Difference between revisions of "Journal of Morphology 21 (1910)"

From Embryology
Line 53: Line 53:
===No. 1. MARCH===
===No. 1. MARCH===
Robert E. Coker. Diversity in the scutes of Chelonia. Fourteen plates. ... 1 Edwin Chapin Starks. The osteology and mutual relationships of the fishes belonging to the family Scombridae. Three plates 77
Robert E. Coker. Diversity in the scutes of Chelonia. Fourteen plates. ... 1  
J. Thomas Patterson. Studies on the early development of the hen's egg. 1. History of the early cleavage and of the accessory clea\age. Thirty-two figures 101
Edwin Chapin Starks. The osteology and mutual relationships of the fishes belonging to the family Scombridae. Three plates 77
J. Thomas Patterson. Studies on the early development of the hen's egg. 1. History of the early cleavage and of the accessory cleavage. Thirty-two figures 101
===No. 2. JULY===
===No. 2. JULY===
Line 82: Line 84:
CiiDEON S. DoDDS. Segregation of the germ-cells of the Teleost, Lophius. Thirty-four figures 563
CiiDEON S. DoDDS. Segregation of the germ-cells of the Teleost, Lophius. Thirty-four figures 563
J. Paksons Schaeffer. The lateral wall of the cavum nasi in man, with especial reference to the various developmental stages. Fifty figures 613
J. Paksons Schaeffer. The lateral wall of the cavum nasi in man, with especial reference to the various developmental stages. Fifty figures 613
Line 88: Line 92:
J. ¥. Gvdernatsch. The thyreoid gland of the Teleosts. Twenty-one text figures and five plates 709
J. ¥. Gvdernatsch. The thyreoid gland of the Teleosts. Twenty-one text figures and five plates 709

Latest revision as of 20:24, 21 May 2020

Embryology - 7 Jun 2020    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)

A personal message from Dr Mark Hill (May 2020)  
Mark Hill.jpg
I have decided to take early retirement in September 2020. During the many years online I have received wonderful feedback from many readers, researchers and students interested in human embryology. I especially thank my research collaborators and contributors to the site. The good news is Embryology will remain online and I will continue my association with UNSW Australia. I look forward to updating and including the many exciting new discoveries in Embryology!
Historic Disclaimer - information about historic embryology pages 
Mark Hill.jpg
Pages where the terms "Historic" (textbooks, papers, people, recommendations) appear on this site, and sections within pages where this disclaimer appears, indicate that the content and scientific understanding are specific to the time of publication. This means that while some scientific descriptions are still accurate, the terminology and interpretation of the developmental mechanisms reflect the understanding at the time of original publication and those of the preceding periods, these terms, interpretations and recommendations may not reflect our current scientific understanding.     (More? Embryology History | Historic Embryology Papers)

Journal of Morphology 21 (1910)

J Morphol. : 1 - 1887 | 2 - 1888-89 | 3 - 1889 | 4 - 1890 | 5 - 1891 | 6 - 1892 | 7 - 1892 | 8 - 1893 | 9 - 1894 | 10 - 1895 | 11 - 1895 | 12 - 1896 | 13 - 1897 | 14 - 1897-98 | 15 - 1898 | 16 - 1899-1900 | 17 - 1901 | 18 - 1903 | 19 - 1908 | 20 - 1909 | 21 - 1910 | 22 - 1911 | 23 - 1912 | 24 - 1913 | 25 - 1914 | 26 - 1915 | 27 - 1916 | 28 - 1916-17 | 29 - 1917 | 30 - 1917-18 | 31 - 1918 | 32 - 1918 | 33 - 1919-20 | 34 - 1920 | 35 - 1921 | 36 - 1921-22 | 40 - 1928 | 47 - 1929 | 51 - 1931 | 52 - 1931 |
Historic Journals: Amer. J Anat. | Am J Pathol. | Anat. Rec. | J Morphol. | J Anat. | J Comp. Neurol. | Johns Hopkins Med. J | Ref. Handb. Med. Sci. | J Exp. Zool.
Links: Historic Journals | Historic Embryology Papers

Founded BY C. O. Whitman



Tufts College, Mass.

with the collaboration of Gauy N. Calkins T. H. Montgomery

Columbia University University of Pennsylvania

VV. M. Wheeler William Patten

Bussey Institution, Harvard University Dartmouth (IJollege

Edwin G. Conklin

Princeton University





By Thio Williams & VVilkins Company

Baltimore, U. S. A.



No. 1. MARCH

Robert E. Coker. Diversity in the scutes of Chelonia. Fourteen plates. ... 1

Edwin Chapin Starks. The osteology and mutual relationships of the fishes belonging to the family Scombridae. Three plates 77

J. Thomas Patterson. Studies on the early development of the hen's egg. 1. History of the early cleavage and of the accessory cleavage. Thirty-two figures 101

No. 2. JULY

Harry Lewis Wieman. A study in the germ cells of Leptinotarsa signaticoUis. Seventy-three figures 135

C. W. AND G. T. Hargitt. Studies in the development of Scyphomedusae. Forty-nine figures 217

James Homer Wright. The histogenesis of the blood platelets. Twenty-one figures 26.3

X. M. Stevens. Further studies on reproduction in Sagitta. One hundred and two figures 279


Robert J. Terry. The morphology of the pineal region in Teleosts. Twenty figures 321

H. H. Newman and J. T. Patterson. The development of the nine-banded armadillo from the primitive streak stage to birth; with especial reference to the question of specific polyembryony. Fifteen text figures and nine plates 359

Leland Griggs. Early stages in the development of the central nervous system of Amblystoma punctatum. Twelve text figures and one plate 425


H. li. WiEMAN. The degenerated cells in the testis of Leptinotarsa signaticoUis. N iuo figures 485

H. S. Jennings and G. T. Hargitt. Characteristics of the diverse races of Paramecium. Twenty-four figures 495

CiiDEON S. DoDDS. Segregation of the germ-cells of the Teleost, Lophius. Thirty-four figures 563

Schaeffer JP. The sinus maxillarus and its relations in the embryo, child, and adult man. (1910) Amer. J Anat. 10: 313-

J. Paksons Schaeffer. The lateral wall of the cavum nasi in man, with especial reference to the various developmental stages. Fifty figures 613


J. ¥. Gvdernatsch. The thyreoid gland of the Teleosts. Twenty-one text figures and five plates 709



With Text Figs. A to Q and Plates I to XIV.



Introduction 2

Character of the Diversity 2

Some Recent Views Regarding the Phylogenetic Significance of Normal and Abnormal Scutes 5

Use of Terms 9

Part I. Malaclemmys 13

Section 1. Observations 20

Section 2. Review of Observations 20

Inframarginals 20

Interplastrals 21

Plastrals 22

Marginals 23

Nuchal 24

Costals 24

Neurals 28

Section 3. Adjustment of Neurals and Costals 33

Section 4. Age, Sex, Symmetry 43

Summary 43

Part II. TlialassochcJys carctta (L) 46

Section 1. Introduction 46

Material 46

Conditions of Development 48

Explanation of Tables 49

Section 2. Observations on Diversity 50

Section 3. Review of Observations 61

Marginals 61

Supra-Marginals 62

Nuchal 62

Costals 62

This paper was completed in June, 1906. While unavoidable circumstances have prevented its earlier appearance, the text has had no material alteration.

The .Touenal of Morphology. — Vol. 21, No. 1.

2 Robert E. Coker.


Neurals G3

Adjustment of JMeurals aud Costals G4

"Incomplete Division" 65

Part III. Significance of the Abnormalities G7

Introduction G7

External Conditions G8

Inheritance G9

Atavism 69

Summary of Part III 73

Note 73

Literature cited 74

Explanation of Plates 75


At the begiuning of an inquiry regarding the diamond-back terrapin, Malaclemmys centrata (Latr.), undertaken in 1902, raj attention was attracted bj the frequent instances of striking deviation from the normal" number and arrangement of the horny scutes of the carapace. The anomalies seemed of sufficient interest to justify a record of the facts, and accordingly were included in the notes made in connection with the economic study then in progress. Later, the study was extended in some lines, as it was seen that the anomalies might have some significance for theoretical interpretation or experimental study.

For kind permission to use in this paper such data as was obtained while employed in the economic study of the terrapin, my acknowledgmient is due to Hon. George M. Bowers, United States Commissioner of Fish and Fisheries, and to Professor J. A. Holmes, State Geologist of North Carolina. My thanks are also due to the officials of the Bureau of Fisheries and to Dr. Caswell Grave, Director of the Fisheries Laboratory, for many courtesies extended to me while occupying a research table in the Laboratory.

I wish to make grateful acknowledgment of my indebtedness to Prof. W. K. Brooks, under whose guidance the study has been prosecuted.

Character of the Diversity.

At first I was particularly impressed by the comparatively great prevalence of apparent "abnormalities" in the number of horny

Diversity in the Scutes of Chelonia. 3

scutes of the carapace. Longer acquaintance with a number of individuals of this species that were kept under observation at the Fisheries Laboratory at Beaufort, ISForth Carolina, and with the eggs, embryos, and young of the loggerhead sea-turtle Thalassochelys caretta, made apparent a wide diversity in many respects. Among the characters in which the turtles and eggs manifested striking individual differences, we may mention shape and size of eggs and of young, shape and size of head, shape of carapace, depth of body, color pattern, boldness, habits of feeding and of hibernating and of moulting, rate of growth, etc. It seemed that a diversity in respect of scutes and plates was no more than one would expect in view of the diversity shown in so many other respects. The aphorism that no two individuals are exactly alike would seem to apply with preeminent fitness to certain species of turtle.

JSTevertheless, there are certain features of these "'abnormalities" which the observer cannot but be impressed with, and which seem to suggest some special significance. These features are:

1. The frequent recurrence of certain more or less regular scutes in definite positions.

2. The striking correspondence of recuri'ing abnormal elements of one species to those of another, and sometimes to normal scutes of other species. Thus —

(a) In 31 specimens of new-born gTeen turtles* 44 different abnormalities were noted, but these were reducible to 11 types; or, to omit variations which occurred not more than twice and are possibly coincidences, we have 39 abnormalities of 7 types. The most anterior scute of the plastron, the normally unpaired intergular, was in 6 specimens represented by a pair of scutes, and in 9 other specimens was partially divided. This scute occurs in few genera, but in the normal conditions of Macroclemmys and of Chelys (Gadow, '01, p. 325), there is found in this position a pair of scutes.

(b) In 3 specimens of the same lot a rectangular scute appeared in the neural series between the normal fourth and fifth shields. An almost exactly similar scute occurs twice as the only abnormality of dorsal scutes noted in 4 specimens of Thalassochelys (Colpochelys)

See author's previous paper, '05a, p. 23.

4 Robert E. Coker.

kempii Garman ; it is noted in several specimens of T. caretta, and in other species J and a practically identical scute is iigured by Boulenger for the remote species Chelodina novce-guinece Boulenger (Boulenger, '89, PI. 5), but the author does not state whether or not the presence of this scute is normal. There seems some ground for the belief that such a definite recurrence of a scute of fairly regular shape and position has some special significance.

(c) In 3 of the above 31 turtles, the nuchal was represented by a pair of scutes. The sanje abnormality occurred in 10 of 243 specimens of Malaclenwnys (Tables I-IV) and in 9 others the nuchal, though unpaired, was marked by a median longitudinal groove. This shield occurs in paired condition in several specimens of Thalassochelys. In some species, notably in Chrysemys guttatus, the nuchal often shows a distinct notch in the anterior margin.

(dj In most genera of land and fresh-water turtles, axillary and inguinal scutes are found anterior and posterior, respectively, to the bridge and just beneath the marginals (Fig. B). These scutes are regarded as the remnants of an ancestral series of infraonarginals separating the pectoral, abdominal^ and femoral scutes of the plastron from the marginals. Malaclemmys centrata has normally" only the axillary, yet over 21 per cent, of 244 specimens examined, possess inguinals, of varying size, on one or both sides.

A number of other cases might be mentioned of recurring scutes in definite positions in the neural, costal, and marginal series, but these instances are sufiicient to illustrate the nature of the conditions that have led some recent wl:"iters to assume that these abnormal scutes are atavisms and that, as such, they have a comparative value, similar to that of normal scutes, but of much more sigTiificance, since they may point more directly to remote ancestral forms.

Finally, it must be said that many other scutes are noted which are not of these definite types, but which are perhaps not less significant.

In the present paper I will present my observations on the scutes of two species of turtles, and then, in the light of these and other observations, will inquire into the basis for an atavistic interpretation. It is necessary first to have in mind the phylogenetic sig

Diversity in the Scutes of Chelonia, 5

nificance attached to normal scutes. Some representative views arcgiven in the next section.

Some Recent Views regarding the Phylo genetic Significance of

Normal and Abnormal Scutes.

O. P. Hay's view as to the origin of the carapace (Hay, '98) is quite important in this connection, for he seems to have been the first to realize the probable phylogenetic significance of the epideiinal scutes. Previous discussions had centered chiefly about the dermal and periosteal bony skeleton.

Hay distinguishes three kinds of bone in the carapace: (1) cartilage bone; (2) true dermal bone, developed in the skin itself, and comparable to the osteodermal plates of Dermochelys and of the crocodiles; (3) fascia bone, originating subcutaneously by ossification of the fascia below the skin. The present carapace owes its phylogenetic origin to the complete fusion of cartilage and fascia bone. Tnie dermal bone has probably completely or almost completely disappeared from the carapace of modern Thecophora. But, according to Hay, the ancestors of all turtles, Thecophora, as well as Athechffi, have possessed an armor of true dermal bony plates, probably mosaically arranged, and with twelve well differentiated keels. Of the keels, five were dorsal (one median and two pairs lateral), two were marginal, and five ventral (one median and two pairs lateral). The plates of this armor were adapted to previously formed epidermal scutes of the same mosaic pattern. This osteodermal armor is retained in Dermochelys with essentially the primitive pattern, but the epidermal shields are lost in the adult and indicated only in the young.

In Thecophorous turtles the mosaic 12-keeled dermal armor underwent important modifications partly in correlation with the greater development of the internal skeleton. The plates became very much reduced in number, a few of the plates of the keels, growing in size, and assuming the whole function of the protective ai-mor, crowded out many of the keel plates, and all of the smaller plates between the keels. Some of the keels too were lost. As the result, we may suppose a dermal carapace of twelve series of scute-covered plates, with

6 Eobert E. Coker.

a comparatively small number of miits in eacli series. This armoi overlapped, or broke joints," with the yet imperfectly developed cartilaginous carapace, but with the further development of the latter, and the consequent loss of the usefulness of the outer armor to the animal, the dermal skeleton became reduced and finally disappeared. The epidermal scutes, corresponding to these plates, remain, however, and from their present arrangement in the carapace of Thecophorous turtles we are enabled to infer something of the modifications which the dermal armor underwent. Probably the last remnants of the dermal plates persisted as small ossicles at the keel prominences that are noticeable in adults of some species, but especially in the young of certain species (e. g., of Thalassochelys, cf. Fig. 74:). In the further course of evolution these remnants became either completely reduced or merged into the deeper plates underlying them; but Hay has observed at least one, and probably two such ossicles in the neural series of a fossil specimen of Toxochelys.

Thus, in Hay's view, the series of scutes of the carapace are directly homologous to the series of epidermal areas overlying the plates of the keels of a young Dermochelys. The neural series corresponds to the median dorsal keel. Of the two lateral dorsal keels, one is represented by the costals, while the other is lost in most turtles, but preserved as a short series of supra-marginals in Macroclemmys. Marginal keels correspond to marginal scutes. Of the two lateral ventral keels, the internal gave rise to the plastral scutes, while the external is more or less reduced, but still survives in most turtles, either as a continuous series of inframarginals (sea turtles, etc.), or as isolated axillary and inguinal shields. In some land tortoises this series is entirely unrepresented. The median ventral is almost entirely lost, but remnants are seen, for example, in the characteristic unpaired intergular of Chelodina, and in the occasional occurrence in other species of small unpaired scutes in the median line of the plastron. Such a scute is most commonly found just at the apices of the gulars and is referred to as an intergular," or, better, as an "interplastral." An hypothesis advocated by J^ewmann follows in logical order


Diversity in the Scutes of Chelouia. 7

upon Hay's view; but a view advanced by Gadow should be referred to here, as it is not only the next in chronological sequence, but is the first in which the atavistic interpretation was applied in a comprehensive way to the abnormalities of scutes.

The two striking features of Gadow's paper are: (1) the attempt to classify the variations and interpret them as reversions to ancestral conditions; and (2) the hypothetical explanation of these atavisms as stages in ontogeny or arrests of development. In his own concise words, the abnormalities are viewed as "simply ontogenetic stages, passing reminiscences of earlier phylogenetic conditions" (Gadow, '05, p. 638).

Originally, according to this author, there was a scute for each dorsal plate. Thus, as there are eight transverse series of dorsal plates, each series consisting of a median neural and a pair of lateral costals, so there were originally at least eight transverse series of dorsal scutes, scute and plate coinciding. But a process of reduction ensued. First, by the reduction of a pair of costals (probably the second), the scutes of neural and costal series came to dovetail into one another, and this dovetailing plan was subsequently retained throughout all stages. Gradually the number of scutes was reduced by the suppression — first, of the second costals, then of a neural; then fifth costals, fifth neural, and finally by the fusion of the last two costals; thus was attained the present typical condition of Thalassoclielys with six neurals (including the nuchal), and five pairs of costals. ^\lien turtles are found with more than this number of scutes the condition is to be regarded as "reminiscent" of one of these phylogenetic stages. The order of suppression of scutes given above is inferred from the relative frequency of recurrence of "supernumerary" scutes in the several positions. Thus far, Gadow's interpretation, though open to criticism, is interesting and suggestive. Wlien he goes further, and regards these atavisms, when found in adults, as instances of arrested development, or, when found in younger turtles, as proper stages in ontogenetic recapitulation of the phylogenetic stages, his position seems untenable on the basis of any facts now in hand, as the writer has previously sho^vn ('05 a), and as I^ewmann's observations also indicate (ISTewmann, '06. p. 92).

S Eobert E. Coker.

Xewmann ('06), who has made the most extended, study of the abnormalities of scutes and plates, lays much stress on the phylogenetic significance of scutes, but departs materially from Hay in that he regards Dermochelys as out of the line of descent of Thecophorous turtles, as "an abnormal and perhaps highly specialized forin." This is a position which Baur held and zealously defended in a series of papers from 1886 to 1896 ('86, '88 a and b, '89 a and b, '90, '96). 1 The view was adopted by Case ('97).

]S[ewmann bases his view primarily on a study of the assumed atavisms and of the color patterns of the carapace, but in part also on the comparison of the different scute-plans normal for different species of turtles. The keynote of his paper is expressed in the words: "Careful study has convinced me that these abnormalities are to be considered not as meaningless anoiualies but as examples of systematic atavism in the sense of de Vries. From this standpoint it seems possible to throw some light on the phylogeny of CJielonia."

These abnormalities as ISTewmann regards them are reversions, not to the keel plan of Dermochelys-like turtles, but to a plan essentially similar to that found in the "tail-trunk" of modern Chelydra, where he believes the primitive condition of the scutes is most nearly preserved. There he finds seven principal rows of scutes and, alternating with them, seven subordinate rows of smaller or less regular scutes. To homologize these with the series of scutes in the carapace and plastron, — the seven principal rows correspond to neural (one), costal (two), marginal (two), and plastral (two) ; these rows are found in all carapaces (except those of the Trionychoidea) . The seven subordinate rows correspond to (a) paired neuro-costals (lost in all turtles) ; (b) paired supramarginals (preserved normally only

^Newmann's statement ('06, p. 99) that Baur with Hay regarded Dermochelys as the ancestral form, seems based on a preliminary note by Baur dated October 6, 1886, and appearing in the American Naturalist for January, 1887 Subsequent to the writing of this preliminary note, even prior to its appearance, Baur had discarded the old and generally accepted view, so that his complete paper, dated October 26, and appearing in the Zoologische Anzeiger for November, 1886, announced unmistakably the view which he thereafter maintained.

Diversity in the Scutes of Chelonia. 9

in Macroclemmys) ; (c) paired inframarginals (preserved in sea turtles and, as inguinals and axillaries, in most other turtles, but entirely lost in a few land tortoises) ; and finally, (d) unpaired interplastrals — preserved normally only as a single intergular in a few species, as Chelodina. In general, the principal rows are retained in all turtles, but the subordinate rows are largely lost. Remnants of these latter series, however, are retained in the normal condition of some species, and, further, reappear as atavistic abnormalities in individuals of species that do not normally possess them. Thus^ inframarginals were found "abnormally" in Graptemys geograpliica and Clirysemys marginata, etc. ; interplastrals, in specimens of Chrysemys, Chelydra, and Graptemys. IsTewmann does not seem to have found either neuro-cosials or siipramarginuls as individual variations.

Furthermore, ISTewmann believes that the number of scutes of the neural and costal series was formerly about twice as great as at present, and that alternate scutes have been forced out and lost, but that these, again, recur in individuals as atavisms.

There has been, then, a reduction both in the number of rows and in the number of scutes in a row; but atavistic occurrences of the lost elements are met with in abnormal individuals of many species. From a systematic study of these atavisms, we may infer something of the evolutionary history of the Chelonian carapace and plastron.

ISTewmann differs from Hay in supposing fourteen original rows as opposed to Hay's twelve; in seeking the ancestral type not in Dermochelys. but in the disposition of the scutes of the base of the tail of Chelydra ; in supposing but a comparatively small number of original scutes in each series of the carapace ; and in seeking a basis for his views in the systematic study of atavisms.

The question of the value of the anomalous scutes as evidence of phylogeny will be discussed in a later portion of this paper.

Use of Terms.

The term "scute" will apply invariably to a horny shield of the epidermal carapace; "plate" to an element of the bony carapace; "seam" refers to the line of separation of adjacent scutes, as "suture" to that of adjacent bones.

The nomenclature of the scutes will be clear from Text-figs. A, B and C.


, Marginals

Fig. a. Normal carapace of Malaclemmys centrata (Latr.) (No. 253 of Table V). A median series of unpaired scutes composed of a small anterior nncliah followed by 5 large neurals; 4 pairs of costals; 12 pairs of marginals. Position of ^

"Irrterplastral A"> when present.!


(ln|uinal usually wanting.)

Fig. B. Normal plastron of Malaclemmys centrata (Latr.). Dotted lines indicate the positions of inguinals and interplastrals, normally absent.

Diversity in the Scutes of Chelonia.


(See also "abbreviations," below.) The nuchal is not, for present purposes, included in the neural series. As the term "intergular" is commonly applied both to the unpaired scutes, anterior to the gulars (typical of certain species) and also to a smaller or larger median scute at the posterior-median angles of the gulars (of other species), I will use the name intergular only in the former sense, for a median scute on the anterior margin between the gulars (Fig. C) ; interplastral is applied to other median shields of the plastron (Fig. B). The terms "inframarginal" and "submarginal" must be distinguished. Infrmnarginal applies to a series of scutes separating pectoral and abdominal shields from the marginal series (a typical series in Chcloniidae, etc. (Fig. C), and, presumably, represented in other species by the axillaries and -Uu/iiiiKih) (Fig. B) ; siihmarginal is used, as applied by Baur COO), to anomalous scutes found between the inframarginals and marginals (Fig. C).


Infra -'/ Marginals

Sub ,' Mar|inals

, Marginals

Fig. C. Plastron of a specimen of Thalassochelys caretta (L.). Note the anterior unpaired intergular, the series of inframarginals, and the small anomalous suhmarginals.

It seems impossible to escape the use of words that could be taken to imply more than is intended. It may, for instance, be perfectly normal for one turtle to have a scute not possessed by most of its fellows of the same species, but, for our present purpose, the number and arrangement typical of the species is termed "normal." A scute not found in the typical plan will be described as an "abnormality," an "anomaly," a "supernumerary" scute, or a "variation." The collective term least liable to mislead is "diversity," and this is used whenever practicable.

12 Kobert E. Coker.

Explanation of Tables.

Abbreviations: "L," left; "R," right; "N," neural; "C," costal; "M," marginal. Thus "Nl" denotes the first (most anterior) neural, LC4, the fourth (most posterior, tj'pically) left costal, etc.

In the "inguinal column, "L" denotes the presence of a left inguinal, "R" of a right, "L R" of both. If the left inguinal is distinctly larger than the right, this is indicated thus "L > R," or, if the reverse is the case, "L < R."

In the "Remarks" column "Supn'y" is used for supernumerary.

In the "nuchal" column, "pr" indicates that the nuchal is represented by a pair of scutes, "f," that it is marked by a longitudinal furrow, not distinguishable as a seam.

The letter "x" in any column refers the reader to the "Remarks" column.

In the "figures" column, letters refer to text figures, numbers to the figures on the plates.

Regarding the identification of scutes, it is not always clear to which series in the carapace an anomalous scute should be referred, or in fact whether there is reason in ascribing it to any normal series. But for convenience of description scutes will, when possible, be referred to normal series.

In dealing with forms of so high a market value as the diamond-back teri'apin (Tables I to V) it was not always possible to retain the specimens for further study. But in regard to scutes alone, revision of judgment is necessary in a comparatively small number of cases, and a field sketch frequently helps to remove this difficulty.

Regarding the figures, some are photographs, some tracings from photographs, and some camera sketches, while others are field sketches (see Explanation of Plates," etc.). No greater accuracy is claimed for the field sketches than that they represent, somewhat diagrammatically, the position, shape and proportionate size of the scutes depicted.

As diversity manifests itself in so many respects, we must of necessity disregard most individual differences, and consider only certain defined "abnormalities." In the tables that follow, a turtle is "normal," unless it possesses more or less than the number of scutes typical of the species, shows instances of partial division or partial fusion of scutes, has a median plate marked with a distinet and complete median groove, exhibits such asymmetry as with one scute of a pair ttcice as large as its mate, or has scutes not firmly united to the bone.

Diversity in the Scutes of Chelonia.



Ualaclemmys centrata (Latr.), Beaufort, N. C, 1902.

gth of PlasDn in inches.

-^ i










A ■























H-l « 


(— (

I— 1





4 4












4 .




5 4




Mil broad, evidently representing normal 11 and 12; minute scute anterior to LCI.








R. inguinal small.






(See remarks on Mil of No. 4)





4 5


Supn'y costal small, anterior, L axillary ^ R in size.

f 3 1 4






12 1 15 f













181 191




20). 23]




24) 25/




261 27 J









291 34)



Normal. No. 29 -with 8 rings; Nos.


30 and 34 with 5.

351 37/












Pair of scutes between N5, C's 4 and M's 12 and 13 ( = 5th pr. of costals, or 6th neural divided?) Supn'y marginal posterior.









411 43/









Deep notch between last pair of marginals.





Deep notch between RMIO and 11 (Scar?).




L. axillary wanting,






RMll and RM12 continuous distally, divided proximally.

481 49 j 50




Normal, except that 2d, 3d, and 4th

pairs of plastral scutes overlap, respec

tively, the scutes immediately posterior.

511 54/

Normal. (No. 53 with 4 rings of




growth. Seam between N4 and N5










4 5

Minute 5th R C. about 1-40 size of 4tli. Seam between N4 and N5 oblique.















L. axillary wanting.

601 61












5 4

LC4 reduced.

(Continued on next page.)


Robert E. Coker.


ontinued from preceding page.)









Ci4 O

u O















« a

02 tf


1-1 rt


64 >




66) 67,

9 6


Normal. No. 68 with 4 rings.

69 J 70



Anterior rings of N4 divided by rne


dian seam Median furrow on anterior


part of N2.


71) 72 J






9 4



Sagittal furrow on nuchal and N2, and, less distinct, on Nl and 4.








Small Interplastral A.



9 4







9 6




N2, 3 and 4 marked with sagittal



9 5



Normal, except that RM12 is small and axillaries are small.



9 7


Normal, faint median line on nuchal.







9 8

5 5





9 7

9 6


851 86 J 87


9 9








RM12 represented by 2 scutes; abdominal and femoral scutes continuous mesially, and mid-line of plastron^distorted to left posteriorly.

12 M








5 5

Costals small, anterior.









Diversity in tlie Scutes of Chelonia.



Malacleminys centrata (Latr.), Beaufort, N. C, 1903. (A few from Brunswick County, N. C.)

A. — Males.













eft. ight







^ « 

H-l « 

1— 1





94) 95 j"





R axillary ^ of L in size.


98) 104/



Axillaries wanting.



105) 107/





13 13

M13 small.

109) 113/





Axillaries not fixed to bone beneath, but movable; L<R.







117) 120/





R axillary twice as large as L.













B. — Femaues.



13 13

2 marginals in place of normal Ml.










12 13

2 scutes in place of normal RM12.



5 4

Small LC5.



5 4

12 13

2 scutes in place of normal RMl. Small L costal anterior to normal first.




Supn'y plastral scute between femoral and anal of R side.




5 4

Small 5th. LC.







R axillary twice L in size.

136) 137 /•






L inguinal small.




Small scute between Nl and 2, similar in size and shape to supn'y scute of No. 150, but entirely to left of median line and without keel. (Cf. fig. 6 of No. 150.)




141) 142/







LMll is less than i LMIO in size; RMllislessthan JRMIO.




(Continued on next page.)


Eobert E. Coker.

(Continued from preceding page.)





a •3





Left. Right.

Left. Right.






Supn'y M posterior to LM3, barely

showing dorsally. L abdominal and R

1 37

femoral meeting for more than i mesial

r 38

border of femoral.

1461 148] 149

4.4 4.4





2 scutes occupying place of N5. Ing.

1 11




Small scute anterior to L side of N4, extending barely over median line and bearing keel prominence.





4 5

12 13

Region of N3, N4 and N5 occu lied by

) 7 j- 15

6 scutes, the last incompletely divided.

L axillary wanting, R small.





Seam between N4 and N5 distorted slightly.

Normal, except that LM12 and RM




12 were very narrow proximally (*

1541 156/

width of 11) but of usual width distally.





L axillary wanting.




159) 160)







Inguinals small. Seam between N4 and N5 distorted.





13 13

Seam between N4 and N5 oblique.



Normal, but axillaries small.

164 1651 166 j



L inguinal small.





5 4

N3 incompletely divided, N4 represented by 2 scutes.







2 scutes in place of N5.



















12th margi'ls asymmetrically placed, LM12 having a position in median line.








RM's 2 and 3 fused proximally.


Diversity in the Scutes of Clielonia.


TABLE III. Malaclemmys centrata (Latr.). Observed at Crisfield, Md., 1903.

All Females.

Length of Plastron in inches.


Serial Number.








a 1


Left. Right.








13 '

LM12 and RM12 each represented by 2 scutes.


1791 180 i 181





5 4

2 scutes in region of N5, one quite



minute. Small interplastral.










RM12 and 13 equivalent to LM12.











12th marginal small, unequal.

186) 187 j










Normal, except for slight distortion of


posterior scutes of carapace.

192) 194/








Region of N3 to N5 occupied by 5 scutes.









Region of N5 occupied by 2 scutes.






Inguinals small.




? ?



3 scutes between N5 and marginals.







Region of N5 occupied by 3 scutes.






201) 206)











Furrow a little to right of median line of nuchal.






13th marginals small, with deep notch in R. Broad notch in outline of carapace in region of RMll.






? 5


Very abnormal carapace: see photo







Median furrow on nuchal and on anterior halves of Nl and 2.





















2 scutes posterior to N5. RC5?


|41 144




6 6








L inguinal small.







N5 reoresented by 2 scutes.





N4 elongated antero-posteriorly; N5 reduced, pair of minute triangular supn'y scutes.




225) 226 j




Robert E, Coker.


Embryos and Newborn — Malaclemmys centrata (Latr.),




















t-i a m



d f

£, M

to —

H-1 M

i 1













13 13

3 scutes in the place of N5.















Inguinajs small; slight notch in nuchal anteriorly.









4 5

12 12

Very abnormal in shape but scutes adhere to typical series arrangement; neural scutes most abnormal (v. fig. 25). Supn'y scute in L series of plastron posteriorly. L axillary wanting. A R inframarginal?












Misshapen: nuchal pushed over to R side; 12th L marginal represented by 2 scutes, LC4 reduced, N6 wider on L side. Posterior part of plastron shortened and distorted; large umbilical scar. (About 7 months old. Hatched from eggs laid in confinement.)








4 5













About 7 months old. Hatched in pound at Crisfield, Md.








Normal. (Cf . note on No. 240.)





(Cf. note on No. 240.)

Diversity in the Scutes of Chelonia.



Selected Abnormal Terrapin — Malaclemmys centrata (Latr.)



244 245

246 247

248 249


251 252

253 254



^ s

C o


3.5 3.4




3.6 4.

4.3 4.3






pr pr



1-1 rt

5 5 5 5

5 5

6 4 5 4

5 5

5 4







12 13

13 12


13 12


Median furrow on anterior portions of Nl and N2; supn'y costals anterior.

Supn'y costals anterior; LMll nearly equivalent to RMll and 12. (Carapace distorted in drying, and broken at "x".)

Supn'y costals anterior and small; seam between N4 and 5 distorted slightly.

N5 represented by a large and a small scute. (Note that, with seam removed there would be a scute of nearly normal size, and that the areola together would make a normal areola.)

One medium, one small, and one minute scute occupying place of LM9 — (Wound?).

Asymmetry of scutes and distortion of seams in posterior region of carapace (Neurals, Costals and Marginals).

V. fig. 50, also description p. 26.

7 scutes in neural series (2 in place of 4th).

5th neural renresented by 2 scutes of unequal size; RM12 incompletely divided.

Small L inguinal.

Large inguinals; small scute immediately anterior to 1j inguinal.

Of abnormal shape; 12th L marginal represented by 2 scutes; only 12 R marginals, but between RMll and RM12, a small gap occu|)ied by soft skin. Supn'y costals posterior: also a very small scute between LC4, LC5 and marginals. Nuchal partly divided; inguinals. (Vertebral column distorted to left posteriorly. L side of carapace flattened, somewhat concave posteriorly. Shell suggestive of injuries rec'd in embryonic stage. E\'idence of much shedding, horny covering thin and smooth.)

RM12 equivalent to LM12 and 13. 4 scutes occupjang a space almost exactly coinciding with that of normal N5. Small inguinal.

LC3 followed by two scutes, which together are almost exactly equivalent to a normal C4 (cf. RC4).

3 tlO



19 24

40 36


14 49




20 Kobert E. Coker.

Part I. Malaclemmys. The observations are presented as made on individual specimens in the tables of Section 1, and are treated in classified form in Section 2. The classification of abnormal neurals and costals leads to the discussion of adjustment of neurals and costals which forms the subject of Section 3.

1. Observations.

The first 243 specimens that could be observed carefully are included in Tables I to IV. These tables may, therefore, give a fair idea of the proportion of "abnormal" individuals in this species.

Table V includes a few specimens that were selected from a considerable number of others that have come under observation.

2. Review of Observations.^ Inframarginals.

The inguinals are by far the most common scutes not typical of the shield of this species. JSFo less than 21 per cent of the 243 specimens possessed one or both inguinals. In 33 individuals inguinals were present on both sides (v. Figs. D and E), but in 4 of these a distinct difference in size was noted between the scutes of the pair. Seventeen specimens had inguinals on only one side (Fig. F). Axillaries were wanting in only one terrapin, but were quite small in two others. In 5 specimens one axillary was wanting (Fig. L), and in 4 others one scute was at least twice as large as its mate of the opposite side (Fig. 4). Another individual (jSTb. 114) had axillaries of normal appearance, but they were found to be loosely attached to the bone beneath, the skin connecting them with the adjacent scutes permitting a certain freedom of movement when the axillaries were pressed with the finger. It may be noted from the tables that a right inguinal occurred alone 12 times as opposed to 5 instances of the left alone, and the right was distinctly larger than the left 3 times as against a larger left once ; also, while the right axillary occurred without the left 5 times, the left without the right was not noted. The right axillary was twice as large as the

"Numbers and percentages refer to Tables I to IV, unless otherwise stated.

Diversity in the Scutes of Olielonia.


left in 3 specimens, while the left was three times as large as the right in one terrapin. In Table V the left inguinal occurs twice without the right (Fig. F) and both are found in E"os. 254 (Fig. G) and 255. An additional scute in this "inframarginal series occurred in one or two specimens: in 254 and possibly in N^o. 233, (Figs. G and L.)

Inter plasirals.

The shield of Malaclemmys has normally no interplastral scute^ and such a scute w^as found to be a very rare abnormality, as it occurred only in ISTos. Y5 and 181. In both cases it was small and

Fig. F. ^---\-^ Fig. G.

Fig. D. Plastron of No. 149, showing unsyminetrical pair of inguinals. Fig. E. Plastron of No. 161, showing paired inguinals. Fig. F. Plastron of No. 253, showing left inguinal.

Fig. G. Plastron of No. 254, showing pair of large inguinals and a small scute anterior to left Inguinal.

situated at the meeting-point of gnlars and brachials (Fig. H, cf. Fig. B). In some species of turtles a large scute in this position is a normal feature of the shield (Fig. I). I do not know of any species in which characteristic interplastrals occur elsewhere. JSTew


Kobert E. Coker.

marni has observed interplastrals, as abnormalities, at other points, but always at the meeting-point of four plastral scutes.


The plastral scutes (cf. Fig. B) are remarkably constant; only two supernumerary scutes were noted. No. 132 showed an anomalous scute between the femoral and the anal of the right side (Fig. K), and in No. 233, two scutea occupied on the right side a space equivalent to that occupied by the femoral on the left (Fig. L). No. 88 presented an interesting abnormality discussed in a former paper ('05 a, pp. 14-18 and Fig. 6). In this specimen, the abdominal and femoral scutes were entirely separate to a certain point, but the last three rings of growth were perfectly continuous between the two scutes internally or mesially (Fig. M).

Fig. H.

Fig. I,

Fig. H. Anterior portion of plastron of No. 181, showing small "interplastral."

Fig. I. Part of plastron of a species of Chelodina, showing large "interplastral."

In at least one species of turtle Eretm-oclielys imhricata, the scutes of the shield (Car. and Plas.) are imbricate and are said to be added to anteriorly as they wear away posteriorly. In the carapace of Malaclemmys the scutes are not imbricate, but they grow more in the anterior and lateral directions than in the posterior direction. In the plastron, however, there are often traces of overlapping posteriorly, especially in young shells. The posterior part of the rings of growth, usually very narrow, may be particularly so in such specimens, or even entirely wanting, the mesial segments of the

Diversity in the Scutes of Clielonia.


rings ending abruptly at the posterior margin which overlaps the next scute. The overlapping was particularly striking in ISTo. 50,

ISTo. 145 was characterized by marked asymmetry of abdominal and femoral scutes (Figs. 37, 38, PI. IX).

Mm^ginals. Except in two malformed shells, supernumerary marginals occur only in the regions anterior to the normal position of M 3, and posterior to that of M 10. This means that none occur except anterior or posterior to the region of the doraal ribs. The exceptions are l^os. 145 and 248 (see Figs. 37 and 40, PI. IX, and Remarks" in tables). Anterior supernumerary scutes were found in ISTos. 126 and 131 (Figs. 13 and 9). In each of these the first two marginals (R and L in Xo. 126, R in Xo. 131), taken together, are about equivalent to a normal first marginal. In Xo. 210 (Figs. 5 and 17) the first right marginal was partially divided.

Fig. L.

Fig. K. Fig. M.

Fig. K. Plastron of No. 132, showing anomalous soute on right side posteriorly. Fig. L. Plastron of No. 233, supernumerary scute on left side. Fig. M. Plastron of No. 88, left femoral and abdominal continuous mesially.

Abnormalities are more common in the posterior region, additional marginals being observed on the left side in Xo. 235, on the right in Xos. 88 (Fig. 12), 129, 151 (Fig. 7), 183 (Fig. 33) ; on both sides in Xos. 35, 36, 37, 38, 108, 162, 178 (Fig. 30), 196 (Fig. 42), 198 (Fig. 45), 209 (Fig. 32), and 228 (Fig. 22); cf. also Table V, Xos. 250 (R), 252 (R), 255 (L) and 256 (L) and fieures.

24 Kobert E. Coker.

Sometimes the marginal series shows a reduced number of scutes. Thus ISTos. 4 (Fig. 20) and 8 had only eleven pairs of marginal scutes, the eleventh marginals being very long. No. 245 had oidy eleven on the left side (Fig. 18) and the two posterior scutes of the left side, taken together, are equivalent in size to the last three on the right. Partial fusion of 11 and 12 (K) was noted in ]^o. 47 and of 2 and 3 (R) in No. 176 (Fig. 34). In Nos. 153, 174 and 185, the twelfth marginals were reduced in size, and, in the latter two specimens, were asymmetrical (Figs. 26 and 29).


The most interesting abnormalities of scutes are those shown by the nuchal and the neurals. The most anterior of the median scutes of the carapace is always given the name nuchal, but is generally included with the other median scutes in the neural series. It resembles the other neurals in being median and usually unpaired, but differs from them in some evident respects. Its direction of predominant growth is the reverse of that of the other median scutes. It is absent in some existing species and in the fossil turtles of the genus Pleurosternum Owen, belonging to the sub-order amphichelydia Lydekker, a group which von Zittel regards as ancestral to the modern Pleurodires and Cryptodires ('02). I find it occurring in paired condition or marked with a median funwv much more frequently than any other commonly unpaired scute of the carapace. In this connection the frequent evagination of this scute in Clemmys guUatus is of interest.

The nuchal was represented by a pair of scutes in ten specimens: Nos. 44, 74, 78, 97, 124, 168, 196 (Figs 31), 200, 210 (Figs. 5 and 17), and 228 (cf. also Table Y, Nos. 244, 245 and 249, and figures). A distinct median furrow, not distingiiishable as a seam, occurred in nine individuals: Nos. 3, 73, 116, 125, 182, 185, 208, and 211 ; a faint median furrow was observed in No. 80.

Costals. It will be observed that abnormal scutes occur much more frequently in the posterior costal region than in the anterior.

Diversity in the Scutes of Chelonia. 25

Typically, each costal series consists of four scutes, which dovetail between the five neurals. The seam between CI and C2 meets the fifth marginal near its anterior end, and the succeeding costal seams meet alternate marginals (seventh, ninth and eleventh). These typical relations of the costals to the shields of the other series, may aid in the diagnosis of supernumerary schutes (v. Fig. A).

In Fig. 9 the first large scute in the L. costal series shows essentially the relations of the normal' first costal, so that the small scute anterior to it may fairly be considered the supernumerary element. Sush anterior supernumerary costals may occur on both sides, approximately symmetrically (IsTos. 244, 245, and 246, PI. I, Fig. 2, and PL VII, Figs. 18 and 19) ; or on both sides but quite asjanmetrically (No. 218, PI. IX, Fig. 41), or on only one side (No. 9, right side, PI. II, Fig. 3, and Nos. 4 and 131, left side, PI. VII, Fig. 20, and PI. V, Fig. 9). In each case referred to above, the extra scute was small in comparison with the other costals. In No. 218 supernumerary scutes were observed both anteriorly and posteriorly (Figs. 41 and 44).

Perhaps some of the cases to be discussed below belong in one of the above classes. In Nos. 210 (Figs. 5 and 17) and 151 (Figs. 7 and 15) the supernumerary scute in the costal series may be the large CI ; but it would be difficult if not impossible to determine in these instances whether the first or second scute is to be regarded as supernumerary. We must always recognize the possibility that in some cases we are not presented with the four normal scutes as individuals plus an extra, "supernumerary," individual, but simply with a five-scute series instead of a typical four-scute series. Furthermore it may be quite immaterial whether an anomalous scute is termed a neural or a costal ; the designation may be without real significance. Of course, however, if supernumerary scutes are attributed a morphological value, their proper classification is of the first importance (cf. below, p. 42).

The difficulty of identification of scutes is enhanced in the posterior region, where it is often difficult to decide whether a given anomalous scute is to be classed as a neural or a costal [cf. Nos. 210 (Figs. 5 and 17) and 252, Fig. 49]. It was found necessary to adopt certain rules of classification which may be' in 'some measure

26 Eobert E. Coker.

arbitrary. They are based on the fact that the seam between C4 and ]Sr5 typically meets Mil at a point near the middle or anterior part of its inner border (Fig. A, etc.).

