) and the uniformly large size of the intramedullary bipolar sensory cells (p. 40). As indicated by the earliest development and difierentiation, the primitive neurons should be capable of carrying impulses at an even younger age than either the spinal ganglion cells or the permanent motor neurons. Certainly, at the age at which human fetal activity has first been observed, the primitive neurons are equally able to function. The rapidity with which development occurs in the human embryo, however, may result in a telescoping of activity (p. 40) just as appears to be the case in structural develop ment.

Literature Cited

AaIiiNs Knrrms, C. U., G. CARL Hvsns AND E. C. Cxosnv 1936 The comparative anatomy of the nervous system of vertebrates, including man. The Macmillan Co., New York.

BEARD, J. 1889 On the early development of Lepidosteus osseus. Proc. Roy. Soc., London, vol. 46, pp. 108-118.

1892 The transient ganglion cells and their nerves in Raja batis. Anat. An7.., Bd. 7, S. 191-206.

1896a On the disappearance of a transient nervous apparatus in the series: Scyllium, Ac-anthias, Mustelus and Torpedo. Anat. Anz., Bd. 12, S. 371-374.

1896b The history of a transient nervous apparatus in certain Ichthyopsida. An account of the development and degeneration of ganglion-cells and nerve-fibres. Part I". Raja batis. Zool. Jahrb., Abt. f. Anat., Bd. 9, S. 319-426.

BECCARI, N. 1909 Le cellnle dorsali o posteriori dei Ciclostomi. Ricerche nel Petroymzon marinus. Monitorc zool. ital.', vol. 20, pp. 308-323.

BODIAN, D. 1936 A new method for staining nerve fibers and nerve endings in mounted paraffin sections. Anat. Rec., vol. 65, pp. 89-97.

1937 The staining of parafiin sections of nervous tissues with activated protargol. The role of fixatives. Anat. Rec., vol. 65, pp. 89-97.

Box, 8. T. 1928 Das Riickenmark. In “Handbuch der mikroskopischeu Anatomic des Menschen” by W. v. Mtillendorff. J. Springer, Berlin.

BUCKLAND, K. R. 1889 Histologische Untersuchungen am Riickenmark der Tritonen. Arch. f. rnikr. Auat., Bd. 34, S. 131-156.

Buns, H. S. 1928 The central nervous system of Orthagoriseus mola. J. Comp. Neur., vol. 45, pp. 33-128.

Coghill, G. E. 1913 The primary ventral roots and somatic motor column of Amblystoma. J. Comp. Neur., vol. 23, pp. 121-143.

1914 Correlated anatomical and physiological studies of the growth of the nervous system of Amphibia. I. The afferent system of the trunk of Amblystoma. J. Comp. Neur., vol. 24, pp. 161-233.

1916 Idem. II. The afferent system of the head of Amblystoma. J. Comp. Neur., vol. 26, pp. 247-340. 1924 Idem. IV. Rates of proliferation and differentiation in the central nervous system of Amblystoma. J. Comp. Neur., vol. 37, pp. 71-120.

1926a. Idem. V. The growth of the pattern of the motor mechanism of Amblystoma punctatum. J. Comp. Neur., vol. 40, pp. 47-85.

1926b Idem. VI. The mechanism of integration in Amblystoma punctatum. J. Comp. Neur., vol. 41, pp. 95-152.

1929 Anatomy and the problem of behavior. The Macmillan Co., New York.

1934 New anatomical relations and the probable function of Mauthner’s fibers. Psychiat. en Neurol. BI., Bd. 38, S. 386-391.

COWDRY, E. V. 1924 General cytology. The Univ. of Chicago Press, Chicago, Illinois.

Caossr, E. 0., AND T. HUMPHREY 1939 Studies of the vertebrate telencephalon. I. The nuclear configuration of the olfactory. and accessory olfactory formations and the nucleus olfactorius anterior of certain reptiles, birds and mammals. J. Comp. Neur., vol. 71, pp. 121-213.