1. (a) 111 a specimen with the usual number of twelve marginals, any scute in order with the costals which does not extend positively beyond Mil is considered a costal. This rule certainly becomes arbitrary when applied to small scutes which extend neither anteriorly nor posteriorly to Mil, and which may, therefore, be as much within the neural region as within the costal region (PI. VII, Fig. 21). It is convenient, however, to class such elements with the costals.

Compare :

much within llic neural region as within the costal region (PI. IV,

ig. 21, small scute, right side

ig. 10, small scute, left side

Ig. 16, large scute, left side*

ig. 46, large scute, right sidef

Ig. 44, medium scutes, both sides

ig. 50, small and medium scutes, left side

ig. 14, medium scute, left side

ig. 39, small scute, right side

ig. 23, medium scute, right side

ig. 27, LC4 represented by two scutes

(b) Sometimes extra marginals occur in association with extra costals; in such cases, the costal region may be supposed to extend to the twelfth marginal, which is the next to last, as is the eleventh in the typical series. Therefore, if there are thirteen marginals, a scute in order with the costals, which does not extend posteriorly to M12, is classed as a costal.

Compare :

No. 151, Pl{vn}^^S-{l5'} ^^^'^^ ^^"*^' ^"^^^* ^^^^*

No. 196, PL IX, Fig. 42, ' medium scute, left side

No. 198, PI. X, Fig. 45, small scute, each side

No. 255, PI. IX, Fig. .39, small scute, left side

No. 256, PI. VII, Fig." 23, small scute, left side

2. If, in a specimen with twelve marginals, a scute in order with the costals extends barely over on M12, but in its anterior

In the left costal series of this specimen, the supernumerary costal may, of course, be one of the more anterior scutes.

fNo. 217 is incorrectly included in this list. It may properly pertain to the next list (b), but is a very doubtful case.














































No. 257, PI. VIII, F

i i

Diversity in the Scutes of Chelonia. 27

part is in contact with MIO, it may fairly be regarded as a terminal costal that encroaches slightly on the neural region. Compare :

No. 195, PI. Ill, Fig. 8, medium scute, left side

No. 199, PI. X, Fig. 43, medium scute, left side

No. 21u, PI. jyJj'JFig. j jM medium scute, left side*

aud, possibly, No. 181, PI. X, Fig. 47, medium scute, left side

No. 149, PI. IV, Fig. 11, large scute, left side

3. A scute which overlaps the last marginal and is not in contact with any marginal anterior to the next most posterior (Mil, in typical marginal series), is within the neural region and must be classed as a neural.

Compare :

No. 222, PI. X, Fig. 48. No. 2.52, PI. XI, Fig. 49. No. 256, PI. VII, Fig. 23, small scute ou right side.

It is doubtful whether the extra scutes on the left in ISTos. 149 (PI. IV, Fig. 11) and 181 (PI. X, Fig. 47) should be included in this or the preceding class. The small scute on the right in ISTo. 247 (PL VIII, Fig. 24) would seem to belong in the first class, but the shape of the areolse of this and of 'N5 and its manner of growth as indicated by the rings suggest clearly that it is but the separated lateral end of ]Sr5.

It may be that this scheme of classification, as applied to such specimens as ISTo. 198 (PI. X, Fig. 45) and 256 (PI. VII, Fig. 23) is quite arbitrary. This leads to the general question whether the reference of many of the abnormal scutes to nonnal series has any other merit than that of convenience of description. This question will be referred to in a later section, but it is well to make clear the difficulty of identification because, if abnormal scutes are to be regarded as of morphological value, rational and exact classification is essential.

The difficulty of diagnosing the costal series of cases like Xo. 151 (Figs. 7 and 15, right side), 167 (PL VII, Fig. 16, left side) and 210 (Figs. 5 and 17, both sides) has been referred to. In Xo. 210,

Cf. footnote (*) preceding page.

28 Kobert E. Coker.

for example, the seam between the second and third shields of the right costal series, meets M5, and this is the usual position of the seam between normal Cl and C2. Considering this seam as corresponding to the normal seam between Cl and C2, we note that the following costals (C3 to C5 in this shell) have exactly the relations to the marginals normal for costals 2, 3 and 4. We have only to assume that nonnal Cl is represented by two scutes in this carapace, or that one of the first two scutes is supernumerary ; the right costal series is thus made perfectly intelligible hj considering its relations to the marginal series, and disregarding its relations to the neural series. Turning to the left costal series, the four large scutes present a perfectly normal appearance, and have nearly normal relations to the lateral apices of the neural series ; yet not one of the four has normal relations to the marginals, though the marginal series of each side is nearly normal. This series would be intelligible if we disregarded the marginal relations. Either series is interpretable, provided we do not apply the same criteria to both sides. It is clear that the exact identification of supernumerary scutes must often be impossible or quite arbitrary. The significance of such cases will become clearer after the following section.


Excluding the luichal, for present pui-poses, the neural series consists typically of a row of five median scutes, which dovetail laterally with the costals. The first four usually bear marked prominences, which together form a low^ serrated keel. The fifth, sometimes, especially in younger specimens, shows a very low and inconspicuous continuation of the keel (cf. PI. YI, Figs. 12 and 13). This keel is suggestive of the serrated dorsal keel of the tail of Chelydra, and Hay and ISTewmann regard the neural scutes as serially homologous with the scutes of the dorsal keel of the tail of Chelydra. In view of the accredited phylogenetic significance of the keel prominences they may aid in the interpretation of supernumerary scutes in the neural series. Such scutes are almost always placed not symmetrically, in longitudinal sequence, or alongside of each other, but asymmetrically and wedged in together. The photographs

Diversity iii the Scutes of Cheloiiia. 29

are most iustructive. In Fig. S, PL III, there are at least seven elements in the neural series, of which the most posterior five fit in together as if crowded out of a condition of longitudinal sequence, each scute extending across the median line. It will be noted that while the rings of gro\vth indicate that each scute has gTown most in the antero-external direction (toward the antero-lateral costal), jet each scute shows comparatively broad rings toward its anteromesial neighbor of the same series, although growth in this direction must cause further distortion of the keel by pushing the prominences further and further away from the median line. Nevertheless the prominences still make a continuous but crooked carapace keel. No. 151 (Fig. 15) presents a different condition. Numbering the scutes from Nl posteriorly, (cf. PI. IV, Fig. 7, of the same carapace) N3 and N5 are almost parallel in position to Nl, and the keel of N5 extends anteriorly half-way by the keel of N4, to which it is exactly parallel. Clearly, in this case the keel of the carapace branches and is double for a part of its course.

No. 150 (PI. IV, Fig. 0) may offer a clear case of longitudinal sequence, though the supernumerar}^ scute is largely on the left side. No. 139 had a very similar anomalous element, but entirely to the left of the median line and without the keel prominence.

Nos. 167 (PI. VII, Fig. IG) and 251 (PL VII, Fig. 14) are comparable to the first illustration given above.

In the cases cited, which are representative of a number of others, the keel prominences throw no definite light on the question at hand. Is it to be assumed that such scutes are primarily in longitudinal sequence but are crowded out of position? Or that they are really paired scutes, asymmetrically placed ? Or is their significance something still different ? Newmann has decided in favor of the first explanation, but suggests that Gadow would probably consider such types as evidence of the original paired character of the neural row.

It is possible hypotheficaUy to regard many of the asymmetrical neural anomalies as illustrating only a secondary asymmetry, a modified longitudinal sequence; but some cases can hardly be referred to such a condition, especially when a portion of the keel parallels another portion, as in No. 151 described above. Now, is

30 Eobert E. Coker.

the keel necessarily a single structure which we must not expect to find divided 'i So far as I know, there is no phylogenetic evidence of the neural series of turtles ever having been paired, or, of a double dorsal keel in primitive turtles (cf. Newmann, '56, p. 92). If we must regard these abnormalities as atavisms, we can hardly conceive of the atavism taking so often the form of paired supernumerary neurals; we would be led to assume, with jSTewman, that we were presented with a modified longitudinal sequence. JSTow, disregarding atavism for the moment, we will consider the anomalies as we find them.

No case of unmistakably paired neurals has been observed in the terrapins of North Carolina or Maryland, if we exclude the nuchal, which is, however, usually included in the neural series; but there are interesting abnormalities that are significant in this connection. The nuchal, as has been seen, is sometimes paired, sometimes marked by a median furrow in the position of the seam, and the furrow may be so marked as to make it difficult to distinguish from a seam, or there may be a seam on the anterior half of the scute, continued posteriorly by a furrow (Thalassochelys, PL XIII, Fig. 89, and PI. XIV, Fig. 94) . Apparently the difference between the furrow and the seam is that the furrow divides incompletely, while the seam divides the scute completely as far as the seam extends. Now in four specimens of Malademmys (Nos. 73, 78, and 211 ; also No. 244 of Table V), some of the neurals were marked by a median furrow similar to the furrows observed on nuchals, and in each of these cases the nuchal was either marked with a furrow, or paired. Especially suggestive is No. 70 (PI. I, Fig. 1) where N2 shows a short furrow on the anterior rings,"^ and N4 has all of its rings intersected anteriorly by a seam. Compare also PI. I, Fig. 2, of a terrapin with nuchal paired and with furrows on Nl and N2.

In logical sequence with these specimens come some terrapins (M. littoralis) from Texas. In a small number of specimens placed by Mr. W. P. Hay in the United States National Museum, we observe all stages from very incomplete to complete division of the

'Owing to the manner of reflection of ligtit the furrow on N2 is not apparent in the photograph, though very evident in the shell.

Diversity in the Scutes of Chelonia.


neural series (Figs. 53 and 54).^ (See also W. P. Hay, '05,

^These specimeus are of peculiar interest on account of tlie precise median longitudinal division of neural scutes (partial or complete) — a rare abnormality, though common, apparently, in the specimens from this particular region.

Nine Specimens of Malaclemmys littoralis Hay.


Length of


in inches.

Median longitudinal division of scutes, as follows:



N2 with seam complete except in anterior portion of areola (oldest portion of scute), and in first (most posterior) part of anterior ring of growth.

N3 similar to N2, except that seam, anteriorly, arises a little later.



Nuchal paired.

Nl partially divided posteriorly. N2 completely divided.

N3 partially divided anteriorly and posteriorly. Fig. 53.



N2 and N3 divided except for areolae. Fig. 54.



Nl with furrow anteriorly.



N3 partially divided anteriorly and posteriorly.



Nuchal paired.

N2 divided except areola.

N3 partially divided anteriorly and posteriorly.



N2 partially divided anteriorly and posteriorly. N3 divided except areola.



N3 slightly divided anteriorly — that is, just beginning to show division.


• •

Only third and fourth neurals present; N3 marked anteriorly by faint furrow.

It appears that only a few of these scutes (possibly only N2 of 259, Fig. 53) were split at birth; some others (as N2 and N3 of 258) were divided only in the posterior portion of the scute (of course, the areola of the older stage represents the entire scute of the newborn turtle) ; while in others the division, anteriorly and posteriorly, appeared at various subsequent stages. In

32 . Robert E. Coker.

regarding the abundance of such forms.) A young specimen of Chelydra in the zoological museum of the Johns Hopkins University has its keel marked with a distinct sagittal furrow which extends from the anterior margin of JSTI to the posterior margin of 'N5, and is continuous through all of the keel spines except that of N5 ; such a furrow is seen less distinctly in the shells of older turtles.

Ill the light of these median furrows and incomplete and complete seams, and of the parallel keel prominences of Xo. 151, there would seem no inherent improbability of duplication of the median keel or of neural scutes, nor any necessity for explaining away appearances of duplication. To jump to the other conclusion, that the asymmetrical neurals are primarily paired scutes which are crowded out of a symmetrical plan, would be equally unwarranted. Crowding" may be, at best, but a descriptive term referring only to the appearance presented, rather than to any organic phenomenon.'^

May it not be that we have to do here neither with scutes in sequence nor with paired scutes, but with a real asymmetry of scute plan ? Symmetry is a normal feature of the carapace, but the cases in question are admissably abnormal, and, on the face of it, the scutes are asymmetrically disposed. The question is thus : Is the visible asymmetry secondary and due to the crowding out of line of elements that are primarily symmetrical, or is there in such cases a real primary, though abnormal, asymmetry, which is perhaps correlated with some other asymmetry ? The almost invariable association of asymmetry in number or arrangement of the costal scutes is strongly suggestive of the latter conclusion. Granting some primary abnormality arising either as a germinal variation or in consequence of environmental conditions, it may be that the conditions of gro'wth cause the development of the neural scutes neither in linear sequence nor in pairs, but in essentially such an asymmetrical plan as is presented by the cases in question.

these nine terrapin fifteen neural scutes, besides nuchals, sliow some degree of division. Hay states that the longitvidinal division of scutes was "so common that it was really difficult to pick out a full-grown specimen which did not show it in some degree."

"Extra marginals may be small or large but do not seem to be "crowded out," and the same may be said of extra costals.

Diversity in the Scutes of Cheloiiia. 33

The asyimnetrical neurals lead us to the subject of the interadjustment of neurals and costals, a subject of sutficient significance to justify its treatment in a separate section.

3. Adjustment of Neurals and Costals.

In the typical carapace, neural scutes and costal scutes have an alternating relation. The neural series dovetails on each side into the costal series and a neural seam extends transversely from the apex of each costal scute. Extra scutes may occur in different parts of the costal series and often only on one side. In cases where the abnormality is such as seriously to disturb the symmetry of the costal series, the adjustment of the neurals to the costals might be accomplished in one of several ways.

1. The sequence of neurals might remain normal, presenting some such appearance as is represented in Fig. IST, I.^ Such an arrangement of neurals without regard to the costals of one side has not been observed.

2. I^eural scutes might have normal relations to costals on each side (cf. Fig. IST, II). There would have to be one supernumerary neural in correspondence with the supernumerary costal, and some of the median scutesi would occupy oblique positions. In such a carapace, the antero-posterior extent of a neural scute measured from the plane of its most anterior point to that of its most posterior point would be unusually great. The antero-posterior extent of ISTS in the figure is indicated by the line x y. Disregarding temporarily some rare and partial exceptions, this plan of adjustment is not observed.

Asymmetrical costals ofteji occur in such form as to give opportunity for plans something like one of the above, and these plans seem to offer the simplest schemes of adjustment for neurals. Since with the partial exceptions that will be mentioned later, neither of those plans is observed, we may assume that these hypothetical schemes do not accord with the laws of growth.

3. Assuming that a better coordinated carapace results when,

'In Fig. N the costal series is traced from an actual specimen (No. 151) and the same series is used in all three plans.

3-i Eobert E. Coker.

(a) on each side there is practically the usual relation between costals and neurals, and (bj neural scutes have not an antero-posterior extent that is relatively unusually great, we may imagine such a plan of neurals as is represented in Fig. jST, III. On each side neural and costal scutes alternate in position, and the two series dovetail together in usual fashion, while cross seams so divide the neural area that no individual scute has an excessive anteroposterior extent. This plan is not a hypothetical one, as were the others, but is an inference from the observations. Compare with it the plan of scutes illustrated by Xos. 151 (Figs. 7 and 15), and 195 (Fig. 8).

The first two types of adjustment assumed above (hypothetical) may thus be defined:

1. Neurals with normal adjustment to costals on one side, but not on the other. Individual scutes show this plan very rarely.

2. jSTeurals with normal adjustment to costals on both sides, but with the two sides of the neural series related in a way that is rarely observed.

The third type (observed) may be defined thus:

3. Neurals with normal adjustment to costals on each side ; the two sides of the neural series unsvmmetrical and so related to each other that the plan of each side is largely restricted to that side.

In this manner of gTowth, then, we have a neural plan (or lialfplan) on one side, a different neural plan on the opposite side, and each plan is adapted to the costal plan of the same side (cf. the diagram, Fig. O, III a. The mesial region of the carapace is supposed to he covered hy a strip of paper). The plan of each side does not usually terminate abruptly in the median line as represented in the diagram, III, b (cf. however, ISFewmann's Fig. 6),* but the scutes necessary to the plan of one side extend a little over the median line (III, c), sometimes almost or quite across to the opposite costals (III, d). In consequence, we get, not a plan on one side independent of the plan on the other side, but on each side a plan that is mora or less modified hy the over-extension of the scutes necessanj to iJw, plan of fJi.r opposite side.

Fig. P, XTI.

Diversity in the Scutes of Clielonia.




Fig. N. See text.



Fig. O. See text.

36 Robert E. Coker.

In the endeavor to make clear the inferred law of growth, I have, in a measure, proceeded in anticipation of the evidence from which the inference was made. The body of the evidence may now be best presented, not by detailed discussion, but by reference to textfigure P, in which are indicated the neurals and costals of a number of specimens. I have drawn freely on Newmann's figures, as his observations include more specimens bearing on this particular point than do my observations on Malaclemmys. My observations on another species need not be introduced at this point.

Each carapace figured may be compared, on the one hand, with the types Fig. ]^, I and II, and on the other hand, with that of Fig. K, III, or Fig. O, III c, and III, d.

It is readily observed that there is no correspondence between the number of extra costals and number of extra neurals, but that the region of "abnormality" in the neural series corresponds in anteroposterior extent with the region of asymmetry in the costal region. If asymmetry of costals is confined to a small region, the adjustment of neurals may require only one or two supernumerary elements (cf. Fig. P — V, VI, VII, etc.) ; but when asymmetry marks a larger part of the costal region the adjustmient of the neural ins^olves two, three, or more elements in excess of the normal number (cf. Fig. P— I, II, III, IV).

A few cases which may be classed as partial or complete exceptions require discussion. These are included in Fig. Q, to which refer the Roman numerals in the following paragraphs:

The carapace figured in VI illustrates in part the adjustment supposed. The adjustment would be complete only if a seam united the apices of RC4 and LC5. On any hypothesis this is a remarkably abnormal shield. (From l^ewmann's Fig. 9.)

In VII there are 5 costals on each side, but the number of neurals is normal. However, the mesial region of the two costal series are

'The nuchal and the marginals are omitted in most cases, since they have no bearing on the point in question, and would only complicate the figures. In the subscription full references are supplied, so that the original figures may be consulted if it is desired.

Diversity in the Scutes of Chelonia.








Fig. p. See text.





38 Robert E. Coker.

almost exactly symmetrical, and the second pair of costals are so small at their mesial ends as to be of little significance with reference to the neurals. It may be noted that the second neural is not longer antero-posteriorly than any other scute of the series. (From l^ewmann's Fig. 25.)

The four next following have an especial interest.

In I the adjustment appears in the posterior region, but more anteriorly are found two oblique neural elements. The most abnormal, ISTo, shows an incomplete seam, which, were it complete to the point X, would be a decided step toward perfecting the adjustment in the manner assumed. (Specimen ]^o. 167, above.)

II presents a very complicated appearance, but the adjustment is more complete than at first appears. Thus, the two interrupted seams, y and z, are completed by furrows through the areolae. The adjustment would conform perfectly to the usual plan, had these seams been complete, and had there been a seam in the position of the broken line x. The broken lines represent lines of depression of uncertain significance that radiate from the areolae across the rings of growth (cf., the photograph of this carapace, PI. II, Fig. 5). (Specimen ISTo. 210, above.)

III represents a carapace that is very abnormal on any hypothesis. Even here, though, there is found an incomplete seam which makes a step in the direction of the assumed plan of adjustment. (From ISTewTnann's Fig. 7).

In the carapace represented by IV, the law of gTowth" in question would be fully expressed if the incomplete seam were complete to the point x. (From ][Sre^\anann's Fig. 14.)

Recalling that the growth of the scutes is accomplished by the addition of peripheral rings, and observing that the incomplete seams are in each case in the peripheral or newer part of the scute, one may be justified in making the tentative inference that they indicate post-natal attempts to perfect a previously inadequate adjustment of neurals and costals. Certainly they alter the plan of adjustment, even if it be a coincidence that the alteration in these few cases is in the direction noted. This occurrence of apparently post-natal divisions in cases of imperfect correlation has been observed in

Diversitv in the Scutes of Cheloiiia.






VI] Mil

Fir. Q. Pee text

40 Eobert E. Coker.

the relations of marginal scutes and plates and will be seen again in embryos and young of Thalassochelys.

V violates the principle of adjustment in the anterior region. The second neural, exceptional in being in contact with three pairs of costals, is incompletely divided, but the significance of the incomplete seam is not evident. (Newmann's Fig. 4.)

VIII presents a very unexpected neural series ; but this specimen was an embryo and we do not know how its co-ordination would have stood the test of life, nor whether post-natal seams would have appeared to alter the plan of adjustment. Several of the specimens mentioned above evidently would have shown less conformity if they had been obser\^ed in the embryo stage. (JSTewmann's Fig, 38.)

Another of NeMauann's specimens is of particular interest (IX). The carapace is perfectly symmetrical, but there are five pairs of costals and no additional neurals. ISTS is in contact with three pairs of costals (C2, C3 and C4). Here is, therefore, a case of mal-adjustment. However, l^ewmann found that the fourth costals were growing forward underneath the third, and he interprets this as a stage in the suppression of the third ("sixth," in his terminology). If this interpretation is correct, we have the mal-adjustment in process of correction, not, as in other cases, by the partial division of the neural, but by the renioval of the supernumerary costal. I do not say, however, that the imperfection of adjustment is the cause of the squeezing off.

Hence the exceptions are chiefly partial exceptions, and, on the whole, even these specimens favor much more than they o]>pose the assumption that the proper adjushneni of neurals and costals is of more vital importance than mere number of scutes. Of course, it is to be expected that turtles will show abnormalities in adjustment as well as in number of scutes; but observation of the relative frequency and degree of deviation from the usual numerical relations, on the one hand, and from the usual adjustment, on the other hand, may be a means of estimating the relative value of adjustment as compared with numerical relations. Supernumerary scutes occur comparatively frequently, while mal-adjustment is rare, and may it not be of some significance tliat in turtles with imperfect

Diversity in the Sci;tes of Chelonia. 41

adjustment, there is evidence of post-natal partial division of scutes by seams that lessen the abnormality in adjustment while they increase the abnormality in niHnbev of scutes ?

There are two classes of costal abnormality that might seem to be exceptional and which, therefore, should be referred to here, (a) When small supernumerary elements occur at the extreme anterior or posterior ends of the costal series (Figs. 10, 21, 23, etc.), without interfering with the symmetrical plan of the normal scutes we would expect no change in the neural series — the adjustment of the series is already practically perfect, (b) Another illustration of asymmetry in number only is offered by the division of the 5th costal (Fig. 27). Two scutes may occupy just the position of this shield, without effect on the general plan of the series. On the other hand, of course, asymmetry of costal s may occur w^ithout supernumerary costal scutes.

To sum up —

In the normal symmetrical carapace neurals and costals have an alternating relation, and the neurals dovetail between the costals. In the imsymmetrical carapace, this relation prevails on each side. It follows that one side of the neural series may be formed according to a different plan than that of the other side. The two opposing plans are independent of each other, but never entirely so. The scutes necessary to one plan extend more or less over into the other side, sometimes even to the opposite costal series, but always in reduced form. The two sides of the body show, at the same time, a degree of mutual independence, with a degree of mutual dependence.

In consequence of these conditions of correlation it is rare to find neural scutes of abnonnally great antero-posterior extent, and there is noted a correspondence not between numbera of neurals and of costals, but between the respective regions of abnonnality of the two series.

The value of these observations would be lessened if the suggestion of the general applicability of the principle inferred would conflict with previous observations, or with hypotheses which for other reasons we must accept. Previous observations on land and

42 Eobert E. Coker.

marsh turtles have been used above) ; those on sea-turtles will be used in a later portion of this paper.

The only hypothesis which has a bearing is that which regards these abnormal scutes as atavisms. I believe that the general truth of these observations would conflict with that explanation, unless it could be supposed that the reversion was to a stage when neurals displayed the asymmetrical condition observed. In any event, I believe that the upholders of the hypothesis as applied to the scutes in question should account for the following facts :

1. That, generally speaking, more supernumerary neurals are observed in specimens with supernumerary costals on one side than in those with symmetrical supernumerary costals.

2. That, while supernumerary scutes may occur in any one series without additional scutes in any other series, thus indicating the partial independence of each series, yet, when a Siupemumerary costal occurs in such a position that the symvnetry of the costal plans is seriously interfered with, the number of supernumerary neurals will depend on the extent of the region of asymmetry.

3. That we do not find asymmetry of costal plan without supernumerary neurals. (See qualification above, p. 41).

There is nothing in these observations to conflict with a supposition that the primary variation, the resulting adjustment of which we see, may be an atavism.

Referring again to the question of the identification of supernumerary scutes — as costals or neurals, etc. — if the adjustment of the whole carapace in accord with the laws of growth is the thing, it is of little real significance whether a given abnormal scute be termed "costal" or "neural," however convenient such a classification may be for the practical purposes of description.

We have vet to consider the neural and costal scutes in relation to the bony plates beneath, but it is desirable to defer this until after the study of the scutes of Tlialassoclielys, which must conclude the present paper.

Diversity in the Scutes of Chelonia. A'.)

4. Age^ Sex^ Symmetry.


It was shown in a previous paper ('05 a) that my observations do not indicate any significant difference in the proportions of abnormality at different ages. The cases of incomplete division and incomplete fusion noted all seem to tend toward increasing the abnormality in 7iumher of scutes. Newmann ('06), after examination of nearly 500 specimens of Graptemys, including a number of embryos, finds that "abnormalities are no more common in one size than in another." lie notes, however, some cases interpreted as stages in the squeezing off of scutes (cf., my specimens No. 50, p. 22, 23).


The proportion of abnormality in the females of tables I to IV

is noticeably greater than that in the males. Considering all the

abnormalities of carapace and plastron, as defined on p. 12, above,

24 out of Yl males are abnormal, or. 34 per cent, and 71 of 135

females, or 52.6 per cent. Of 37 the -sex is not known. Of the

entire 243, 109, or 45 per cent are abnormal. If, however, w^e

consider only the possession of more or less than the typical number

of scutes in the carapace, the proportion of abnormality is about

20 per cent.


51 specimens show only symmetrical abnormalities.

44 specimens show only non-symmetrical abnormalities.

14 specimens show both kinds of abnormalities.

Thus, of the 109 abnormal turtles of tables I to IV, 51 are symmetrical, 58 unsymmetrical, in the respect that we have taken into consideration.

Before taking up the embryos and young of ThalassocJielys it will be well to give a summary of the observations on Malaclemmys.

Summary. 1. Observation of diamond-back terrapin in nature and in confinement reveals a marked degree of diversity in habit, disposi

44 Eobert E. Coker.

tioii, and structure. Terrapins display mucli individuality in these respects.

2. The diversity in number and arrangement of the scutes accords with the general diversity. Of the first 243 specimens carefully observ^ed, 20 per cent had either more or less than the typical number of scutes in the carapace. In all, 45 per cent showed such differences from the typical number, arrangement, and character of the scutes of carapace, plastron and inframarginal series, as to be classed for present purposes as "abnormal."

3. The "abnormalities" consist in : the possession of a greater or less nimiber of scutes than 38 in the carapace, 2 in the inframarginal series, and 12 in the plastron; of instances of incomplete fusion or incomplete division of normal or abnormal scutesi; of asymmetry in size and plan of scutes ; of imbricateness of scutes ; of the loose attachment of scutes to the bone beneath (one case).

4. Asym/metry manifests itself in various ways, and is at least as common in "abnormal" terrapins as symmetry.

5. Females show a much greater degree of diversity than 7nales. This is true as regards both percentages of abnormal individuals and degree of abnormality in the average individual.

6. 'No evidence was noted of difference in the proportions of abnonnality in young and old terrapins, but the data are inadequate for definite conclusion.

1. The most variable scute is the inguinal. Typically absent, it was present, on one or both sides, in 21 per cent of the 243 specimens. Its approximately symmetrical occurrence was noted in 13 per cent of the 243. Karely, one or both axillaries were wanting. Sometimes they were reduced in size or asymmetrical.

8. The most variable scute of the carapace is the nuchal. Always present, it was sometimes paired, sometimes marked by a median groove.

9. Omitting the inframarginal series, the plastron is far less variable than the carapace. Diversity manifested itself by the presence of extra scutes in the plastral series and of inter-plastrals at the meeting-point of gulars and brachials, by partial fusion of scutes, and by asymmetry. Each abnormality is of rare occurrence.

Diversity in the Scutes of Chelonia. 45

10. Except in misshapen specimens, supernumerary marginals seem to occur only anterior or posterior to the region of the dorsal ribs. They are more common in the posterior region. A reduced number of marginals results from the lack of one or a pair of the posterior scutes. The alternating relation of scutes and plates is well preserved in several specimens.

11. Supernumerary costals may occur at the extreme anterior or posterior ends of the neural series, without disturbing the general plan of the series, or they may occur in such a way that the plan of the series is altered. The two series of costals may be asymmetrical even when the number of costal scutes is normal.

12. Neural scutes may be partially or completely divided by a median seam. This abnormality, though rare in the terrapins of ISTorth Carolina and Maryland, is very noticeable in those of Texas (Hay).

13. Asymmetrical scutes in the neural series are of not uncomnion occurrence. There is observed an adaptation between asymmetrical neurals and asymmetrical costals that strongly suggests that the explanation of these scutes is to be sought in the adjustment of scutes consequent on some more primary asymmetry. I do not regard them as belonging in linear sequence and crowded out of position.

14. I believe that the asymmetrical neural scutes cannot be regarded, individually, as atavisms. Their adaptation to an unsymmetrical carapace seems too clear to permit of explaining them by reversion to any earlier symmetrical plan of scutes.

15. In the asymmetrical scutes we may find an illustration of the fact that we may best interpret a single variation not by regarding it as an isolated unit, but by viewing it in its relations to the associated structures with which it helps to form a more or less well co-ordinated whole.

4G Robert E. Coker.



1. Inteoduction,


Only embryos and young were observed and these were obtained from eggs laid on the ocean beach near Beaufort. In studying the abnormalities it is important to know the conditions under which the embryos developed. The laying ground is inconveniently distant from the laboratory and the nests could be visited only by a sail and a walk on the beach that consumed the greater part of a day. The main object at the time was the collection of embryological material, and, as it was important to have the eggs conveniently accessible, it was necessary to remove them to artificial nests on the island on which the laboratory is situated. Another condition making it advisable to transplant the eggs was the difficulty of protecting the natural nests from depredation. Turtle eggs have a local value as food and are eagerly sought by fishermen.. It was observed too that hogs root up the nests and destroy the eggs.

One nest was left undisturbed. A wire screen, placed over it and well under the sand, served to prevent the escape of the young turtles and to protect the nest from hogs. Traces of the nest were obscured as far as possible to prevent molestation by fishermen. The eggs hatched successfully (see Table VIII).

Most of the eggs were removed from the nests within two days after they were laid, usually on the following morning, and transr ferred in a bucket or box, in which they were covered with moist sand and seaweed. Usually care was taken to keep the eggs right side up. The eggs were then replanted either in artificial nests on the ground or in a sand-box or "incubator." The artificial nests in the ground were not successful. The soil, though chiefly sand, was of a different composition, and a higher temperature obtained than at the same depth on the beach. Other environmental conditions seemed unfavorable, and, in consequence, but a small proportion developed to a late stage. Unfortunately, the first observations indicated that nests in the ground would be more satisfactory than nests in an incul)ator,

Diversity in die Scutes of Clieloiiia, 47

ao that most of the eggs were placed in the ground. Later it was found that a proper incubator would yield better results than the soil of the island.

The incubator used was very simple, it consisted of a box with four shelves, each holding 72 eggs. The shelves were large enough for each egg to be entirely surrounded by a small amount of sand. The light cushions of sand around and above the eggs served to maintain a comparatively uniform condition of humidity, while at the same time permitting the eggs to expand in size without crowding. This growth of the eggs, by the tilliug out and distension of the shell, commonly occurs to a greater or less extent during the development of the embryo. The sand was spriid^led with water as often as necessary. It was not necessary to apply artificial heat since the chief problem was to keep the temperature as low as it would ordinarily be at the depth of the natural nests. To obtain fairly uniform conditions of humidity and temperature, this box w^as placed within another box so much larger than the first that there was the space of 6 inches between the side walls (except on one side), the bottoms and tops, respectively, of the two boxes. The space between the two boxes was packed with moist sand. On the fourth side the inner box opened by a thick door containing a six-inch thickness of sand. Outside of this was the door to the outer box. Thus the inner chamber was well protected from such rapid changes of temperature or moisture as might take place outside. At the same time it could readily be opened at any time and one or more eggs removed without disturbance of the others. If the temperature seemed rising too high, moistening the outside of the box and the top layer of sand would, through evaporation, lower the temperature within a day ; or, if necessary, the doors could be left open, when the evaporation would low'er the temperature very quickly. In this way it was not difficult to keep a tolerably uniform temperature of 26° to 28° C.

The observations to be given below wall at least suggest that the eggs of the loggerhead sea-turtle would lend themselves well to experimental study, by varying the conditions of incubation. During the past summer (1905), however, the experiments were made subservient to the obtaining of embryological material and it can

48 Eobert E. Coker.

not be said that tliis paper includes experimental observations of more than suggestive value. The observations given below were made on such of the embryos (and nev^^born) as were old enough for the scutes to be clearly distinguishable.

Before proceeding to the tables a word should be said as to the normal conditions of development.

Observations on Conditions of Development. .

The loggerhead sea-turtle makes its nest in the region of Beaufort, at or near the base of the sand-dunes that line the beach at a short but varying distance from the water. The nest is subspherical, somewhat flattened on top, and packed with eggs usually to the number of 120-150. The top eggs are 12 to 15 inches below the surface, and, as the nest is 10 inches or more in diameter, the bottom eggs are 22-24 inches below the surface. At this depth, the bottom eggs may be below the level of the high tides, or as much as 4 feet above it, according to the elevation of the ground at the foot of the dunes.^

As the shell of the egg is soft and, in its new-laid condition, not completely filled, it displays a characteristic movable dent, like a rubber ball incompletely filled with air. In the course of development the contents increase in bulk and the shell becomes filled out and spherical : it may even be tightly distended. Agassiz says, "The older the egg the more distended does the shell appear." I have not found that the distension occurs invariably or to a uniform degree. The degree of distension varies with external conditions. Distension generally takes place, and it may occur to an extreme degree. With a large number of eggs massed together deep under the sand, the swelling of the eggs must cause great crowding and coYisiderable intei-pressure. In one nest, for example, the eggs were so distended that upon a single puncture of a shell with a needle, a fine stream of fluid would squirt out a distance of several feet and the shell burst widely in the hand. Even though many of the eggs that failed to develop shrunk in size so as to be almost flattened

See also my paper, '00. pp. 01 to 05.

Diversity in the Scutes of Clieloiiia. 49

against the sand, yet the pressure among those that developed was such that they lost their spherical shape and were somewhat flattened on the sides where they pressed against other eggs. Some of this flattening was lost before the photograph was taken ('06, PI. XX., Fig. B). The possible influence of such a factor as this interpressure of eggs in the production of abnormalities of various kinds is not to be ignored. The indirect results of localized pressure on the developing embryo of other animals is well known, especially through such work as that of Spemann in the production of doubleheaded embryos and cyclopean defects." It may not be a coincidence that two ^'cyclopean" embryos developed in the nest just referred to and that almost all of the turtles were abnormal in scutes, and some in still other respects.

Explanation, of Tables.

There are four main tables (VI to IX). Table VI includes embryos from a single nest obtained in 1903. No further observations were made until 1905. Table VII includes tlie embryos and newborn from various artificial nests in the ground, and this table has several subdivisions (A-E) in or(k»r to keep separate the turtles from difterent original nests. In this way, the degree of diversity in turtles of the same brood may be noted. Table VIII is based on new-born turtles from a natural nest that was not disturbed. Finally, in Table IX are turtles which developed in the incubator, where crowding was provided against.

The tables have essentially the same form as those in the preceding part of the paper. The number of scutes in a series is indicated only where abnormal, or, if normal in number but abnormal in plan, the normal number is \^Titten in italics. The number of marginals, however, is always indicated since it cannot be said that either 12 or 13 is abnormal. Thirteen on each side, 12 on each side, or 12 on one side, 13 on the other — each of these plans is common. For brevity they are referred to as the 13-13 plan, 13-12 plan, 12-13 plan, or 12-12 plan, the number given first being in each case that of the left side. These plans may be seen, respectively, in Figs. «3, 58, 72, and 52.

50 Kobert E. Coker.

The length is the full length of the carapace measured over the dorsal curvature and expressed in millimeters. The day of the embryo is indicated, when the embryo was taken from the egg alive, but the stage of development is perhaps best inferred from the carapace length, since the length of the incubation periods varies within wide limits (73, or less, to 90 days, or more). It was longer for eggs in the incubator than for those in the ground. "B" in the "day" column, indicates that the specimen was new-born,

2. Obsekvations on Diversity.

The carapace differs from that of Malaclermnys in shape — it has a more aquatic form. It is broad in the anterior region, where the long stout swimming flippers are, and tapers to a comparatively narrow but rounded posterior end. The margin is indented over the anterior flippers, and sometimes slightly so over the posterior. As if in adaptation to the shape, with broad anterior end, the nuchal is very wide and the costal series terminates anteriorly in a small scute not represented in Malaclemmys, and others. There is often one more marginal in this region than is found in land and marsh turtles.

In Table VI are presented the observations on 34 embryos from the nest transplanted at the laboratory in 1903. Besides the fact of removal in the first instance, the conditions of development were othei-wise abnormal. Only a small proportion developed successfully and most were removed from the nest during the period of incubation. Only one (ISTo. 32) actually hatched.

Twelve of the 3-1: are abnormal in number or arrangement of scutes or in showing partial division of a scute. The abnormality frequently manifests itself in an asymmetrical plan of neural and costal scutes (Nos. 1, 23, 24, 26, 30, 31). In other cases the costal series of the two sides are unsymmetrical in number, without effect on the symmetry of the general plan of series (Kos. 6, 10, 21, 25). The proportions of these embryos with the several marginal plans referred to on p. 49, above, are interestingly uniform. Of the 29 specimens, in which the marginals could be counted with certainty, ten have the 13-13 plan, ten the 12-12 plan, and nine either 13-12 or

Diversity in the Scutes of Chelonia. 51

12-13. The most abnormal carapace is that of ^o. 26 (Fig. 63), with the large scute between neural and costal series, the small area in the mesial posterior region without a horny covering, and the minute scute that appears as if it were cut out of the left twelfth marginal. These three abnormalities are each unique among my observations.

The only striking abnormality noted, apart from scutes, was in jSTo. 22, the fore feet of which were somewhat rudimentary.

Three small embryos of this nest are not included in the tables. One of these seems to have fi^e neurals, a second six, and the third seven. The other scutes can not be distinguished with certainty.

Table VII, x\, includes 21 embryos from two artificial nests, the eggs of which came originally from a single natural nest. In the table the first ten numbers are from one artificial nest. At the time of removal it was noted that the eggs were much distended, and, from the consequent crowding, more or less misshapen. One-half of the eggs started development, but not more than one-fourth were living on the 44th day. Two of the embryos were peculiarly abnormal. JSTo. 36 shows an approach to the "cyclopean defect" (cf. 91, 92), but its scutes are normal. ISTo. 35 (Figs. 6Q and 67) besides having a harelip, was characterized by a remarkable foreshortening of the body; both heart and lungs are outside of the umbilical opening; the carapace, which measures 10 mm. in length and 17 in width, is almost regular and symmetrical, though its marked deficiency in length is accompanied by a corresponding reduction in the number of scutes in longitudinal series. As a transverse series of scutes may be said to consist of one neural, a pair of costals and two pairs of marginals, the shield is deficient by two complete transverse series, less two marginals on the right side. The inframarginal series are somewhat reduced, but the plastron has the full number of scutes, including an intergular, often wanting, and even one supernumerary scute posteriorly.

Three other specimens are abnormal in respect of scutes of the carapace, and one of these (jSTo. 43) possessed minute "supramarginals."

The last 11 numbers in the table are from a different artificial nest. Most of these eggs developed to some extent, but only half

52 Eobert E. Coker.

(11 out of 24) were living on the 49tli day. The striking feature of this small group of embryos is that more than half displayed small scutes just above the marginals and always at the meeting point of three scutes (Figs. 73, 74). For convenience these are termed in the tables "supramarginals," though it is not intended thereby to imply any homology between these anomalous scutes and the typical supramarginals of Macroclemmys. The prevailing plan of marginals is 13-13.

Table VII, B, is based on 14 embryos from a nest transplanted at the laboratory. Less than one-third of these eggs developed. Four of the 14 are abnormal, in having supernumerary marginals (three specimens), a symmetrical supernumerary neural (No. 60), or two scutes in the place of normal RC5 (No. 60). The marginal plan, 13-13, prevails. The large number of marginals (15-14) in No. 60 is noteworthy.

Table VII, C, includes 2 embryos and 7 new-bom turtles from a nest transplanted at the laboratory. Three are abnormal, two of these having the costal series more or less asymmetrical and, in correlation with the costals, asymmetrical neural scutes. In one specimen (78) two neural scutes, in another (73) three are not completely separated from one another.

Table VII, D. This is -a rather remarkable collection of embryos. Of the 22 specimens 18 are "abnormal," and these include several that are abnormal in scutes to a high degree.

Four are markedly deformed. No. 93 (Fig. 89) is asymmetrical and somewhat misshapen, and the number of marginals on the right side is unusually small. Two of these (Nos. 91 and 92) have an interesting deformity of the head. We are not concerned in this place with the details of anatomy, but the external characters of these specimens may be sketched briefly. The anterior mesial region of the face is reduced. The eyes are thus brought close together, or fused, and are enclosed by single upper and lower eyelids. The single nostril opens at the end of a short snout above tlio upper eyelid. In one specimen the snout ]ioints anteriorly. In the other it is, as it were, rolled back on the top of the head, the nostril appearing on the dorsal side of the posterior end of the flattened

Diversity in the Scutes of Chelonia. 53

backward pointing snout. In accord with the reduction of the parts just above it, the upper jaw is much shortened and iiattened from the front so that it extends across from side to side in a more direct way than usuaL The lower jaw, however, has about its usual form; so that, instead of being, as usual, enclosed mthin the upper jaw, it protrudes well beyond the upper. A somewhat similar embryo has already been alluded to (No. 36), Table VII, A). All three are further characterized by the very slight development of pigment.

The most remarkable defonnity noted is that of Xo. 100 (Figs. 93 and '.).")). The lower jaw forms a pointed horizontal projection from the ventral part of the ht ad, and on the dorsal aspect of this projection is a longitudinal slit representing the mouth. iVbout halfway up the anterior aspect of the head is another pointed projection (snout {). There is no external evidence of eyes. The body is characterized by a reduction of its dorsal part. The carapace consists of two scutes with a smaller ovoid scute anterior to these (Fig. 95). On each side a bridge of one scute connects the carapace with the plastron, which possesses the full number of scutes. Posteriorly the plastron (Fig. 93) bends sharply dorsally in consequence of the reduction of the dorsal region. The dotted line in the figTire indicates the region of the angle. Between the upper posterior end of the plastron and the posterior end of the carapace is the tail bearing the anus on its dorsal aspect. The heart, lungs, stomach and intestine lie external to the umbilical opening. The limbs are situated dorsally.

Among the smaller embryos wath scutes undeveloped and, therefore, not included in the table was a fifth deformed embryo from the same nest. It needs only an allusion here. The body is reduced, and rounded, and the limbs appear rudimentary.

In view of the excessive proportion and degree of abnormality observed in this lot of embryos, the conditions of development should be described. The original lot of eggs was brought to the laboratory by a fisherman. According to his statement they were a mixed lot from two nests taken the preceding day. Some of them had dried somewhat and the lot was regarded as unpromising. However, they were placed in a single nest on the island made as usual in imita

54 Eobert E. Coker.

tion of a natural nest. The spot chosen proved to be rather laore moist than usual. During development the eggs swelled to an unusual degree. As a result they were very much crowded and were misshapen (cf. remarks above:, p. 48-49).

Many of these abnormalities are such as we would naturally attribute to the abnormal conditions of development, and the question naturally suggests itself: can the conditions of development be responsible for the remarkably large proportion of abnormalities of scutes in this nest?

The abnormalities take these forms chiefly :

1. Costals symmetrical in number, asymmetrical in plan, neurals asymmetrical (88, 97), or symmetrical (92).

2. Costals asymmetrical in number and plan, neurals asymmetrical (82, 90, 94).

3. Costals asynmietrical in number (through absence of normal small first costal), symmetrical in plan neurals symmetrical (80, 81, 87, 98, 99). Also, LC5 represented by two scutes (91).

'No. 82 is further abnormal in that the first costal forms part of the margin (Fig. 84).

The nuchal is frequently paired or furrowed. The marginals are usually 12-12.

One specimen possesses supramarginals (ISTo. 88, Fig. 85).

Table VII, E. The eggs were brought to me by a fisherman, who stated that they were a mixed lot from two nests taken two days before. They were replanted in the gTOund. The fourteen embryos and new-born make a comparatively normal lot. Though more than half (8) are abnormal, the variations are comparatively slight, consisting chiefly of paired nuchals, supernumerary costal posteriorly, supernumerary neural posteriorly, or incomplete division of ]Sr4 or 5, and slight asymmetry in the anterior region (104, 107). The marginal plan 12-12 is most frequent.

In Table VIII are presented the observations on 76 turtles from an undisturbed natural nest (see p. 46, above). The young turtles were found under the wire some days after hatching (88th day from the beginning of incubation). Only one unhatched egg remained in the nest.

Diversity in the Scutes of Chelonia. 55

As there were very few abnormal turtles, and since all are of approximately the same age, the table may conveniently be abbreviated. First the abnormal turtles are listed, and then the normal turtles in several- classes according to the plan of the marginals.

The abnormalities are: C5 represented by 2 scutes (4 specimens), nuchal paired (119), JSTS incompletely divided transversely (120), and, supernumerary marginal (121).

All of the common marginal plans occur, but 13-13 largely predominates (40 specimens).

As the nest was never disturbed during the period of incubation (except at the start) it cannot be stated whether or not there was much swelling of eggs with consequent pressure between them. This undisturbed natural nest yielded a far greater proportion of normal turtles than any nest transplanted into the ground at the laboratory.

Table IX. On account of the possibility of the pressure in natural and artificial nests having some effect on the method of growth of the scutes it was desired to remove some eggs from such conditions. Therefore, a number of eggs from two or three nests were placed on the shelves of the incubating box described above. The eggs were surrounded and covered by a very small amount of moist sand, but not sufficient to offer any considerable resistance to the expansion of the egg in any direction. On account of the poor success of the eggs transplanted into the ground, the incubator was frequently drawn on for embryological material, especially to fill the gaps in the earlier stages. Table IX, therefore, includes only 18 embryos — all from a single original nest. The number is entirely inadequate for conclusion, tut it was interesting to find that only a smgle specimen was abnormal (by the possession of a 6th neural — cf., Fig. Y9). The marginals display exceptional uniformity. In the three youngest specimens they were not distinct, in one the plan is 13-12, but in all the others the plan is 12-12. It is possible that the eggs from this original nest would have developed with as much uniformity under other conditions.

The table can conveniently be abbreviated.


Robert E. Coker.


Thirty -THREE Embryos and One Newborn. Eggs from one original nest transferred to one artificial nest in tbe ground.

Carapace meters.













































N5 partially divided transversely.

















Marginal series encroaching on nuchal.


















































Normal except that LCI is rather large at expense of LC2. LM2 very small.












N5 divided obliquely; seam distinct on R side, faint on L posteriorly.











Normal except R marginals. Cf . figure.












Supn'y marg. on left side posteriorly.





Normal. (Fore-feet reduced).



















L costal series normal except for minute scute at posterior end.










Difficult to diagnose; cf. figure..








Faint furrow on Nl and N2, extending into spine; abrupt back side of spines furrowed.



















Faint furrow on Nl and N2 and, more distinctly, on N3, Supn'y marginal posteriorly on left side.













Normal. Nuchal and Nl and N2 show very faint traces of furrow. (Hatched.)











Diversity in the Scutes of Cheloiiia.


TABLE VII. Embryos and Newborn — From Artificial Nests in the Ground.

A. 21 Embryos — Originally from One Natural Nest.

Eggs transferred to two artificial nests in the ground. The tirst 10 embryos below (Nos. 35-44) came from one of these nests, the remainder from the other.







Day. i










bC S





3 3



Remarkable foreshortening of body;

66 ■ 67

compare figures. Width 17 mm. (Head

with hare hp defect.) (Pla.stron with

one supn'y scute.)





Scutes normal. ("Cyclopean defect" of head.)







RC3 much reduced.




4 I
























RC5 incompletely divided.








RC4 somewhat reduced, so that RC5 occupies a position more anterior than normally — cf. relation to marginals. Minute supramarginals observable with lens.










49 th





2 scutes occupying place of RC5; also 1 supn'y scute posteriorly.











1 supramarginal, left side, between C2 C3 and MS.






3 supramarginals on each side; one is minute.

) 73 / 74










Supramarginal between C2, C3, and M8, right side.





Minute supramarginal between C3, C4, and MS, left side.










Small supramarginal, between C3, C4, and MS, left side.






N5 is completely divided transversely.






2 supramargmals on R side, 3 on L. Fig. would apply to R side, with smallest scute omitted; applies to L side.

B. l-'t Embryos. Originally from one natural nest.





























? 45th









13 13

12 13

13 13 12 13

15 14






















Supn'y marginal posteriorly, right side. Malformation of posterior margin right side.

Regular scute following N4. Marginals with 14-13 plan anteriorly but with supn'y scutes posteriorly, each side.






RC5 represented by 2 scutes.

Normal, except that nuchal is notched and shows slight median furrow.


LM2 and LM4 very much reduced.

80 79




Robert E. Coker.

TABLE \n.— {Continued.)

C. 2 Embryos and 7 Neivl)orn.

Originally from one natural nest.










C h-1





















Cos tals.





















13 12

13 13

12 12

12 12

12 12

13 12 13 12

12 12

13 13

Normal. Trace of median furrow on nuchal.

RC5 represented by 2 scutes. RM2 much reduced. 83


Neural series of 7 elements, but 3 of 81 these (N4, 5 and 6) are incompletely separated from one another. Normal.

Norm.al. I

Normal. I

Normal. f

Neural series of 8 elements, but N3| 82

and N4 are not completely separated

mesially. Nuchal with distinct furrow

to left of median line, anteriorly. LCI

wanting; LC5 represented by 2 scutes.


Embryos and New'born.

Eggs originally from 2 natural nests — but transplanted Into one artificial nest.









5 6





5 6









5 4





































6 6





12 ? 12 I {Normal.) L marginals not thoroughjly clear anteriorly.

12 12 1 Supn'j' costal on each side, but LCI wanting. No supn'y neural, but N4 with beginning (?) division on each side so as to give usual adjustment.

IS 13 Symmetry of LCI and 2 with RC2

and 3 suggests that there is a supn'y

costal on each side while LCI is wanting.

Very abnormal: see text, p. 54.

Normal. Slight median furrow on

nuchal posteriorly.


RCl merged in 2.

Nuchal divided anteriorly. N6 showing trace of seam, distinct on left posteriorly. 5 ^'supram-arginals", 4 of which are very minute.

Much of the distortion observable in the figure is due to contraction in the preservative.

12 12 LC5 represented by 2 scutes. "Cyclo pean".

12 12 Scutes abnormal anteriorly. "Cy ^clopean".

12 10 Somewhat asymmetrical and mis shapen. Nuchal with median seam I anteriorly, continued posteriorly by a I furrow.




















77 84



88 87 89

Diversity in the Scutes of Chelonia,


TABLE YIl.— {Continued.)

. t o



03 O .^

















Left. Right.




95 96


98 99



54 52 53











1 i


8 8

6 6 6

4 4

12 13

12 12

12 12

13 13

12 13 12 12

N2, 3 and 4 incompletely separated mesially.


Supn'y scute in each costal series. Probably LC6 is supn'y. since the costals anterior to it ha^-e the usual relations to marginals. On the right side the scutes posterior to RC3 have the usual relations to marginals of C2 — 5; hence RC2, or 3 is probably supn'y (contrast with Nos. 80 or 92).

RCl wanting, LCI reduced. Supn'y marginal posteriorly, right side.

Like the above, except for the supn'y

Remarkable carapace of 3 scutes, v. text, p. 53. Plastron with normal number of scutes.

90 91


1 93 [ 95

E. I'l Embryos and Neivhorn. From one artificial nest. Eggs from two natural nests mixed.



















































6 6

12 12 12 12

13 13

13 13

12 12

12 12

13 13 12 12

12 12

13 12 12 12

14 13 12 13 12 12


Nuchal divided symmetrically by a seam anteriorly, continued as a furrow posteriorly.

N4 partly divided transversely.

Asymmetry of nuchal, first neural, first costal and first marginals.


Nuchal seam oblique.

Normal. Trace of furrow on nuchal.

Normal. Normal.

, Normal.

N5 partly divided. LC5 represented by 2 scutes. Supn'y costal on R side.


96 98

97 99

103 101


Robert E. Coker.

TABLE VIII. 76 Newborn Turtles from a Natural Nest.









Left. Right.

Left. Right.












165 J


178 [







6 6 6 6

13 13. 13 13

12 12

13 13

12 12

13 13 i_ 14

13 13 13 12 12 13 12 12

LC5 represented by 2 scutes. LC5 represented by 2 scutes. LC5 represented by 2 scutes. RC5 represented by 2 scutes.

N5 divided on R side

36 normal turtles. 8 normal turtles. 13 normal turtles. 12 normal turtles.



Eighteen Embryos and Newborn. Transplanted from a single natural nest to an incubator.













a S


. 4^



D rt



£ ^





a bO










191) 193 J 194

12 12

32d 34 th




3 embryos normal as far as observable. Normal.



35 th










Supn'y neural posterior.


Normal embs. Car. of 196 and 197 :

197) 200 J




19 mm.

Car. of 198 and 199:

20 mm.







202) 207/ 208




6 Normal turtles.





Diversity in the Scutes of Chelonia. 61

3. Review of Obsekvations. Marginals.

In strong contradistinction to most turtles there is no definite number of marginals that can be considered normal. In the 12plan (see above, p. 49), the first costal is in contact with only two marginals (Ml and M2), the second with four (M's 2, 3, 4 and 5) — Fig. 74. In the 13-plan, the first costal is in contact with three marginals, the second with M's 3, 4, 5 and 6. The differences between the two is therefore, the presence in the one case, the absence in the others, of a scute between Ml and tlie marginal opposite the seam posterior to CI (marked Mx in Figs. 76 and 79). This scute varies in size from a very minute scute entirely surrounded by the scutes preceding and following it (Fig. 76) to a size as large as the others, Fig. 79 right. Compare Fig. 79 left, Fig. 76 right, and Fig. 83, left and right. Sometimes, with the 12-plan anteriorly, there occurs a posterior scute, making the number really 13. This is spoken of as the 12-plan with a supernumerary scute posteriorly.

Sometimes an extra marginal appears in the region of C2. This scute, 3Iy, Fig. 76, leads to a 14-plan if Mx is also present. Again, a scute, Mz, may intervene between the most posterior marginal and the marginal opposite the seam posterior to nonnal C5. In this event the series contains 13, 14, or 15 scutes, according as the 12-plan, 13-plan, or 14-plan prevails anteriorly. Compare Fig. 92 right— 13 scutes (Mz), Fig. 79,' right— 14 scutes {3fx, Mz), Fig. 79, left — 15 scutes (Mx, My, Mz).

As to the relative frequency of the common marginal plans, 12-12 and 13-13 occur approximately in equal proportions, 12-13 and 13-12 occur each about one-third as often as either of the symmetrical plans. The scutes My and Mz are of rare occurrence. Supernumerary scutes at other places were not observed. The stippled area of MIO, Fig. 84, represents an infolded area. The posterior limb is folded over the carapace in this region, and the slight malformity of the margin noted in this region in this and other specimens (Figs. 80 and. 76) undoubtedly results from undue pressure of the limb against the margin, and this suggests the same explanation for the malformed margins of Ma.laclemmys (Figs. 32, 37, 39, 40).

62 Eobert E. Coker.

Such crude direct effects of pressure are not to be confused with the possible indirect adaptive results suggested above (p. 49).

Less than 12 marginals Avere observed only in malformed specimens. In No. 35 (Fig. 66), with marginals 8-10, the right marginal series shows less reduction than any other series of the carapace. In No. 93 (Fig. 89) the marginals are 12-10.


Small scutes, just above the margin and at the meeting-point of three scutes, are noted only in No. 88 of Table VII D (Fig. 85), and in several embryos of Table VII, A (Nos. 43, 47, 48, 50, 51, 53 and 55, Figs. 73 and 74). The costals do not seem quite to meet the marginals in early stages, and if the presence of scutes between costals and marginals be attributed hypothetically to arrest of development, these observations may lend some support to Newmnim's hypothesis as to the number of primary series of scutes in the carapace. But would not such an assumption as to the significance of supramarginals place them in a distinct class from other supernumerary scutes for which arrest of development is not to be hypothecated ?


As was the case in Malaclemmys^ so here the nuchal is frequently found in paired condition (eleven instances. Fig. 103, etc.), or marked by a median furrow, which may be quite distinct (three instances) or very faint (four instances). In one case an anterior seam was continued posteriorly by a furrow (No. 93, Fig. 89).* The furrow is incomplete in No. 78, Fig. 82, and the seam incomplete in 88, Fig. 85. The division is usually symmetrical but not always so.


Each cosital series consists of five scutes, the most anterior being very small, and having no analogous element in Malaclemmys. It may be noted that, when there are twelve marginals, the first two costals are in contact with the first five marginals. In Malaclemmys, which has, typically, twelve marginals, the first costal is in contact

Also No. 102, Fig. 94.

Diversity iu tlie Scutes of Chelonia. 63

with tiie second to iiftli margiiials, the narrowness of the nuchal bringing the lirst marginal into a position anterior to the iirst neural. Where an anterior supernumerary costal occurs in Malaclemmys, the relation of the hrst two costals to the marginal corresponds to that in Tlialassochelys, Fig. 9, left. Sometimes the small first costal of llialassochelys is wanting (seven instances) and then the condition approximates that of Malaclemmys (cf. Figs. 57 and 92). Symmetry in this abnormality was never noted; in fact, in two of the seven specimens a supernumerary scute was present on the opposite side, Figs. 52 and 64. This suggests that the absence of this scute is attributable to some primary asymmetiy rather than to reversion.

Another not infrequent abnormality of costals consists in the presence of two scutes in the place of C5. The two scutes, together, have practically the usual relations of C5 to marginals on the one side and to neurals on the other. Cf. Figs. 70, 83, right, and 56, 82, 88, left. Nos. 10, 66, 115, 116, 117, 118, 114 left (Figs. 52, 101) have the same abnonnality. Contrast wdth these the posterior supernumerary scute of ±^o. 69^ Fig. 76, and 114, right. Fig. 101, also the incomplete division of LC5 in iSTo. 42 (Fig. 69).

The abnormalities of costals so far mentioned, while manifesting asymmetry so far as the numbers of scutes on the two sides goes, leave the tw-o series symmetrical in general plan. But the plan is disarranged in such sj^x'cimens as ^STos. 94 (Fig. 90) and 97 (Fig. 91) and, as we shall see below, the number of extra neurals occurring in association with such abnormalities of costals depends on the extent of the asymmetiy.

Asymmetry may occur without supernumerary costals. Figs. 71, 72, etc.). On the other hand, supernumerary costals may occur symmetrically; Fig. 78 illustrates this, if it be assumed, as seems apparent, that LCI is wanting while the next to last costal of each side is supernumerary,


Paired neurals or neurals incompletely divided in the median line were not found, but a slight median furrow Avas sometimes observed.

A very symmetrical supernumerary neural sometimes appears in sea-turtles just posterior to ]Sr4. This scute is almost perfectly

64 Eobert E. Coker.

rectangular (Fig. 79). Such a shield was found in two of four shells of Colpochelys kempii, and in three of thirty-one specimens of Chelone mydas. In several cases there is a horizontal seam confined to one side, which, if complete, would isolate such a scute. (Nos. 3, 54, 103, 114, 120, Figs. 75, 96, 101 and 102.) Hence the seam which cuts off this scute from !N^5 occurs more frequently on one side than on both. The partial independence in variation shown by the two sides is nicely illustrated in this case, as in many others.

A seam may start from the usual point of origin of the seam bounding the rectangular scute posteriorly, but bend posteriorly to meet the last costal distally or the marginals (Nos. 37, 73, 109, 196). Cf. Figs. 71, 99, and 81. A short seam completed by a fiirrow is sho\\ai in Figs. 55 and 85.

In Fig. 96 N4 is partly divided by a seam on the right side which is completed on the left l)y a furrow.

Adjustment of Ncurals and Costals.

JSTo. 81 (Fig. 77) illustrates well this adjustment: neurals and costals in alternating relation. In this specimen I infer that LCI is wanting and that a pair of supernumerary scutes is present ; but, disregarding this inference, there are, on each side, five costals in contact with neurals posterior to ISTl, while in the normal carapace there are only four. As the costals are nearly symmetrical from this point posteriorly, a single nearly symmetrical extra neural serves to maintain the usual alternating relation. In 31 (Fig. 65) the supernumerary costal is on the left side alone, and here again we find on each side the usual a'lternating relation of neurals and costals. Counting down the left side there are six neurals, counting down the right side there are five. The supernumerary neural is not absolutely restricted to the left side nor unmodified on that side, but tapers to a point. It seems that in Thalassoclielys, the asymmetrical scutes' are not in general as much restricted to one side as in Malaclemtvmys and Graptemys. Shells, such as l^^o. 94 (Fig. 90) and 97 (Fig. 91) are very suggestive of the types noted in the first part of this paper (p. 34, above), but more commonly the asymmetrical elements are of the type of those of 43 (Fig. 72) and 90 (Fig. 86),—

Diversity in the Scutes of Chelonia. 65

comparatively broad ou both sides but widest on the side where they are in proper alternation with the costals. The subject has been so fully discussed in a previous section that I need only refer to the figures : Figs. 61, 62, 6-i, 72, 70, 84, 85, 90, 91, 100, and a few others that need special discussion.

In No. 80 (Fig. 78) there is the normal number of neurals with, evidently, a pair of supernumerary costals ; N4 is a large scute where we would expect two scutes to complete the alternation ; but we find on each side the beginning of division opposite the apices of costals.

In IsTo. 210 (Fig. lO-lJ the incomplete seam on the left side is to be noted.

In 73 (Fig. 81) the correlation is complete except in so far as two of the seams are incomplete mesially.

'No. 78 (Fig. 82) presents an unexpected scute in the oblique fourth neural. The incompleteness, mesially, of its anterior seam suggests that this scute was still more abnormal at an earlier stage. But if this carapace is to be in harmony with the others, we would expect the development of a seam from the point x extending toward the left end of the third neural or toward the second costal.*

E"o. 94 (Fig. 90) has already been referred to; but it is to be noted that two of the seams are completed in the mesial region only by furrows.

No. 92 presents a clear case of "mal-adjustment" (Fig. 87). N2 is an exceptionally large scute and there are not even the beginnings of seams from the apices of the third costals. It is not surprising, however, that this turtle should fail to manifest the usual laws of development in scutes or in any other organs. Its "cyclopean defect" alone gives it a hopeless deformity.

"Incomplete Division."

It may be said that it is "begging the question" to assume that the incomplete seams represent division rather than fusion. Against the latter explanation it may be said that such seams are always in the peripheral or growing region of the scutes, and that, where

The letter is omitted in tlie plate. The poiut x is on the fourth neural seam near the apex of the third right costal.

66 Hobert E. Coker.

they shade off into furrows, the furrow is mesial, and the seam most ' distinct at the periphery. Further, in Malaclemniys, as has already been shown ('05a, p. 14 ff.), the successive concentric rings of growth give a sure means of determining that the incomplete seams noted in that species are of subsequent development. This seems sufficient to make "partial division" acceptable. It may be added that the incomplete seams would be quite inexplicable on the supposition of fusion. Gadow's hypothesis may presuppose fusion of scutes, but these cases could not fall in line with his hypothesis.^ We would have scutes fusing that were not supposed by the hypothesis to fuse (cf. Figs. 81, 82, 91). On the other hand, as expressions of laws of growth that are not complied with until these seams appear, the late appearance of the seams is quite explicable. It is evident that, instead of suggesting arrest of development as the explanation of supernumerary scutes, they would point in a contrary direction. If they have a significance as to ontogenetic development they point to perfection of adjustment as opposed to perfection of number of elements. See also Fig. 101 (specimen 210, not listed).

Finally, whether the blind seams represent division in process, as suggested, or merely broken seams in stable condition, their positions and their peculiar association with other abnormalities are too suggestive to go unremarked.

We speak of supernumerary scutes, although it may be more a

matter of seams than of scutes. We have to do merely with the

subdivision by seams of a given horny area. What determines the

plan of the seams, to what extent it is dependent on vasculation or on

the relations to the bones and other organs beneath, we do not know,

but it is quite conceivable that the laws of growth and of adaptation

to diverse environmental conditions are adequate to account for the

diverse plans of scutes. The evidence points very directly to the

conclusion that the development of asymmetrical seams in the neural

region takes place in accord with laws of growth as distinguished

from laws of heredity, and this may well be the case with other


'Gadow's figures are not reviewed liere, since, though showing a high degree of multiplication of scutes in the several series, the supernumerary scutes are hardly comparable to those dealt with in this section ; supernumerary scutes

Diversity in the Scutes of Chelonia. 67

Salient features of the observations on Thalassochelys :

(1) The very high proportion of abnormality in these turtles, which developed chiefly under abnormal conditions.

(2) The general unsymmetrical aspect of the abnormalities, the marginal, Mx, being an exception, since it occurs on both sides more frequently than on one side alone.

(3) The occurrence of small "supramarginals" — not previously recorded as an abnormality ; these are always at the meeting point of three other scutes.

(4) The development of an embryo with only 3 scutes in each dorsal series (Fig. G6), and another with 3 scutes forming the entire carapace.

(5) The appearance of a number of "monstrosities" — "cyclopean" embryos, etc. ; and the association of a high proportion of abnormality of scutes with the occurrence of these monstrosities.

Supernumerary scutes occur in the neural and costal series with much irregularity and without symmetry. The only symmetrical recurring scute in either series is the rectangular scute posterior to N4. At least two abnormalities in the costal series show a regularity in their recurrence: these are — the absence of the small Cl, and the presence of two scutes in the place of C5 ; but neither of these abnormalities appeared on both sides and symmetrically.

Part III. Significance of the Abnormalities.


The specimens of Thalassochelys observed display an even greater degree of diversity in number and arrangement of scutes than did those of Malaclemmys. While Newmann found that one-tenth (48) of 476 specimens of Graptemys had supernumerary scutes in the carapace, and only about one-twentieth of 188 specimens of Chrysemmys, about one-fifth of the first 243 specimens of Malaclemmys observed had more or less than the typical number of scutes. But

are intercalated, generally in association with intercalated costals, sometimes independently, but not with marked asymmetry. In fact, the only abnormalities in my specimens that resemble those of his are the presenece of 2 scutes in place of C5, and the interposition of a small neural posterior to N5.

68 Kobert E. Coker.

in 208 specimens of Thalassochelys, almost one-tliird were abnormal. The diversity in Thalassochelys is even greater than this proportion indicates, for, in estimating that proportion of abnormality, both twelve and thirteen marginals were considered as normal. The variable M2 is absent almost exactly as often as it is present and in one-fourth of the total number of specimens it is absent on one side while present on the other. It seems remarkable that there should be such diversity in forms that have been so little modified throii^'h several geological ages. Tlialassoclielys is at least as old as Eocene times, and turtles of the Upper Jura have essentially the same number and arrangement of scutes as have modern turtles. Erom one point of view the arrangement of scutes is exceptionally plastic or variable, while from another, the paleontological point of view, it is exceptionally persistent and fixed.

It would be interesting to know the explanation of the "abnormalities," such a large proportion of which show asymmetry, but the data that we have in hand are, comparatively, very scant. It is advisable, however, to consider the explanation advanced hitherto (atavism) and other possible explanations in the light of the data in hand.

External Conditions.

We have not sufiicient evidence as to the influence of the environmental conditions of the eggs during development in the production of abnormalities of scutes. A large proportion of abnonnalities was found in each of the nests transplanted into the ground at the laboratory, all of which were under more or less unfavorable conditions. The nest which showed by far the largest proportion of abnormal specimens was one in which very great pressure from the distension of the eggs was noted. The smallest per cent of abnormality yielded by any nest in the ground was found in the turtles from eggs left undisturbed to hatch under natural conditions; but the most normal lot of turtles of all was that obtained from eggs which developed in the "incubator," with the element of pressure eliminated. The series is inadequate for positive conclusions, however, as the number of embryos is too small, but it suggests the importance of further experiments.

Diversity in the Scutes of Chelouia. 69

_ Inheritance from Immediate Ancestors.

The diversity in some of the lots of turtles, each of which groups came from one original nest (Tables VI and VII, A, B, and C) certainly does not suggest that the abnormalities of the parents had a great influence on the abnormalities of the young, but on this point, again, the data are inadequate.


It has been seen that the prevailing interpretation of the abnormal scutes is that they are atavisms ; ^'reminiscences of earlier, phylogenetic conditions" (Gadow, '05, p. 638) ; examples of systematic atavism in the sense of de Vries" (Ne^vTiiann, '06, p. 69). In studying this diversity of scutes, then, one is confronted at the outset by these questions : are all of the abnormalities to be regarded as atavisms ? and, if not all, which anomalies are due to reversion ? Comprehensive use of the conception of "systematic atavism," as applied to these variations, has been made for morphological purposes, and as the observations given in this paper might be taken some to confirm, some to modify, others to negative such morphological views, I consider it necessary to inquire with some fulness regarding the basis of the atavistic interpretation of supernumerary scutes.

It has already been made clear that asymmetrical neurals are not to be regarded as instances of reversion. Further, from a glance at such figures as Figs. 63, Q>Q>, and 95, w^e infer that some others of the abnormalities are surely not referable directly to atavism. But, are most of the supernumerary scutes to be explained by reversion ? and, hy tvliat criterion shall it he decided if a given scute is atavistic or coenogenetic in character?

By atavism we understand ordinarily the reappearance of a character, not manifest in the immediate ancestors, but typical of some remote ancestral form. It is assumed that this character, though disappearing from view, has never been actually lost, but, having assumed a latent condition, has been transmitted from generation to generation until finally, on the proper occasion, it reasserts itself. Atavism is, therefore, much more than a descriptive term referring

70 Eobert E. Coker.

to the resemblance of a new to an old character; it is a positive assumption of the unbroken continuity of the old quality in latent condition through succeeding generations and its final reappearance as a variation.

The need for this theory of reversion has been felt asi an explanation of the observed fact that the offspring often manifests some character not possessed by its immediate parents, but peculiar to its grandparents or to some more remote ancestors. Especially in the latter case do we, in Darwin's words, "feel a just degree of astonishment," and, whether the resemblance is to a near or to a more remote ancestor, does there seem a need for the theory of latent characters and reversion. It would seem, however, that the theory was one to be invoked only in cases where there is strong reason to believe that the resemblance is not superficial or accidental, but so vital and unmistakable as to be explicable only as a direct inheritance. Yet the theory of reversion has been overworked to the point of being applied to cases of the most superficial resemblance, even to variations that bear resemblance only to a purely constructive ancestor, the hypothetical existence of which is based largely on the arbitrary assumption that the anomaly in question is an atavism.

The subject of polydactylism in higher vertebrates not only offers an excellent illustration of the wholesale use of the theory to explain all kinds of anomaly, but it is one which has been so much longer and more thoroughly studied in its various aspects (anatomy, heredity, etc.) than has the subject of supernumerary scutes, that we may be allowed to draw some lessons of caution from its history, if we do not seem to imply that the same principles must apply to scutes as to digits.

Darwin himself, "with much hesitation," attributed polydactylism in man to reversion, but soon retracted ("76, p. 459), with the explanation that he had been misled by the statements of observers. The bifid rays of Selachians, and "constructive" Y-toed ancestral mammals, have been called up to account for modern polydactylism (Albrecht, Bardeleben, etc.). Other writers, as Gegenbaur, and, especially, Bateson and Prentiss have put the matter in a better light. Certainly the following simple principles^ self-suggesting, would seem demonstrated in regard to polydactylism.

Diversity in the Scutes of Chelonia. Yl

1. External resemblance is not ground for the assumption of reversion (since the presence of two digits in the horse is sometimes due perhaps to the development of the vestigial second digit, or, in other cases, clearly to the duplication of digit three. (Prentiss, '03, p. 299).

2. Recurrence of a similar abnormality in numerous individuals of the same or different species is not ground for the assumption of reversion. (Cf. duplication of hallux or pollex in man, cat, dog and fowl. Bateson, Prentiss, etc.)

3. AA^iere extra elements occur which are, presumably, of vestigial origin and possibly attributable to reversion, the atavism may be by no means true to the ancestral condition reverted to ; for Prentiss ('03, p. 291) finds that, in a large number of cases, the supernumerary vestigial digit appears in a condition of partial or complete duplication — that is to say, in a condition directly misleading if used for inference as to ancestral form.

4. Supernumerary elements have a negligible value as basis for phylogenetic hypothesis.

Without regard to Polydactyly, the above considerations suggest themselves a priore against the attribution of definite morphological significance to supernumerary scutes. I refer to Polydactyly because these a priore objections seem to gain weight from their demonstrability in another field of variation.

In regard to scutes the following points may be noted :

1. We have no evidence as to the value of external resemblances — as to whether the same appearance might not be due in one individual to one cause, in another to a different cause.

2. The assumption that anomalies of scutes are atavisms rests on the recurrence of certain more or less characteristic scutes in different individuals of the same species, and in different species, and the correspondence of such scutes in a few instances to normal scutes of still other species. But recurrence of certain very definite anomalies in the same or in different species is often noted in cases where atavism w^ould not be suggested. Thus the cyclopean defect occurs in man, and in an essentially similar form in amphibia (Spemann, '04). I have had apparently the same sort of defect to

72 Kobert E. Coker.

appear in three embryos of the loggerhead turtle, taken from artificial nests, and the resemblance to the same abnormality in man and in Triton is striking. Here is the same abnormality occurring in different and widely removed species, and with remarkable definitions of form.

The matter of recurrence need present little difficulty. If the organization of one turtle is much the same as that of another of the present time or of a considerable period of past time (and observations so indicate), I do not know why turtles may not be subject to similar anomalies ; nor why there may not occur now the same variations that once, in connection with others, characterized the ancestors of a diverging species ; nor why, if any of the variations now occurring should characterize a future subspecies or species, they might not continue to occur in other turtles without acquiring a new significance.

3. The use of the supposed atavisms as morphological data implies not only the transmission of the primitive characters in latent condition from generation to generation since remote geological ages, but also that they are now seen practically in pure form and unmixed with any appreciable number of new inheritable variations, occurring first during these ages since the primitive characters became latent. The ground for such an assumption is not apparent.

4. From the phylogenetic point of view, how far back nuist the atavism of scutes point ? The whole matter of the phylogeny of turtles is obscure, but, from paleontological data, it is evident that the carapace of Thecophorous turtles acquired at quite a remote period very nearly the present form as regards arrangement of scutes and plates. Existing genera may be traced back to Eocene {Tlialassochelys) and Upper Cretaceous periods (Chelone). More significant still are such fossil forms as Plesiochelys solodurensis Rutimeyer, and Platychelys oherndorferi Wagner, from the Upper Jura, with carapaces showing essentially the saine plan of scutes and plates as is found in turtles of to-day, except that the dovetailing of neural and costal scutes is less noticeable. I should be slow to draw inferences from the variations in modern turtles regarding the types of periods more remote than the Upper Jura.

Diversity in the Scutes of Chelonia. 73

Finally, in tlie abnormalities we have to do not merely with ■ supernumerary scutes but with many instances of turtles with less than the typical number of elements, which cases it would be diificult to expilain primarily by reversion. If turtles may have less than the typical nimiber (even to only three scutes in each dorsal series, in 'No. 35), otherwise than by reversion, why may not they have moi^e in the same way ?

Summary of Part III,

1. The experiments in incubation are suggestive of the possible significance of external conditions in the causing of abnormalities of scutes.

2. I can atribute to supernumerary scutes no value for deductions as to the phylogeny of turtles.

3. AMiether or not atavism may have a remote or indirect connection with the abnormalities (for which I see no evidence) I do not regard individual scutes as atavisms.

4. Many of the abnormalities are clearly not atavisms.

Note. As my conclusions in regard to the significance of abnormality of scutes is in opposition to that with which IsTe^vmann starts out, and may, therefore, seem inimical to his hypothesis, I think it proper to state that the value of ISTewmaun's view of the evolutionaiy history of the carapace appears to me to l)e largely independent of the question of atavism since the more essential features Avere really based on comparative anatomy, not systematic atavism. He supposes fourteen primitive series of scutes; of the five not present in most turtles, only one was found in his ahnornudities (interplastrals), while three were represented by normal series of certain species (interpl astral and supramarginals, two) and another only by normal scutes of the tail of Chelydra. My obsen^ations might, it is true, show in abnormality representatives of the supramarginal series, but, just as well, they would give reason to infer primitive division of neurals and the primitive presence of a pair of siihnarginal series, neither of which inferences would be in hamionv with his view. Hence the

74 Kobert E. Coker.

number of series involved in bis view seeks its basis, not in supposed atavisms, but in comparative normal anatomy.


Agassiz, Louis, '57. Contributions to tlie Natural History of the United States. Vol I, Part II, North American Testudinata, and Vol. II, Embryology of the Turtle.

Bateson, W., '94. Materials for the Study of Variation. London and New York.

Baub, G., '80. Osteologische Notizen iiber Reptilien. Zool. Anz., 9.

, '87. Morphology of the Carapace. Amer. Nat., XXI, Jan., 1887.

, '88. Osteologische Notizen iiber Reptilien, Fortg. Ill, Zool. Anz.,

pp. 417-424.

, '88a. Unusual Dermal Ossifications. Science, XI, p. 144.

, '89. Osteologische Notizen. Fortg. VI, Zool. Anz., 298, pp. 40-47.

, '89a. Die Systeinatische Stellung von Dermochelys Blaim. Biol.

Centralblatt, IX, pp. 149-153.

. '90. The Genera of the Cheloniidae. Am. Nat'l., XXIV, May.

, '90a. Classification of Testudinata. Am. Nat'l., XXIV, Jvme.

, '96. Bemerkungen liber die Phylogenie der Schildkroteu. Anat.

Anz., Bd. XII.

BouLENGEE, '89. Catalogue of the Chelonians, Rhyncocephalians and Crocodiles in the British Museum. London, 1889.

Case, E. C, '97. On the Osteology and Relationship of Protostega. Journ. Morph., XIV.

Coker, R. E., '05. Diversity in the Scutes and Bony Plates of Chelonia.

(Abstract.) Science, N. S., XXI, 532. , '05a. Gadow's Hypothesis of "Orthogenetic Variation" in Chelonia.

Johns Hopkins Univ. Circ, No. 178, May.

, '05b. Orthogenetic Variation? Science, N. S., XXII, No. 574.

— , 'OG. The Natural History and Cultivation of the Diamond-back Ter

rapin. N. 0. Geol. Surv. BuL, No. 14. Darwin, Charles, '70. The Variation of Animals and Plants under Domestication. Two vols. Second edition.

De Vries, Hugo, '05. Species and Varieties. Their Origin by Mutation. Chicago, 1905.

Gadow, H., '99. Orthogenetic Variation in the Shells of Chelonia. Willey's Zool. Results, Part III, pp. 207-222, May, 1899.

—■ , '01. Amphibia and Reptiles. Camb. Nat'l. Hist, Vol. VIII, London, 1901.

, '05. Orthogenetic Variation. Science, N. S., .

Diversity in the Scutes of Chelonia. 75

Gegenbaur, 'SO. Kritische Bemerkungeu iiber Polyclactilie als Atasivmus.

Morph. Jahrb., Bd. VI, pp. 584-596.

, '88. Ueber Polydactylie. Morph. Jahrb., Bd. XIV, pp. 394-406.

GoETTE, A., '99. Entwicklung des kuochernen Riickenschildes der Schild kroten. Zeitschr. f. wiss. Zool., 66 Bd., 3 Heft., pp. 407-434. Hay, O. p., '98. On Protostega, the Systematic Position of Dermochelys, and

the Morphogeny of the Chelonian Carapace. Amer. Nat'l., XXXII. ■ , '01. The Composition of the Shell of Turtles. (Abstract.) Science,

Vol. XIII, p. 624, April. Hay, W. p., '05. A Revision of Malaclcmmys, a Genus of Turtles. Bui. Bur.

Fisheries for 1904. HoLBROOK, J. E., '42. North American Herpetology, Vol. I. Turtles. Philadelphia, 1842. Newmann, H. H., '06. The Significance of Scute and Plate "Abnormalities"

in Chelonia. Biol. Bui., X, Nos. 2 and 3. , '06. Correlated Abnormalities in the Scutes and Bony Plates of

Chelonia. (Abstract.) Science, N. S., XXIII, p. 588, April. Pabkee, G. H., '01. Correlated Abnormalities in the Scutes and Bony Plates

of the Carapace of the Sculptured Tortoise {CJteloinis insculptus). Am.

Nat'l., Vol. 35. Contr. Zool. Lab. Mus. Comp. Zool., Cambridge, No. 118. Prentiss, C. W., '03. Polydactylism in Man and the Domestic Animals, with

especial reference to Digital Variations in Swine. Bui. Mus. Comp. Zool.,

XL, No. 6. Spemann, '00. Experimentelle Erzeugung zweikopfiger Embryonen. Sitzber.

d. Phys. Med. Gesell., Wiirzburg, 1900. ZiTTEL, Karl A. von, '02. Textbook of Paleontology. Transl. and ed. by C.

R. Eastman, London.


Figs. 1-13, plates I-VI, photographs of shells of diamond-back terrapin.

Figs. 14-54, plates VII-XI, sketches of plan of scutes of diamond-back terrapin ; when only a portion of the carapace is shown, the drawing is from a field-sketch; Fig. 51 omitted; Figs. 53 and 54 from terrapin of Texas (see p. 31).

Figs. 55-104, plates XI-XIV, drawings of plan of scutes of embryos and newborn of the loggerhead sea-turtle.

In the first eleven plates the figures in parentheses folloicing the number of the illustration refer to the serial number of the specimen in tables 7-7; in the succeeding plates (and in plate XI if preceded by the letter "T") the numbers refer to the specimens of Thalassochelys of tables VI-JX,




1 (70)

2 (244)

The Journal of Morphology. — Vol. 21, No. 1.





3 (9)

4 (9)

The Journal of Morphology. — Vol. 21, No. 1.




5 (210)

8 (195)

The Journal of Morphology. — Vol. 21, No. 1.




6 (150)

7 (151)

11 (149)

The Journal of Morphology. — Vol. 21, No. 1.




9 (131)



lU (133)

The Journal of Moephologt. — Vol. 21, No. 1.




12 (88)

13 (126)

The Journal of Morphology. — Vol. 21, No. 1.




15 (151)

14 (251)

16 (167)

17 (210)

21 (56)

18 (245)

20 (4)

19 (246)

22 (228)

23 (256)

The Journal of Morphology. — Vol. 21, No. 1.




24 (247)

25 (223)

26 (174)

27 (257)

28 (191)

29 (185)

30 (178)

32 (209)

33 (183)

34 (176)

31 (196)

35 (200)

36 (249)

The Journal of Morphology. — Vol. 21, No. 1.




37 (145)

38 (145)

39 (255)

40 (248)

41 (218) The Journal of Morphology. — Vol. 21, No. 1.

42 (196)




43 (199)

44 (218)

45 (198)

47 (181)

46 (217)

48 (222)

The Jodenal of Morphology. — Vol. 21, No. 1.




49 (252)

52 (T 10)

50 (250)

53 (259)

54 (260)

58 (T 15)

59 (T 19)

55 (T17)

57 (T6)

61 (T23)

56 (T14)

62 (T 24)

60 (T 25)

The Journal of Morphology. — Vol. 21, No. 1.




63 (26)

64 (30)

65 (31)

66 (35)

67 (35)

68 (66)

69 (42)

70 (45)

71 (37)

72 (43)

73 (48)


74 (48)

75 (54)

76 (69)

The Jouenal op Moephologt. — Vol. 21, No. 1.




88 (91)

87 (92) The Jourxal of Moephologx. — Vol. 21. No. 1.

89 (93)




90 (94)

91 (97)

92 (98)

94 (102)

97 (107)

93 (100)

95 (100)

96 (103)

98 (104)

99 (109)

100 (209)

101 (114)

102 (120)

103 (112)

104 (210)

The Journal of Morphology. — Vol. 21, No. 1.




This investigation of the osteology of the family Scombridge was not undertaken to decide any question of obscure relationship of the groups within the family to each other, but to serve as a foundation for future work on the relationship of the many forms that have from time to time been placed in families supposed to be more or less closely related to the Scombridse, and known collectively as the group Scombroidei — the mackerel-like fishes.^

Though the family Scombridse probably does not contain the most primitive of the Scombroids it has served as a center around which the more or less aberrant forms have been arranged, and is consequently a convenient basis from which to work.

The family may be characterized as follows :

Bones all light and fibrous.

Supraoccipital crest formed anteriorly by f rentals ; usually extending to the ethmoid.

Supraoccipital not separating exoccipitals or epiotics, though more or less completely covering the epiotic suture on the surface of the skull, and sometimes the exoccipital suture.^

^This paper will be followed by others, each treating of a family, or of a few related families, of the Scombroid fishes. Only mutual relationships within tlie family will be discussed, though relationships between families will be touched upon when it seems advisable. This latter question, however, as well as the relationships of the group to other groups will be more fully discussed in a final paper.

-As this character can be seen only by bisecting the cranium it has scarcely been reported upon. When a bone on the surface of the cranium covers a suture between two other bones they are erroneously said to be separated by that bone. Of the few Percoids examined for this character none were found with the epiotics in contact with each other.

The Journal of MonniOLOGY. — Vol. 21, No. 1.

78 Edwin Chapin Starks.

Exoccipitals (except in Scombrinse) broadly meeting over basioccipital.

Prefrontals broadly united at the median line ; each pierced for the passage of the olfactory nerve.

JSTo auditory bulla apparent externally.

Temporal crests well developed, and reaching at least to above middle of eyes.

Myodome large and opening posteriorly to the exterior.

Opisthotic (except in Scombrinse) interposed between the pterotic and exoccipital, and the lower limb of the posttemporal attached to its upper surface.

Basisphenoid present, and with a process descending to the parasphenoid.

Eye with a bony sclerotic case.

Suborbital ring incomplete, and without a sensory tube (except in Scombrinse) .

Nasal bones well developed.

Process of premaxillary short, heavy, and triangular, and abutting immovably against the ethmoid.

Maxillary with an auxiliary element on its upper edge.

Preopercle unarmed with spines, and all of the opercular elements j&tting closely together to fonn a broad smooth plate.

Head of hyomandibular where it articulates with cranium divided into two parts.

Teeth of jaws fitting into alveoli.^

Pour pairs of superior pharyngeals present ; the third and fourth on each side sometimes joined (not anehylosed) to form a more or less complete single plate.

Three basibranchials present ; the first remote from the hypohyal of the first arch, and hooked under the glossohyal.

Clavicle not developed much above pectoral fin, placed very oblique and sloping forward, beyond the hypocoracoid.

Actinosts four in number ; short and broad ; two and a half of them on the hypercoracoid.

This character is appreciable only in the forms with moderate or large teeth.

The Osteology of tlie Scombrid?e. 79

Postclavicle in two parts.

Parapoplajses never developed far anteriorly; when they appear they soon become attached to form hiemal arches and spines at about the middle of the abdominal cavity.

Eibs posteriorly at the tips of the abdominal haemal spines, each pair with their bases in contact.

The epipleurals never borne by the ribs except in Acanthocybium, • Baseosts expanded at base of dorsal spines to form broad bony bucklers.

First interhsemal never much enlarged, nor strongly attached to the first haemal spine.

United parapophyses in no way differentiated from haemal arches, so that abdominal vertebrae are only distinguished from caudal vertebrae by the attachment of ribs.

Caudal rays deeply divided to receive hypural plate, which they usually completely cover.

Dorsal spines weak and flexible ; dorsal and anal with some posterior rays detached to form finlets.

Caudal peduncle very slender; the caudal rays very divergent.