DAHDGREN, U. 1897 The giant ganglion cells in the spinal cord of the order Heterosomata Cope (Anacanthini P1euronectoideiGuenther). Anat. Anz., Bd. 13, S. 281-293. 1898 The giant ganglion cell apparatus. J. Comp. Neur., vol. 8, pp. 177-179.

DETWILER, S. R. 1937 Observations upon the migration of neural crest cells, and upon the development of the spinal ganglia and vertebral arches in Amblystoma. Am. J. Anat., vol. 61, pp. 63-94.

DoNAm>so1$‘, H. H. 1937 The nervous skeleton. Tr. Am. Neurol. Assoc., pp. 1-9.

Frmonnann, J. E., AND W. F. Wmnnr. 1942 Some observations on early human fetal movements. J. Comp. Neur., vol. 76, pp. 159-167.

VAN GEHUCHTEN, A. 1897 Contribution a l’étude de la moelle épiniére chez les vertébrés (Tropidonotus natrix). La Cellule, vol. 12. pp. 115-165.

1898 La moelle épiniere des larves des Batraciens (Salamandra maculosa). Arch. de Biol., vol. 15, pp. 599-619.

HARRISON, R. G. 1901 Ueber die Histogenese der peripheren Nervensystem bei Salmo salar. Arch. 1?. mikr. Anat., Bd. 57, S. 354-444.

HERRICK, C. J., Am) G. E. COGHILL 1915 The development of reflex mechanisms in Amblystoma. J. Comp. Neur., vol. 25, pp. 65-85. Host}, 1. D. 1

944 The development of the nucleus dorsalis (Clarke ’s column). J. Comp. Neur., vol. 81. pp. 69-96.

HOOKER, D. 1936 Early fetal activity in mammals. Yale J. Biol. and Med., vol. 8, pp. 579-602. .

1939a Fetal behavior. Res. Publ. Assoc. Nerv. Ment. Dis., vol. 19, pp. 237-243.

1939b A preliminary atlas of early human fetal activity. Published privately by the author.

1943 The origin of overt behavior. University lecture delivered at the University of Michigan, April 30, 1943. Univ. of Mich. Press (in press).

VAN DER I-Ioasr, C. J. 1918 Die motorischen Kerne und Bahnen in dem Gehirn der Fische, ihr taxonomiseher Wert und ihre neurobiotaktische Bedeutung. Tijdschr. (1. Ned. Dier. Vereen. D1. 16, Afl. 2, S. 168-270.

1936 Table on p. 173 in Ariéns Kappers, Huber and Crosby, The comparative anatomy of the vertebrates, including man. The Macmillan Co., New York.

HUBIPHREY, T. 1940 The development of the olfactory and the accessory olfactory formations in human embryos and fetuses. J. Comp. Neur., vol. 73, pp. 431-468.

INGALLS, N. W. 1907 Beschreibung eines menschlichen Embryos von 4.9 mm. Arch. f. mikr. Anat., Bd. 70, S. 506-576.

JOHNSTON, J. B. 1900 The giant ganglion cells in Catostomus and Coregonus. J. Comp. Neur., .vol. 10, pp. 375-381.

Keibel F. and Mall FP. Manual of Human Embryology I. (1910) J. B. Lippincott Company, Philadelphia.

Konmm, W. 1905 Zur Kenntnis des Riickenmarks von Ammocoetes. Anat. Hefte, Bd. 29, S. 163-214.

KOLSTER, R. 1898 Ueber bemerkenswerthe Ganglienzellen im Riickenmark von Perca fluviatilis. Anat. Anz., Bd. 14, S. 250-253.

von LaNHoss15:K, M. 1892 Beobachtungen an den Spinalganglien und dem Riickenmark von Pristiurusembryonen. Anat. Anz., Bd. 7, S. 519-539.