The principal sub-family and generic differences are indicated in the following key :

A. Pterotic not excluded from brain cavity by a deep pit behind prootic. No bony caudal keel except in Sardiune.

B. The opistbotic situated on lower surface of cranium ; not interposed between the exoccipitals and pterotic ; the lower limb of posttemporal attached to its posterior edge. The myodome open behind in a small transverse slit. The ethmoid produced to a strong angle in front ; nasals narrow and projecting far beyond them. No lateral caudal keel either bony or membranous. Suborbital ring complete with a sensory tube. Scombrin.e.

1. Gill rakers moderately long; metapterygoid not supporting pterygoid. • Scomber.

2. Gill rakers extremely long ; metapterygoid assisting quadrate to sup port pterygoid. Rastrelliger.

BB. Opistbotic as much on superior as inferior surface of cranium, and interposed between the pterotic and exoccipital ; lower limb of posttemporal attached to its superior surface. Suborbital ring incomplete, and without a sensory tube.

80 Edwin Cliapin Starks.

C. Ethmoid concave in front. Nasals broad and attaclied for their full length to frontals and ethmoid. Myodome opening directly to exterior through a longitudinal foramen. D. Temporal crests reaching straight forward to near front of cranium. Caudal keel composed of membrane only.* The last two or three vertebrai normal in size.^ E. Supraoccipital crest reaching to ethmoid ; the full length of cranium concave on each side of it. Preorbital part of cranium not produced. The opening between alisphenoids to brain chamber wide. The vertebrje number 49. ScoMBf3joMORiN.E. Scomberomorus.

EE. Supraoccipital crest ending at front of eyes, anterior to which the broad preorbital part of cranium is transversely and evenly rounded and noticeably lengthened. The alisphenoids nearly divide the anterior opening to the brain case into two parts. The vertebrne number G6.

AcANTHOcYBiN^. Acauthocybium.

D-D. Temporal crests slanting obliquely to supraoi'bital rim. Caudal with a wide bony keel. The last two or three vertebriB abruptly decreased in length. All of which characters as in the ThunniniB, differing only as in division "A." Sardine. Sarda.

AA. Pterotic excluded from the brain chamber by a deep pit or infolding of the bone behind prootic. Caudal peduncle with a bony lateral keel.

CC. Ethmoid produced to a medium angle in front ; the nasals slender and much projecting beyond it. Myodome opening to a more or less specialized chamber in the parasphenoid.


1. Inferior vertebral processes normal. Thunnus.

2. Haemal arches enormously developed and close to the centra.


3. Haemal arches not enlarged but carried far away from the

centra by a long bony pedicle. Auxis.

If we could eliminate the genus Scomber the family W'Ould be much more compact, as it stands farther from the other genera than they do from each other. The characters of Scomber may be for convenience here summed ujj a little more fully than in the foregoing key. It differs from all of the others in having the superior

I am indebted to Mr. Barton A. Bean of the National Museum for investigating the condition of the caudal keel in Acanthocybium. ^This character has not been verified in Acanthocybium.

The O^lcitloiiv of llio S(,'(tiiil)ri(la\ 81

crauial crests peculiar; the opistliotic, and the attaclmieiit of the posttemporal to it normal ; the suborbital ring complete and carrying a sensory tube ; the auxiliary maxillary very small ; the exoccipitals not meeting above the basioccipital, and with their condyles small and peculiarly placed, and in having neither a bony nor a membranous caudal keel.

Eastrelliger, Jordan and Starks, has departed but slightly from Scomber, differing chiefly in the way in which the metapterygoid assists the quadrate in supporting the pterygoid ; in the high knifelike ridge formed by the basibranchials ; and in the great development in the length of the gill-rakers. It is othermse as in Scomber.

Certain characters indicate that Scomberomorus has come more directly from the Scombrina? than have any other of the genera here considered, though its evident alliance with Acanthocybium — possibly the most aberrant of its family — shows hoAV far it lias departed from the Scomber type. Consequently the fact that Scomberomorus is here most closely approximated to Scomber does not necessarily mean that it is the most closelv related to that ffenus, but that its descent is more directly traced. It has no bony caudal keel, but a membranous one shows a development in the direction of one ; its cranial crests are directed straight forward, though they are not interrupted as in Scomber ; its auxiliary maxillary is small ; and its last two or three vertebra? are normal, or not abruptly decreased in length.

Acanthocybium naturally comes next to Scomberomorus, though as is intimated above it certainly does not deserve a position so close to Scomber. It shows, as was long ago pointed out, an apparent divergence towards the sword fishes.

The exact position of Sarda is a little obscure. It has the concave ethmoid and non-projecting nasals of Scomberomorus and Acanthocybium, but has the cranial crests arranged almost identically as in the Thunninse. On the lower surface of the cranium is a slight depression showing a development towards the deep pit of the Thunninse and the caudal peduncle has a lateral keel. Consequently it must have sprung from somewhere betAveen the Scomberominse and the Tliumiina3 to have such marked characters of both groups. It

82 Edwin Cliapin Starks.

shows, however, a much closer alliance with the latter sub-family than with the former.

Thnnnus (the genus Germo is not here recognized as distinct) is plainly close to Auxis and Euthynnus, as is shown by the exclusion of the pterotic from the brain cavity, and in nearly all characters but the condition of the infra-vertebral processes. It is rather astonishing to find genera running so close together as these three do, and yet differing so radically in the condition of the hsemal arches.

Liitken*" calls the iieculiar condition of the inferior vertebral processes in Auxis a modification of the condition of these processes in Euthynnus. This can scarcely be so, as the modification, though as extraordinary in Auxis as in Euthynnus, is of a different character. It is difficult to imagine either condition as being a modification of the other. It seems probable that these two genera left their parent stem at about the same place, or in other words that they are both a modification of some similar condition in the common ancestor from which the development has been divergent.

In Euthynnus the inferior vertebral foramina and the h?emal arches are enormously developed, the latter springing almost directly from the centra of the vertebra^, while both the postero and anterozygopophyses equally form long slender processes between the arches.

In Auxis neither the haemal arches nor parapophyses are enlarged, and they are borne far away from the body of the vertebrae each by a solid bony pedicle formed (at least in part) by the antero-zygopophysis ; the postero-zygopophysis taking no part in this formation.

So to consider the condition of Auxis as a modification of that of Euthynnus we should have to eliminate the postero-zygopophyses together with the laminse of bone that incloses the inferior foramen behind each arch, join the antero-zygopophyses and the bone surrounding the front of the inferior foramen into a solid pedicle, and restrict the great arch to a small opening at the distal end of the pedicle.

The foregoing may be summed up by the following diagram showing the supposed origin of the genera.

'Spolia Atlantica, p. 596, 1880.

The Osteology of the Scombriclse.



■iy/ ACAjrrHocYBnrM


A. The characters of Scomber.

B. The acquirement of the interposed opisthotics.

C. The Thnnnius type of cranial crests, and the inferior cranial pit indicated, with the Scomheromorus type of ethmoid and nasals.

D. The inferior cranial pit excluding the pterotic from the brain cavity, and the condition of the ethmoid and nasals of Scomber.

E. The condition of the infra-vertebral processes from which Auxis and Euthynnus have diverged.

F. The Scomheromorus type of cranial crests, elongate form, concave ethmoid, and nonprojectlng nasals.

In the following pages the osteology of the genera is described in greater detail.

SCOMBER.' A specimen of Scomber japonica, Ilouttuyn, 11 inches in length, from the Canary Islands, and a skull of a slightly larger specimen of the same species from Peru.

The supraoccipital crest is developed backwards in a long spatnlate process, which is free from the exoccipitals below. On top of the craninm it is only slightly developed, and is carried forward by the frontals to opposite the beginning of the posterior third of the orbital

'For descriptions in detail of the osteology of Scomber see the beautifully illustrated work by Edward Phelps Allis, Jr., entitled "The Skull, and the Cranial and First Spinal Muscles and Nerves in Scomber scomber." Jour. Morph., Vol. xvii, 1903.

84- Edwin Chapin Starks.

cavity, or coterminous with the temporal crests. From the front of the supraoccipital a broad low rounded ridge runs obliquely across the frontal outward and forward toward the middle of the orbital cavity where it becomes lower and broader and merges into the general level of the frontal. Against this ridge the pterotic and temporal crests end a short distance from the orbital rim. A thin high, auxiliary crest is developed between the anterior ends of the temporal and pterotic crests. The myodome is large, but posteriorly its opening is closed all l)ut a small transverse slit by the parasphenoid. The supraoccipital widely separates the parietals and is developed a little anterior to them. Posteriorly it covers the suture between the extreme upper ends of the exoccipitals, but it nowhere separates them. The exoccipitals do not meet over the basioccipital at the mouth of the long tunnel-like foramen magnum, though some distance in the foramen they are in contact for a short distance. The vertebral condyles of the exoccipitals are very small. They are anterior to the basioccipital condyle instead of overhanging it as usual, and they slope outward and forward away from the median line, rather than outward and backward or toward the median line. The pterotic ends posteriorly in a sharp spine and there is no deep pit between it and the prootic and sphenotic. There is no pit on the lower part of the cranium behind the prootic, and the pterotic is not excluded from the brain cavity. The opisthotic is as in the majority of spiny-rayed fishes: covering the suture between the pterotic and exoccipital, but not at all separating them. It is wholly on the ventral surface of the skull, and the lower limb of the posttemporal is attached to its posterior edge. The lateral process of the parasphenoid does not reach to the upper edge of the prootic. The vomer is broad and thick and on. its outer anterior edge is an oblique facet, which fits snugly against the inner surface of the maxillary. The vomer and palatine bear small teeth.

The prefrontals broadly meet at the median line behind the ethmoid ; they are pierced by the olfactory foramina. There is an articular facet on the posterior end of each, and another on the anterior end of the ethmoid for the attachment of the palatine. The ethmoid is but little posterior to the front of the vomer, and it is broadly rounded

The Osteology of tlie Scombridse. 85

anteriorly; the short premaxillary processes scarcely reach to it. A . basisphenoid is present and is separated from the roof of the myodome as nsual ; it has a process descending to the parasphenoid. The alisphenoids do not meet each other.

A well developed preorbital is present and the suborbital ring is complete, consisting of thin plates along the anterior edge of which runs a small sensory tube. There is no suborbital shelf, and the eye has a bony sclerotic case as do all the other members of the family. The nasals are attaclied to the sides of the frontals and project far forward beyond the ethmoid.

The head of the hyomandibular, by which it is attached to the cranium, is divided into two parts, a round knob in front and an elongate portion behind, well separated from each other. The metapterygoid is channeled on its posterior edge to receive the edge of the hyomandibular, and behind the hyomandibular it sends a long triangular process backward half way across the preopercle. It has a large articular facet on its posterior edge to support the opercle. The symplectic is long and slender, somewhat broadened behind the metapterygoid, and extending behind the quadrate in a channel. The opercle bones are wide and fit smoothly together without ridges ; the preopercle in particular is broad. The wedge-shaped process of the articular does not nearly fill the usual deep notch of the dentary, thus leaving a considerable space between these two bones along the upper edge. A well developed angidar is present. The premaxillary and dentary teeth are small and set in alveoli. On the inner anterior edire of the maxillary is a large -articular facet for attachment to the side of the vomer. On the posterior upper edge is a small auxiliary maxillary. The maxillaries are not in contact with each other anteriorly. The premaxillary processes are heavy, short, and triangular, and they do not project backward to reach the ethmoid.

The hypohyals are very large and are paired on each side ; a wide flat glossohyal is present ; the broad urohyal has scarcely any lateral wings developed along its lower edge. Four branchiostegal rays are attached to the ceratohyal and three to the epihyal ; the anterior ones are attached to the lower edge of the arch, but they creep up to the outer surface posteriorly. A short interhyal is present.

86 Edwin Chapin Starks.

Three bony basibrancliials are present; that of the fourth arch is cartilaginous. The second and third arches join the third basibranchial, the first arch joins the second, and the first basibranchial is wholly in front of the arches, and projects forward under the glossohyal. As usual there is no hypobranchial to the fourth arch. The inferior pharyngeals are wide and are thickly covered with long brush-like teeth. There are four superior pharyngeals present on each side. The third and fourth together form a single elongate plate though they are not anchylosed to each other. The second is small and narrow and lies beside the anterior part of the third rather than in front of it. • The first is as usual toothless ; the others are covered with teeth similar to those on the lower pharyngeal, but smaller.

The clavicle is a long evenly curved bone with a broad wing projecting backward over the pectoral for the support of the postclavicle. The hypercoracoid foramen is large and just below the center of the bone. The hypocoracoid is long and slender and arches away from the clavicle to rejoin it again a considerable distance above its point. From the upper edge of the hypocoracoid a wing is developed backwards to the tip of the lower actinost. The four actinosts are broad and short with a small pore between each pair ; the third one from the top is supported equally by the coracoid elements. The postclavicle is in two parts, the upper broad and thin, the lower broad above but tapering to a long point downward. A short wide supraclavicle is present. The posttemporal is forked ; the upper limb lies broadly over the epiotic extending slightly onto the supraoccipital ; the lower limb is attached to the posterior edge of the opisthotic as in the majority of fishes.

The pelvic girdle is rather complex in shape, consisting of a thin horizontal plate, which meets its fellow of the opposite side at the median line. From the horizontal plate is developed downward a subvertical plate. From the union of the opposite sides of the girdle a pair of long thin processes are developed forward, and a pair of spine-like processes are developed backward between and above the ventral fins.

There are 15 abdominal vertebrae, and 15 caudal vertebrae, or a

The Osteology of the Scombridce. 87

total of 31 with the hypural. The parapophyses are not developed anteriorly. On the eleventh abdominal vertebra they first appear, but are here joined to each other to form a hsemal arch. Anterior to this on two or three vertebrae the edges of the socket into which the rib fits is slightly produced into a crater-like rim, but no developed process is present. On the first three hsemal arches no spine is present, but on the arch of the last abdominal vertebra the spine is long and similar to those on the anterior caudal vertebrae. Consequently aside from the fact that the abdominal vertebrae bear ribs, and the first caudal vertebrae bear the anal fin there is nothing to differentiate the abdominal vertebrae from the caudal. The inferior foramen is present in the base of the haemal spine of the last abdominal vertebra, and in most of the caudal vertebrae. The last two or three vertebrae are not abruptly shortened. On the anterior caudal and posterior abdominal vertebrae each antero-zygopophysis reaches forward and forms a semi-inclosed space behind and below each inferior vertebral foramen. The first two vertebrae bear epipleurals only; the first ribs being on the third vertebra. The ribs are flattened and are directed backward so that they lie close to each other. When the haemal arches develop, the ribs drop down to their tips and the bases of each pair of ribs are thus brought in contact. The epipleurals are attached directly to the vertebrae, and anteriorly their bases are in contact with those of the ribs, but posteriorly they hold their position at the bases of the haemal arches, while the ribs drop down to the haemal spines. The epipleurals are present back to opposite the posterior part of the anal fin. The intemeurals of the spinous dorsal are much expanded at the upper end, but the baseosts, one of which is in front of each spine, are so broad they extend laterally far over the interneurals, and form a bony buckler that is visible under the skin of the undissected specimen. A long baseost extends between each of the finlets behind the dorsal and the anal fin. The first interhaemal is not enlarged. The caudal rays are deeply divided and so fit over the hypural plate that they cover it more than half from sight.

The elements not mentioned here are as they normally are in the great majority of spiny-rayed fishes,

88 Edwin Chapin Starks.


A specimen of Rastrelliger braucliysomus, Jordan and Starks, 12 inches in length, from the Fiji Islands.

The cranium is less depressed than in Scomber, though it does not differ in the crests and ridges of the cranium from that genus. The epiotics appear to meet very broadly posteriorly, but close examination reveals a slender spur from the supraoccipital extending down between them to the exoccipital suture. The top of the craniuin in front of the oblique ridge that runs from the supraoccipital to the supraorbital rim is finely sculptured and thickened by a network of fine ridges where in Scomber the bone is smooth. The foramen magnum forms a long tunnel of the exoccipitals as in Scomber, and the condition of the exoccipitals over the basioccipital and their condyles is the same.

The mandible and maxillary elements are much weaker than in Scomber.' The premaxillary is a long slender bone from which the maxillary arches widely away, being attached to it only at each end; the auxiliary maxillary is small. The most striking difference between this genus and Scomber lies in the arrangement of the lateral bones of the skull and the basibranchials. The pterygoid normally (as in Scomber) is attached along the anterior edge of the quadrate, at the upper end of which it bends at an angle forward to support the palatine; the metapterygoid is behind and a little above the quadrate. In Kastrelliger the metapterygoid is abOve and somewhat in front of the quadrate, and the pterygoid borders the entire front of both the quadrate and metapterygoid turning at an angle at the upper edge of the latter.

The basibranchials form a high, sharp, knife-like ridge, while the hypobranchials are deep and compressed and help to elevate the basibranchials still higher. The second and third superior pharyngeals are joined into a single plate a little more firmly and completely than in Scomber. The branchial arches are crowded backwards against and between the shoulder girdles ; and in fact all of the bones of the head give the impression of having been drawn downward and backward and compressed.

There are 14 abdominal vertebrae and 16 caudal, or a total of 31

The Osteology of the Scomhriclse.


with the hypiiraL The vertebral elements are arranged as in Scomber, as are the other elements with the above exceptions.

SCOMBEROMOKUS. A speciiiien of Scomberomorus sierra, Jordan and Starks, 24 inches in length, from Mazatlan, Mexico, and tlie liead of a siiecimen of S. maculatus (Mitchill), 18 inches in length, from Chesapeal-ce Bay.

This genus differs from Scomber in having the supraoccipital crest carried forward by the ffontals to the ethmoid, and in having the cranium deeply concave for its full leng-th on each side of the crest. The temporal crests are directed straight forward as in Scomber, but they are not interrupted above the eyes by a transverse ridge. They nearly reach to the ethmoid, and anteriorly between them and the pterotic crests are developing small auxiliary crests as in Scomber, Imt situated farther forward. ' The pterotic crest extends forward to above the middle of the eye. The myodome opens posteriorly through a wide longitudinal foramen.

The supraoccipital crest extends down over the exoccipital suture more broadly than in any other genus except Acanthocybium, though it is not at all interposed between them. The exoccipitals broadly meet for their full length above the basioccipital. Their vertebral condyles are large and slope back over the basioccipital as usual. The pterygoid docs not end in a sharp spine posteriorly. The ethmoid is forked or concave in front to receive the premaxillary processes. The opisthotic is interposed between the exoccipital and pterotic, so that as much of it is on the superior surface of the cranium as on the inferior. The lower limb of the posttemporal is attached directly, or without the intervention of a ligament, to its superior surface rather than to its posterior edge.

The suborbital ring is incomplete; the preorbital is well developed and there are two small suborbital plates, the second developed as a small thin suborbital shelf. The rest of the ring is made up of the thick scales that cover the cheek, which are slightly turned inward at the border of the eye ; they carry no sensory tube. A small thin Y-shaped supratemporal bone is present bearing a branched sensory tube just under the skin. The nasals are wide bones attached for their full length to the side of the frontal and ethmoid, and do not project at all beyond the latter.

90 Edwin Cbapin Starks,

There is no process sent backward from tlie metapterygoid across the inner surface of the hyomandibular. There is no space left between the articular and dentary. The maxillary bears a small auxiliary maxillary as in Scomber. The short premaxillary processes project into the concavity in the front of the ethmoid. The teeth in the dentary and premaxillaries are large and laterally flattened ; the vomer and palatine bear small granular teeth.

The third and fourth superior pharyngeals are not nearly so closely attached to form a single plate as in Scomber, and the teeth on both the superior and inferior pharyngeals are smaller, stiffer, and less brush-like. The hypercoracoid foramen is large and through the center of the bone. The pelvic girdle is more slender and the vertical plate is but little developed.

The vertebrae number 19 abdominal, 29 caudal, or a total of 49 with the hypural. The condition of the parapophyses, zygopophyses, ribs, epipleurals, and interspinous bones is similar to those of Scomber. The haemal and neural processes are more slender, fragile, and fibrous than in any other genus of the Scombridse. The caudal rays lap over the hypural plate even farther than in Scomber, or until the bases of the opposite rays meet on the median line and hide the plate almost completely. The urostyle of the hypural is better developed.

Aside from these characters Scomberomorus is as described for Scomber.


A heart of Acaiitlioc-yhiniii solaiirtri. Cuvier anrt Valeiifiennes, iuchidiiig: the upper elements of the shoulrter girdle and the first three vertebrie, from the Hawaiian Islands.

In this genus the cranial crests extend straight forward as in Scomberomorus, not obliquely towards the suborbital margin as in the Thunninte and Sardinae; there is no auxiliary crest as in Scomberomorus. Both the temporal and supraoccipital crests stop, however, at the front of the eye, and the top of the cranium anterior to them is evenly rounded transversely unlike any other genus of its family. The supraoccipital covers the exoccipital suture rather broadly as in Scomberomorus. The preorbital part of the cranium is noticeably

The Osteology of the Scombridne. 91

produced. In Scomberomorus the brain chamber is open widely between the alisj)henoids, while in this genus the alisphenoids nearly meet at their middle and almost divide the opening into two parts. The vomer bears a patch of small teeth.

The nasals and suborbital ring are as in Scomberomorus, but the sclerotic case is thicker and denser than in any other genus. The teeth on the jaws are flat and saw-tooth-like with very sharp finely serrate edges. The anterior part of the premaxillaries project very much in front of the maxillaries to form a sharp beak, though only to a greater degree than in Scomberomorus ; the auxiliary maxillary is well developed.

The lateral head bones including the hyoid and branchial elements exhibit no departure from the other genera.

The first two vertebrae bear epipleurals only, the first rib being on the third vertebra. This rib bears on its side, some distance from its base, an epipleural (the succeeding vertebrae are missing). In this respect Acanthocybium differs from all of the other genera, as in the others all of the epipleurals are on the centra. The vertebrae are said to number 32 -|- 34 = 66.


A specimen of Sarda chileusis, Cuvier and Valenciennes, 30 inclies in length, from Puget Sound.

The cranium is broad and depressed with moderately high thin crests. The supraoccipital crest is carried forward by the frontals to the ethmoid. The temporal crest bends outward from the epiotic and reaches the supraorbital margin above the middle of the orbit. The pterotic crest forms the posterior lateral margin of the cranium. All of the crests are nearly identical with those of the Thunninse. The supraoccipital crest dips down a little farther to the exoccipitals than in Thunnus, but not so much as in Scomberomorus. It does not at all separate the exoccipitals and barely covers the suture between the posterior end of the epiotics. There is a very slight indication of the pit on the lower surface of the cranium that is so pronounced in the subfamily Thunninse, but it in no degree excludes the pterotic from the brain cavity. There are no openings into the brain cavity at the end of the frontals. The myodome opens directly

92 Edwin Chapin Starks.

to the exterior posteriorly tliroiigli an elongate foramen, at the anterior end of which is a slight cavity in the parasphenoid indicating the chamber that is developed at this place in the Thnnnin?e.

The exoccipitals meet broadly above the basioccipital and have very large condyles, which slope normally over the basioccipital ; they are each nearly as big as the basioccipital condyle. The condition of the opisthotic is as described for Scomberomorns. The vomer has an obliqne articular facet on each side for attachment to the maxillary. The prefrontal is a much swollen bone broadly meeting its opposite fellow at the median line behind the ethmoid; it is pierced l)y the olfactory nerve. The palatine is attached to the prefrontal and the ethmoid as described under Scomber. The ethmoid is a thick Avide bone forked or concave in front to receive the blunt premaxillary processes; its most anterior part is not at all posterior to the front of the vomer. The alisphenoids reach forward nearly to the prefrontals, and meet each other at their anterior ends above the opening to the brain case, but do not at all obstruct the opening.

The suborbital ring is incomplete. The nasals are thick, wide bones attached for their full length to the frontal and ethmoid and do not at all project beyond the latter. The symplectic from the outer surface of the skull appears to be long and cylindrical, but behind the metapterygoid and quadrate it spreads out to a broad triangular shape, and is attached to the inner surface of the metapterygoid by a deeply dentate suture. It extends downward in a channel in the quadrate. The auxiliary maxillary is well developed, and the premaxillary teeth are large and set in alveoli. The third and fourth superior pharyngeals have such a deep constriction between them on each side that they can no longer he said to fonn a single plate as in Scomber. The other head liones do not differ materially from those of the Thunninse.

The shoulder girdle is essentially as in Scomber, l)ut the pelvic girdle differs in having the subvertical plate below the horizontal plate turned outward, and a similar plate developed above the horizontal plate.

There are 45 vertebrse ; it is impossible to distinguish abdominal from caudal vertebrse in the specimen at hand, as the ribs and anal

The Osteoloffv of the Scombridse. 03


fin are detached and the two portions of the vertebral column are diiferentiated only by the attachment of these elements. The parapophyses are nowhere large, iind the first one large enough to consider occurs on the ninth or tenth vertebra, though there are three or four bony tubercles developed anterior to this. The parapophyses bend down and join to form haemal arches at about the middle of the abdominal cavity. The first two vertebrae bear epipleurals only. The anterior ribs fit in sockets in the vertebrae, but posteriorly where the parapophyses are developed the ribs are attached to their tips, so when the former unite to form haemal arches the bases of the opposite ribs are brought in contact. The ribs are all directed backwards so that they lie close together. The epipleurals are developed back to the caudal keel, and are all on the centra of the vertebrae, so are remote from the ribs where the ribs drop down to the tips of the parapophyses or hiemal spines. The zygopophyses are well developed along the greater part of the column, but toward the tail they become very small and neural and haemal processes take their place. These posterior spines are broad and flat, and each are laid firmly down over the next succeeding vertebra apparently'restricting the vertical motion of the tail. A wide keel is developed on the side of the caudal peduncle, and a urostyle is present on the hypural plate. The forked caudal rays nearly cover the hypural plate from sight. The baseosts are much expanded laterally at the base of the dorsal spines. The interspinous rays of the vertical fins are crowded together and have well developed lateral wings, but they are not connected to each other by bony wings.


A specimen of Thunnus alaluuga (Gmelin), 35 inches in lengtti, from San Diego, California, and a liead of a large specimen of Tlinnnus thynnns (Linnaeus), QYj inches long, from the tip of the vomer to the end of the basioccipital, from San Francisco.

In this foi-m the ethmoid is produced to a median angle in front, and the nasals project far anterior to it. The supraoccipital crest projects back in a spatulate process unattached below to the exoccipitals. The epiotics meet broadly, but a slender spur of bone sent down from the supraoccipital to the exoccipitals covers the suture

94 Edwin Chapin Starks.

between them. There is a deep pit or infolding of the bone on the lower surface of the cranium between the pterotic, the prootic, and the sphenotic, just inward from the condyle of the hyomandibular. It extends upward, so that the prootic is in contact with the epiotic on the inner surface of the cranium, and tlie pterotic is thus completely excluded from the brain chamber. There is a large, triangular, smooth opening into the brain chamber just behind the frontal on each side of the supraoccipital crest. The alisphenoids are joined together in such a way at the median line that the anterior opening into the brain chamber is divided into a small one in front of their union and a little larger one behind. The myodome opens through a long slit in the side of a very large conical chamber in the parasphenoid. This chamber in a cranium measuring 5I/2 inches in length is an inch long and nearly half an inch broad across its mouth. The opening in its side from the myodome does not extend quite to its tip. The vomer bears an elongate patch of teeth. The suborbital bones are similar to those of Scomberomorus, but the cheek scales that border the eye are much thicker.

The vertebrae number 40 with the hypural plate ; in this specimen as in our specimen of Sarda the ribs and anal fins are detached and it is impossible to distinguish the caudal from the abdominal vertebrae without them. On the fifth vertebra the first parapophysis appears ; on the next three it has increased very rapidly in size, and extends straight out laterally; on the ninth vertebra its ends bend abruptly downward, and on the tenth it has united with its opposite fellow to form a broad round haemal arch without a spine. On the eleventh vertebra a hsemal spine appears and the bases of the opposite ribs are brought in contact with each other as in the other members of its family. The inferior postero and antero-zygopophyses are equally developed, arching towards each other each as a sharp spur. There is no inferior foramina in the base of the haemal arches. The superior zygopophyses, the ribs and epipleurals, and the other elements not mentioned are as in Sarda.

The skull of Thunnus thynnus differs from that of T. alalunga in having the parasphenoid developed upward in a broad wing to meet the descending wing from the basiphenoid, and a descending wing developed from the union of the alisphenoids.

Tlie Osteology of the Scombridse.


EUTHYNNUS. A specimen of Eutbynnus pelamis (Linnaeus), 19 inches in length, from Japan.

In this form the cranial crests, the condition of the ethmoid and nasals, the chamber in the parasphenoid into which the myodome opens are all nearly identical with these characters in Thunnns and Anxis. It differs in having the alisphenoids separated, and the vomer toothless. The large pit behind the prootic has nearly broken through the top of the craninm where the bone is very thin and





•J f



A. Front view of 15th abdominal vertebra.

B. Lateral view of 11th and 12th abdominal vertebriB.

C. Lateral view of 15th and IGth abdominal vertebrae.

D. Lateral view of 3d and 4tb caudal vertebrae.

E. Lateral view of 8th and 9th caudal vertebra?. iz., inferior zygopophysis.

if., inferior foramen.

h. c hipmal canal. ^

p., parapopbysis.

h., hremal arch.

1'., rib.

pierced by several small holes. On top of the craninm jnst behind the frontal at each side of the snpraoccipital crest, where in Thunnns is a large smooth opening to the brain cavity, the bone is irregularly broken through ; the opening being much wider on one side of the cranium than on the other in the specimen at hand. There is no infolding of the bone between the prootic and alisphenoid, and there

96 Edwin Cliapin Starks,

is no opening into the myoclome along the posterior edge of the lateral process of the parasphenoid as in Anxis.

The greatest difference between Euthynnns and Thnnnus lies in the condition of the inferior vertebral processes. There are 20 abdominal vertebrae, and a like nnmber of candal, or a total of 41 with the hypnral, the abdominal and caudal regions being more evenly divided than in Auxis. The lower processes of the vertebrse differ from those of Auxis in that the inferior foramen that is typically through the base of the hsemal arch is here enormously developed, while the haemal arches themselves have developed to even a greater degree. The postero and antero-zygopophyses share equally in forming long slender processes between the haemal arches. The arches spring almost directly from the body of the vertebrae, and the largest are as wide as the length of a vertebra and over twice that long. The longest diameter of the largest of the inferior foramina is considerably greater than the length of a vertebra. The first parapophysis appears on the eighth vertebra, and is scarcely developed, but the succeeding ones quickly attain a great length. Only four pairs are developed before they unite in haemal arches. The first six or seven haemal arches are broadly rounded at the lower median line, but posteriorly a haemal spine is developed.

The ribs and epipleurals and their attachment to the vertebrae are as in Auxis. The caudal keel is as wide, but is not developed so far forward. The other vertebral processes, and all of the other elements not here mentioned are essentially as in Auxis.

AUXIS. A specimen of Auxis tliazard (Lacepede), from Japan, 9 inches in length.

The supraoccipital crest is as in Thnnnus or Sarda, but the temporal crests run less obliquely, and merge into the general level of the supraorbital rim without reaching the orbital edge. The pterotic crest runs more on top of the cranium and forms less of the lateral cranial outline. There is a well marked depression running along each side of the supraoccipital crest from the front of the frontal to the epiotic as in Sarda and Thnnnus. The myodome opens posteriorly in a very large round opening where the basioccipital and parasphenoid

The Osteology of the Scombridse. 97

are expanded to accommodate it, though there is no well separated parasphenoid chamber as in Euthynnus and Thunnus. Between the prootic and alisphenoid there is a narrow infolding of the bone to form a deep groove.

The supraoccipital widely separates the parietals, but does not altogether separate the epiotics, which meet broadly behind it. There is no opening into the brain chamber behind each frontal, and the pit behind the prootic shows no tendency to break through the top of the cranium. The lateral process of the parasphenoid attaches to the prootic above but along its posterior edge is an opening into the myodome.

The auxiliary maxillary is as broad as the maxillary and nearly half as long. The teeth on the jaws are very small; there are none on the vomer or palatines. The third and fourth suprapharyngeals are covered with moderate sized teeth, and they are a little less closely attached to each other than in Scomber. On the second pharyngeal the teeth are very small and set in a small patch, which is scarcely differentiated from the other dentiferous plates that cover the inner surface of the arches. The other characters of the skull, and shoulder and pelvic girdles are essentially as in Euthynnus.

There are 22 abdominal vertebrse and 15 caudal, or a total of 38 with the hypural. The abdominal region is very much longer than in Scomber, and a most extraordinary modification has taken place in some of the inferior processes of the vertebrae. This condition will best be appreciated if studied from behind forward. On the fourth caudal vertebra the processes are normal ; a hsemal arch of moderate size springs directly from the lower surface of the vertebra, and terminates in a long hsemal spine. From the front of the arch the antero-zygopophysis meets the postero-zygopophysis of the preceding vertebra; each forming half of a sharp spur. On the next vertebra the antero-zygopophysis and the base of the arch have enlarged into a single, solid, bony pedicle carrying the haemal arch away from the body of the vertebra. The pedicle gradually increases in length until at about the beginning of the posterior third of the abdominal cavity the haemal arch is distant from the body of the vertebra a distance equal to twice the length of a vertebra. The haemal arch is nowhere


Edwin Cliapin Starks.

enlarged but is in the form of a small ovate foramen. The spur of the zygopophysis is carried out with the haemal arch and projects in front of it as it does when in the normal position on the surface of the vertebra. The spur disappears on about the third from the last abdominal vertebra, and in front of this the haemal arch is open above so that the tip of the process is forked; each fork being now






Fig. 2. — Vebtebkae of Auxis.

A. Lateral view of 17th abdominal vertebra. A' Front view of the same.

B. Lateral view from the 1st to the 5th caudal vertebrae. B'. Front view of the 1st caudal vertebra.

B". Front view of the fifth caudal vertebra.

iz., inferior zygopophysis.

p., parapophysis.

r., rib.

he, hfemal canal.

the homolog of a parapophysis. The process or pedicle holds its length to about the middle of the abdominal cavity, anterior to which it gradually grows shorter and disappears on the eighth abdominal vertebra. There are no parapophyses present that are not anchylosed with their fellows of the opposite side at the base. When the antero

The Osteology of the Scombridse. 99

zygopophysis begins to enlarge the postero slightly enlarges to meet it, but it at once begins to grow smaller again and does not help to form the pedicle. The superior postero-zygopophyses are moderately developed, but the antero-zygopophyses are very large and reach far over them.

The ribs, epipleurals, and other elements of the trunk are as described for Euthynnus.


e, ethmoid. eo, exoccipital. ep, epiotic. fr, frontal, n, nasal. op, opisthotic. p, parietal, p/, prefrontal. pt, pterygoid. soc, superoccipital. S'p, sphenotic. V, vomer.




(Errata. Paste this sheet to page 100, facing Plate I, Journal of Morphology, Vol. 21, No. 1, March, 1910.1




Thii Jouenal of Morphology. — Vol. 21.



t) e


eo op

Thh Journal of Morphology. — Vol. 21.



Thg Jouexal of Moephologt. — Vol. 21.



I. History of the Eakly Cleavage and of the Accessory


J. THOMAS PATTERSON. With 32 Figures.

I. Introduction.

The period extending from ovulation to the laying of the egg is a most obvious gap in our knowledge of the development of the hen's egg. It has been the writer's desire to fill in this break, and he is indebted to the trustees of the Elizabeth Thompson Science Eund" for a gTant which made it possible to undertake the work. If the problem contained no possibilities other than that of merely filling in a gap, it is doubtful whether the work would have been undertaken, since the results could not have been commensurate with the labor involved. But it was felt that certain points, brought out in a study of the pigeon's egg by several students at the University of Chicago (Harper, '04; Blount, '09; Patterson, '09), deserved further investigation. Among these were fertilization, accessory cleavage, and gastrulation.

On account of the importance centering in gastrulation and the accessory cleavage, their discovery in the hen's egg would be of the greatest interest ; for a true gastrulation has never been found in this egg, and the accessory cleavage has been neither figured nor described. We have not even known whether fertilization in the hen's egg is monospermic or polyspermic.

^Contribution from the Zoological Laboratory of the University of Texas, No. 103.

The Journal of MoKPHOLOGy. — Vol. 21, No. 1.

102 J. Thomas Patterson,

It is not necessary to enter into an extensive discussion of the literature on the subject of the early development of the hen's egg, for the several papers touching on this subject are well known. The studies of Duval, '84, have, perhaps, received more attention than those of any other investigator, and yet it has been demonstrated that his fundamental conclusions are incorrect, and that he was probably misled in his interpretations through the use of pathological material (Kionka, '94; Barfurth, '95; Schauinsland, '99; Patterson, '09). Kionka, '94, although figuring stages throughout the greater part of the period to be considered in these studies, does not give us a good idea of the character of the very early cleavages. I^either of these workers, nor any one of the others who have investigated these stages, has had anything like a complete series from which to draw his conclusions; consequently it is not surprising to find that the majority of the interpretations do not accord with the principles of vertebrate development, and that the more fundamental points are obscure.

The recent discovery by Guyer, -'09, of an "accessory chromosome" in the male germ cells of the chicken lends unusual interest to the study of fertilization in the hen's egg, for it ought to be possible to demonstrate from the study of the mitoses of the supernumerary sperm nuclei whether or not such nuclei are dimorphic.

It was the writer's intention at first to publish the entire history, from ovulation to laying, in a single paper, but the slow rate at which material naturally accumulates makes it desirable to publish the part already completed; the remaining parts, one on maturation and fertilization, and the other on late cleavage and gastrulation, will appear later.

II. Methods.

Since the methods employed are essentially the same as those used in handling the pigeon egg, they need be mentioned but briefly here. The picro-sulphuric-acetic mixtures, which were found to be so excellent for fixing the pigeon egg, do not work well, for they render the yolk too hard. The picro-acetic fluid, however, although not entirely satisfactory, gives fairly good results. Por preparing

Early Development of the Hen's Egg. 103

surface views, a weak solution of Flemming's chromo-osmic acid is excellent and has but one disadvantage, viz., that after its use the egg usually can not be sectioned.

III. Some IN'otes o:^ the Laying Habits of the Hen.

The behavior of the hen during the breeding season would make an interesting topic for the student of animal behavior; for while one sees many evidences suggesting that domestication has wonderfully influenced the behavior of the hen, yet there are continually cropping out certain habits that evidently have been derived from her wild ancestors, and which even centuries of domestication have not completely eradicated. One of the most noticeable of these is seen in connection with the nest building. The hen never carries building material to the nest, but she often stands in the vicinity of the proposed site and makes a futile effort to get straws and feathers into the nest by tossing them over her back. In several of the other Gallinse this same habit is observed. Many species of this group of birds are accustomed to building their nests in tufts of grass, where an abundance of material is ready at hand, and its building is a comparatively simple matter, consisting in the arrangement of the grass. Occasionally, however, other birds (e. g., the quail) will engage in exactly the same futile effort as that cited above for the hen, only in a more pronounced manner.

The writer's study of the habits of the hen was not carried on with any intention of winting a paper on its behavior, but rather in order to find out if there is any regularity in its laying habits. If one is to collect eggs for the purpose of obtaining a close series of stages, it is of the greatest importance to be able to tell just when to kill the hen in order to secure a desired stage. It is only in this way that one can hope to obtain sufficient data for a correct interpretation of the history of development.

It is commonly supposed that the hen lays very irregularly, and while the writer finds this to be true for some few hens, yet in most cases he was soon able to predict the time of laying to within a few minutes. This is especially true of that class of hens laying daily. Such hens are found to lay slightly later each day, and


J. Thomas Patterson.

ttie difference between any two succeeding days is sometimes exactly

one hour (hen 1).

Hen 1.

Laid April 1, 8 :30 a.m.

" 2, 9:30 "

" 3, 10:30 "

" 4, 11:30 "

" 5, 12:30 "

When a hen is not laying daily the matter of determining the time is not so simple, and yet even here one can approximately predict the exact time of the laying, as can be demonstrated in the

following case:

Hen 2.

Laid June 12, 2 :00 p.m.

" 14, 10:00 A.M.

" 15, 2:00 P.M.

" 17, 10:00 A.M.

It is evident from these data that the hen was laying at 10 :00 a.m. and 2 :00 p.m. on succeeding days and then was missing a day. It was, therefore, predicted that she would lay early in the afternoon of the 18th, and since an early cleavage stage was desired, the hen was killed at 4:00 p.m. on the 17th. The stage secured is shown in Fig. 15.

There are some hens that apparently do not lay at any regular intervals, and in such it is quite impossible to predict the time of laying. As an example, I may cite the following case:

Hen 3.

Laid July 25, 11:00 a.m. " 28, 11:00 " 30, 11:45 " Aug. 1, 11:00

6, 9:30

7, 3:00 p.m. 9, 1:00

" 11, 12:30

" 13, 2:00

" 16, 1:30

" 18, 1:00

Killed August 19, 5:00 p.m. (secured an early cleavage stage).

Earlj Development of the Hen's Egg. 105

There are but few hens that ever lay before 8 :00 a.m. or after 4 P.M. It is evident, therefore, that a hen in laying daily will eventually come to the 4 o'clock period, and will then miss a day (sometimes more) before beginning a new set. Indeed, this is true for all hens whether laying regularly or irregularly, for they lay in a sort of rhythm. In the case of irregular laying, cited above (hen 3), the eggs laid from the 25th of July (when the observations were begun) to the 1st of August, are the last of a set in which the hen was laying approximately every other day; while those laid from August 6th to 18th constitute another set. Evidence that the eggs will be laid in sets can be obtained by an examination of the ovary, which shows several graduate series of ovarian eggs.

It will be evident from the above considerations and data that if one is to secure a close series, it is necessary to study each hen individually and while this involves a great amount of labor, yet it is the only way in which one is able to meet with any success.

The collecting of the above data has another advantage besides that of aiding in securing a close series, for it makes possible the. determination of the rate of development of the different stages. The time occupied by the egg in passing down the oviduct has been variously estimated at from eighteen to twenty-four hours, and even •as high as thirty-six hours.- This seems like a wide variation, and in taking up this work, the writer was prepared to find the normal time more constant than is indicated in such estimates.

The writer finds that in a hen kept under normal conditions, the egg traverses the entire length of the oviduct in about twenty-two hours. The time occupied in the different portions of the oviduct is as follows : Glandular portion, three hours ; isthmus, two to three hours; uterus and laying, sixteen to seventeen hours.

As just stated, these estimates were made on hens kept under normal conditions ; that is, hens that were given the freedom of the barnyard. It is possible to lengthen the time beyond the twentytwo hours by disturbing the hen when she is about to lay, and on one occasion the writer was able to delay the laying of the egg for twenty hours. When it was finally deposited an examination revealed the fact that it was in a stage of development equal to about twenty hours of incubation. This would account for the high esti


J. Thomas Patterson.

mates of other writers, and perhaps also for those cases reported in the literature, where a freshly laid egg is said to contain an embryo with a well-developed vascular system. The writer is convinced that any appreciable extension beyond twenty-two hours is not due to an increase in the length of time that it takes the egg to traverse the reproductive passage, but rather to a rentention of the egg in the lower part of the oviduct, on account of some influence inhibitory to laying. The writer has found, however, slight variations in the length of time, but these are probably to be correlated with the variations in length of the oviduct in different hens.

When once it has been established that the time occupied by the egg in its passage through the different parts of the oviduct is practically constant, we are then in a position to determine the rate of development; because we need only to note the stage of development in eggs taken from the different parts, and from the data thus collected determine the time elapsing between any two stages. The following table will give the reader an idea of these estimates, which were determined from a study of hens laying daily (about one hour later each day). The table also gives the exact time of each of the stages described in the rest of the paper.




Last egg laid.

Succeeding taken from the oviduct.


Position in oviduct



No. I

1:30 A. M. May 28

4:00 P.M.. May 28


11 inches from the infundibulum



No. 2

8:30 A.M., Aug. 10

2;30 P.M., Aug. 10

3 "

just entering the isthmus



No. 3

8:30 A.M., Apr. 13

2:30 P. M., Apr. 13

31 "

in the isthmus



No. 4

10:00 A.M. Sept. 1

5:00 P.M. Sept. 1

4 "

(1 tt (1



No. s

11:00 A.M., .\ug. 4

6:30 P.M., Aug. 4

4i "

ti t» t(



No. 6

12:30 P.M., July 31

8:15 P.M., July 31

4i "

(1(1 (1

thirty-two celled

No. 7

8:45 A.M., .July 30

4:45 P.M., July 30

5i "

in shell gland

64 cells in surface view.


No. 8

9:30 A.M., Aug. 7

7:30 P.M., Aug. 7

7 "

i( i( (1

154 cells in surface view.



8:00 A.M., Aug. 27

7:00 P. M., Aug. 27

8 "

11 t( «•

346 cells in surface view.


Early Development of the Hen's Egg. 107

IV. Functions of the Oviduct.

The whole reproductive apparatus is a most delicately adjusted mechanism. The primary function of its oviducal portion is to transmit the egg from the body cavity to the exterior, but in the course of its evolution it has taken up several other functions, such as transmitting and storing sperms and the secreting of the accessory layers around the egg. The co-ordination between the infundibulum and the ovary is often very exact and delicate, but it is in the birds that we see this co-ordination reaching its highest degree of perfection. Coste describes the infundibulum as actually embracing the ovum in its follicle at the time of ovulation, and the writer has been able to confirm his statement by several observations. Coste believed that the infundibulum exerted some pressure on the follicle, and it may be that this is the direct cause of ovulation. Indeed, it is highly probable; for while ovulation may take place without the direct assistance of the oviduct, as in the lower vertebrates, yet the weight of evidence supports the opposite view. We have been able to show (in a paper not yet published) that the follicular orientation is preserved in the oviduct, and furthermore that this preservation probably occurs only when ovulation is directly caused by the activities of the infundibulum.

This explanation of ovulation also gives us the key to the solution of another problem, viz., why it is that normally but a single egg is found in the oviduct at a time. If we examine the oviduct of a hen that is laying daily, some time before the deposition of an effS, it will be found to be inactive ; but an examination shortly after laying, reveals the fact that the oviduct is in a state ol high excitability, with the infundibulum usually clasping an ovum in the follicle. In one case it was embracing a follicle containing a half-developed ovum, and with such tenacity that a considerable pull was necessary to disengage it. It seems certain, therefore, that the stimulus which sets off the mechanism for ovulation is not received until the time of laying, or shortly thereafter. So long as there is an egg in the lower part of the reproductive passage the infundibulum apparently does not clasp the ovum, and a second egg is thus prevented from entering the oviduct.

108 J. Thomas Patterson.

V. Feetilization.

Since it is intended to describe in detail the process of fertilization, the writer wishes here to make only a brief statement concerning the time of its occurrence. Harper, '04, has shown that fertilization in the pigeon's egg takes place immediately after ovulation, when the egg is in the region of the infundibuliim. The writer finds the same to be true of the hen's egg also. Eggs taken from the upper part of the oviduct at distances varying from one to twelve inches from the infundibulum, give all the stages of development from maturation to the formation of the first cleavage spindle.

We have shown above that the egg is about twenty-two hours in passing down the oviduct, and since fertilization takes place immediately after ovulation, it is obvious that it occurs approximately twenty-two hours before the time of laying. Throughout this paper we shall, therefore, determine the "age" of any particular stage from this estimated time of fertilization.

VI. The Two-celled Stage, — Three Houes.

The first cleavage furrow makes its appearance just as the egg is entering the isthmus, about three hours after the estimated time of fertilization, and by the time the inner-shell membrane can be recogiiized as an extremely thin sheet covering the albumin, the furrow is well developed and covers a distance equal to about onethird the diameter of the area of primary cleavage. The furrow is usually situated in the central part of the disc (Fig. 1).

Any mention of the first cleavage furrow calls to mind the question of the relation of the plane of this furrow to the longitudinal axis of the later embryo. Of the five cases so far obtained, in which it was possible to determine absolutely the plane of the first furrow, only one showed it coinciding with the long axis of the future embryo. This would seem to indicate clearly that the plane of the first furrow does not necessarily parallel the median axis of the embryo. It may be, however, that in the case of each of the eggs mentioned, we were dealing with one in which the axis of the later embryo would be abnormal; that is, it would not meet the chalazal axis at right angles. Duval, '84, has pointed out that a

Early Development of the Hen's Egg. 109

certain number of ben's eggs sbow abnormal relations existing between the two axes, and the writer has found a similar condition in the pigeon egg. In each of these species the percentage of abnormal axes was found to be small. It seems highly improbable, therefore, that four out of five eggs taken at random, and in the twocelled stage, would later show abnormal axes. The final answer to the question, however, could only be obtained by studying a twocelled stage and noting the plane of the furrow, and then after incubating the egg until the axis of the embryo became visible, it would be possible to determine the point in question. This would necessitate a much more extensive study than the object of this paper calls for. The problem, furthermore, has lost much of its earlier significance, inasmuch as it has not proved to be a fundamental law of development.

In sections taken transverse to the first cleavage furrow (Fig. 2) the membrane is seen to cut almost through the fine granular portion of the disc, and is peculiar in that it arises from a membrane plate, which, at both sides, is continuous with the perivitelline space. In section the membrane does not extend down from the membrane plate as a straight line, but is wavy (Fig. 3) ; and while this condition may be the result of unequal contraction of the materials, caused by the fixing and hardening fluids, yet it obtains for all of the earlier membranes.

Another point of interest brought out by the section (Fig. 2) is the depression in the surface of the disc just above the cleavage membrane. This is, of course, the cleavage furrow, which in this egg stood out with remarkable clearness in the living condition, but in most eggs it is practically wanting (Fig. 11). The lack of a furrow is the cause of the indistinctness of the early cells. In this respect, the early cleavages of the hen's egg differ greatly from those of the pigeon's egg, for in the latter their clearness is such as to permit photographing the living cells, while in the former, photographs are impossible, except in a few cases.

After the division of the first cleavage nucleus the daughter nuclei migrate peripherally in a line lying at right angles to the cleavage membrane, and are always elongated in the direction of motion. In

no J. Thomas Patterson.

this egg they showed signs of preparation for the next division when they had reached a distance from the membrane equal to 0.175 mm. Polyspei^my. A close study of surface views of two-celled stages has failed to reveal any trace of the "accessory cleavage," which is such a characteristic morphological feature of the early pigeon blastoderm. It would be a great mistake, however, to conclude from this that fertilization in the hen's egg was monospermic, for a study of the sections brings to light the fact that ordinarily five or six supernumerary sperm nuclei are in the egg, and, as we shall see later, some of these may migrate to the periphery of the area of primary cleavage and there give rise to a rudimentary accessory cleavage.

In one egg (Fig. 5), which is in a j^recleavage stage of development, twenty-four extra sperm nuclei are fouud. This high number is very unusual, and led the writer at first to assume that the egg was abnormal. All the evidence, however, is against this assumption. In the first place, it can not be said that the physiological condition of the egg was such as to favor the multiplication of the sperm nuclei soon after their entrance into the egg; for if this were the case there ought to be evidences of nuclear multiplication, but not a single sperm nucleus gave any sign of undergoing division. .In the second place, the egg in all probability was normal in so far as undergoing normal development is concerned, for the cleavage nucleus was in the act of producing a spindle at the time when the egg was fixed, and other eggs from this hen underwent normal development when incubated.

In the light of these facts it seems evident that in some few eggs a comparatively large number of sperms may enter. In such eggs this may be due to a greater attraction between the protoplasm of the disc and that of the sperms than ordinarily exists; or to a failure of the inhibitory agencies to become operative quickly enough after ovulation to prevent their entrance.

In the egg from which Fig. 1 was made only five supernumerary sperm nuclei are present, and none of these had reached the margin of the disc at the time when the egg was fixed, but all are located centrally. Two of the nuclei are situated quite superficially in the disc, while three have passed down deep and are resting on the coarse granular yolk (Fig. 9). One of the nuclei has undergone division

Early Development of the Ilen^s Egg. Ill

twice, and produced a "nest" of four nuclei (Fig. G, sn). It is difficult to determine whether or not such nuclei later migrate from the nests to the margin of the primary area and there participate in the production of the accessory cleavage. The writer believes not, because nests containing many small fragments of nuclei are found in slightly later stages, and such nests are located in the same position as the earlier ones. This would seem to indicate that the nuclei, after sinking down into the coarse granules, continue to divide and fragment, finally disappearing altogether. If this be true, then we see the beginning of the degenerative ag(mcy which will cause all of the supernumerary nuclei to disappear.

An egg showing a case of fragmenting nuclei is outlined in Fig. 10. It is a three-celled stage, and the degenerating nucleus is close to the margin of the primary area. All of the sperm nuclei, excepting one, are located much more peripherally than in the preceding egg (cf., Figs. 9 and 10). The difference in position of the two sets indicates the distance traversed by the nuclei during the time intervening between the two stages.

A section of the egg figured above gives one a good idea of the character of the blastodisc in the early stage (Fig. 11). It also demonstrates the point made above, that very often the first cleavage membranes are not accompanied by a cleavage furrow.

This egg furnishes still other points of interest, for in fixing it, not all of the albumin was removed, and the thin chalazipherous layer adheres to the vitelline membrane. Embedded in this layer and next to the membrane are about ninety sperm heads, none of which is located peripherally to the terminal ends of the cleavage membranes (Fig. 8).

The presence of the sperm heads in such large numbers, together with their position at the central part of the disc, is important. It can not be said that the scarcity of sperm nuclei in the disc (as compared with the number found in the pigeon egg) is to be accounted for by the "lack of sperms in the oviduct. Their location in the central part of the disc only, shows that they must be attracted there by a force which is strongest at the central point, and which gradually diminishes toward the periphery. The attraction between the protoplasm of the disc and that of the sperms is evidently neu

112 J. Thomas Patterson.

tralized suddenly, because sperms are found lodged in the vitelline membrane, as though they had been stopped in the very act of entering the egg (Fig. 7).

VII. The Four-celled Stage.- — Three and One-fourth Hours.

The four-celled stage is produced by a vertical division in each of the blastomeres of the tw^o-celled stage, and the two furrows meet the first one approximately at right angles (Fig. 12). While the division in one blastomere may slightly precede that of the other, usually they occur simultaneously. It has been stated that the point where the second furrows meet the first is situated eccentrically, the displacement being toward the posterior border of the blastoderm. It is not uncommon to find eggs with the center of the cleavage eccentric, but the displacement may be in any direction from the center. The writer does not believe, therefore, that any importance can be attached to the eccentricity of cleavage.

The rudimentary accessory cleavage makes its aj^jDearance in the four-celled stage, and in the egg shown in Fig. 12 there are three such furrows present. These cut across the margin of the area of primary cleavage, and their planes are approximately radial. In most eggs the furrows lie entirely without the margin. This blastodisc shows, in addition to the three accessory furrows, two other small ones lying well within the margin, but their position in the anterior blastomeres makes it clear that they are the approaching divisions of these two cells.

VIII. The Eight-celled Stage. — Four Hours. In the formation of the eight-celled stage, the third division furrows, at least in some cases (Fig. 16), tend to remain regular; that is, each of the third furrows is vertical and meets the second furrow at right angles. There is thus produced two parallel rows of four cells each. In the majority of eggs, however, the form of the cleavage in this stage apparently is not regular, though probably if it were possible to follow the divisions in the living cells it would be found that there was considerable regularity. The cells do not always divide simultaneously, and since there is a tendency for the

Early Development of the lien's Egg. 113

early blastomeres to flow together, it may be that the original relationship between the cleavage planes is modified. If this is not the case, then the variation in the form of cleavage which characterizes the later stages is anticipated in the eight-celled stage.

As in the two- and four-, the cells of the eight-celled stage are not true cells, in the sense that they are not completely delimited by cell walls, but are open to the periblast both below and peripherally. Occasionally, however, one of the blastomeres may be surrounded (in surface view) by cell walls (Fig. 13). We have, therefore, two regions of cleavage, in which the cells are usually designated as central and marginal.

The connections between both the central and marginal cells with the periblast are very clearly brought out in the section (Fig. 18), which also gives one a clear idea of the beginning of the horizontal cleavage planes. These planes not only separate the blastomeres from the underlying or central periblast, but also mark the position of the future segmentation cavity. Such an interpretation for the origin of the cleavage cavity is not in accord with the account of Duval and others. Duval contends that a very narrow space situated between a single superficial layer of cells and the deeper cells represents the segmentation cavity, and, furthermore, that the deeper cells are derivatives of the deeper parts of the disc, and have arisen by additions upward to the parts already segmented. Duval's view has been shown Vo be untenable for the pigeon's egg, and I find it is also incorrect for the hen's egg; but exactly the same thing occurs in the latter egg as described for the pigeon's egg by Miss Blount. The increase in the number of cell layers in the disc is not brought about by the addition upward of cells from the underlying material, but by the appearance of horizontal cleavages, which occur between the segmentation cavity and the surface of the blastodisc and thus the large central cells are cut up into a number of cell layers.

The accessory cleavage remains distinct up to the eight-celled stage (Figs. 14, 16, 13), and in the reconstruction of a series from the seven-celled stage there are shown ten supernumerary sperm nuclei (Fig. 14). Seven of the nuclei can be arranged into three groups, indicating that originally but five sperms entered the egg and not ten. Four of the nuclei are in the last stages of degenera

114 J. Thomas Patterson.

tioiij while two are associated with a rudimeutary accessory cleavage furrow. The furrow was clearly visible in the living egg, where it appeared as a shallow groove. In section it is characterized by having a broad shallow depression, at the bottom of which is found a distinct membrane plate of exactly the same appearance as that of an early primary cleavage furrow, and differs from the latter only in the absence of a membrane (Fig. 19, a. f.). It is, therefore, rudimentary, and the presence of two supernumerary sperm nuclei in its immediate vicinity leaves no doubt as to the interpretation \ that it is an accessory furrow. One of the nuclei (Fig. 19, s. p.,

nucleus on right) has passed down into the coarser granular yolk and undergone almost complete fragmentation, and this may be the reason the furrow never proceeds to the point of forming a membrane.

In addition to the rudimentary cleavage this egg also presents an interesting case of horizontal accessory cleavage (Fig. 4, c). In reality this is not a cell division, because there is associated with the cleft but a single nucleus, which lies just below the cleft. Such cases would seem to be attempts at cell formation without the accompanying nuclear division. They are not of any great importance, as but two examples have been found.

So far the writer has not observed the accessory cleavage after the ten-celled stage, and in all probability it completely disappears shortly after this period. It seems highly jDrobable that many of the sperm nuclei degenerate soon after their entrance into the egg. The degeneration occurs when they pass down into the coarse granular yolk; but so long as they remain superficially situated they seem to possess the power of migration. Undoubtedly, there are not a few eggs in which all of these nuclei degenerate before reaching the margin, and hence such eggs would at no time show an accessory cleavage. The writer does not believe, however, that this would account for the failure of previous investigators to discover the accessory cleavage of the hen's egg. Only a few of them have described the four-celled stage, and, so far as we are aware, none have figured the eight. The brief period during which they would most likely observe this peculiar form of cleavage is, therefore, the one to which they have given least attention. Furthermore, the acces

Earlj Development of tlie Hen's Egg. 115

scry cleavage, except in a few cases, is extremely difficult to detect in the living egg; and it is only after the use of Flemming's fluid that it becomes clearly demonstrable. Its difficulty of demonstration is further increased by the fact that the furrows most often occur in radial planes, thus leading the observer to believe that they are only the terminal ends of the primary cleavage furrows. We have been unable to find a satisfactory explanation as to why the clefts take this radial direction, unless -it be that they follow the course of least resistance. I^ot all of the accessory cleavages, however, have their furrows lying in radial planes, for the writer has found several cases in which they were lying in various planes; and, furthermore, he has observed two very clear cases of completely formed accessory cleavage cells (Figs. 13 and 17).

The contention of Blount that the accessory cleavages in the pigeon's egg completely disappear and, therefore, these cells take no part in the formation of the embryo, receives a full confirmation from these studies. The demonstration of that conclusion is even clearer in the hen's egg than in the pigeon's egg. In the hen's egg the marginal cells never become closed, thus indicating that their closing in the pigeon's egg must be in response to a stimulus received from the numerous accessory cleavage cells. Probably the closed margin cuts off influences which emanate from the accessory cells, and which might interfere with the normal development of the blastoderm.

At best the accessory cleavage in the hen's egg is but a weak attempt at cell formation, which has become inhibited, shortly after its initiation, by the degenerative tendency of the accompanying nuclei.

IX. Fifteen to Seventeen Cells. — ^Foue and One-half Houks.

In the stage consisting of approximately sixteen cells there are four or five central and eleven or twelve marginal ones (Figs. 17 and 20). The central cells increase by the cutting off of the inner ends of the marginal cells, while the latter multiply by the formation of radial furrows. In some cases one can still see a tendency in the form of the cleavage to remain regular. In Fig. 20 the anterior

IIG 'T. Thomas Patterson.

half shows exact regularity, there being just eight cells, but the jDosterior half is slightly irregular, with nine cells.

There are no accessory cleavages in this stage, but a large number of short radial furrows lie just inside the margin of the primary area (Fig. 20). These furrows can be seen in the living egg, but are brought out more clearly in osmic acid preparations. Under the higher power of the microscope they are seen to be due not so much to the formation of furrows as to the arrangement of the granules. In the median lines they are composed of fine granules, while to either side are several rows of larger granules. Occasionally one can detect a membrane, which appears as a delicate thread running through the median streak of fine granules.

At first sight it might seem that these furrows represent an abundant accessory cleavage, but such is not the case. In the first place, the furrows are situated entirely within the primary area, while the accessory furrows lie without the area. In the second place, there are no nuclei directly associated with these short furrows. They are simply the beginnings of the peripheral extensions of the primary cleavage furrows ; for it will be noted that for the most part they lie either in the same radial planes with the primary furrows (marginal), or in j^ositions where the next divisions of the marginal cells will soon occur. This interpretation is fully substantiated when slightly later stages were studied, when it was found that not only are the short furrows greatly diminished in numbers, but that the primary furrows now reach the margin of the primary area (Fig. 15), and at the same time the marginal cells are practically double in numbers.

These short furrows do not occur in all blastoderms, but when they do appear it is at the sixteen-celled stage, though they may last until comparatively late cleavage stages (Fig. 30). In significance these furrows indicate that the cytoplasmic division of the marginal cells is felt at the margin of the primary area earlier than in regions lying somewhat more centrally; and the attempted division at the margin probably immediately follows that of the marginal cell nucleus, while the intervening space awaits the approach of the central portion of the primary cleavage membranes.

Early Development of the Hen's Egg. 117

In median sections of the fifteen-celled stage the horizontal cleavages have progressed to the point of effecting a complete separation of the central cells from the central periblast (Fig. 22), and these cells, are therefore, completely delimited by cell walls. The segmentation cavity is expanding laterally beneath the marginal cells by the extension of the horizontal clefts (Fig. 22, on left), and v^ill eventually increase in depth by the accumulation of fluid within it.

X. Thirty-two to Thirty-five Cells. — Four and Threefourths Hours.

In this stage (Fig. 23) the central cells have begun to multiply more rapidly than those of the margin, and the two kinds are now of equal numbers. The form of the cleavage is irregular, and, therefore, very variable in different eggs. At the posterior border of the blastoderm are seen two short radial furrows (Fig. 23, m), and while it is possible that there were more at an earlier period, yet probably this is an egg in which there were never many ; because if numerous at an earlier period of development, the primary cleavage furrows should give some evidence of having extended to the margin of the primary area, and it will be noted that they fall quite short of reaching the margin.

The median section of a blastoderm, which showed thirty-two cells (sixteen marginal and sixteen central) in the living egg, is seen in Fig. 24, A. The cleavage cavity is especially well developed, and the one-layered condition of the blastoderm is unusually clear. The vertical cleavage separating the two cells which lie slightly to the left (anterior) of the center has not completely cut through to the cavity. Such a connection between adjacent cells is very slender, and, in this particular case, is absent in a few sections lying to either side.

While the planes of the early cleavages are, for the most part, vertical (that is, meeting the surface of the blastoderm at right angles), yet some of the planes take an oblique course. Under such conditions it is not difficult to find many places where the blastoderm appears to be two cells thick ; especially is this true of sections that are taken some little distance to either side of the median line (Fig.

118 J. Thomas Patterson.

24, B). Here, the lower portions of the two cells (on right) appear to bo "buds" or outgrowths from the floor of the cavity; but if the succeeding sections on each side are carefully examined, it will be found that such "buds" are nothing more nor less than the lower ends of cells whose upper portions reach the surface of the blastoderm in another section — and vice versa, the cells that appear to form a free upper layer are found to have portions extending obliquely downward to, and connecting with, the floor.

This point is of considerable interest, because a failure to appreciate its import probably has been the source of error on the part of some embryologists (e, g., Duval), who have stated that cells are cut off from the floor and are added upward to the blastoderm, thus contributing to its increase in thickness. The deception arising from the appearance of these buds is all the more striking in the cases where the nucleus of the cell concerned lies deep enough to be included in the "bud" (e. g., Fig. 24 B, n). In some sections both the upper and lower portions of the cell are included (Fig. 24 B, c), and in such the connection of the cell with the floor is clearly shown. In later stages the connection will be severed by the extension of the horizontal cleavage planes.

The accessory cleavage has entirely disappeared by the thirtytwo celled stage, but occasionally a supernumerary sperm nucleus will be found. The sections of the egg considered just above were subjected to a very careful examination for the purpose of determining how many such nuclei were present. This study gave the following results : In the sixteen central cells eighteen nuclei were found, but three of the cells had two nuclei each; that is, the cytoplasmic division of these three cells had not yet taken place. In the sixteen marginal cells, together with the surrounding periblast, nineteen nuclei were found, and twp of the marginal cells had two nuclei each. The extra nucleus was found far out in the periblast — so far out, indeed, that there is no possibility of its being considered as a derivative of a marginal cell nucleus. It must be, therefore, a sperm nucleus. This could have been determined without counting the other nuclei, for the one in question is in an advanced stage of degeneration, and rapidly disappearing.

It is certain, therefore, that not only does the accessory cleavage

Early Development of the Hen's Egg. 119

disappear at an early cleavage stage, but the extra sperm nuclei also disappear early. The writer has never found sperm nuclei after the thirty-two-celled stage. __

XL SixTY-FOUK Cells in Surface View. — Eive and One-half


This stage brings out more clearly than any we have so far considered the method by which the region of central cells increases. At the inner ends of many of the marginal cells are large central ones, which have been just recently cut off; and located centrally to these are smaller cells (Fig. 25). The number of central cells increases, therefore, in two ways: First, the central region grows at the expense of the marginal cells, and in this manner the central region gradually extends peripherally ; second, the central cells thus formed multiply inter se. If we compare this stage with the preceding one (Fig. 23), in which there were seventeen cells in each region, it will be seen at once that while the number of marginal cells has increased but six, the central ones have increased twenty-four (in surface view). This comparison becomes all the more striking when it is stated that the central region averages two cells in depth, due to the formation of horizontal cle'fts, so that there are probably . a total of some eighty central cells, or an increase of sixty-three. The central cells multiply, therefore, more than ten times as rapidly as the marginal ones.

The manner in which the central region becomes more than a single layer deep is made clear in a study of a section of a blastoderm in this same stage of development (Fig. 26). The segmentation cavity is very distinct and above it the original one-layered disc (see Fig. 22) is now two cells deep; and this condition has been brought about not by the addition of cells from that portion of the disc lying beneath the cleavage cavity, as supposed by Duval, but entirely by the formation of horizontal clefts in the cells lying above the cavity. As we might have expected, the method of increase in cell layers is identical with that of the same process in the pigeon blastoderm. At this stage there is no possibility of cells being added to the disc from the central periblast, because that region is entirely void of nuclei.

120 J. Thomas Patterson.

On account of the obliquity of the so-called horizontal clefts, the central cells are characterized by great irregularity both in shape and size. Many of them are shaped like squamous epithelium, and often give the appearance of stratified epithelium in section.

A detailed drawing of the anterior end of a section will show the manner in which the marginal cells add their products to the central area (Fig. 27). The cleavage membranes cut down deep into the disc, and at their terminal points the horizontal clefts begin to spread beneath these large cells, thus separating them from the underlying j)ortion ; when this is accomplished, the large cells are split up into smaller ones by vertical and horizontal divisions (on right of Fig. 27).

XII. Ojs'e IIujxdeed and Fifty-four Cells in Surface View. —

Seven Hours.

During the next hour and a half the marginal cells undergo but few divisions, but the central ones multiply very rapidly, and since they receive but few additions from the marginal cells, their increase must be due to their own activity in division. This results in producing a large number of small cells in the central area, and the smallest cells lie at the very center of the blastoderm (Fig. 30), It would seem, therefore, that the early cleavage of the hen's egg follows the rule which states that the time occupied between any two successive cleavages grows shorter and shorter as the volumes of the cells decrease.

The average depth of the central part of the blastoderm at this period is about three cells (Fig. 31). The cleavage cavity, although not so distinct as in the preceding stage, is, nevertheless, clearly recognizable and there are no connections between its floor and the lower cells of the blastoderm. The anterior and posterior ends of the section show different conditions in the character of the cells. At the posterior end there are three large cells, in addition to the marginal one, which have not been broken up by horizontal clefts. At the anterior end, on the other hand, all of the cells, except the marginal, have undergone division. This difference is probably only a local condition, and, therefore, is not fundamental.

Early Development of the Hen's Egg. 121

The periblast still remains free of nuclei, but the marginal cell nuclei are beginning to show a tendency to migrate farther peripherally than usual (Figs. 28 and 29).

XIII. Three Hundred and Forty-six Cells in Surface View. —

Eight Hours.

The final stage that we shall consider in this paper is shown in Fig. 32. During the hour intervening between this and the previous stage the marginal cells have added many more cells to the central area than at any former period of like duration, and consequently the (radial) length of the marginal cells has greatly decreased, and their furrows now are beginning to cut out into the periblast.

This stage is one of the most important of all the early cleavages, because it represents the transitional period between the "unorganized" and "organized" periblast; but we shall not consider the sections at this time.

XIV. General Summary.

The absence in this paper of comparisons between the development of the hen's egg and that of other vertebrate eggs is not due to a lack of appreciation of the importance of such comparisons, but rather to the fact that in the main these have been pointed out for the corresponding stages of the pigeon's egg. In this connection the writer wishes, therefore, to confine himself to emphasizing the close similarities between the development of the hen's egg and that of the pigeon, although to those who have followed closely the work on the latter egg this may seem unnecessary.

Exact agreement in all details of development, even in the eggs of two species as closely connected as those of the hen and the pigeon, is not to be expected, but the fundamental processes should certainly agree. And such has proved to be the case. The minor differences in development of these two forms have to do primarily with time relations ; for although the eggs of these two species are in about the same stage of development at the time of laying, yet the pigeon's is forty-one hours old and the other's twenty-two. The holomogous processes, therefore, necessarily do not occur at exactly the same time after fertilization.

122 J. Thomas Patterson.

The more important comparisons are as follows:

1. The process of fertilization (that is, the entrance of the sperm) in each egg occurs immediately after ovulation, when the egg is in the region of the infundibulum.

2. At the time of fertilization in the pigeon's egg, from twelve to twentj-five supernumerary sj^erm nuclei enter the egg. In the hen's egg only five or six such nuclei are found (except in one case where twenty-four were present).

3. Upon their entrance into the egg these sperm nuclei, in each egg, migrate toward the periphery of the disc. In the pigeon's egg the nuclei, on reaching the margin, become active, divide and give rise to an accessory cleavage, which disappears between ten and twelve hours after fertilization. In the hen's egg some of the supernumerary nuclei pass down into the deeper portions of the disc and there undergo complete fragmentation ; others may succeed in reaching the margin, and there give rise to a rudimentary accessory cleavage, which disappears shortly after the eight-celled stage, or between four and five hours after fertilization.

4. In the pigeon's egg the marginal cells become closed and remain so throughout the period occupied by the accessory cleavage. In the hen's egg the marginal cells always remain open to the periblast both below and peripherally. This would seem to indicate that the condition of a closed marginal cell in the pigeon's egg is to be correlated with the presence of a large number of accessory cleavages. Perhaps it is for the purpose of cutting off some influence emanating from the accessory sperm nuclei.

5. In neither egg does the direction of the first cleavage plane, or the eccentricity of cleavage, if present, seem to bear any constant relation to the axis of the future embryo.

6. Immediately after the disappearance of the accessory cleavages and their accompanying nuclei in the pigeon's egg the marginal cells open to the periblast, and their nuclei divide and some of the daughter nuclei migrate into the periblast and "organize" it. In the hen's egg there is a period of from two to three hours after the disappearance of the accessory sperm nuclei during which the periblast is void of nuclei of any kind.

Early Development of the Hen's Egg. 123

7. In each egg the first horizontal cleavage plane marks the position of the segmentation cavity.

Austin, Texas, November S, 1909.


Babfubti^ D., 1895. Versuche iiber die Parthenogenetiscbe Forschung des Hiibuereies. Arcbiv f. Entw. Mecb., Bd. 2, pp. 303-351.

Blount, Mary, 1909. Tbe Early Development of tbe Pigeon's Egg, witb especial Reference to Polyspermy and tbe Origin of the Periblast Nuclei. Journal of Morphology, Vol. 20, No. 1, pp. 1-64.

Duval, M., 1884. De la Formation du Blastoderme dans I'ceuf d'oiseau. Annales des Sci Nat, 6 Series, Vol. 18, pp. 1-208.

GuYER, M., 19<J9. The Spermatogenesis of tbe Domestic Chicken {Gallus gallus (lorn.). Anat. Anz., Bd. 34, pp. 573-580.

Harper, E. H., 1904. The Fertilization and Early Development of tbe Pigeon's Egg. The American Journal of Anatomy, Vol. 3, pp. 349-386.

Kionka, H., 1894. -Die Forschung des Hiihnereies. Anat. Hefte, Bd. 3, pp. 395-443.

LiLLiE. F. R., 1908. Tbe Development of the Chick. New York, Henry Holt & Co.

Patterson, J. Thomas, 1909. Gastrulation in tbe Pigeon's Egg — A Morphological and Experimental Study. Journal of Morphology, Vol. 20, pp. 65-123.

ScuAuiNSLAND, H., 1899. Beitrjlge zur Biologie und Entwickelung der Hatteria nebst Bemerkungen iiber die Entwickelung der Sauropsiden. Anat. Anz., Bd. 15, pp. 309-334.


J. Thomas Patterson.


Fig. 1. — A two-celled stage, which was drawn from a free-haud sketch and measurements of the living egg.= For the history of this egg see Table 1, hen No. 2. X 18.


Fig. 2. — ^A median section taken transverse to the furrow of the egg illustrated in the preceding figure. The nucleus of Pander is very poorly developed in this egg. c.n., cleavage nucleus, x 39.

= In this, as in the succeeding figures of surface views, the anterior margin of the blastodisc is toward the top of the page, and hence the median axis of the later embryo will parallel the sides of the page. In sections the anterior end is always toward the left. In the surface views the area occupied by the primary cleavage, together with the first "ring"' of periblast, are shown.

Early Development of the Plen's Egg.


9 •

• oO "O o o OQ o • . • ; ' ., o .°'o 0O„t O °' ■■:■':

e • s a e © -©r ■ ' e • ®» ••

» ®.«"g°o©©

« 9 ^


« e " . © .


® *

® © © ® • ••


Fig. 3. — Enlarged di'a\Ying of the cleavage memlirane of the preceding figure, cm., cleavage membrane; in. p., membrane plate; v., vitelline membrane. X 525.

Fig. 4. — Enlarged drawing of the extreme left end of the section shown in Fig. 18. This shows a supernumerary sperm nucleus, s.p., al)out which cell formation is attempted, as evidenced by the horizontal cleft situated just above the nucleus. X 525.

Fig. 5. — Diagram of an unsegmented blastodisc showing the distribution of twenty-four sperm nuclei. This egg was removed from the oviduct about two and a half hours after the estimated time of fertilization, when it was eleven inches from the infundibulum. See Table 1, hen No. 1. X 18.


J. Thomas Patterson.

r innini i' 'iii M" ii'i , ii i ii' ii :i r


Ytg. 6.— a nest of four supernumerary sperm nuclei. X about 600.

Fig. 7.— a sperm head lodged in the vitelline membrane. The sperm had evidently been stopped in the act of entering the egg. X about 600.

Fig:* 8.— Three sperm heads embedded in the chalazipherous layer of albumin and located next to the vitelline membrane. Both of these figures (7 and 8) are taken from the central region of the egg shown in Fig. 10. X about 600.

Pig. 9._A diagram of the blastodisc shown in Fig. 1, showing the distribution of the supernumerary sperm nuclei. The black dot indicates that the nucleus is located more or less superficially in the disc; while a dot marked prime one shows the location of a nucleus that is situated deep in the disc, and one marked prime two. a nucleus that is undergoing fragmentation. The broken line, A— B, is the plane of the section shown in Fig. 2. X 18.

Fig. 10.— Diagram of a blastodisc of an egg taken from the oviduct three hours after fertilization, shortly after it had entered the isthmus. This shows six supernumerary sperm nuclei. Line A— B is the plane of the section shown in Fig. 11. X 18.

m p

Early Development of the Hen's Egg.





3? «-. r

",';■."■.■.•■;■ ■•.■•••■.■..", . .o'»»«'#r .

>»«■■•■;•■• ■.■■•■.•,•■.•■■•.•••»•♦•• *• ■•

.^.;*;vv.{; :.:::-;::;:.vv • •/:.;%«••

'■-,•:■;■ ■■•■■.•..•■.. .•'!.* •« ^* <

•vr. >•:■■••••. ••*»^'*'* **


• - V*'

» » * .• '•■'it-'

» »*^ •

J * « /


•*^ ..»*



Fig. 11. — A section through line A — B of Fig. 10. vi.p., marginal periblast ; c.n., cleavage nucleus; /, neck of the latebra. X 52.




Fig. 12.— The four-celled stage, drawn from a whole mount preparation. The egg as taken from the isthmus (see Table 1, hen No. 3). a.c, accessory cleavage furrows. X 18.

Fig. 13.— An eight-celled stage, drawn from a whole mount preparation. The history of this egg is given in Table 1, hen No. 4. It shows one central and seven marginal cells, a.c, accessory cleavage furrows ; s.a.c, small accessory cleavage cells. X 18.


J. Thomas Patterson.



16 17

Fig. 14. — A diagram of a seven-celled stage, showing the distribution of the cleavage and supernumerary sperm nuclei. The egg was taken from the isthmus about three and three-fourths hours after the estimated time of fertilization. A— B, plane of the section shown in Fig. IS, and C — D that of Fig. 19. X 18.

Fig. 15.— An interesting blastodisc showing a large number of marginal cells in the process of formation. The egg was taken from the isthmus. X 18.

Fig. 16.— a blastodics showing a comparatively regular form of cleavage in the eight-celled stage. The egg was taken from the isthmus about four hours after fertilization. The letters indicate the cells that are probably homologous ; and the numbers, the first and second cleavage planes. Three accessory cleavage furrows are shown. X 18.

Fig. 17. — An early stage showing an eccentric cleavage, with the displacement toward the anterior, g.a.c, a group of five small accessory cleavage cells. The egg was taken just as it was passing into the shell-gland.

Early Development of the Hen's Egg.



Pig. 18.— Section through plane «— ?>, Fig. 14. m.n., marginal coll nucleus; s.ii., supernumerary sperm nucleus with a cleft lying just above it. X 73.

-::;.V.••^:.■^■:::^•^:■;•"•.v:^•.U'^ .s;£-.:.-.*:..>.-..: .••,;.:;•. /X/V;'.!;;'


Fig. 19.— Anterior portion of a section through c— rf, Fig. 14. s.n., sperm nucleus; a.f., accessory cleavage furrow; p.f., terminal portion of a primary cleavage furrow with membrane. X 66.


J. Thomas Patterson.



Fig. 20. — A seventeen-celled stage, drawn from a whole mount preparation. The egg was taken from the isthmus between four and five hours after fertilization. The blastoderm is remarkable in that it has many short radial furrows situated near the margin of the primary area (see text for a description of these furrows). X IS.

Fig. 21. — A fifteen-celled stage, drawn from the living egg (see Table 1, hen No. 5). a — h, plane of section shown in Fig. 22. x 18.


o •;■•.■; ■' »,»«»,» A ■••'•.••.!•.•*• ■.•■.•. .••■.■-■.■■:


Fig. 22. — Section through plane oX 55.5.

-Z>, Fig. 21. S.G., segmentation cavity.

Early Development of the Hen's Egg.



Fig. 23. — A tbirty-four-celled stage, drawn from a whole mount preparation. The egg was taken from the shell-gland. There are seventeen marginal and seventeen central cells. At the posterior margin are shown two of the short radial furrows which were noted in Fig. 20. X IS.


>• ;;.-;;; -;\:;;;^:;i;>**'




Fig. 24. — A. Median section of a blastoderm showing thirty-two cells. B. A section taken seven sections to the left of the preceding, n, nucleus; c, cell showing a connection with the floor of the segmentation cavity (see text for a description of these figures) . Both X 59.


rl. Thomas Patterson.


Fig. 25. — Blastodisc showing sixty-four cells in surface view — forty-one central and twenty-three marginal (see Table 1, hen No. 7). X 18.



Fig. 26. — Median longitudinal section of a blastoderm in a stage of development corresponding to that of the preceding figure, s.c, segmentation cavity. X 55.

Early Development of the Hen's Egg.






•• • ••• « ■•• Too o o • o *o ■-■"-.

'::]'.-:'.\°.\\'^°l<,c-.v:°°y' ^^m^^

, - ^ o a * o p 9 • • „ , „_„&<* -J o*>a ••• •••

mc n


Fig. 27. shown iu gin of the

Fig. 28. shown in

Fig. 29. sistei* nuc X 139.



The anterior end of a section located to the right of the one the preceding flgnre. This gives the details of structure of the mardisc. Some of the nuclei are taken from adjacent sections. X 139. — Section of a marginal cell from the same series as the section Fig. 31. m.c.n., marginal cell nucleus, undergoing division, x 139. — Another marginal cell from the same series, showing how the lei have migrated apart, m.c.n., sister nuclei of the marginal cell.


Fig. 30. — Surface view of a blastoderm of an egg taken from the shell-gland seven hours after fertilization (see Table 1, hen No. S). There are 31 marginal and 123 central cells, or a total of 154 in the surface view, m., short radial furrows at the margin. X 18.


J. Thomas Patterson.





Fig. 31. — Median longitudinal section of a blastoderm in a stage of development coiTes]»onding to that shown in Fig. 30. iii.c.n., marginal cell nucleus; S.C., segmentation cavity. X 65.


Fig. 32. — A blastoderm showing 346 cells in the surface view, 34 marginal and 312 central (see Table 1, hen No. 9). Note that the marginal cell furrows are beginning to cut out into the periblast. X 18.




From the Hull Zoological Laboratory, University of Chicago

With Seventy-three Figures


Introduction 135

Methods 136

I. The development of the ovary 138

The cell elements of the ovary 138

The formation of the egg-strings 149

The nutrition of the egg 151

The nurse cells 1^8

II. The development of the testis 167

III. The chromosomes 1'^

IV. Summary 194

The ovary 194

The testis 197

Amitosis 198

The chromosomes 198

Bibliography ^^1

Explanation of figures 205


Leptinotarsa signaticollis, a chrysomelid beetle found in the northern and eastern tributary valleys of the Rio Balsas system in Mexico (Tower, '06), is a favorable form for cytological investigation, in as much as the long breeding season, the slow development of the reproductive organs, together with the fact that the insects can be successfully bred and reared in breeding cages, enables one to obtain any stage in development with relative ease.


136 Harry Lewis Wieman.

The present paper embodies a study of important stages occurring for the most part in the pre-maturation period of the germ cells. Recent discoveries in chromosome behavior have brought the reproductive cells of insects into considerable prominence; but most of this work has been prosecuted along rather narrow lines. The chromosomes are undoubtedly important elements, but they are far from being the only factors to be considered in a study of the mechanism of heredity. The cytoplasm, though long neglected, is coming more and more to be recognized as of equal if not greater importance in this regard ; and in the present undertaking the so-called nucleo-cytoplasmic relationship has been studied by following the morphological and chemical transformations undergone by the cellular elements of the ovary and testis during critical stages of development.

The ovum offers a more extended field for observation than the spermatozoan ; since spermatogenesis is a relatively simple matter compared with the complex changes involved in the production of a mature ovum.- As a result I have given more attention to the reproductive organs of the female than to those of the male.

To Professor CO. Whitman, at whose suggestion this work was taken up, and to Professor W. L. Tower, who kindly supplied me with material from his pedigreed stock, I am greatly indebted for much valuable advice and criticism.


Two distinct methods of killing and fixing were followed depending upon whether chemical or morphological differentiation was the object sought after. Experience showed that Flemming's strong solution was far superior to any other reagent for faithful preservation of morphological detail; but the presence of osmic acid greatly interferes with the action of stains employed in the study of chemical changes. For the latter purpose I used a saturated aqueous solution of picric acid to which sufficient acetic acid was added to make a 10 per cent solution. This does

Germ Cells of Leptinotarsa Signaticollis. 137

not give as perfect fixation, but enables one to study chemical transformations more satisfactorily. This mixture was used almost exclusively for eggs in later stages of development. The eggs were allowed to remain for about ten minutes in the reagent which was maintained at a temperature of 55°C., and then transferred to 70 per cent alcohol. At the end of a day or two the egg envelopes stand out from the egg, and can be readily dissected off with the aid of needles. Good results were also obtained with Hermann's platino-aceto-osmium mixture. The killing fluids were used in the cold except in the case of the eggs as noted above. The most satisfactory stains were found to be safranin (basic), and lichtgriin (acid) . Others were used in various combinations, but without such good results. Even for the study of chromosomes I have found these dyes superior to Heidenhain's ironalum-haematoxylin. Grubler's Safranin 0" was made up according to the following formula:

Safranin 1 gram

Anilin water (4cc. anilin oil + 90 cc. water) 90 cc.

Alcohol (95 per cent) 10 cc.

Sections were left in the safranin for four to six hours, passed through graded alcohols and immersed for a few seconds in the acid stain. After washing in 95 per cent alcohol, the material was transferred successively to absolute alcohol, clove oil and xylol ; and then mounted in balsam. These two stains work together perfectly, and give sharp and clear contrast at every stage.

Iron-alum-haematoxylin with or without counterstain (orange G or lichtgriin) and Gram's gentian violet method were also employed, though to a less extent.

For embedding, Johnston's parafiin-asphalt-rubber mixture was used in various degrees of dilution with pure paraffin. This method is especially helpful in working with mature eggs, which show a great tendency to crumble when cut in pure paraffin. Sections were as a rule cut 61^ in thickness.

138 Harry Lewis Wieman.

1. The Development of the Ovary

The cell elements of the ovary

The insect ovary has long been a favorite object for microscopical study, yet some of the most important aspects of the problem it presents are unsatisfactorily answered. The organ exists in a wide variety of morphological types, but however great its lack of constancy in macroscopical structure, it always shows in the eggtube the presence of three elements: germ cells, nurse cells and epithelial cells. One of the questions that has engaged the attention of many investigators is the origin of these cells.

Following the contributions of Dufour ('33, '41) and others on the gross anatomy of the insect ovary. Stein ('47), in his monograph on the female genital organsof a large number of Coleoptera, published the results of the first thorough investigation of the histology of the ovary, and thus laid the foundation for all subsequent work in this field. He showed that the terminal thread {Endfaden) is not a blood vessel as had been stated years before by Johannes Miiller ('25), but that in all probability it serves as a suspensory ligament which binds the ovarioles together and fixes them to the dorsal wall of the thorax. The eggs, he believed, arise from the large cells in the lower or proximal part of the terminal chamber, and that the cells in the other part of the tube are the yolk-building elements, i.e. , nurse cells. He correctly described the two sheaths of the ovary, the outer "peritoneal Hillle" and the inner structureless "tunica propria.^' To the germinal vesicle he attributed the morphological value of a cell, and judging from more recent work, he also erred in considering the chorion of the egg a result of the fusion of follicle cells, instead of a secretion product of these cells, as later research has demonstrated.

H. Meyer ('49) described two kinds of so-called "nuclei'^ in the ovary of Lepidoptera, small ones of an epithelial nature, and large ones which develop into germinal vesicles, while the nurse cells were regarded as aborted ova. Here then is the first claim for a common origin for reproductive and nurse cells, the epithelial cells having a different origin.