PENFIEILD, W. 1932 Cytology and cellular pathology of the nervous system. Paul B. Hoeber, Inc., New York.

RAMON Y CAJAL, S. 1909 Ilistologie du systéme nerveux de l’homme et des vertébrés. T. l. A. Maloine, Paris.

RANSON, S. W. 1943 The anatomy of the nervous system. W. B. Saunders Co., Philadelphia and London. 7th’edition.

RETZIUS, G. 1894 Die embryonale Entwicklung der Riickenmarkselemente bei den Ophidiern. Biol. Untersuch., N. F., Bd. 6, 8. 41-45.

1898a Weiteres uber die embryonale Entwicklung der Riickenmarkselemente der Ophidier._Biol. Untersuch-., N. F., Bd. 8, S. 105-108.

1898b Zur Kenntniss der Elemente des Riickenmarks von Anguis fragilis. Biol. Untersuch., N. F., Bd. 8, S. 109-113.

SARGENT, P. E. 1898a The giant ganglion cells. in the spinal cord of Ctenolabrus coernleus. Anat.‘ Anz., Bd. 15, S. 212-225.

1898b The giant ganglion cells in the spinal cord of Ctenolabrus adspersus (Wal'b.:Goode). J. Comp. Nenr., vol. 8, pp. 183-194.

STREETER, G. L. 1920 Weight, sitting height, head size, foot length, and menstrual age of the human embryo. Contrib. to Embryol. (Carn. Inst. Wash. Publ. no. 274), vol. 11, pp. 143-170.

STRONG, O. S., AND A. ELWYN 1943 Human neuroanatomy. The Williams and Wilkins Co., Baltimore.

STUDNICKA, F. K. 1896 Ein Beitrag zur vergleichenden Histologie und Histogenese des Riickenmarkes. (Ueber die sog. Hinterzellen des Riickenmarkes.) Sitzungsb. d. k.-biihm. Gesellsch. d. Wissensch., Math.-naturwis. K1., Druckjahr

1896, no. 33, S. 48-89. Reviewed in J. Comp. Neur., vol. 6, Literary Notices, pp. 48-49.

TAGIJAN1, G. 1899 Ueber die Riesennervenzellen in Ruckenmarke von Soles. impar. Anat. ' Anz., Bd. 15, s. 234-237.

TERN1‘, T. 1922 Ricerche istologische sul midollo spinale dei rettilli, con particolare riguardo

ai componenti spinali del fascicolo longitudinale mediale = (Osservazioni in Gongylus ocellatus Wagl.). Arch. ital. di anat. e di embryol., vol. 18, Suppl., pp. 183-243.

TRETJAKOFIF, D. 1909 Das Nervensystem von Ammocoetes. 1. Das Ruckenmark. Arch. 1'. mil-rr. Anat., Bd. 73, S. 607-680.

Wmnmz, W. F. 1934 Correlation between the development of local reflexes and reflex arcs in the spinal cord of cat embryos. J. Comp.-Neur., vol. 59, pp. 487-503. 1940 Physiology of the fetus. W. B. Saunders Co., Philadelphia and London.

Wmnns, W. E, AND J. E. FITZGERALD 1937 Development of the spinal reflex mechanism in human embryos. J. Comp. Neui-., vol. 67, pp. 493-506.

YOUNGSTROM, K. A. 1938 Studies on the developing behavior of Anura. J. Comp. Neur., vol. 68, p. 351-379.

1940 A primary and a secondary somatic motor innervation in Amblystoma. J. Comp. Neur., vol. 73, pp. 139-151;

Cite this page: Hill, M.A. (2019, June 17) Embryology Paper - Primitive neurons in the embryonic human central nervous system. Retrieved from

What Links Here?
© Dr Mark Hill 2019, UNSW Embryology ISBN: 978 0 7334 2609 4 - UNSW CRICOS Provider Code No. 00098G