Germ Cells of Leptinotarsa Signaticollis. 139

During the next decade or more, most investigators were occupied with the study of the structure and development of the various envelopes of the egg and ovary, but in 1864, Claus, returning to the question of the origin and significance of the three recognized cell elements, came to the conclusion that they were all of common origin, being derived from the primordial germ cells. This conclusion was shortly afterward confirmed by Leuckart ('65) and Landios ('67), but Leydig ('66), on the other hand, questioned this result, and held that only egg and nurse cells are '4n ihrer Wurzel identisch," and Epithel hingegen, besteht fiir sich, und es findet kein Uebergang zu dem Keim-und Ei-Zellen statt." Likewise, Metschnikoff ('66), in his investigation of the embryonic development of Cecidomyia arrived at a similar conclusion namely : that the egg and nurse cells are derived from the "Polzellen der Geschlechtsorgangen," while the epithelial cells have an entirely different history.

From a study of Nepa and Notonecta, Will ('85), described an entirely new as well as unique method of egg formation, in which both nurse and germ cells arise inside of the large "nuclei'^ filling the terminal chamber of young insects, and w^hich he called " Ooblasten.'" Later, the rupture of the membrane permits the contents of the nuclei to pass out, when the remaining part of the ooblast reconstructs a membrane and becomes a germinal vesicle. Similar processes have been described by Sabattier ('86) and Perez ('86). The latter states that three kinds of nuclei, representing egg, nurse, and epithelial cells, are at first enclosed in a mother cell.

Will's theory has been severely criticized byKorschelt ('86), who has shown a different origin for these cells in a large number of species. Korschelt supports Claus's idea of a common origin for all three kinds of cells.

However, Leydig ('89) again produced evidence of a separate origin for germ and epithelial cells, and this view has in latter years been steadily gaining ground. Thus Heymons ('91) showed that in Phijllodromia {Blatta) germanica, the primitive germ cells appear even before the somites are established, while the cells of the terminal thread and the epithelial cells are derived from the

140 Harry Lewis Wieman.

dorsal wall of the mesoblastic somites. Later, Heymons ('95), extended his work to the Dermaptera and Orthoptera, and confirmed this conclusion. Wheeler ('93) in Xiphidiurn ensifermis was unable to detect the germ cells until the somites are formed when they appear as metameric cell clusters, each of which is confined to the median portion of the splanchnic w^all of the somite. These two authors do not agree as to the exact time of appearance of the primordial germ cells, but they show very clearly that certain cells, differing from the ordinary mesodermal or epithelial cells are established at an early period in ontogeny, and constitute the material from which the germ cells arise. Carriere und Burger ('98), came to similar conclusions for Chalicodoma muraria.

In a study of the embryology of Donacia crassipes, Hirschler ('09) states, die Geschlechtsanlage bei Donacia schon vor der Entwicklung desselben an der ganzen Eioberflasche als histologisch differenzierte Zellenanhaufungen auftritt. Die Genitalzellen sind also ontogenetisch alter als die Keimbatter" (p. 637). More recently, Hegner ('09), published the first connected account of the Keimbahn in insects, which established beyond a doubt the early differentiation of the germ cells, thus excluding the possibility of the presence in the egg-tube of indifferent cells which might give rise to either epithelial or sexual cells.

Between these early embryonic stages and the adult condition, there is a wide gap in our knowledge of the developmental process. Fairly extensive comparative studies based on adult and slightly younger stages have been made (Gross ('03), Kohler ('07) etc.), but in order to study the history of a continuous developmental process such as the differentiation of the elements of the ovary, it is necessary to have a complete series of normal stages taken from a single type. One can not tell from an examination, however intensive, of the completed adult structure, how that structure has been brought about. Inspection of a large number of sections taken from later developmental stages, especially larval and pupal, indicated to me that a study of this period of the formation of the ovary would yield interesting and important data, and throw some light on the histology of the adult organ.

The adult female reproductive organs of L. Signaticollis con

Germ Cells of Leptinotarsa Signaticollis.


sist of a median vagina with a somewhat funnel-shaped oviduct on either side. To the broad distal ends of each oviduct are attached from 45 to 46 ovarioles or egg tubes, which are of a type common to a large number of Coleoptera. Each tube is divided into three more or less distinct regions, namely; the ovariole stalk (Rohrstiehl) , which is the part proximal to the oviduct, the terminal chamber (Endkammer) , in the lower part of which the egg passes through its early development, and the terminal thread {Eiidfaden), which is dilated at its base into a broad cap-like

Fig. W. a, ovary; B, single ovariole represented in longitudinal section; e, egg; g. c, germ cells (young ovocytes); I. m., limiting membrane; n. c, nurse cells; 0. St., ovariole stalk; ov., oviduct; t. t., terminal thread.

structure marked off from the terminal chamber by a definite membrane (Fig. W.). In the terminal chamber are found three different kinds of cells, the egg cells in the lower proximal region, the nurse cells occupying the distal part (|) of the chamber, and the epithelial cells scattered between those of the other two groups. In seeking for a stage at which to begin this account, I found the close of the larval period to be the most advantageous, for here very simple conditions prevail which permit one to distinguish,

142 Harry Lewis Wieman.

without the slightest doubt, between germ and epithelial cells; while the nurse cells have not yet been differentiated. Furthermore, the germ cells are practically unaltered in appearance from early embryonic stages. Figure 1, represents a longitudinal section of an egg tube characteristic of the larval condition, in which one of the striking features is the direct continuity between the cells of the terminal thread (tt) and those of the terminal chamber. In the region of the terminal chamber two kinds of cells can be readily made out, the large germ cells (g.c.) with deeply staining cytoplasm and with nuclei showing an irregular chromatin reticulum, and the small pale epithelial cells (ep.c.) continuing outward into the terminal thread. The ovariole stalk is composed of +all columnar cells (o.st.c). The peritoneal sheath is in the process of formation, and is represented by the large epithelial cells (p.sh.) which are flattened against the sides of the tube. Beneath this outer sheath lies the tunica propria (t.pr.) bounding the terminal chamber and thread. The ovarioles are united by their terminal threads coming together on either side into a single bundle which is inserted into the dorsal wall of the body cavity. (Fig. W, A).

The epithelial cells scattered between the germ cells are what the older authors designated as the small nuclei surrounding the ovocyte, to which they supply nutrition. In the figures of these authors, these cells are never shown with cell boundaries, but always as nuclei lying in a homogeneous matrix. At best it is not easy to make out cell boundaries, since the cytoplasm shows such a weak affinity for the stain, but a careful study of many sections at different stages has convinced me that cell walls are present and that each nucleus belongs to a single cell, except under certain conditions as noted below. At a.c. is a large cell which I have designated the apical cell, the significance of which I have been unable to determine. It is also seen in later stages (Figs. 2,3).

The fact that the epithelial cells show such a slight affinity for the stain, seems rather strange in view of what appears to be their function; namely, to supply nourishment to the germ cells. Ordinarily, cells having a secretory function show in their cytoplasm more or less granular masses which stain deeply. Are they

Germ Cells of Leptinotarsa Signaticollis. 143

really nutritive in function? The evidence for believing so rests on their close relationship to the germ cells, and on the changes undergone in size and form after coming into contact with the germ cells. It seems to be the general concensus of opinion that these cells later form the follicle of the egg by surrounding the latter as it passes from the lower region of the terminal chamber; but I believe it can be readily shown that the follicle is not formed in this manner. The evidence for this will be given in its place a little later.

Closer examination of the germ cells in the stage shown in Fig. 1, reveals the presence of one or two basic-staining nucleoli, each surrounded by a clear non-staining area (Fig. 34) . This condition proves to be the result of the division of a single nucleolus, and this is the first step in what appears to be a process of amitosis. The division later extends to the nucleus as shown in Fig. 35. A fuller discussion of this process and its significance will be considered in connection with the germ cells of the male.

In Fig. 2, which is from a very young pupa, we find a considerable increase in the number of cells in the ovariole, brought about by mitotic divisions which are very abundant at this and succeeding stages, especially among the germ cells. Mitotic figures among the epithelial cells are very rare, but as they do occur, it is inferred that mitosis is the method of cell multiplication, for I have never observed authentic cases of amitosis among these cells at this stage. The region of the terminal thread is much larger in diameter than in Fig. 1. The cells at the base of the thread show a tendency to flatten out, and the lower margin of these flattened cells represents the position of the future limiting membrane which will definitely separate the terminal chamber from the terminal thread.

Fig. 3 is from a slightly older pupa, and shows at l.m. more distinctly the place where the limiting membrane will form. Just below this is seen a double row of epithelial cells, which compared with those surrounding the germ cells lower down, show considerable differences in form and size. The latter are smaller, and the nuclei are bean-shaped. These transformations suggest that these cells are in some way concerned with the metabolism of

144 Harry Lewis Wieman.

the germ cells. However, as will be shown later, the accumulation of these cells at the lower end of the terminal chamber seems to be involved in the differentiation of the germ cells, and it may be therefore, that these form changes are simply due to mechanical pressure as the cells make their way down the tube. Fig. 4 is a cross section of an ovariole showing the same relationship.

During the next few days the terminal thread undergoes a remarkable growth, as a result of which its diameter is increased enormously. In Fig. 5 is seen a longitudinal section of it at maxmum development, when it greatly exceeds the terminal chamber in volume. The same figure shows another stage in the formation of the peritoneal sheath, which for the sake of simplicity was omitted from several of the preceding drawings. Fig. 10 is from a pupa about two days older, in which itisan be readily noted that the terminal thread has diminished very considerably in volume. The process continues until in the adult it has the appearance shown in Fig. 13, where throughout the greater part of its length there is evidence of degeneration, leaving at its base a cap-like mass of cells which is practically all that remains of the structure shown in Fig. 5.

Fig. 10 shows at the base of the terminal chamber the first appearance of the limiting membrane, which comes into existence first at the periphery, so that the center is the last part to be closed off. The principle evidence for this is the fact, that for a considerable time after it can be definitely made out at the sides, groups of epithelial cells can be seen in the central region entering the terminal chamber. The membrane does not show any cellular structure, and is probably a product of secretion of the epithelial cells at the base of the terminal thread.

Wagner ('36, '37), Siebold ('71), Will ('85), and others early opposed the idea of the terminal thread functioning in a purely mechanical manner as a suspensary ligament for the ovariole, and maintained that it is the place where the "Keimstatte" are produced, w^hich in the lower part of the egg tube surround the germ cells. Brandt ('78) thought this interpretation might apply in forms where there was direct continuity between the terminal chamber and thread, but where the two regions are separated by

Germ Cells of Leptinotarsa SignaticoUis. 145

a limiting membrane, he believed the terminal thread served as a ligamental structm-e. Kramer ('69) regarded the thread as the solid continuation of the tunica propria of the egg tube.

As a matter of fact an almost unlimited variety of functions, from that of a blood vessel to that of a piece of connective tissue, have been ascribed to the terminal portion of the ovariole. This is largely because most workers have confined their studies to the adult organ which represents the end-product of a developmental process. Furthermore, this structure does not present the same appearance in all species, and all gradations are to be found from a condition where it is a mere rudimentary appendage of the egg tube, to cases where it is a direct continuation of the terminal chamber. Agreement as to its significance can scarcely be expected as long as observations are based almost entirely upon the adult structure.

Heymons ('91) was perhaps the first to study the development of the terminal thread, and he showed in B. germanica that its cells, as well as the epithelial cells, are derived from the dorsal wall of the primitive mesoblastic somite. According to this author the terminal thread does not contribute epithelial cells to the terminal chamber. In the adult he found the thread ending freely, so that if a supporting function is to be attributed to any part of the apparatus it is concerned with the peritoneal sheath, and not with the cells filling the interior. However, he considers this of no great importance, since the fat-bodies, connective tissue, and tracheal tubes serve to hold the organs in place. The end threads are of functional significance only in embryonic and larval periods of development, when they are concerned with orientation of the ovarioles which, as I understand him, consists in directing the growth of the egg tubes backward and upward toward the dorsal wall of the primitive segments.

The appearance of the terminal thread in the adult of L. signaticollis would justify the conclusion that it is a rudimentary appendage of the terminal thread from which it is separated by a definite membrane (Fig. 13 I. m.). The peritoneal sheath has become invested with striated muscular tissue and air tracheae, which form a thick covering grading off into the region of the egg

140 Harry Lewis Wieman.

chamber, into a thin epithelial layer closely applied to the tunica propria. Everything indicates that the peritoneal sheath with its muscular investment, which is inserted into the dorsal body wall, serves to support the ovarioles.

However vestigial its structure in the adult, in the larval and pupal periods the terminal thread is of considerable importance in producing the epithelial cells of the end chamber that later play an important part in the development of the egg. Whether or not it has, during the larval and pupal periods, the function ascribed to it by Heymons, I am not prepared to say, since this is a point that can not be readily demonstrated one way or the other.

It will be noticed in Figs. 5 and 38, that the chromatin of the germ cells has a peculiar granular appearance that is quite different from the reticular structure of the nuclei of the germ cells at the lower end of the tube (Fig. 6) . This is the first indication of the differentiation of the germ cells into functional sexual cells that will develop into eggs, and nurse cells.

The question immediately arises as to what causes this differentiation. Giardina ('01) described in Dytiscus marginalis a process of differentiation in which one of the daughter nuclei, resulting from the division of a primordial germ cell, receives in addition to the usual number of chromosomes, a certain amount of chromatin, and develops into a functional germ cell ; whereas, the other daughter cell lacking this extra chromatin becomes a nurse cell. Nothing of this kind is to be observed in L. signaticollis.

At just about this time in the developmental history, there occurs at the junction of the terminal chamber and tube stalk a transformation that is significant in this regard. Fig. 6 shows the condition immediately preceding the change; the boundary between the egg chamber and the tube stalk is sharp, and the cells of the latter can be distinguished from the epithelial cells not only by their shape, but by the capacity of the cytoplasm to stain more deeply. Fig. 7, which is a few days older, shows a pale, lightly staining, semi-fluid mass that blends with the epithelial cells on the one hand and the cells of the ovariole stalk on the other. It is very difficult, impossible in some cases, to make out cell bound

Germ Cells of Leptinotarsa Signaticollis. 147

aries, which suggests that a process of liquefaction is taking place, as a result of which the walls of the epithelial cells and to a less extent those of the stalk cells are dissolved, producing a matrix in which the nuclei of these cells are suspended.

This diffuse effect is not a chance occurrence to be observed in a few odd preparations, a result which might be attributed to an artifact, but is an event that takes place regularly at what appears to be a critical period in the development of the ovum.

As this condition comes on at the time when the chromatin configuration enables one to distinguish between egg cells and nurse cells, it seems that there might be some causal relation between the two phenomena. A study of the epithelial cells demonstrates that they filter down from the upper portion of the terminal chamber to its lower end, where they accumulate, and the result is aninteraction between them and the contiguous cells oftheovariole stalk. The semi-fluid matrix formed might then be regarded as exerting a specific effect on those germ cells coming under its influence, enabling them to develop into ova, while the more distant germ cells become nurse cells.

For some time preceding this stage the germ cells multiply so rapidly that the daughter cells do not accumulate much cytoplasm; and a" section at this time shows the cells as large clear nuclei, containing chromatin in the form of a spireme, around which is a very narrow margin of cytoplasm (Fig. 8). These cells have completed their division period and therefore represent the last generation of ovogonia. As they enter upon the growth period they move down into the diffuse region. Fig. 9 is from an adult several days older, when the eggs appear with a considerable quantity of cytoplasm which increases steadily in amount from now on. ^ It is also to be noted that mitotic divisions continue in other parts of the egg tube long after they have ceased in this region. This fact tempts the suggestion that the energy which is preserved in the ovocytes for the future maturation divisions and perhaps the development of the embryo, is expended in mitotic divisions by the nurse cells, the sister cells of the ovocytes. The occurrence of tripolar spindles (Fig. 19) and abnormally large single spmdles that result in the production of gigantic cells, are manifestations

148 Harry Lewis Wieman.

of irregular cell activity among the future nurse cells that lend additional support to the idea.

Do the egg cells cease to multiply because they have passed through a fixed number of cell generations; or does the cause lie in the action of some substance produced in the egg tube that prevents further cell division, and turns the direction of cell activity into different channels, the result of which is growth? Much evidence favors the latter view, for we are learning more and more that development and differentiation are largely a matter of correlation of mechanical and chemical forces. Thus Spemann ('01), Lewis ('04), Le Cron ('07), and others have shown that the development of the lens of the eye in certain amphibia depends upon a stimulus set up in the ectoderm in the region where the outgrowing optic vesicle touches it. In what is known as hormone action" we have an example of the secretion of one organ reacting upon another organ in such a manner as to cause secretion in the latter, or to affect its metabolism in other ways. The phenomena accompanying the oncoming of puberty and the results of castration are too well known to need mention.

Such facts indicate that the interaction, whether by mechanical contact or through the production of chemical combinations of tissues or their secretions upon each other is an important factor in developmental processes. Therefore, I believe this semiliquid condition occurring in the terminal chamber at a definite time in the history of the organ, is to be regarded as a physical and chemical reaction in which the epithelial cells of the egg chamber and those of the ovariole stalk are involved, and that it is an important, if not the causative, factor in the differentiation of the primordial germ cells into egg cells and nurse cells.

From this it follows that the epithelial cells, which come chiefly from the terminal thread, are very intimately connected with the development of the germ cells, but I do not believe they take any part in forming the follicles of the egg. These are formed from the columnar cells of the tube stalk. In the first place, the epithelial cells are never found below the lower limit of the diffuse area, while the follicles are always formed below this level (Fig. 11). In the second place, the folhcle cells (Figs. 11, 12) bear an

Germ Cells of Leptinotarsa Signaticollis. 149

unmistakable resemblance to those of the stalk in early stages (Figs. 1, 2), whereas the epithelial cells are of a totally different appearance.

Both kinds of cells have their origin in the mesodermal somites (Heymons '91, Wheeler '93), but the cells of the ovariole stalk early undergo a differentiation which distinguishes them from the epithelial cells. The former are columnar in outline and stain deeply, while the latter are round or oval and show no affinity for the stain.

In the differentiation of the germ cells into egg and nurse cells, we have a very good example of the general conclusion which Whitman ('93), F. R. Lillie ('06), and others have reached ; that morphogenic processes can not be conceived as merely the sum total or the resultant of the individual cell activities, but that the organism (the ovary in the present case) operates as a unit without respect to cell boundaries. There is no reason for believing that the primordial germ cells differ in nuclear contents ; neither is there any evidence of a differentiating division of the chromatin which would predetermine which of them were to become functional ova and which nurse cells. On the contrary, the process seems to be the result of the activity of several distinct cell elements which operate together as a whole.

The Formation of the Egg Strings.

The development of the egg string was first studied by Leydig ('66), who regarded it as a portion of the protoplasm of the egg. Its relation to the nurse cells does not seem to have been very clear, especially in telotrophic ovarioles such as are present in L. signaticollis. For a long time it was supposed that the eggs of these forms were without egg strings and that the follicle cells supphed the ovum with nutrition. Will ('85) gives a diagram (Fig. 11) showing the relation of the strings to the eggs in Nepa and Notonecta that portrays the condition in my material, but he says nothing about their formation or the nature of their connection with the nurse cells.

150 Harry Lewis Wieman.

In L. signaticoUis, at the appearance of the semi-fluid condition described at the junction of the egg tube with its stalk, the germ cells exhibit signs of amoeboid movement and move away from each other; the first step in the process being shown in Fig. 9. The young eggs are just entering the growth period, and are taking positions in a linear series. As they separate they leave behind a strand of protoplasm which comes into relationship with the nurse cells through the medium of the matrix. Fig. 11 is a rather fortunate section showing a number of egg strings in various degrees of development. It should perhaps be stated that in Fig. 11 the distal end of the ovariole is toward the bottom of the page, whereas in Figs. 6, 7, 8, and 9 the distal end is toward the top of the page.

These nutritive strings are very delicate structures composed of thin strands of the cytoplasm of the egg, drawn out like pseudopodia, and owing to their transparency, are very easily overlooked. At the points where they connect with the nurse cells, the strings blend with the intercellular region.

The nurse cells as shown in Fig. 11, are large polynuclear cells, between which the egg strings terminate. The groove-like spaces between the nurse cells can be compared to ducts into which the nutritive material from the nurse cells is secreted, and from which it is taken by the egg strings into the egg.

As the eggs pass from the semi-fluid region they come into contact with the columnar cells of the tube stalk which form the follicles. After the eggs have taken up their positions, one behind the other in their respective follicles, we find that each string leaves its egg laterally and finds its way back through the follicle cells to the nurse cells (Fig. 12). This figure shows the egg string at the height of its development. Later it disappears and no trace of it can be found in the mature egg.

The real growth period of the egg is initiated with the formation of the nutritive string, and shortly afterward the egg moves down into the tube stalk probably by amoeboid movement or peristaltic action of the ovariole, or both. Korschelt ('86) observed peristaltic movement when the egg tube of Dytiscus marginalis was placed in physiological salt solution. I have examined fresh ovaries

Germ Cells of Leptinotarsa Signaticollis. 151

of Leptinotarsa dissected out in saline solution, but have been unable to detect peristalsis. If it does occur, the action is so slow that it is not readily perceptible to the eye. The irregular outline of the egg at this time points to amoeboid movement as the motive force, although fresh preparations revealed no such movement, but here again it may be so slow as to elude detection.

The Nutrition of the Egg

At the end of the division period (Fig. 8), the ovogonium is nearly all nucleus: the cytoplasm being very small in amount, and pale and almost transparent. With the beginning of the growth process, the cytoplasm undergoes a complete change, taking on a granular appearance and staining deeply. After safranin and lichtgriin, a purplish tint is produced, which is due to the combined color effect of a green acid staining reticulum, through which are scattered more or less uniformly small red granules taking the basic stain. From this point the egg increases rapidly in size, and shows no striking changes in the cytoplasm until it reaches the stage shown in Fig. 11.

The nuclear contents, in the meantime, have undergone considerable transformation. The chromatin passes from the spireme stage of Figs. 9, 41, 42, into a delicate irregular thread staining with safranin (Fig. 43), which gradually loses its sharp outline and its affinity for the basic dye, until finally (Fig. 44) it becomes a filmy, feathery, green staining mass, of irregular outline. At the same time a number of rounded basic staining nucleoli make their appearance.

While the above changes have been taking place a steady stream of nutritive material has been pouring into the egg by way of the egg string. I agree with Kohler ('07) where he states (p. 378) that: Die Follikel-epithelzellen leisten keinen Betrag zur Ernahrung der Oocyte, dagegen liegt ihnen die Produktion des Chorionbildung materials ob." This conclusion that the follicle cells are concerned with the production of the egg envelopes is in keeping with the view of their origin from the cells of the ovariole stalk, since the latter from the very beginning show in their



Harry Lewis Wieman.

deeply staining cytoplasm the evidence of a secretory function that is entirely absent in the epithelial cells. This view of the origin of the follicles is different from that prevailing in the literature, but I am inclined to believe that a similar method of follicle formation may be found to be of much wider occurrence in other species of insects than is at present supposed.

Fig. X. Photograph of a longitudinal section of a half-mature ovum showing the form of the nutritive stream, n, nucleus; n. s., nutritive stream.

The red granules appearing with the beginning of the growth period, enter the egg by way of the egg string, and are at first evenly distributed throughout the cytoplasm. As the egg string increases in size, the granules are supplied in larger amounts than can be disposed of, and the result is an accumulation of

Germ Cells of Leptinotarsa Signaticollis. 153

the nutritive material extending from the mouth of the nutritive string to the nucleus and even slightly beyond, as shown in Fig. 11. This is from a preparation killed with Flemming's solution, and owing to this method of fixation, the food stream is not differentiated as clearly as in material prepared in other ways. The fact that the nucleus is at this time, composed largely of an acid-staining ground substance, while the granules of the food stream are basic-staining, points to the existence of a chemical attraction between them.

Preparations made with picro-acetic acid, and stained with safranin and lichtgriin were found to be very valuable in studying the nutritive process, and the following account is based on such material.

Fig. X represents a longitudinal section of an egg at the height of the functional activity of the nutritive string. The dark area shows the region where the red basic-staining granules are thickest, and the lighter area where the acid-staining material predominates. The configuration is quite different from that described as typical of the early growth period.

The nucleus (n.) shows a green acid-staining ground work in which are embedded a number of more or less vacuolated basicstaining nucleoli of various sizes. The egg string (n. s.) is seen leaving the egg at the lower end of the follicle, only a short portion of it showing because of a bend in the tube. This figure shows clearly that the nutritive material is a basic-staining substance; that is, a compound containing an organic acid resembling the nucleic acid of the nucleus in its ability to unite with the dye. Nucleo-albumens have been known for a long time to occur in the yolk of eggs, so that it is very likely that the acid constituent of the nutritive stream is nucleic acid in one form or another. As might be expected this material shows important differences in staining reaction from the chromosomes, since the latter, if they are to be considered identical with the contents of the heads of spermatozoa, yield only phosphoric acid and xanthin bodies as splitting products, while the nucleo-albumens (pseudonucleins) yield protein and phosphoric acid, but no xanthin bodies. Thus while the chromosomes always stain deeply with safranin regard

154 Harry Lewis Wieman.

less of the killing agent employed, the food stream does not show the same constancy in behaviour. After Flemming's solution it is almost impossible to demonstrate the nature of the food stream by means of basic dyes. More satisfactory results are obtained after picro-acetic acid, but even with this, the food stream can be made to take the acid stain. Similar staining reactions have been observed by many workers in the case of other nucleins, and Mann ('02), p. 339, in his criticism of A. Fischer's views, gives a very good explanation of this behaviour. ' ' When therefore, Fischer observes nucleins to stain readily with basic dyes and only after some delay with acid dyes, it means that the basophil an-ion, nucleic acid, has its basic tendencies incompletely satisfied by the kat-ion albumen radical, and for this reason it readily absorbes some more kat-ions, namely the color base. Conversely, if the nucleic acid is not sufficient to satisfy the demand of the kat-ion albumen, as is the case in those compounds which contain only a little nucleic acid, then some more an-ions are attracted, namely the color acid an-ions." The variations in staining reaction therefore, are not to be taken as indicative of a purely physical or rather mechanical, as opposed to a chemical union between the dye and the substance dyed.

The nutritive stream has undergone considerable change in form since the preceding stage (Fig. 11). On entering the egg (Fig. X) it now divides so as to enclose a more or less pearshaped portion containing the nucleus, leaving a narrow free margin at the periphery of the egg. The granules are therefore distributed in the form of an oval shell enclosing the yolk which has begun to form in the center of the egg.

Closer inspection of the cytoplasm shows a green-staining reticulum having much coarser meshes than in earlier stages and interspersed with granules from the food stream. In the photograph the lighter areas show the regions where the reticulum is practically free of granules. Fig. 33 represents a section where the two zones adjoin. The food stuff spreads out along the lines of the reticulum toward the center (to the left in the figure) and the periphery (to the right) of the egg in the process of yolk formation. In the course of this process, the granules disappear

Germ Cells of Leptinotarsa Signaticollis. 155

and the large polygonal masses of acid-staining yolk are produced in the meshes of the reticulum. In the mature egg, the reticulum is represented by the cytoplasm of the interdeutoplasmic spaces.

This reticular structure may not be present as such in the living egg, but granting that it is an artifact resulting from re-agents used in fixing, the structure is one that varies in size and form at different periods in the development of the egg ; and thus may be regarded as representing regions of varying chemical or physical consistency in the living ovum, that indicate the paths of distribution of the food stream.

Korschelt ('89) observed in the egg of Dytiscus marginalis that the granules from the nurse cells enter the ovum and migrate in a broad stream toward the nucleus, which actually exhibits amoeboid movements, sending out pseudopodia-like processes toward the granules. These form changes, which were observed in both living and fixed material, are regarded as manifestations of an attractive force exerted between the nucleus and the granules. Somewhat similar processes were observed in the eggs of Carahus, Bombus and Apis.

I have not found such pronounced evidence of nuclear movement in either the living or fixed egg of Leptinotarsa. Fig. 11 represents conditions comparable to what Korschelt has described in Dytiscus, although the nucleus does not show any change in form. However, the facts are such as to indicate the presence of an attractive force of some sort, probably of a chemical nature, between the nucleus and food stream.

At the periphery of the mature egg and enclosing the yolk, is found a narrow layer of protoplasm continuous with the interdeutoplasmic cytoplasm . Since the yolk is elaborated from within out, being first formed in the center of the egg, it appears that the outer layer of cytoplasm together with the reticular part represent regions where the cytoplasm of the primordial germ cells has remained undifferentiated.

The evidence all goes to indicate that the substance of the food stream is identical with what has been called the yolk nucleus" by a large number of writers, (Stuhlmann '86, Jordan '93, Balbiani '93, Calkins '95) in the eggs of spiders, myriapods, amphibia,

156 Harry Lewis Wieman.

insects, and earthworms. In all these cases the action of differential stains indicates that the substance of the yolk nucleus is nearly related to chromatin, as I found to be the fact with the nutritive material in the egg of Leptinotarsa. The close proximity of the yolk nucleus to the germinal vesicle has led many workers to regard it as derived from the chromatin. Possibly when more complete data are at hand, it may be shown that in these forms, as in Leptinotarsa, the yolk nucleus is nothing more nor less than a nutritive stream that has its origin in cells outside of the ovum; that is, in the follicle or nurse cells, as the case may be.

It is rather interesting that the chromatin and the granules of the nutritive stream should show similar reactions toward the basic dyes. The granules are used in the formation of the yolk, and the product resulting from this transformation takes the acid stain. In the development of the embryo there is a reversal of this process. At fertilization the germ nuclei unite in the center of the egg, and the cleavage nuclei become scattered about in the yolk. These nuclei, which take the basic stain as intensely as the nutritive granules or the chromatin, then migrate in part to the pheripheral layer where they form the blastoderm, and in part, remain in the yolk as the yolk nuclei. Through the activity of the latter, the yolk is converted into a form which can be assimilated by the growing embryo. Thus the agencies which are concerned with the conversion of the inert yolk into living protoplasm, the yolk nuclei, are somewhat similar in chemical make-up to the material from which the yolk is elaborated, the granules of the food stream. The entire process bears considerable resemblance to a reversible chemical reaction.

It is well known from the work of Loeb ('02) and Mathews ('07) that the eggs of the starfish, Asterias Forhesii, if allowed to mature in sea water in the presence of oxygen soon die, unless fertilized. Mathews has shown that the early death of the egg after maturation occurs only when free oxygen is present : from which it is concluded that death is brought about by the oxidation of the cytoplasm. Furthermore, this takes place much more rapidly after the contents of the germinal vesicle have been discharged into the

Germ Cells of Leptinotarsa Signaticollis. 157

cytoplasm than before. If the nuclear wall remains intact, the egg does not become opaque even in the presence of oxygen. According to Mathews the sperm brings into the egg cytoplasm, which already contains an oxydase, two substances: a reducing agent, the centriole, which counteracts the action of the oxydase of the cell cytoplasm, and a very active nucleus which growls rapidly and forms more reducing substance, and possibly some oxydase. ' ' By the entrance of the sperm there is set up that extraordinary series of opposite actions of oxydations and reductions which accounts for the sudden bursts of respiratory activity which probably underlies many of the most important syntheses and chemical transformations in protoplasm" (p. 107).

It has been shown by Fischer ('99) and others that the granules in the living starfish egg take the basic stain and are therefore electro-negative. It is also well known that a region of intense reduction will act as a negative electrode, from which it follows that granules staining with basic dyes are to be regarded as reducing substances.

The conditions in the growing egg of L. signaticollis are of course somewhat different from the material on. which the above conclusions were based; but there are a number of very suggestive points of resemblance that should be considered.

In the first place, if one may judge at all from staining reactions, the food stream coming from the egg string consist of particles of a reduced substance which, through the activity of the oxydase in the cytoplasm, is converted into an inert oxidation product, the yolk. After fertilization every cleavage nucleus represents a region of intense reduction which reacts with the oxydized yolk converting it into living protoplasm.

This hypothesis rests on the assumption that the staining reaction of a substance is an indication of its chemical nature. A large number of investigators (Ehrlich '91, Mathews '98 and others) have demonstrated very clearly that staining with aniline dyes, depends upon a chemical union between the dye and the substance dyed. However, it often happens that in the process of fixation, the chemical reaction of the tissue is made opposite from what it was in the living condition, and this is especially true when the

158 Harry Lewis Wieman.

salts of heavy metals are used. For this reason I selected picroacetic acid as being least objectionable.

Any process of fixation produces chemical changes in protoplasm which make the fixed material quite different from the living. Whether or not a complete reversal in chemical reaction occurs depends upon the amount of nucleic acid present. If this is present in more than sufficient quantity to counter-balance the tendency of the killing fluid to make the protoplasm positive, the staining reaction would, of course, be the same as that in the living material. This is undoubtedly true of the chromosomes, and the evidence goes to show that it is probably true of these granules, although the proportion of nucleic acid in the latter is much less than in the former.

In any event the presence of a cyclical process with alternating phases of acidity and basicity is perfectly evident, since the same killing agent is used at all times, and its specific action may be assumed to be the same at all stages. The variable factor is the protoplasm; the changing proportions of the basiphil anion, nucleic acid, and the kation albumen radical, being responsible for the different staining reactions obtained.

Furthermore, these staining reactions are in keeping with what one would expect from the theoretical side of yolk formation. Yolk is regarded as an inert chemical substance, which implies a low degree of chemical activity. Its reaction to the stain shows it to be in an oxidized condition. The nutritive stream which is the product of living cells shows an opposite reaction, indicating chemical properties opposite to those of yolk. In the process of forming yolk the basic staining granules of the food stream disappear, and the acid staining yolk comes into existence as a product resulting from the interaction of nucleus, cytoplasm and food stream.

The Nurse Cells

The nurse cells are descendants of the primordial germ cells whose reproductive function has been lost, and therefore they are to be regarded as aborted eggs. The first result of the differen

Germ Cells of Leptinotarsa Signaticollis. 159

tiation of the nurse cells has already been noted, and consists in the chromatin of the nuclei taking on a granular appearance, whereas in the young ovocytes the chromatin is in the form of a spireme. It has also been pointed out that the nurse cells undergo mitotic division for some time after the ovogonia have stopped dividing, but before the formation of the egg-string and the beginning of the growth period, mitotic divisions cease in those nurse cells situated next to the ovocytes. As these cells are the first to be called upon to supply nutrition to the egg, they are the first to take on the characteristic appearance of the nurse cells as found in the adult (Fig. 11), The first step in this process is brought about by the grouping of the cells into cyst-like structures, as shown in Fig. 15. These elongated cysts of polynucleated masses do not maintain their enormous size, but gradually break up into smaller pieces, even into parts containing a single nucleus, and therefore, morphologically equivalent to a single cell. After the cysts have appeared, the epithelial cells are found in the spaces between the cysts, where their relation to the latter is the same as their relation to the primitive germ cells of early stages (Figs. 4, 6, etc.). This similarity of relationship suggests that the cells of each cyst are descended from a single mother cell, and examination of intermediate stages bears out this idea. In earlier stages as in Fig. 1 the definite relationship between germ and epithelial cells does not exist. However, as development continues each germ cell becomes surrounded by its own set of epithelial cells bringing about the condition shown in Fig. 37, which may be regarded as the first step in cyst formation. From this point, the germ cells, or rather the nurse cells as they now may be called, undergo amitotic divisons of the nuclei, and the resulting polynucleated mass remains enclosed by the epitheUal cells. At first the lines of demarcation of the cysts are not sharp, but as the spaces which develop between them widen, the cysts become more rounded in outline and distinctly separated from one another (Fig. 16, which is a transverse section of an ovariole throughthe region of the nurse cells) . This process begins among the lowermost nurse cells and gradually extends throughout the distal portion of the egg chamber. As the cysts appear at the

160 Harry Lewis Wieman.

beginning of the growth period of the ovocytes, their formation is undoubtedly connected with the functional activity of the nurse cells.

Those of the primordial germ cells that develop into functional ova are not grouped into cysts, and epithelial cells are scattered among them very irregularly (Figs. 6, 7, 8, 9, etc.).

The formation of cysts does not have anything to do primarily with the differentiation of the primordial germ cells into nurse cells on the one hand and egg cells on the other, for in the testis where a similar process of cyst formation occurs, the cells thus inclosed develop into functional spermatozoa. Therefore the process of encystment can not be fundamentally antagonistic to the development of functional germ cells. In the case of the ovary, the differentiating factor has already operated before the cysts are formed.

In the course of cyst formation in the more distal part of the tube, one frequently sees mitosis and amitosis going on side by side (Fig. 13). It is interesting to note that here mitotic figures are never found inside of definitive cysts, where the multiplication of nuclei always takes place by the amitotic mitotic method. The products of mitotic cell division are single cells. The occurrence of irregularities, such as shown in Fig. 19, has been alluded to.

Aside from the absence of the mitotic division apparatus, the amitotic division figure is characterized by the fact that the nuclear membrane persists throughout the entire process (Figs. 20, 21, 22, 23). Further, the chromatin of the nucleus, which is in the form of rounded granules embedded in a reticular network, is separated from the nuclear wall by a clear space that disappears when, the nucleus comes to rest after division (cf. Fig. 16).

The first authentic case of amitosis in the insect ovary seems to have been noted by Mayer (75), who observed the occurrence of doubly-nucleated cells in the follicular epithelium of Pyrrochoris apterus. A little later, Brandt ('78) described in the nurse chamber of Septura rubratisturea" hisquii formige Kerneauftreten die auf Theilungen hindeuten." Will ('85) described similar divisions of the nurse cells in Nepa and Notonecta, and used the facts to good advantage in developing his "Ooblasten" theory.

Germ Cells of Leptinotarsa Signaticollis. 161

At about the same time, Carnoy ('85) noted amitotic divisions in the follicle cells of Gryllotalpa and a little later, Korschelt ('87) made similar observations in the case of Hydrometra locustris. As a result of studying the ovaries of Nepa cinerea and Locustra viridissima, Preusse ('95) described amitosis not only of the nucleus, but of the entire cell, and therefore claimed for the direct method of cell division an active part in the multiplication of cells. De Bruyne ('99), working in part with the same material, maintained that in no case does the division extend to the body of the cell, and that degeneration inevitably follows amitosis. Gross ('03) interpreted amitosis as he found it in eleven species of hemiptera in a slightly different manner. Its occurrence among the nurse cells is regarded as an indication of degeneration, but in the case of the epithelial cells of the egg follicles it has a deeper significance, for here it serves to enlarge the activity of the nucleus by increasing the area of contact between it and the cytoplasm.

The Ziegler-vom Rath theory (Ziegler '91, Ziegler and vom Rath '91, vom Rath '95) briefly stated is as follows: Amitosis appears in old and used-up tissue and consequently also in cells which have a transient significance. It occurs principally in cells which through very marked specialization take on the function of intense secretion or assimilation.

With the exception of Preusse, the unanimous opinion seems to be that amitosis as it occurs in the insect ovary is confined to division of the nucleus and is therefore not to be regarded as a method of cell multiplication. Thus Kohler ('07) says; "Die Kerntheilung der Nahrzellen und Follikelzellen die amitotisch verlaufen, fiihren nie zu Zelltheilung. Diese Teilungen sind keine Vermehrungsteilungen sondern Differenzierungsteilungen. Sie bezwecken gar nicht eine Zellvermehrung und konnen deshalb in keinen Weise gegen die Moglichkeit eine propagativen Beweiskraft besitzen." This conclusion may be accepted as the general trend of opinion. Preusse's figures are far from convincing, and his conclusions regarding this point have been severely criticized.

In Leptinotarsa it has been shown that the large polynucleated masses eventually break up into single cells, and if this is to be looked upon as the last step of the amitotic process that started

162 Harry Lewis Wieman.

in the nuclei, amitosis is here certainly concerned with cell multiplication. This protracted form of cell division is quite different from Preusse's descriptions, nor is it necessarily at variance with the observations of other authors ; for unless a continuous series of stages are studied, it would never be suspected that the amitotic divisions which are at first confined to the nucleus, later extend to the cytoplasm.

The striking feature of the amitotic figure is the fact that the nuclear membrane does not disappear, but remains intact throughout the entire process of division. This fact together with the staining reactions serves to throw some light on the fundamental difference between mitosis and amitosis.

It has long been recognized that the dissolution of the nuclear membrane is in some way associated with a well defined alteration in the capacity of the egg for further development. From the observations of Delage ('01), it would appear that the essential feature of maturation is not so much the separation of the polar bodies as the removal of the barrier between the nuclear and cytoplasmic areas. The critical event which causes a change in the cytoplasm is the passage of nuclear constituents into it. The result may be a change in the osmotic pressure of the cytoplasm or in its rate of oxidation. Delage also observed that enucleated egg fragments of Asterias are incapable of fertilization before the germinal vesicle has broken down, but that very soon after the membrane shows signs of dissolution, merogonic cleavage becomes possible.

In his study of the karyokinesis of the Crepidula egg, Conklin ('02) arrived at conclusions which are important in this connection. The nuclear membrane appears to permit the passage of materials inward, but not outward during the resting period, whereas the escape of nuclear material into the cell is brought about by the disappearance of the membrane during karyokinesis. " He also determined cy tologically ' ' a very extensive exchange of material between the nucleus and cytoplasm. A large part of that most characteristic nuclear substance, the chromatin, passes into the cytoplasm during every cell cycle, while a relatively small part is reserved for the purpose of reproducing the daughter

Germ Cells of Leptinotarsa Signaticollis. 163

nuclei." The passage of the nuclear material into the cytoplasm is regarded as a fundamentally important condition to the subsequent changes undergone b}^ the latter.

Wilson and Mathews ('95) have shown that by far the greater part of the chromatin is set free in the cytoplasm in the first maturation division of the starfish egg, and F. R. Lillie ('06) states that in Chcetopterus the greater part of the germinal vesicle consists of a ^'residual substance" which is set free in the cytoplasm of the first maturation division, and plays an important part in the future development.

If these phenomena are characteristic for mitosis in general, as they seem to be, the rupture of the nuclear membrane permits the escape of some substance into the cytoplasm that is essential to the changes which follow. Therefore, if the membrane remains intact during division, a difference in cell metabolism is certain to take place.

Lyon ('04) has shown that the production of carbon dioxide by the dividing egg follows a rythm parallel with that of the nuclear division, and Loeb ('06) has connected these oxidations with the synthesis of nucleins from the compounds of cell metabolism — a process which likewise undergoes a rythm parallel with that of the mitotic process.

Mathews ('07) suggests that the periodic dissolution of the nuclear membrane in mitotic cell division might have the significance of providing for the distribution of the oxydases (synthesized in the nucleus) through the cytoplasmic area which would naturally result in a periodic acceleration of oxydative processes in the cell.

R. S. Lillie ('08) points out that as certain enzymes exhibit the properties of nucleoproteins, it is reasonable to regard the so-called oxy chromatin" or "residual substance" as consisting, at least in part, of ferments concerned in the chemical processes — largely oxidative in nature as shown by the condition in the starfish egg — that determine the later characteristic changes in the cytoplasm. In this way the physiological data can be readily reconciled with cytological observations.

The failure of the nuclear membrane to dissolve in the course

164 Harry Lewis Wieman,

of amitotic divisions in the nurse cells would result in the cytoplasm being deprived of the proper amount of oxidative ferments, and owing to the diminished rate of oxidation in the cytoplasm there would be an accumulation of unoxidized substances. The evidence derived from staining reactions justifies such a prediction.

At about the time of the fragmentation of the large polynucleated masses, a change occurs in the staining properties of the cell constituents toward aniline dyes. Up to this point the nucleus takes the basic dye while the cytoplasm takes the acid; but now a complete reversal is to be noted, the nuclei staining a deep green, while the cytoplasm, filled with the granules of the food stream, stains deeply with the red (Fig. 24).

The change is not an abrupt one, but begins gradually, shortly after the cessation of the amitotic divisions of the nuclei when the latter pass into a kind of resting state. Just before the change, one or more large basic-staining granules about the size oi a nucleolus appear in the cytoplasm either closely applied to the nuclear membrane or at various distances from it (Figs. 14, 24, gr.) In younger stages these granules are found inside of the nucleus. As conditions between the intra- and extra-nuclear position of these bodies are not wanting, it seems clear that they arise in the nucleus. Whether or not these granules represent part of the chromatin contents of the nucleus that is being cast out into the cytoplasm as a result of degeneration or intense secretory activity, is a matter of speculation. It is certain that the chromosomes never appear subsequently. On reaching the cytoplasm the granule either breaks up into smaller particles or dissolves gradually without first disintegrating.

The smaller basic-staining granules of the food stream are found at the nodal points of a reticular network (Fig. 24, etc.). The latter is possibly an artifact, but the granules are probably present as such in the living egg, being comparable to the zymogen granules of the pancreas.

Thus, as far as the staining reactions are concerned, the nucleus and cytoplasm have exchanged places, and this peculiar inversion of the natural order of things seems to be the result or accompani

Germ Cells of Leptinotarsa Signaticollis. 165

ment of amitosis when continued for a number of cell generations. It has already been pointed out that the basic-staining property of the cytoplasm is probably due to incomplete oxidation of substances which consequently remain in a reduced condition. Inside of the nucleus on the other hand, the retention of the oxydases results in a high degree of oxidation; hence the capacity of this region to stain with the acid dye.

If these changes in chemical properties are due to the failure of the nuclear membrane to disintegrate periodically, we can readily see how amitosis in the nurse cells is related to the preparation of the basic-staining granules of the food stream. Furthermore this conclusion is in perfect keeping with the physiological observations of other workers which support in every way the interpretation put forth here.

We now come to a series of phenomena which represent a most peculiar form of cell activity. Shortly after the red nucleolar granules are extruded from the nucleus, one notices here and there a basic staining spot in the green nuclei of these cells (Figs. 27). Ordinarily, this does not occur in a singly-nucleated cell, but in one of the nuclei of a polynucleated mass (Fig. 25). This spot increases in size and forces the green area to the periphery (Fig. 28) . Finally as a result of the continued expansion of the central part, the original nucleus and cytoplasm become reduced to mere shells (Fig. 31); but before it has reached its largest size, a green area appears in the center of the central red area as shown in Figs, 25, 26, 29, 30, 31. About this time, or even before (Fig. 28), the cell with its concentric layers usually becomes pinched off from the polynucleated mass (Figs. 26, 31). Fig, 32 shows a peculiar condition in which the central part (nucleus?) is dividing.

This remarkable series of changes begins in the distal region of the ovariole at about the time the first batch of eggs are halfgrown. Since all of the nurses cells do not pass through this cycle of changes, the phenomena can not be regarded as degenerative processes, nor would they seem to be concerned with the elaboration of nutritive material, for the reason that so few cells are involved.

The fact that in the final stage as shown in Fig, 30, the central

166 Harry Lewis Wieman.

area with its green nucleus bears some resemblance to an egg in early stages, led to the suggestion that this might be a part of the regular development of the egg. However, I soon found that most of the eggs do not undergo such transformations, but develop as has been described, from the germ cells at the base of the chamber.

Inasmuch as these peculiar configurations are of no significance as manifestations of the general degeneration associated with specialized function of the nurse cells, or in the production of the nutritive stuff for the eggs, or lastly, in the normal process of eggbuilding; I have reached the conclusion that the process is one of phylogenetic significance, in which certain of the nurse cells are passing through alternating conditions of oxidation and reduc

FiG. Y. Semi-diagrammatic drawing of the testis, one of the lobes being turned through an angle of ninety degrees, c, cap-like region opposite the sperm duct; /, follicle; sp. d., common sperm duct of one side.

tion similar to those undergone by the egg (for it is to be remembered that these cells are really germ cells that have lost their reproductive function). There is no evidence that functional eggs result from this process. It can be readily shown that the eggs that are differentiated at the base of the chamber do not exhibit these changes. The possibility of two methods of egg formation might be considered, but there is no indication of dimorphism in the offspring, such as would be expected. On the other hand, the possibility of reproductive cells from either source producing identical offspring, meets the objection that there are more than enough germ cells differentiated at the base of the egg chamber to account for all of the eggs laid.

Germ Cells of Leptinotarsa Signaticollis. 167

To my knowledge, the literature is without data bearing on these later transformations in the nurse cells. Korschelt ('86) described conditions in the young eggs of Dytiscus which resemble these changes to a certain extent, but his descriptions are too meager to admit of much comparison. I am inclined to think that most workers have regarded such appearances as the result of degeneration, and have, therefore, passed over them as of no particular interest.

II. The Development of the Testis

The testes of the adult consist of a pair of bean-shaped lobes on either side of the body (Fig, Y). Two ducts, one from each lobe, unite to form a common duct (sp. d.) on one side, which in turn joins with its fellow from the other side to form the median ejaculatory duct. Internally each lobe is made up of radiating follicles, containing cysts of germ cells in various stages of maturation, while in the center is a lumen filled with mature spermatozoa during the breeding season. No suspensory ligament comparable to the end thread of the ovariole is present, the organ being supported by the fat bodies, tracheae etc., which are packed about it.

In the early stages of development the organ bears but slight resemblance to the adult, each lobe being somewhat spindle-shaped and looking very much like a single ovariole at a corresponding stage (Fig. Z, A).' Histologically, as in the ovary, two kinds of cellular elements can be distinguished — germ cells and epithelial cells (Fig. 17, 18). Here too, there is the same tendency for several of the latter to be grouped around a single germ cell (Fig. 36) . Fig. 17 shows the first stage in cyst formation in which, as in the ovary, the epithelial cells exhibit the same relationship to the completed cysts as they do to the single germ cells of earlier stages. The contents of each cyst are the descendants of a single mother-cell.

1 Each lobe terminates in a cap of epithelial cells resembling the expanded base of the terminal thread of the ovariole of the adult female. A consideration of the significance of these cells together with a more complete description will be given in a future publication.


168 Harry Lewis Wieman.

Sections of the late larval testis show that cyst formation has already occurred in the proximal part of the testis, i. e., the part nearest the sperm duct, while in the more distal region they have not yet formed.

The spindle shape of each lobe persists during the greater part of the larval stage. The first change consists in an increase in diameter, while the length remains practically the same, which results in the production of a pear-shaped body (Fig. Z. B). During the pupal period the growth at right angles to the original axis of the lobe continues, but not equally in all directions, being inhibited at certain points, and producing a structure which in section resembles the hub and spokes of a wheel. The process continues in this manner until the radiating follicles characteristic of the adult are formed (Fig. Y) .

At the point opposite from where the sperm duct leaves each lobe, a button-like cap exists (Fig. Y. c) in which the cells are very much 3-ounger than in any otlier part of the testis. This region represents the distal end of the embryonic organ, as can be seen from the method of development.

The epithelial investment covering the organ is at first very loose and does not extend in between the follicles, and when the testes are removed from the body at this time, the delicate epithelium falls away and the naked organs are obtained. In late pupal stages, the epithelial cells become more closely applied, and adhering between the follicles, soon produces the appearance characteristic of the adult, in which the folh'cles are separated from each other by a thick layer of these cells. Through increased number of cell divisions and the growth processes accompanying the maturation of the germ cells, the follicles become greatly enlarged and press tightly against the epithelial partition causing the latter to appear as a definite part of the follicle.

In connection with the early stages in the development of the ovary, the occurrence of amitotic divisions in the primordial germ cells was noted. At a corresponding stage the same phenomenon is to be seen in the testis. Three steps in the process are shown in Figs. 67, 68 and 69, the appearance of which closely resemble Figs. 34 and 35 from the ovary.

Germ Cells of Leptinotarsa Signaticollis. 169

When about to divide, the nucleolus appears as an abnormally large basic-staining body surrounded by a perfectly clear area that serves to intensify the vividness of the dye. Fig. 68 shows the separation of the daughter halves during which process the clear area takes the shape of an hour-glass. The constriction then extends to the nucleus itself (Fig. 69), and eventually to the entire cell.

The amitotic period for any given cells is not of long duration, not more than perhaps several cell generations; although this could not be determined very precisely. It begins in the larval period at about the time of cyst formation, with the early development of which it is concerned. At any rate, in the earliest stages at which the cysts can be recognized, they are filled with cells undergoing amitosis. A little later these same cysts are filled with mitotic spindles, after which no evidence of direct division is to be found. It is thus seen that the early stages in the building of cysts in the testis are practically the same as those described for the formation of the analogous structures among the nurse cells of the ovary.

While the appearance of the amitotic figure here is entirely different from that found in the nurse cells of the ovary, there are several points of resemblance. In the first place, the nuclear membrane remains intact in both cases. Secondly, the chromatic part of the cell is surrounded by a clear region; extra-nuclear in the case of the nurse cells, and intra-nuclear in the cas'e of the primordial germ cells of both ovary and testis. Thirdly, in both, the process is connected with cyst formation, except in the primordial germ cells of the ovary.

Perhaps the most striking differences are that in the nurse cells direct division continues indefinitely, resulting in certain chemical changes in nucleus and cytoplasm, and that it is not followed by mitosis; while in the germ cells it is, of relatively short duration and is followed by mitosis: the cells developing into functional germ cells. Evidently the process does not have the same significance

the two cases.

The fact that the amitotic divisions occur in the germ cells of the ovary for only a very brief period makes it somewhat difficult


Harry Lewis Wieman.

to detect the process. In the testis it is more readily discovered, for here, even in the adult, when the organ is packed full of spermatozoa, amitosis can be seen among the spermatogonia lying in the cap-shaped region opposite the point where the sperm duct leaves. Many of these cells have not yet been grouped into cysts, and all graduations between this condition, which is really characteristic of the larval stages, to well defined cysts, can be found.

sjig da

Fig. Z. Diagram of longitudinal sections showing four stages in the development of the testis. A, larva; B, larva two days older; C, pupa; D, adult, d. a. area of degeneration; sp. d. sperm duct; spg. spermatogonia, filling cap-shaped region opposite the mouth of the sperm duct; spz. spermatozoa.

In the examination of a large number of sections of the adult testis, I was surprised to find in all of them certain definite regions in which degeneration of cells occurs. Further study showed that the phenomenon is a regular event in the development of the organs. (I have yet to discover a testis in which it is not found.)

Germ Cells of Leptinotarsa Signaticollis. 171

When adult stages such as represented m Fig. Z, D, are examined, the degenerated area {d. a.) is seen to lie directly beneath the button-shaped region (spg.) and to be encapsulated in a definite covering of epithelial cells, which sharply mark it off from the surrounding tissues.

The process begins in the larva and extends throughout the pupa and adult stages. The first step is an accumulation of epithelial cells in the center of the testis, reaching from the opening of the sperm duct back to the cap-shaped area (Fig. Z, C). The region becomes filled with irregular cell fragments that stain deeply with basic aniline dye or with iron-alum-haematoxylin. The cells remaining intact take on a shrivelled and wrinkled appearance and exhibit indications of amitosis, although the latter could not be made out very distinctly, owing to the shrunken condition of the tissue. As the walls of the sperm duct (sp. d.) grow into the testis, the degenerated area retreats, and finally becomes encysted by the surrounding epithelial cells.

Fig. Z, D, is from a young adult and shows the relation of the degenerated area to the surrounding regions. Above it at spg. is the cap-shaped part filled with spermatogonia, and below it is the general cavity of the testis packed full of spermatozoa (spz).

I have noted this curious condition not only in L. signaticollis which had been reared in breeding cages, but also in wild L. decemlineata, collected from potato vines growing in the open. Therefore the process can not be regarded as a pathological one produced by unnatural conditions accompanying confinement in a cage, but is an event that has a normal physiological significance, which will be considered presently.

Amitosis in the primary spermatogonia of certain amphibia has been described by La Vellette St. George ('85), Meves ('91), Benda ('93), and McGregor ('99), and the descendants of these cells are said to become functional spermatozoa. King ('07) states that in Bufo lentiginosus amitosis never occurs. An irregular outline is characteristic of the nuclei of the primary spermatogonia, but the constrictions never lead to actual division. Nussbaum ('01) describes ' ' maulbeerf ormige Kerne," a condition which he maintains is merely a step in the process of mitotic cell

172 Harry Lewis Wieman.

division and not an indication of amitosis or degeneration. This author observed nuclei whose sinuous outlines simulate stages of direct division in the egg of Rhahditis nigrovenosa and Ascaris megacephala, and in the spermatogonia of Rana fusca. These cells in all cases subsequently divide mitotically.

On the other hand, the work of Gerassimow ('92), Nathanosohn ('00) and Haecker ('00) demonstrate experimentally at least, that mitosis may follow amitosis and vice versa in both plants and animal cells. In these experiments it was found that under the influence of low temperature or narcotics, mitotic divisions cease, and amitosis ensues. When the cells are brought back to normal conditions mitotic divisions reappear.

Child ('07 c) has reported the occurrence of amitosis in the representatives of six different animal phyla, in practically all kinds of tissue including the reproductive cells. In general, his observations indicate that amitosis is more frequent than mitosis in connection with a rapid cell multiplication which accompanies normal and regulatory growth. Child speaks of mitosis as a cyclic process in that the nucleus, starting from the resting stage, undergoes during every mitotic division a series of changes and finally returns to a resting stage similar to the starting point. In amitosis there is no such cyclic movement. Nothing in the visible phenomena indicates the occurrence of reversal in direction of the processes involved." In other words the nuclei of these cells are not in conditions of equilibrium with the cytoplasm.

The most interesting of his observations (Child '07 a, '07 b) deal with, the development of the ovaries and testes of Moniezia, in which the early divisions of the germ cells are amitotic. Then comes a growth period at the beginning of which a spireme is formed, followed by the maturation divisions, which take place mitotically, as do those of the early cleavage. The remaining divisions are amitotic.

Patterson ('08), from observations made on the blastoderm of the pigeon's egg, came to the following conclusions. 1. Mitosis may follow amitosis and vice versa. 2. Amitosis is the result of a physiological stimulus which creates a stimulus to growth. 3. Amitosis exists in regions of rapid growth. Maximow ('08)

Germ Cells of Leptinotarsa Signaticollis. 173

described amitotic cell division in the stellate cells of the body mesenchyme. Like Patterson and Child, he believes it occurs in regions of rapid cell proliferation, and attributes to it a normal physiological significance of some kind.

The occurrence of amitosis in so many different forms and such a variety of circumstances makes it imperative that this type of cell division be recognized as a factor in normal developmental processes. In view of so many accounts, only a few of which are mentioned above, the fact of its existence must be generally recognized, but the question of its significance and relation to mitosis remains without a satisfactory answer.

In the species under consideration, I have described two distinct types of amitosis. In the case of the nurse cells, the process is evidently concerned with the differentiation of these cells for a highly specialized function. Mitosis never occurs afterward. The existence of the direct form of cell division under these ct)nditions has been accepted without much question since it was regarded as a species of degeneration, and further, was in no wise antagonistic to the hypothesis of the individuality and continuity of the chromosomes. In other cases where the claim has been made that mitosis may follow amitosis, so long as only somatic tissue was involved, no very serious objections have been made. However, when we come to instances of amitosis among the germ cells that later develop into functional reproductive cells, the supporters of the chromosome hypothesis have found it very difficult to accept the results of such observations.

Bearing in mind the theoretical interest centering about this point, I have been very careful to examine the ground thoroughly before stating definitely the occurrence of amitosis in the germ cells of Leptinotarsa. I have already pointed out that it is merely transient and inconspicuous in the ovogonia. In the spermatogonia it is more prominent, persists longer and is involved in the formation of the cysts.

The experiments of Gerassimow ('92), Nathansohn ('00), and Haecker ('00), have suggested to me that disturbances in the nutrition may be responsible for the amitotic period; the effect of narcotics and low temperature being similar to what one migh*:

174 Harry Lewis Wieman.

expect to result from a reduced oxygen supply. A very rapid increase in cell multiplication would cause a temporary diminution in the oxygen supply for each individual cell. The stimulus to increased cell division is a distinct factor that is bound up in the process of building cysts of the testis. As soon as the sudden increase in number of cell has been compensated by a corresponding increase in oxygen supply, the conditons required for mitosis are restored.

Child's observations upon the cestode Moniezia offer further suggestions along this line. This form is not only relatively low in the animal scale, but one which, inadditionhas undergone degeneration : a combination of circumstances which would lead one to expect primitive methods in cell division as well as in general metabolic processes. Here amitosis appears to be the regular method of cell division, while mitosis is comparatively rare, occurring only during the maturation period and early cleavage stages.

What causes the change from mitosis to amitosis? It has appeared to me that here likewise a gradual diminution in nutrition is responsible. It might be assumed that the object of the long rest stage or growth period in the development of the ovum is to elaborate food and formative materials for maturation, fertilization and embryonic development. We might further assume that the same process provides for a certain number of mitotic divisions extending through the early cleavage. The direct method of cell division sets in because of a deficiency in the amount of nutritive material (oxygen?) necessary for continued mitotic divisions. This is in keeping with the fact that amitosis occurs usually under abnormal metabolic conditions which are unfavorable to normal metabolic processes. Amitosis might be regarded as a simpler form_ of cell division, not so much because it takes place in the absence of spindle and chromosomes, as for the reason that it can occur under circumstances that make mitosis impossible.

In accordance with this idea one notices that amitosis has been observed most frequently under conditions of rapid growth, where if this explanation is applied ,the cell makes use of the direct method of division rather than follow the slower and more complex mitotic method largely because of limitations in the way of

Germ Cells of Leptinotarsa Signaticollis. 175

nutrition. The same principle can be applied in the case of the nurse cells. When the latter are being differentiated for their specialized function, very rapid and prolonged amitotic divisions occur. In degenerating cells amitosis is to be expected, because in many cases at least, the cells or tissues do not receive their normal oxygen supply.

Amitosis in the higher forms as a survival of a primitive process of direct division from the protozoa was suggested long ago by Strassburger ('82) and Waldeyer ('88), but the view has never met with much favor. This has been largely because of the popular and growing tendency since that time to associate amitosis with processes of degeneration and specialization, reserving for mitosis exclusively the role of cell multiplication in normal processes. However, there is much to support the older view, and the evolution of the mitotic method of cell division is, in a way, the expression of the evolution of a higher type of cell metabolism than that found in the lower forms. The raison d' etre of the mitotic figure must rest upon a physiological basis, and the complexity of the mitotic cycle appears to be associated with a corresponding complexity of metabolism of a higher order than that found in cells that divide amitotically. In this sense amitosis in the germ cells of the testis and ovary of Leptinotarsa can be regarded as a temporary reversion to an ancestral method of division but the direct cause I believe, lies in some disturbance in the cell metabolism which occurs periodically in the ontogeny.

The appearance of the amitotis division figure is by no means the same in all cases, and this is of importance from the phyletic stand point. In nearly every instance, the process indicates a division of the nucleus into two approximately equal parts,but among the germ cells the mechanism is more carefully adjusted for this purpose (cf. figs. 21, 22, 23, 34, 35, 68, 69, etc.) Thus in the ovogonia and spermatogonia division of the nucleus is preceded by a very exact division of a large chromatin nucleolus, and as the halves separate surrounded by the clear area, the appearance reminds one very much of the division of a chromosome on a spindle. In fact, the process suggests a very primitive method of karyokinesis. Meves ('91) and Benda ('93), in Salamadra

176 Harry Lewis Wieman.

have described a mechanism which may have the same significance. Here the direct division is said to be brought about by the constricting power of a ring-shaped centrosome.

The fact that in both the germ and nurse cells, the daughter nuclei are of approximately equal size, indicates the presence of a division mechanism of great precision and accuracy. The power or force involved in karyokinesis therefore, may not lie in the visible structures — centrosome, spindle, chromosomes, etc. — but in some invisible factor that is a property of the living protoplasm.

Mitosis and amitosis are often regarded as representing antithetical conditions, but, as a matter of fact, these two methods of cell division really stand for the extremes of a graded series. A simple type of amitosis is that shown in the nurse cells, where there is no evidence of any division apparatus. The nucleus undergoes a simple constriction and divides. In the case of the germ cells, the presence of the nucleolus, its exact division and the occurrence of the surrounding pale area, point to what might be called a higher type of amitosis. Another advance is seen in the germ cells of amphibia, where the constriction centrosome divides the nucleus (Meves and Benda). In many of the so-called mitotic division figures of the lower forms, the spindle is far from conspicuous and in many cases is represented by only a few strands of achromatic substance. In the division figure of the macronucleus of Spirochona and Actinosphaerium as figured by R. Hertwig ('79, '84), the spindle and equatorial plate are formed inside of the nuclear membrane. In Spirochona a hemispherical ' ' end plate" or pole plate" is situated at either pole of the spindle. Hertwig claims that these arise by a division of a large nucleolus whose behavior reminds one of the large chromatin nucleoli of the germ cells of Leptinotarsa. Keuten ('95) has demonstrated the origin of similar pole plates from nuceoli in Euglena and Schaudinn ('95) in Amoeba. Euglena presents a very primitive type of mitosis, spindle fibers being scarcely recognizable.

Among the metazoa, examples are to be found in which the mitotic figure is equally primitive, as in certain aphids, where according to Tannreuther ('07) no chromatic spindle occurrs in the maturation division of the egg. Definite chromosomes of

Germ Cells of Leptinotarsa Signaticollis. 177

constant size and number are found, but no astral radiations appear.

Furthermore, it is to be noted that among the protozoa and unicellular plants where mitosis may be said to exist in its most primitive form, the nuclear membrane remains intact and does not disappear at any stage of the division. This condition is characteristic of amitosis while on the other hand rupture of the nuclear membrane at some stage is an accompaniment of mitotic division. Again, in the lower forms the arrangement of the chromatin granules to form chromosomes appear to be of secondary importance as compared with the higher forms and the essential feature in nuclear division appears to be the fission of the individual granules.

These facts all point to the origin of mitosis from a primitive amitotic method of divisions, and therefore give considerable ground for a phylogenetic interpertation of amitosis.

Owing to the usual association of amitosis with degeneration, I at first thought that the degenerated area in the testis would furnish an explanation of the direct division in the early spermatogonia. I soon found, however, that only epithelial cells participate in the degeneration, and that so far as I was able to determine, the spermatogonia undergoing amitosis later develop into functional germ cells.

What then is the significance of the degeneration? The phenomenon recalls certain peculiarities in the development of the ovary. In the first place, it might be compared with the changes following the accumulation of epithelial cells at the base of the terminal chamber, which results in the effacement of the sharp line of demarcation between the latter and the cells of the tube stalk; but the appearance of the cells in the two cases is entirely different. In the ovary, the nuclei of the cells involved undergo a kind of liquefaction, whereas in the testis the cells disintegrate into irregular fragments.

Physiologically the process may be analogous to the change which converts certain of the germ cells of the female into nurse cells. The degenerated area of early stages is actually much larger than later on when it is represented by the part enclosed in the cyst,

178 Harry Lewis Wieman.

which contains only cell fragments. These fragments are the remains of the more solid parts of the cells, the liquid constituents having been separated and secreted into the general cavity of the testis where they probably serve as a nutrient medium for the spermatozoa. This is suggested by the fact that the cavity of the testis is at first occupied by the degenerated area, and as the latter retreats toward the distal end of the testis, the former gradually comes into existence as the continuation of the lumen of the sperm duct. As a result then of the process of contraction, the fluid contents of the degenerated area would be expressed into the central cavity of the testis. The degenerated area might of course be regarded as a general source of nutrition for the germ cells in all stages of development, instead of merely the mature spermatoza.

The degeneration then may be of significance in connection with nutritive processes; but it is to be remembered that the cells involved are not homologous with the nurse cells of the ovary, since the latter are derived from the primordial germ cells. This is not necessarily an objection, but is only in accordance with what might be expected. The spermatozoa are produced in far greater numbers than the ova. In order for this to happen it is essential that all or nearly all of the germ cells of the male develop into functional spermatozoa, none, if any being reserved for the function of supplying nutrition. That would throw this function on the only other element of the testis, namely, the epithelial cells. In the ovary where the course of evolution has taken a different trend, a large number of relatively small and simple ova has come to be not so essential as a small number of greatly enlarged and highly complex germ cells, as a result of which a considerable number of non-functional germ cells become the nurse cells of the ovary.

III. The Chromosomes

The chromosomes of L. signaticolKs are not of a size and number to make the material the best for a satisfactory study of all stages of the maturation process, but they present a number of features

Germ Veils of Leptinotarsa Signaticollis. 179

that can be readily examined, and that are of interest and importance in view of recent developments in this field.

A polar view of a first spermatocyte spindle at metaphase shows the presence of sixteen chromosomes of various sizes, whose bivalent condition is indicated by a deep groove (Fig. 56) . A longitudinal section of the spindle at anaphase shows at one pole a thick V-shaped body that stains red in safranin-lichtgrtin, preparations exactly like the chromosomes (Fig. 54 x). Of the chromosomes, at the middle of the spindle, there is always one very much larger than the others, consisting of two L-shaped parts placed end to end. In polar views this chromosome can be readily identified by its large size. From the telephase shown in Fig. 57, it is evident that an unequal distribution of chromatin has taken place, one of the daughter cells receiving the odd body in addition to the sixteen chromosomes. This odd body is never found at metaphase with other chromosomes. In Fig. 55 it is shown at x, but in order to see it the focus had to be changed. This figure shows its usual position relative to the other chromosomes, which as a rule it precedes to the pole of the spindle, although it occasionly lags behind them.

When we come to work out the history of this odd body, it is found to have its origin in the resting stage following the last spermatogonial division, as a deeply basic-staining nucleolus of a bi-partite form (Figs. 48 and 49), which persists throughout the resting period when the ordinary chromosomes are represented by an irregular reticulum.

Spermatogonial equatorial plates show the presence of two homologous series of chromosomes, the number of which can not be determined satisfactorily at this stage by counting. Following the last spermatogonial division, the cells have the appearance represented in Fig. 45, in which the chromatin is arranged in the form of a split spireme that is closely contracted at one side of the nucleus. Generally at this time, a structure bearing an unmistakable resemblance to the odd body of the first maturation spindle can be seen ( as in Fig. 39 although this figure represents a young ovocyte) . From this we go to the stage shown in Fig. 46, where the spireme is becoming disentangled. The double nature

180 Harry Lewis Wieman.

of the spireme can now be made out very distinctly, and the bilobed body is seen at x. In Fig. 47 is shown a sHghtly older stage in which the chromatin is collected in knot-like aggregations that manifest a diminishing tendency to unite with the basic dye, while the odd body or chromatin nucleolus, as it now maybe called, retains its basic-staining capacity and becomes more prominent by contrast.

In Fig. 48, a reticular structure can be made out, at the nodal points of which can be seen small collections of chromatic material. Fig. 49 marks the end of the resting stage and shows the reticulum very jagged and irregular. It will be noticed that the nucleolus is attached to the ends of two parallel threads, and this is the first step in the reconstruction of the spireme of the prophase of the first maturation divison. In Figs. 50 and 51, this spireme is seen to be composed of two homologous parts, each containing the reduced number of segments. From its position in close contact with the nuclear membrane, the nucleolus of the resting stage {x) can be readily distinguished from the segments of the spireme which it greatly resembles, (Fig. 50.) In Fig. 51, its identity is not so certain. Figs. 52 and 53 are prophases in which the spireme is breaking up into the reduced number of bivalent chromosomes plus the nucleolus, resulting in the production of seventeen bodies, each showing a transverse constriction. All except the nucleolus soon show the presence of another constriction at right angles to the first, forming the tetrads of Fig. 53. Here the nucleolus or odd body is shown in its characteristic position even before the spireme has completely segmented. It shows nothing of the quadri-partite condition of the other chromosomes that is indicative of the two subsequent maturation divisions.

The succeeding stages (Figs. 54, 55, and 56) have already been described. In Figs. 59 and 60 are shown the metaphases of the second division. In the former, seventeen chromosomes, including the odd body, {x) are represented, and in the latter, sixteen. No anaphases sufficiently clear for counting or drawing were found, but it is evident from Fig. 58 which is a characteristic second division telophase, that no uneven distribution of the chromosomes takes place, so that it is inferred that the bi-partite body divides, either transversely or longitudinally.

Germ Cells of Leptinotarsa Signaticollis. 181

There is nothing of especial interest in the remaining details of the spermatogenesis except one stage in the development of the spermatozoan, where the head is seen to be made up of sharply defined basic-staining spheres whose number I estimated to be the same as the reduced number of chromosomes (Fig. 61). A little later this configuration disappears, and the head of the mature spermatozoan shows a smooth solid basic staining mass of the usual pointed form characteristic of Coleoptera.

As has already been mentioned the tangled condition of the chromosomes in the spermatogonial plates makes a direct determination of the unreduced number almost impossible. However, it is clear from the maturation spindles that the number must be thirty three or thirty four, depending upon whether the odd body is to be considered equivalent to one or two chromosomes.

At this point, I should like to call attention to certain appearances in the cytoplasm at the beginning of the growth period, that are suggestive of a process comparable to yolk formation in the ovocyte. As seen in sections, this consists of a crescentic acidstaining area traversed by a coarse irregular fibrous network (Fig. 49) that grows until it nearly fills the cytoplasm (Figs. 50, and 51). When the nuclear membrane disappears for the first maturation division, this material becomes diffused throughout the entire cell and loses its distinctness. It is interesting that its staining reaction is similar to that of the yolk of the egg.

The appearance of the resting stage recalls at once the condition first described by Gross ('04) in Syromastes marginatus, in which he observed a bi-partite nucleolus that arises by the synapsis of two spermatogonial chromosomes. This constitutes the accessory chromosome which from its mode of origin is bivalent. It divides longitudinally in the first division, but fails to divide in the second, passing bodily to one pole in advance of all other chromosomes. The result is that all of the spermatid nuclei receive ten chromosomes and half of them, in addition, the accessory. He further supposed only those spermatozoa to be functional that contained the accessory, the others being regarded as comparable to polar bodies. From the numerical relations he believed the somatic

182 Harry Lewis Wieman.

number to be twenty-two in both sexes. The fertilization formula is:

Egg 11 + s-pcrm 11 (10 + accessory) = oosperm 22 (cf or o).

Gross complicated his conclusions by a very fanciful interpretation of the relation between the accessory and the microchromosomes, which I shall not enter into here.

Wilson ('09a) confirmed Gross's observation that in Syromastes, the spermatogonial number of chromosomes is twenty-two, and that in the second maturation spindle there is a bivalent heterotropic chromosome. He also pointed out that the bi-partite nucleolus of the resting stage is composed of slightly unequal parts. Since the accessory arises from two chromosomes, Wilson considers it equivalent to two in the maturation divisions. Therefore the number of chromosomes characteristic of the two classes of spermatozoa are ten and twelve respectively, both classes being assumed to be functional. In a later paper Wilson ('09c), the ovogonial number was found to be twenty-four. The fertilization formula is given as follows:

Egg 12 + sperm 10 = oosperm 22 (cf ). Egg 12 + sperm 12 = oosperm 24 (o).

The fertilized egg then, contains two bivalent accessory chromosomes, which judging from the facts of the spermatogenesis one would expect to appear in the resting stage of the ovocyte as a quadri-partite or quadrivalent body. To determine this point in L. signaticollis, I examined series of sections of young ovocytes, and found shortly after the last ovogonial division, a stage identical with synizesis in the male in which there appears a bi-partite basic staining nucleolus of approximately the same size and equal in all respects to the body noted in the spermatocyte at a corresponding stage. This is represented in Fig. 39. In Fig. 40, the spireme is unwound and is a stage that is comparable to Fig. 47 in the spermatocyte. Figs. 41 and 42 are slightly older. In the latter the nucleolus is seen at the end of the widely separated parts of

Germ Cells of Leptinotarsa Signaticollis. 183

a thin double spireme, which condition persists for quite a while. Fig. 43 is a drawing of a nucleus considerably older, in which the chromatin represented by the beaded strands is beginning to lose its affinity for the stain. Thus Figs. 39 to 43 inclusive in the female, present an interesting parallel to figs. 45 to 48 inclusive in the male, and show that the stages marking the beginning of the growth period are exactly the same in both sexes.

The important point is the presence of a bi-partite nucleolus in the resting stage of the ovocyte that bears a very striking resemblance to the body found at a corresponding stage in the spermatocyte.

Stevens ('06) found in L .decemlineata (Doryphora deceinlineata) , a closely related form, that the spermatogonial number is thirtysix. The nucleolus of the resting stage is described as an unequal pair, the members of which separate in the first division and divide equationally in the second. The great similarity between the telophases of the first division as represented by Stevens in Figs. 175 and 176 of her paper and the corresponding stage in signaticollis led me to examine the ovaries and testes of decemlineata. I found the nucleolus of the primary spermatocytes to accord with Stevens 'description as far as the resting stage is concerned, but that its unequal components separate in the first division does not seem to be the case, and in this regard I cannot agree with her observation. Figs. 62, 63, and 64 show various stages in the first division, in which the behaviour of the accessory body is exactly the same as in signaticollis.

Stevens claims that the V-shaped body seen in the first spermatocyte spindles is the larger component of the unequal pair and that owing to its smaller size the lesser component is often concealed by the '^ V, " when the group has the appearance of an Orthopteran accessory".

In signaticollis there can not be the slightest doubt as to the accessory not dividing in the first division, and since the two species are so closely related (cf. Tower '06), some similarity in the behavior of this body is to be expected. That Stevens should have placed a different interpretation on this point is not strange in view of the cytological difficulties in the way of a satisfactory


184 Harry Lewis Wieman.

study, for which purpose the material is much less favorable than signaticollis. With the conditions in the latter relatively clear to guide me, I am convinced that the odd body does not divide in the first division in decemlineata. Telophases like those shown in Figs. 62 and 64 would seem to settle this point very definitely. Furthermore, the accessory body even in the late telophases does not mingle with the ordinary chromosomes but occupies a position at one side. If a separation of the components occurred, one would expect both parts to behave similarly in this regard and remain apart from the other chromosomes at their respective poles, but such is not the case. At only one pole is an eccentric body to be found and this bears an unmistakable resemblance to the nucleolus of the resting stage (Figs. 64, 66).

If, with Stevens, we regard the unequal pair as made up of two somatic chromosomes, it follows that the equal pair of signaticollis is likewise equivalent to two chromosomes. In this event the spermatogonial number is thirty-four (2x16+2).

The evidence from signaticollis would perhaps favor the idea that the bi-partite body represents a single chromosome, since its appearance and behavior in the maturation divisions recalls the unpaired accessory of the Orthoptera and certain Hemiptera. On the other hand, the fact that in decemlineata the homologous components are of unequal size practically compels one to regard the accessory body as two chromosomes in both cases. The chromatin nucleolus of the ovocyte of decemlineata is composed of two parts of unequal size (Fig. 65).

It has already been pointed out that the odd body in all probability divides in the second spermatocytic division, but whether transversely or longitudinally can not be readily determined. At first I was inclined to believe that the division took place so as to separate entire chromosomes but this view leads to serious difficulties, since it would follow that three classes of spermatozoa would be produced. In signaticollis one-half of the total number would contain sixteen chromosomes; one-fourth of the total number, sixteen plus one component of the odd body ; the other fourth, sixteen plus the other component of the odd body. The same kind of difficulty would be met with in decemlineata.

Germ Cells of Leptinotarsa Signaticollis. 185

The other alternative, namely, that the odd body divides longitudinally in the second division is more in accord with well established observations in other forms, and furthermore meets with no objections in the appearance of the various division stages. If we accept this interpretation, the case is exactly similar to Syromastes except that the order of division is reversed, the bi-partite dividing in the second division instead of the first, thus giving two classes of spermatozoa, one with sixteen and one with eighteen chromosomes. An identical instance would be seen in Phylloxera (Morgan '09), where the odd body is equivalent to two chromosomes which divide only in the second division, and then longitudinally.

On this basis the somatic tissues of the female would contain two more chromosomes than those of the male, and the fertilization formula for signaticollis would be:

Egg 18 + sperm 16 = oosperm 34 (c?). Egg 18 + sperm 18 = oosperm 36 (9).

However, the tangled condition of the chromosomes in dividing somatic cells, prevented verification of these points by direct observations.

My observations of the maturation spindles of the egg are very incomplete, and I am unable to say whether or not there is an unequal distribution of chromosomes such as occurs in the first spermatocytic division. It is possible that while the bi-partite nucleolus appears in the ovocyte, at the same time as in the spermatocyte, its subsequent behavior in the maturation divisions is entirely different.

Is the nucleolus of the ovocyte the homologue of the nucleolus of the spermatocyte? The transformations accompanying the long growth period of the ovocyte make it a very difficult matter to follow the history of this body through to the maturation spindles. The last stage at which it can be satisfactorily made out is shown in Fig. 43, which represents a section of the nucleus of an ovocyte. From this there occurs a gradual transition to the condition seen in Fig. 44, where in place of one, five rounded basic

186 Harry Lewis Wieman.

staining nucleoli are present. The chromatin has passed from the condition of beaded strands into a formless mass that takes the acid stain. Finally an acid staining ground substance is produced in which a number of basic staining nucleoli are embedded. What relation these nucleoli bear to the nucleolus of the resting stage I am at present unable to say.

There are a number of facts that argue in favor of the nucleolus of the resting stage being homologous in the two sexes. In the first place in pre-reduction stages in both cases, a nucleolus is present in the vegetative nucleus between mitotic divisions, but it is entirely different in appearance from the one characteristic of the growth period. It is often jagged in outline, and may be single or multi-partite, while the bi-partite condition of the nucleolus is strictly a feature of the resting stage.

Secondly, the bi-partite nucleolus appears in both sexes at the time of synizesis (Figs. 39 to 46). Wilson ('06, p. 22-23), speaking of Anasa and other Hemiptera states: In the female no trace of a chromosome nucleolus can be found in the contraction stage of the synaptic period .... I can, therefore, only state that no chromosome nucleolus is present in the contraction period synapsis or in the early growth period, and even though it be present in later stages, which I think very doubtful, a wide difference between the sexes would still exist in respect to the earlier period." Evidently the conditions in Anasa and Leptinotarsa are not the same, for the presence of the nucleolus in the latter can be demonstrated without any difficulty.

Thirdly, the behavior of this body in relation to the other chromosomes in the early part of the growth period is the same in both sexes (Figs. 39, 40, 41, 42 46, 47, and 48).

The evidence all tends to show that this body is in some way bound up with the processes connected with the early part of the growth period. It persists in the ovocyte as long as the development of the latter is parallel with that of the spermatocyte, but as soon as the ovocyte begins to show signs of the highly specialized metabolism involved in its enormous growth, the nucleolus can not be distinguished from a number of other bodies that make their appearance at this time.

Germ Cells of Leptinotarsa Signaticollis. 187

A knowledge of whether this bi-partite nucleolus represents one or two chromosomes may be of no great importance in itself, but it has a very important bearing in determining the relation of the nucleolus to the segments of the spireme in synapsis. The manner in which synapsis takes place in insects is scarcely at all understood. The critical stages are just the ones that are omitted in descriptions or passed over so hurriedly that the reader is usually left to follow his own choice in the matter.

Wilson ('09 a) in a admirable series of photographs, has produced evidence against the idea supported by Montgomery ('00), Sutton ('02), Stevens ('03) and Dublin ('05) and others, that synapsis occurs in the closing anaphases of the last spermatogonial division. He has shown that inPyrrochoris the number of chromosomes in the postphases following the last spermatogonial division is not the reduced number, but approximately the somatic number. Synizesis does not follow immediately but is separated by a long resting period.

The extreme synizesis stage in the male of signaticollis is shown in Fig. 45, where the chromatin is in the form of a tightly wrapped split spireme. The double nature of the spireme is seen to better advantage in Fig. 46, where it is unfolding. As the number of its segments is greater than the reduced number of chromosomes, synapsis in the sense of an actual fusion has not yet taken place.

In the ovocyte, following the last ovogonial division, practically the same condition occurs. It happens that in the particular instance shown in Fig. 39 the two halves of the spireme are separated; which indicates very clearly the double form of this structure. Similar examples are not wanting in the spermatocyte. As one studies the succeeding stages shown in Figs. 41, 42 and 43, it is clear that no reduction has yet occurred.

Actual conjugation of chromosomes does not take place in the last stages of the spermatogonial or ovogonial divisions, although the chromosomes at this time do seem to pair into two homologous series which in synapsis have the appearance of a split spireme.

It is rather questionable whether synizesis indicates an actual contraction of the chromatin; at least in the material under consideration. In Fig. 46 it seems to mark the unfolding or dis

188 Harry Lewis Wieman.

entangling of the chromatin thread which has remained practically unchanged from the telophases of the last pre-maturation division, while the nuclear cavity has been enlarged. Wilson ('09a) speaks of the nuclear reticulum in Pyrrochoris as undergoing contraction, yet the actual contraction of the chromatic material is rather insignificant (as shown by his figures. Photos. 35-42 inclusive) compared with the expansion of the nuclear membrane. Likewise in the case at hand the chromatin has changed but slightly in bulk since the telophases of the pre-maturation divisions, and the apparent contraction seems to be due principally to an enlargement or vacuolization of the nuclear cavity.

Fig. 51 shows the characteristic appearance of the spireme of the prophase of the first division, when the thickened thread is composed of the reduced number of segments, each of which is double. The nucleolus of the resting stage forms one end of the series (Fig. 50), and its relation to the spireme is such as to suggest that each segment of the latter is homologous with it. That is to say, if the nucleolus represents two fused or closely united chromosomes, then each segment of the spireme can be regarded as a double chromosome. In other words, synapsis has taken place by side to side union of the chromosomes. On the other hand, if we regard the nucleolus as a single chromosome, then end to end union of the chromosomes has occurred; and in this case the groove separating the two halves of the spireme represents the line of a longitudinal division ; whereas in the other it represents the line along which the chromosomes had come together in synapsis.

An examination of Figs. 46, 47, 40 and especially 42, shows that during synizesis or at least very shortly after, the split spireme bears a similar relation to the nucleolus. To each lobe of the latter is attached one of a double series of segments, the number of which cannot be counted, but which is more than the unreduced number. In fact, there is approximately twice this number, which means that the chromosomes had united to form a spireme the demarcation of which into segments foretells the two maturation divisions.

Throughout the period preceding the first maturation division.

Germ Cells of Leptinotarsa Signaticollis. 189

the behavior of the two halves of the spireme is such as to indicate two more or less independent parts, as in Figs. 39 and 42, where the halves are widely separated, and Fig. 50, where the separation is partial. These facts, together with the relation of the segments of the spireme to the presumably bivalent nucleolus, point to a separate origin of the two components of the spireme, as opposed to their resulting from a longitudinal splitting of a single series of units united end to end.

On this basis the synaptic process would take place somewhat as follows: In the post-telophase stages of the last pre-maturation division, the chromosomes arrange themselves side by side in homologous pairs to form a long thin double spireme behind the accessory body, which arises from two chromosomes that, instead of elongating as the rest of them do, contract to form the nucleolus of the resting stage. The line of separation between the two homologous series indicates one of the m.aturation divisions. The other division (except in the case of the nucleolus which only divides once) is indicated by a tranverse groove in each chromosome which produces the large number of segments seen in synizesis and stages immediately after (Figs. 39 to 47 inclusive). With the exception of the nucleolus, all trace of chromosome structure is lost during the resting stage. The first indication of a reconstructing process is shown in Fig. 49 by the appearance of two parallel strands upon which the chromosomes take their places as shown in Figs. 50 and 51.

The occurrence of side to side union of the chromosomes or parasynapsis (Wilson '09a), has a bearing on certain theoretical aspects of maturation. It can be readily shown that the two divisions take place at right angles to each other (Figs. 53 to 60 inclusive). Thus in one division whole chromosomes are separated and in the other, each of these is divided transversely. Therefore no equational division occurs; and this to some would be a fatal objection to parasynapsis as outlined above. Before such an objection can be taken seriously it must first be shown that the supposed difference between quantitative" and" qualitative " divisions really exists. The theory rests upon the assumption of the arrangement of the chromomeres of each chromosome

190 Harry Lewis Wieman.

in a linear series so that a division in a plane including the axis of a chromosome, halves each particle in it, while on the other hand a division at right angles to the axis separates whole chromosomes. Further discussion of this question need not be entered into here for it is evident that so long as nothing definite is known of the arrangement or significance of the component parts of a chromosome, the absence of the so-called quantitative division need not be considered of especial importance. Similar conditions in other forms may be of much wider occurrence than is at present supposed.

It is seen from this account that there does not seem to be a true synapsis or complete fusion of chromosomes in the formation of the spireme of the prophase of the first maturation division. The duality of the series is perfectly evident and the frequent instances where the components are more or less separated speaks for the absence of fusion. The only fusion that has occurred is in the disappearance of the transverse furrow which in synizesis divided each chromosome so as to produce twice the somatic number of segments (Figs. 50 and 51) . The disappearance of this furrow is only temporary, for it reappears in each bivalent chromosome after the spireme has broken up into the reduced number of segments (Figs. 52 and 53).

Cases like Fig. 49 showing the first step in the re-formation of the spireme after the resting period, suggest very strongly that the biserial arrangement of the chromosomes on the parallel linin threads connected with the chromatin nucleus persists through the resting stage. The evidence for what has been called prochromosomes" is not very marked; the chromatin being represented by irregular masses whose number could not be determined with any exactness. Evidence along this fine has been urged by many workers in support of the conception that the chromosomes are permanent cell structures. Boveri ('04) in his stating that the chromosomes are individual elementary cell organisms "die in der Zelle ihre selbstandige Existenz fiihren," has perhaps gone to the extreme in developing the individuality hypothesis; and Rabl ('06) has expressed a similar view. ' Harper ('05) has pointed out that it is somewhat questionable

Germ Cells of Leptinotarsa Signaticollis. 191

whether Boveri is justified in combining the conception of the permanence of the chromosomes and the doctrine that they are individual or elementary organisms which lead a relatively independent existence in the cell ; and furthermore that such a conclusion does not necessarily follow from permanence in number, form, and position in the nucleus, any more than that the cytoplasm is an individual organization because it grows and divides. No more is gained in support of the hypothesis of chromosome individuality by regarding the chromosomes as elementary and relatively independent organisms which bear a symbiotic relationship to the cell, than that they are definite permanent parts of a cell mechanism having a permanent, but not necessarily independent existence in the cell.

More recently Boveri ('07, p. 229) has given a definition of chromosome individuality that is not quite so rigid, Was durch den kurzen Ausdruck, 'Individualitat der Chromosomen' bezeichnet werden soil, ist die Annahme, dass sich fiir jedes Chromosoma das in einem Kern eingegangen ist, irgend eine Art von Einheit im ruhenden Kern erhalt, welche die Grund ist, dass aus diesem ruhenden Kern wieder genau ebensoviele Chromosomen hervorgehen, und dass diese Chromosomen iiberdies da, wo vorher verschiedene Grossen unterschiedbar waren, wieder in den gleichen Grossenverhaltnissen auftreben und dass sie dort, wo sie vor der Kernbildung in characterischer Weise orientiert waren, diese Orientierung bei ihrer Wiedererscheinung haufig in gleichen Weise darbieten." This definition illustrates the general tendency to depart from the older idea of the chromosomes being strictly automatic individuals.

It is this conception of automatism against which most of the criticism of the chromosome hypothesis has been directed. The observations of a number of investigators seems to show that the chromosomes are represented in the resting stage between cell divisions. Thus Bonnevie ('08) has shown that in rapidly dividing cells (cleavage stages of Ascaris and root tip of Alium), although the identity of the original chromosomes is lost in the resting nucleus after each mitosis, each new chromosome arises by a kind of endogenous formation from within and from the substance of its

192 Harry Lewis Wieman.

predecessor. In this way the genetic continuity of the chromosomes is preserved in the resting stages, "Eine genetische Continuitat der Chromosomen nacheinanderfolgenden Mitosen konnte in der von mir untersuchten Objekten teils sicher (Alium, Amphiuma) teils mit iiberwiegender Wahrscheinhchkeit {Ascaris) verfolgt werden. Es ging aber auch hervor, dass eine Identitat der Chromsomen verschiedener Mitosen nicht existiert, sondern dass jedes Chromosom in einem friiher existierenden Endogen entstanden ist, um wieder am Ende seines Lebens fiir die endogene Entstehmig eines neuen Chromosoms die Grundlage zu bilden" (p. 54).

Overton ('09) claims that in the somatic nuclei of Thalistrum purpurascens and Calycanthus floridas the chromosomes are represented during the resting stage by definite, visible bodies; the pro-chromosomes, which are arranged in parallel pairs with apparent linin intervals. Prochromosomes are also present in the resting nuclei of the germ cells of these plants and Richardia africana, in exactly the same arrangement and form as in the somatic nuclei. The homologous parental elements are already associated in pairs when they enter the reconstruction stage of the germ nuclei.

The results of these two investigators will serve to indicate the increasing evidence in favor of permanence of the chromosomes, which is to be held distinct from the idea of automatism in these bodies.

In the spermatocyte of L. signaticollis, when the spireme of the synizesis stage enters the rest stage showing a definite arrangement of its segments, and later emerges from that stage showing the same arrangement of parts, it is obvious that the organization which is responsible for this configuration must have persisted in some form through the intervening rest period; even though the outlines of the chromosomes and spireme are temporarily lost to view. In other forms where the evidence for prochromosomes is relatively clear, the existence of this organization is more apparent ; but even here the chromosomes undergo a kind of vacuolization or disintegration, leaving but a mere skeleton to indicate their presence. In other words, each chromosome on entering the

Germ Cells of Leptinotarsa Signaticollis. 193

resting stage passes into a condition of almost complete disorganization from which it is able to recover its individuality and reappear in its characteristic form in the prophase of the first maturation division.

According to the strict individuality hypothesis these transformations are to be explained solely by the remarkable properties possessed by the chromosomes as individual units.

The results of the present study show that the nucleus and cytoplasm of the cell taken together constitute a unit ; while the chromosomes are regarded as the expression of an organization of the protoplasm of which they are a part. The occurrence of the amitotic period in the germ cells of both sexes clearly shows that this organization is a property of the protoplasm as a whole; for it is well nigh impossible to conceive of the chromosomes as independent entities, when during this amitotic cycle no provision is made for their transmission from generation to generation. It might be argued that in amitosis the chromosomes are divided with as great precision as in mitosis; but actual observations do not warrant such an assumption.

One of the strongest arguments that have been used in favor of the individuality hypothesis is the fact that in many cases at least the size of the nucleus is dependent upon the number of chromosomes that it contains. Thus Boveri ('05-'07) states that in sea urchin larvse the surface area is proportional to the number of chromosomes contained in the nucleus, and that nuclei possessing an abnormal number of chromosomes pass on the abnormality to the daughter nuclei. The general truth of this statement is clearly indicated in the primordial germ cells of the ovary and the testis of L. signaticollis where occasional abnormal mitotic figures occur which result in the production of abnormal daughter nuclei.

Is it necessary to use the individuality hypothesis to explain these phenomena? Could they not be as well explained by considering the chromosomes simply as a portion of protoplasm possessing a definite organization?

During the last two or three decades the chromosomes have received much attention and the results of the combined efforts of

194 Harry Lewis Wieman.

a large number of workers have yielded some interesting and important data; but the theoretical deductions drawn from these data have not been in all cases entirely justifiable. Attention has been focussed on the chromosomes largely because they happened to be objects made conspicuous by the readiness with which they can be stained with certain dyes employed in cytological technique, while other factors in heredity have been overlooked simply because they are not expressed so clearly in a morphological form. The chromosomes certainly have their place in hereditary processes, but the results of the present investigation indicate that they are but the manifestation of properties which belong to the cell protoplasm as a whole.

IV. Summary The Ovary

The functional germ cells of the ovary and the nurse cells have a common origin from the primordial germ cells. The configuration of the latter in larval and pupal stages is such as to suggest an amitotic division period of short duration. The nuclei at this time are characterized by the presence of a basic staining nucleolus in various stages of division surrounded by a clear nonstaining area.

The epithelial cells are somatic cells derived from the mesodermic somites. During the larval and pupal stages the terminal thread contributes epithelial cells to the egg chamber. In pupal stages, the two regions become separated by a structureless "limiting membrane" which develops first at the sides and finally becomes continuous all the way across the ovariole. In the erlier history of the ovary, the relation of the epithelial cells to the germ cells is such as to indicate a nutritive function for these cells, but it does not appear that they enter into the formation of the egg follicles. The view is here advanced that the latter structures arise from the cells of the ovariole stalk. In later stages the epithelial cells are concerned in forming the delicate membranes which enclose the cysts. The peritoneal sheath of the ovary,

Germ Cells of Leptinotarsa Signaticollis. 195

especially in the region of the terminal thread, becomes invested with striated muscle fibers, the whole being inserted in the dorsal wall of the body.

There is no evidence of a chromosome basis for the differentiation of functional germ cells from nurse cells. On the other hand the production of a semi-fluid medium from the interaction of epithelial and ovariole stalk cells, at the time when the differentiation is occuring, suggests that it is an important factor in the process. Those of the primordial germ cells coming under its influence are enabled to develop into functional ova ; the remainder become nurse cells.

Mitotic division figures disappear first among the germ cells of the ovariole at its lower proximal end, but they can be seen for some time afterward in the more distal part among the potential nurse cells. The persistence of the mitotic divisions in the latter instance suggests that the energy which is conserved in the functional germ cells for the needs of developmental processes is here being expended in cell divisions. The frequent occurrence of multipolar and abnormally large single spindles is also noted. Mitotic figures are never seen inside of cysts in the ovariole.

The chromatin of the nuclei of the nurse cells appears more highly granular than that of the germ cells; and this is the first obvious result or accompaniment of the process of differentiation of the nurse cells. Amitosis and formation of cysts takes place first among those nurse cells at the base of the chamber and adjacent to the functional ova. The contents of each cyst are the descendants of a single mother cell. At first, only the nucleus divides; the result being large polynucleated masses. Later the division extends to the cytoplasm, producing in some cases mononucleated masses morphologically equivalent to single cells. Amitosis is regarded as an indication of an intense metabolic activity involved in the differentiation of the nurse cells which demands the most rapid and expedient method of nuclear division. This is in accordance with the view taken that amitosis represents a more primitive and relatively simpler method of cell multiplication than mitosis.

The prolonged amitotic period brings on certain chemical

196 Harry Lewis Wieman.

changes which are assumed to be due to the failure of the nuclear membrane to rupture at regular intervals and discharge materials of the nature of oxydases or oxidizing substances into the cytoplasm, and the result is a reversal in the ususal staining reactions of nucleus and cytoplasm. This change is preceded by the expulsion from the nucleus into the cytoplasm of a basic staining granule which gradually undergoes solution and disappears.

In their later history some of the nurse cells pass through a cycle of chemical changes which originate in the center of the nucleus and spread toward the periphery, leaving as a final stage, an acid-staining central area surrounded by three concentric shells of alternating basic and acid-staining regions.

The nutrition of the ovum is derived entirely from the nurse cells by means of an egg-string," which is a pseudopodium-like process of the cytoplasm of the egg, that is left behind as the ovum moves down the tube in the early part of the growth period, and the distal end of which comes into relation with the spaces between the nurse cells. This intercellular region from its appearance in sections has been likened to a system of ducts into which the nutritive material from the nurse cells is secreted ,and whence it passes via the egg-strings into the eggs.

The nutritive stream consists of basic staining granules identical with those found in the cytoplasm of the nurse cells. At first these granules are found evenly distributed in all parts of the cytoplasm of the ovocyte, but later, a definite stream extends from the end of the nurse string to the nucleus. In still later stages the deposition of the yolk in the center of the egg causes a split in the food-stream, compelling the latter to take the form of an oval shell inside of which the yolk masses are embedded.

Attention is called to a rythm of chemical changes in the development of the ovum. The cytoplasm of the primordial germ cells shows the presence of an acid-staining reticulum which contains basic-staining granules apparently identical with those of the food-stream. As yolk is formed these granules of the foodstream disappear in the central part of the egg, being found only in the superficial cytoplasm. Thus it appears that the basicstaining granules from the nurse cells through interaction with

Germ Cells of Leptinotarsa Signaticollis. 197

the nucleus and cytoplasm are converted into the inert acid-staining yolk. After fertilization, the basic staining yolk nuclei react with the yolk and convert it from its inert condition into a form which can be readily assimilated by the living protoplasm of the blastoderm.

Likewise, the nucleus passes through an interesting series of changes. Beginning with the end of the division period, the ground work of the nucleus is a clear, homogeneous, non-staining substance, in which a basic-staining nucleolus and chromosomes are embedded. As the growth proceeds, the chromosomes lose their sharp outline and gradually take on a filmy form in which they stain deeply with acid dye. Finally in the nearly mature egg, all of the nuclear contents appear granular and takes the acid stain, only the nucleoli which have increased greatly in number take the basic stain.

The Testis

In the larva each lobe of the testis is of a cylindrical form resembling a single ovariole of the ovary. As in the latter both epithelial and germ cells can be readily identified. In the ensuing stages the increase in size takes place principally at right angles to the axis, but not equally in all directions, being inhibited at regular intervals which mark the spaces between the radiating follicles. The original apex is represented in the adult organ by a cap-like lobe which lies just opposite the aperture of the sperm duct.

The epithelial investment, at first loosely applied, in late pupa stages enlarges between the follicles and produces the appearance characteristic of the adult in which the follicles are separated from each other by a thick layer of epithelial cells.

Amitotic division figures of the same type observed in the germ cells of the ovary can be seen in the germ cells of the testis of the larva, the pupa and even the adult in the cap-shaped region. All of the germ cells pass through the amitotic cycle which commences at the beginning of cyst formation and persists for a number of cell generations. These cells later divide mitotically.

Areas of degeneration are to be seen in all stages of the testis

198 Harry Lewis Wieman.

beginning in the larva and extending through the adult. The process commences with accumulation of epithelial cells in the region representing the lumen of the general cavity of the testis continuous with that of the sperm duct of the adult. The cells involved fragment, producing irregular masses which stain deeply with the basic dyes. As the degeneration proceeds the area contracts until in the adult it is represented by a concavo-convex discshaped region marked off from the neighboring parts by a connective tissue capsule. The process is regarded as a method of providing nutritive material either for the spermatozoa or for the germ cells during the maturation period. Thus the cells that undergo degeneration are looked upon as the nurse cells of the testis.


The occurrence of amitosis at corresponding stages in the germ cells of both sexes is believed to be due to a periodic fluctuation in the nutritive supply of the cells brought about by a stimulus to a rapid cell division which causes a temporary derangement in the normal metabolism. In the ovary the disturbance is merely transient; but in the testis it is more prolonged for the reason apparently that it is here involved in the formation of cysts, a process that in the species under consideration is always accompanied by rapid cell multiphcation.

In the nurse cells the initial cause of amitosis is probably the same; but in this instance it is carried to an extreme, so that a permanent change in metabohsm occurs.

Amitosis and mitosis are believed to stand for the extremes of a continuous series; the different configurations of the division figures being due to the different types of metabolism represented.

The Chromosomes

The first spermatocyte spindle at metaphase, shows the presence of sixteen bivalent chromosomes that divide in this division, and a bivalent V-shaped body, made up of equal parts, that does not divide but passes entire to one pole.

Germ Cells of Leptinotarsa Signaticollis. 199

In the second division all of the chromosomes including the bi-partite body divide. The latter probably divides longitudinally so that two kinds of spermatids result, containing sixteen and eighteen chromosomes respectively.

The odd body is a basic-staining nucleolus that makes its appearance in the synizesis stage when the other chromosomes are in the form of a doubly segmented spireme whose halves may be more or less separated.

In the rest stage following, the nucleolus persists as a deeply basic-staining bi-partite body while the remaining chromosomes are represented by faintly staining clumps of matter distributed on an irregular reticulum.

The first step in the formation of the spireme of the ensuing prophase consists in the appearance of two parallel strands of linin extending from the nucleolus. These apparently represent the framework of the spireme not only of this stage but of the preceding synizesis.

When the spireme appears, it is composed of a thickened double thread whose halves are more or less closely applied to one another, each of which contains sixteen segments. The nucleolus which is attached to the nuclear membrane forms an additional member in the series.

After the spireme has broken up the sixteen bivalent chromosomes proper immediately shows signs of a second groove at right angles to the first producing tetrad-like bodies. The bi-partite chromatin nucleolus of the resting stage never shows such evidence of a second division.

In the ovocyte at a corresponding time there is a similar synizesis stage and bi-partite chromatin nucleolus. Instead of passing directly into a rest stage as in the male, the chromatin knot becomes disentangled producing the characteristic spireme" stage which persists for quite a while.

In L. decemlineata an odd body appears in the first spermatocyte spindles, but in this case it is composed of two unequal parts. Its behaviour in the maturation divisions is believed to be the same as that described for the homologous body in signaticollis. Synizesis stages of the ovocyte of decemlineata show a bi-partite nucleolus also composed of unequal parts.


200 Harry Lewis Wieman.

It is assumed that in botii species the odd body represents two conjugated somatic chromosomes.

There is much evidence to show that the spireme is formed in the post-telophase stages of the last pre-maturation division by the arrangement of the chromosomes into two parental series which become closely applied to each other (parasynapsis) propucing in this way the double spireme characteristic of synizesis.

The evidence from observation does not justify the view that the chromosomes are independant individual units of a lower order than the cell. On the other hand, they are to be regarded as permanent cell structures in so far as they express a definite organization of the protoplasm. During amitosis this organization is assumed to persist, although the chromosomes which represent it in mitosis do not come to expression. This is believed to be due to differences in cell metabolism in the two cases.

Accepted by The Wistar Institute of Anatomy and Biology, February 10, 1910. Printed August 3, 1910.


Balbiani, E. G. Centrosome et ' ' Dotterkern." Journ. de I'anat. et de la physiol., 1893 T. 29, pp. 145-179.

Benda, C. Zellstriicturen und Zellteilungen des Salamanderhodens. Verhandl. 1893 der Anat. Gesellschaft.

Brandt, Al. tlber das Ei und seine Bildungsstatte. Leipzig. 1878.

BoNNEviE, K. Chromosomenstudien I. Arch. Zellforschung, Bd. 1, pp. 450-514.


BovERi, Th. Ergebnisse iiber die Konstitution der chromatischen Substanz des

1904 Zellkerns, Jena.

1905 Tiber die Abhangigkeit und Zellenzahl der Ausgangszellen. Zellen Studien, Bd. 5.

1907 Die Entwicklung dispermiger See'igel Eier. Ein Beitrag zur Befruchtungslehre. Zellen-Studien, Bd. 7, Jena.

Calkins, G. N. Observations on the Yolk-nucleus in the Egg of Lumbricus. 1895 Trans. N. Y. Acad. Sc, June.

Carnoy, J. B. La cytodierese chez les Arthropodes., La Cellule, T. f, pp. 191-440. 1885 ^

Carriere, J. und O. Burger. Die Entwicklungsgeschichte der Mauerbiene1898 Nova Acta Acad. Leop.-Carol. Bd. 69, pp. 253-420.

Germ Cells of Leptinotarsa Signaticollis. 201

Child, C. M. a — Studies on the relation between Amitosis and Mitosis I. Devel1907 opment of the Ovaries and Oogenesis in Moniezia. Biol. Bulle.

vol. 12, pp. 89-114.

1907 h — Studies, etc., II. Development of the Testes and Spermatogenesis in Moniezia. Biol. Bulle. vol. 12, pp. 191-224.

1907 c — Amitosis as a factor in Normal and Regulatory Growth. Anat. Anz., Bd. 30, pp. 271-297.

Claus, C. Beobachtungen iiber die Bildung des Insekteneies. Z. wiss. Zool., 1864 Bd. 14, pp. 42-54.

CoNKLiN, E. G. Karyokinesis and Cytokinesis in the Maturation, Fertilization

1902 and Cleavage of Cripidula and other Gasteropods. Journ. Acad. Nat. Sci., Phila., Vol. 12., pp. 1-121.

De Bruyne, C. Contribution a I'etude physiologique de I'amitose. Livre juhi1899 lare dedie a Charles van Bamheke. Bruxelles.

Delage, Y. Sur 1' interpretation de la fecondation merogonique et sur une theo"

1899 rie nouvelle de la fecondation normale. Arch. Zool. exper. gen, Ser. 3, T. 7, pp. 511-527.

1901 Etudes experimentales sur la Maturation cytoplasmique et sur la Partheneogenese ai-tificielle, etc. Arch. Zool. exper. gen., Ser. 3, T. 9, pp. 285-326.

Dublin, L. I. The History of the Germ Cells in Pedicellina americana. Ann. 1905 N. Y. Acad. Sci., vol. 15, pp. 1-64.

Dtjfour, L. Recherches anatomiques et physiologiques sur les Orthopteres, les 1833 Hymenopters et les Neuropteres. Memoires pres. par div. sav.

I'acad. roy. d. sc. de Vinst de France., T. 4, Paris.

1841 Ibid, T. 7, Paris.

Ehrlich, p. Histologie und Clinic des Blutes. Berlin.

1891 Fischer. Fixirung, Farbung, und Bau des Protoplasmas.

1899 Gerassimow, J. tJber die kernlosen Zellen bei einigen Conjugaten. Bull. Soc,

1892 Imp. des Natur. Moscow. No. 1.

GiARDiNA, A. Origine dell 'oocite e delle cellule nutrici nel Dytiscus. Internal' 1901 Monatsschr. f. Anat. u. Phys., Bd. 18, pp. 417-484.

Gross, J. Untersuchungen iiber die Histologie des Insektovariums. Zool. Jahb.t

1903 Anat. u. Ontog., Bd. 18, pp. 72-186.

1904 Die Spermatogenese von Syromastes marginatus L. Zool. Jahrb.,

Anat. u. Ontog., Bd. 20, pp. 439-498.

Haecker, V. Mitosen im Folge amitosenahnlicher Vorgange. Anat. Anz., Bd'

1900 17, pp. 9-20.

Harper, R. A. Sexual Reproduction and the Organization of the Nuclei in certain

1905 Mildews. Carnegie Inst., Washington, Pub. no. 37.

202 Harry Lewis Wieman.

Hegner, R. W. The Origin and Early History of the Germ Cells in Some Chry1909 somelid Beetles. Journ. Morph., vol. 20, pp. 231-296.

Hertwig, R. Beitrage zur Kenntniss der Bildung, Befruchtung und Teilung des 1877 thierischen Eies, II. Morph. Jahrb., Bd. 3, pp. 1-86.

1885 Das Problem der Befruchtung und der Isotropie des lies, eine Theorie

der Vererbung. Jen. Zeit., Bd. 18, pp. 276-318, 188.

Heymons, R. Die Entwicklung der weiblichen Geschlechtsorgane von Phyllo1891 dromia (Blatta germanica L.), Z. wiss. ZooL, Bd. 53, pp. 434-536.

1895 Die Embryonalentwicklung von Dermapteren and Orthopteren und

besonderer Beriicksichtigung der Keimblatter. Jena.

HiRSCHLER, Jan. Die Embryonalentwicklung von Donacia crassipes. Zeit. wiss. 1909 ZooL, Bd. 92, pp. 627-744.

Jordan, E. O. The Habits and Development of the Newt. Journ. Morph., vol. 8, 1893 pp. 269-366.

Keuten, J. Die Kerntheilung von Euglena viridis Ehr. Z. wiss. ZooL, Bd. 60, 1895 pp. 215-235.

King, H. D. The Spermatogenesis of Bufo lentiginosus. Am. Journ. Anat., vol. 1907 7, pp. 345-388.

KoHLER, A. Untersuchungen iiber das Ovarium des Hemipteren. Z. wiss. ZooL, 1907 Bd. 87, pp. 337-381.

KoRsCHELT, E. tjber die Entstehung und Bedeutung der verschiedenen Elemente

1886 des Insektenovariums. Z. luiss. ZooL, Bd. 43, pp. 537-720.

1887 tlbereinigeinteressante Vorgangebei der Bildung der Insecteneier. Z. tviss. ZooL, Bd. 45, pp. 327-397.

1889 Beitrage zur Morphologic und Physiologic des Zellkernes. ZooL Jahrb. Anat. u. Ontog., Bd. 4, pp. 1-154.

Kramer, P. Beitrage zur Anatomic und Pysiologie der Gattung Philopterus 1869 (Nitsch). Z. wiss. ZooL, Bd. 19, pp. 452-468.

Landois, L. Anatomic des Hundeflohes. Nova Acta Acad. I^eojp. -Carol., Bd. 33, 1867

Le Cron, W. L. Experiments on the Origin and Differentiation of the Lens in 1907 Amblystoma. Am. Journ. Anat., vol. 6, pp. 245-258.

Leuckart, R. Die ungeschlcchtlichc Fortpflanzung der Cccidomyienlarven.

1865 Arch. f. Naturcjeschichte, Bd. 1, pp. 286-303.

Lewis, W. H. Experimental Studies on the Development of the Eye in Amphibia. 1904 I. On the Origin of the Lens, Rana palustris. Am. Journ. Anat.,

vol. 3, pp. 505-536.

v. Leydig, F. Der Eistock und die Samentasche der Insecten. Nova Acta Acad,

1866 Leop.-CaroL, Bd. 33,

1889 Beitrage zur Kenntniss des thierischen Eies in unbefruchteten Zustande. ZuoL Jahrb., Anat. u. Ontog., Bd. 3, 287-^32.

Germ Cells of Leptinotarsa Signaticollis. 203

LiLLiE, F. R. Observations and Experiments concerninf? the Elementary Phe1906 nomena of Embryonic Development in Cha'topterus. Journ. Exper.

Zool., vol. 3, pp. 163-268.

LiLLiE, R. S. Momentary Elevation of Temperature as a Means of producing 1908 Artificial Parthenogenesis in Star-fish Eggs and the Condition of

its Action. Journ. Exper. Zool., vol. 5, 379-428.

LoEB, J. Maturation, Natural Death and the Prolongation of the Life of Unfer1902 tilized Starfish Eggs (Asterias Forbcsii) and their significance for

the Theory of Fertilization. Biol. Bulle., vol. 3, pp. 295-302.

1906 Versuche uber den chemischen Character des Befruchtungsvorgangs. Biochem. Zeit., Bd. 1, pp. 183-206.

1906 Weitere Beobachtungen uber den Einfluss der Befruchtung und der

Zahl der Zellkerne auf die Saiirebildung im Ei. Biochem. Zeit., Bd. 2, pp. 34-42.

Lyon, E. P. Ryhthms of Susceptibility and Carbon Dioxide Production in Cleav1904 age. Am. Jown. P/i?/sioZ., vol. 11, pp. 52-58.

McGregor, J. H. The Spermatogenesis of Amphiuma. Journ. Morph., vol. 15

1899 (Supplement), pp. 57-104.

Mann, G. Physiological Histology, Methods and Theory. Oxford.

1902. Mathem^s, a. p. a Contribution to the Chemistry of Cytological Staining. Am.

1898 Journ. Physiol., vol. 1, pp. 445-454.

1907 A Contribution to the Chemistry of Cell Division, Maturation and

Fertilization. A^n. Journ. Physiol., vol. 18, pp. 89-111.

Maximow, a. tjber Amitose in den embryonalen Gewebe bei Satigetieren. Anat.

1908 Anz., Bd. 33, pp. 89-98.

Mayer, P. Anatomic von Pyrrhocoris apterus L. Arch. Anat. Physiol., pp. 3091875 355.

Metschnikoff, E. Embryologische Studien an Insekten. Z. wiss. Zool., Bd. 1866 16, pp. 309-500.

Meves, F. tJber amitotischen Kernteilung in den Spermatogonien des Sala1891 manders, und Verhalten der Attraktionssphiire bei derselben. Anat.

Anz., Bd. 6, pp. 626-639.

Meyer, H. Uber die Entwicklung des Fettkorpers, der Trachean und der keim1849 bereitenden Geschlechtstheile bei den Lepidopteren. Z. wiss.

Zool, Bd. 1, pp. 175-197.

Montgomery, Jr., T. H. The Spermatogenesis of Peripatus (Peripatopsis) bal 1900 fouri up to the Formation of the Spermatid. Zool. Jahrh., Anat. u. Ontog., Bd. 14, pp. 277-368.

Morgan, T. H. A Biological and Cytological Study of Sex Determination in

1909 Phylloxerans and Aphids. Journ. Exper. Zool., 7, pp. 239-352.

204 Harry Lewis Wieman.

MuLLER, JoH. tjber die Entwicklung der Eier im Eierstock bei den Gespenstheu1825 schrecken und eine neu entdeckte Verbindung des Riickenge fasses mit den Eierstocken bei den Insekten. Nova Acta Acad. Leop. -Carol; ol., T. 12, Pars 2, pp. 533-672.

NussBAUM, M. tJber Kern und Zellteilung. Arch. mik. Anat., Bd. 59, pp. 6471901 684.

Nathansohn, a. Physiologische Untersuchungen liber amitotische Kerntei1900 lung. Ja/tr6. w;{ss..Boi., Bd. 35, pp. 48-78.

Overton, J. B. On the Organization of the Nuclei in the Pollen Mother-cells of 1909 certain Plants, with Especial Reference to the Permanence of the

Chromosomes. Ann. Bot., vol. 23, pp. 19-61.

Patterson, J. T. Amitosis in the Pigeon's Egg. Anat. Anz., Bd. 32, pp. 117-125.

1908 Pere/,, J. Sur I'histogenese des elements contenus dans les gains ovigenes des

1886 insectes. C. R. Acad. Sc. Paris, T. 102, pp. 181-183, 557-559.

Preusse, F. tJber die amitotische Kernteilung in den Ovarien der Hemipteren. 1895 Z. wiss. ZooL, Bd. 59, pp. 305-350.

Rabl, C. tJber organbildende Substanzen und ihre Bedeutung fiir die Vererbung. 1906 Leipzig.

VOM Rath, O. tJber den feineren Bau der Driisenzellen des Kopfes von Anilocra 1895 mediterranea Leach im Speciellen und die Amitosenfragen im AU gemeinen. Z. wiss. ZooL, Bd. 60 pp. 1-89.

Sabbatier, a. Sur la morphologic de I'ovaire des Insects. C. R. Acad. Sc. Paris, T, 1886 102, pp. 61-63, 267-269, 441-43.

ScHAUDiNN, F. tJber die Theilung von Amoeba binucleata Gruber. Sitz. Ber* 1895 Ges. Naturforsch. Freunde, Berlin.

v. SiEBOLD, C. Tn. Beitrage zur Parthenogenesis der Arthropoden. Leipzig,

1871 Spemann, H. tJber Correlation in Entwicklung des Auges. Verhl. der Anat. Ges.t

1901 Stein, F. Vergleichende Anatomie und Physiologie der Insekten. 1, Die weib"

1847 lichen Geschlectsorgane der Kafer. Berlin,

Stevens, N. M. On the Ovigenesis and Spermatogenesis of Sagitta bipunctata* 1903 Zool. Jahrb., Ariat. u. Ontog., Bd. 18, pp. 227-340.

1906 Studies in Spermatogenesis. 2 A Comparative Study of the Heterochromosomes in certain Species of Coleoptera, Hemiptera and Lepidoptera, with Especial Reference to Sex Determination. Carnegie Inst., Washington, Pub. no. 36, 2.

St. George, La Vallette. Spermatologische Beitrage, 1. Arch. mik. Anat., 1885 Bd. 25, pp. 581-593.

Strasbtjrger, E. tJber den Teilungsvorgang der Zellkerne und das Verhaltniss 1882 der Kernteilung zur Zellteilung. Arch. mik. Anat., Bd. 21, pp.


Germ Cells of Leptinotarsa Signaticollis. 205

Stuhlmann, Fr. Die Reifung des Arthropodeneies nach Beobachtungen an 1886 Insekten, Spinnen, Myriapoden und Peripatus. Ber. NaturJ. Ges.

Freiburg, 1.

Sutton, W. S. On the Morphology of the Chromosome Group in Brachystola 1902 magna. Biol. Bulle., vol 4, pp. 24-39.

Tannrexjther, G. W. History of the Germ Cells and Early Embryology of Cer1907 tain Aphids. Zool. Jahrb., Anat. u. Ontog., Bd. 24, pp. 609-642.

Tower, W. L. Evolution in Chrysomelid Beetles in the Genus Leptinotars a 1906 Carnegie Inst., Washington, Pub. no. 48.

Wagner, R. Prodomus historiae generationis. . Lipsiae. 1836

1837 Beitrage zur Geschichte der Zeugung und Entwicklung. Abhandl. d. mathphysik. Klasse der konigl. Bayerischen Akademie 2.

Waldeyer, W. tjber Karyokinese und ihre Beziehungen zu den Befruchtungsfor1888 gangen. Arch. mik. Anat., Bd. 32, pp. 1-122.

WheeleRjW M. A Contribution to Insect Embryology. Journ. Morph., a'oI. 1893 8, pp. 1-160.

Whitman, C. O. The Inadequacy of the Cell Theory of Development. Journ. 1893 Morph., vol. 8, pp. 639-658.

Will, L. Bildungsgeschichte und morphologischer Wert des Eies von Nepa cine 1885 rea L. und Notonecta glauca L. Z. tviss. Zool., Bd. 41, pp. 311-364.

1886 Oogenetische Studien. 1, Die Entstehung des Eies von Colymbetes

fuscus, L. Z. lui^ss. Zool. Bd 43, pp. 329-368.

Wilson, E. B., and Mathews, A. P. Maturation, Fertilization and Polarity in 1895 the Echimoderm Egg. New Light on the Quadrille of Centers.

Journ. Morph., vol. 10. pp. 319-342.

Wilson, E. B. a — Studies on Chromosomes. IV. The "Accessory" Chromosome 1909 in Syromastes and Pyrrochoris with a comparative review of

the types of sexual differences of the Chromosome Groups. Journ. Exper. Zool., vol. 6, pp. 69-98.

1909 6 — The Chromosomes of Metapodius. A Contribution to the Hypothesis of the Individuality of the Chromosomes. Journ. Exper. Zool., vol. 6, pp. 147-205.

1909 c — The Female Chromosome Groups in Syromastes and Pyrrchoris. Biol. Bulle., vol. 16, pp. 199-204.

ZiEGLER, H. E. Die biologische Bedeutung der amitotischen (direkten) Kern1891 teilung im Tierreiche. Biol. Centralbl., Bd. 11, pp. 372-415.

ZiEGLER, H. E. u. vom Rath, O. Die amitotischen Kernteilung bei den Arthro1891 poden. Biol. Centralbl., Bd. 11, pp. 744-754.

206 Harry Lewis Wieman.


Fig. 1. Longitudinal section of ovariole of larva.

Fig. 2. l^ongitudinal section of ovariole of pupa.

Fig. 3. Longitudinal section of ovariole of pupa one day older than the preceding

Fig. 4. Transverse section of tei'minal chamber of pupa; same age as preceding.

Fig. 5. Longitudinal section of terminal thread at its maximum degree of development ; four days older than the preceding.

Fig. 6. Longitudinal section through the proximal end of the terminal chamber, showing a sharp line of demarcation between the latter and the ovariole-stalk; same age as the preceding.

Figs. 7, 8 and 9. Sections through the same region showing succeeding stages, which illustrate the efTacement of the line of demarcation between the cells of the terminal chamber and those of the ovariole-stalk through the production of a semi-fluid mass in which epithelial nuclei are embedded.

Fig. U. Longitudinal section through the same region as the preceding, showing the formation of egg-strings in the adult.

Fig. 12. Longitudinal section showing egg-string at its maximum degree of development.

Fig. 13. Longitudinal section of the distal end of the ovariole whose proximal end is shown in Fig. 11.

Fig. 14. Longitudinal section of terminal chamber in region of nurse-cells of a somewhat maturer adult.

Fig. 15. Longitudinal section of ovariole showing the formation of cysts among the nurse cells lying adjacent to the young oocytes; very young adult.

Fig. 16. Transverse section through the middle of the terminal chamber showing well developed cysts; mature adult.

Fig. 17. Longitudinal section of larval testis; cysts just beginning to form.

Fig. 18. Later stage of preceding; two days older.

Fig. 19. Tripolar spindle; common among nurse cells before amitosis sets in.

Figs. 20-23 inclusive. Four stages in amitotic cell division as seen in the nursecells.

Figs. 24-32 inclusive. Illustrate peculiar transformations undergone by certain of the nurse cells. For full description see the text.

Fig. 33. Represents a section of the ovum from the region where the yolk and nutritive stream adjoin. The nutritive stream consists of red basis-staining granules (to the right) which spread along a reticular network in the meshes of which the green acid-staining yolk is deposited (to the left).

Figs. 34, 35. Two stages in amitosis in the primordial germ cells of the larval


Fig. 36. Germ cell from the pupal testis showing its relation to the surrounding epithelial cells. Fig. 37. Germ cell from the pupal ovary showing the same relationship.

Germ Cells of Leptinotarsa Signaticollis.


Fig. 38. Nurse cell from the pupal ovary showing the characteristic configuration of the nucleus.

Fig. 39. Synizesis in the ovocyte. Figs. 40-44 inclusive. Stages following synizesis. Fig. 45. Synizesis in the spermatocyte. Figs. 46, 47, 48. Stages following synizesis.

Fig. 49. First stage in the re-construction of the spireme; following the resting period.

Figs. 50, 51. Spireme of the prophase of the first maturation division.

Figs. 52, 53. First spermatocyte prophases.

Fig. 54. Side view of first spermatocyte spindle in early anaphase.

Figs. 55, 56. Polar view of first spermatocyte spindles at metaphase.

Fig. 57. Telophase of the first spermatocyte division.

Fig. 58. Telophase of the second spermatocyte division.

Figs. 59, 60. Second spermaocyte metaphases.

Fig. 61. Stage in the development of the sermatozoan in which approximately the reduced number of chromatin bodies (chromosomes?) appear in the head.

Fig. 62. Telophase of the first spermatocyte division in L. decemlineata.

Fig. 63. Anaphase of the first spermatocyte division in L. decemlineata.

Fig. 64. Polar view of the telophase of the first division in L. decemlineataa.

Fig. 65. Resting stage of the ovocyte of L. decemlineata.

Fig. 66. Resting stage of the spermatocyte in L. decemlineata.

Figs. 67, 68, 69. Three stages of amitosis in the spermatogonia of L. Signaticollis.


a. c. — apical cell

am. — amitosis

e. c— egg cell

ep. c. — epithelial cell

gr. — granule

g. c. — germ cell

I. m. — limiting membrane

m. — mitotic figure

71. c. — nurse cell n, stm. — nutritive stream n. str. — nutritive string (egg string) p. sh. — peritoneal sheath 0. St. c. — ovariole stalk cell t. pr. — tunica propria z.— chromatin nucleolus of the resting stage.

The figures in the plates are camera drawings made at stage level with the following combinations of Zeiss apochromatic objectives and compensating eyepieces.

T^ : Figs. 19, 20-23 inclusive, 33, 34^43 inclusive, 49-69 inclusive. T^ : Fig. 44. V- : Figs. 24-32 inclusive.

Figs. 1-18 inclusive.


Harry Lewis Wieman.


Germ Cells of Leptinotarsa Signaticollis.









f lisa"



Harry Lewis Wieman.

Germ Cells of Leptinotarsa Signaticollis.





Harry Lewis Wieman.


Germ Cells of Leptinotarsa Signaticollis.




Harry Lewis Wieman.


Harry Lewis Wieman













J^/e>na7t^ oU£

Journal of Morphology, Vol. 21, No. 2

Germ Cells of Leptinotarsa Signaticollis.




Harryi^Lewis Wieman.

67 M^/CTVUTi, del



From the Zoological Laboratory, Syracuse University

With Forty-nine Figures


Introduction 217

Material and Methods 218

General Account 219

Breeding Habits 220

Historical 224

Cyanea arctica 231

Cleavage 232

Gastrulation 234

Aurelia flavidula 237

Oogenesis 237

Maturation 238

Cleavage 241

The Germ Layers 242

Later Development 245

The Planula 245

The Scyphistoma 246

The Ephyra 249

Bibliography 252


Several years ago the senior author began a more or less detailed study of the life history and development of Cyanea arctica, devoting to the subject both laboratory research, and such attention to distribution and open sea habits as was afforded by a cruise of several days to the Gulf Stream on the schooner Grampus, of the Fisheries Bureau.

218 Chas. W,. and G. T. Hargitt.

The various features of larval development were followed almost continuously during the entire summer, a brief preliminary report of which was published in the American Naturalist (1902 b, pp. 555-559). Unfortunately the material obtained at this time was deficient in the earlier stages of egg development, especially those relating to maturation, fertilization and early cleavage. Attempts were repeatedly made at subsequent times to secure the desired material for completing the problem, but with only partial success; and it was not till the spring of 1908, in April, that the junior author fortunately succeeded in obtaining a fairly adequate supply for certain of the missing phases. He has accordingly, been associated in the present work of carrying forward the problem to a degree of completion which would have otherwise been more or less impracticable, and is chiefly responsible for the cytological part of the paper.

Material a^d Methods

The earlier material was obtained in July, 1901, from a few specimens of medusae which had apparently, drifted into the harbor at Woods Hole, and were bearing ripe gonads, or rather embryos in early stages of development. The occurrence of the specimens at this time, and in this condition, would seem to be more or less unusual, the usual breeding season for this medusa being in April and early May in this region. But as will be mentioned in another connection, there are known to be many variations and exceptions to this rule. As already intimated, the material was quite abundant so far as certain stages were concerned. Unfortunately, however, the methods employed in fixation of the material, chiefly picro-sulphuric and picro-nitric, reagents then much in vogue, were found to be almost worthless so far as these eggs were concerned, and as a consequence no satisfactory cytological results were obtained. Later material was fixed in various ways several of which proved most excellent, among which Bouin's fluid and Zenker's seemed to give the most uniform type of fixation, and gave no subsequent difficulty in staining. We also obtained during the current summer at Harpswell several ripe

The Development of Scyphomedusse. 219

specimens of Aurelia from which some excellent material was obtained. It was fixed by the same methods as just mentioned. In this case care was taken to secure early stages by removing portions of the ovaries and fixing in mass in Bouin's fluid and this later yielded some very desirable stages which would otherwise have been lacking.

General Account

Comparatively little is known as to details of life history and habits of Scyphomedusse. This is not to be taken as implying lack of knowledge as to many phases of developmental history in certain species. For example, it is well know^n that in the case of Aurelia and Cyanea there is a perfectly clear history from planula to ephyra, involving the intermediate phases of the polyp and its later strobilation to give birth to the ephyra. Further it is also w^ell known that there is direct metamorphosis of the ephyra into the young medusa. But there are matters of detail as to the time involved in certain stages, which are yet uncertain. For example, it was pointed out in the earher paper (1902 b,) that in some instances planulse were early transformed into polyps, and that these in turn early strobilated and gave birth to ephyrse, while others continued in these respective stages for a relatively long time. Then too, it is a matter of uncertainty as to the details of the life habits of the adult medusae ; their length of life, mode of life, rate of growth, etc. In general it seems fairly certain that the spawning season is in early spring, March to May, but with notable exceptions as pointed out in an earlier connection. Hence it may be assumed that while the earlier season is predominant in our latitude it is not restricted to this time. In the case of Aurelia the spawning season is chiefly mid or later summer, but also with probable exceptions as in the other case, a few specimens breeding in early spring along with Cyanea.

Now, as to the further features of life history, range of depth, length of life, etc., there is less certainty. It may be stated as a general fact that the entire life C3^cle falls within the period of one

220 Chas. W. and G. T. Hargitt.

year, often much less. This is especially the case with Cyanea. As shown in the previous paper the entire time from the liberation of the planula to the free-swimming ephyra may be as brief as fifteen to twenty days. The growth of the young medusae is rapid, and sexual maturity may be reached within a few weeks, or perhaps months. Just what may be the range of habits in the case of Cyanea, whether it is to any considerable degree given to inhabiting the deeper waters at any definite period of life, is a matter of uncertainty. There is little evidence, such as may be afforded by the dredge or trawl, in support of such a view; and on the other hand it seems almost certain that they are not predominantly pelagic. Conditions of weather have much to do with this feature. A rough surface invariably drives them downward, and calm weather is the occasion of a reverse movement. And this may be assumed as more or less the case with many other medusae, as well as with other organisms of similar habits of life, such as ctenophores, copepods, etc.

So far as we have been able to observe there is no distinguishable influence of light or darkness in the movements of these medusse. we have found them under all conditions of light — early morning, late twilight, the full glare of mid-day sun, indeed so common is Cyanea in this full sun glare that it is commonly designated by fishermen as the 'sun scald.'

Breeding Habits

The more or less sudden appearance of medusse in a given locality, which has often been observed and remarked upon, has been regarded by many as due to a breeding instinct which leads them at such times to seek each other. The elder Agassiz was quite explicit on this point, stating that at the time of spawning toward the end of July or beginning of August they may be seen gathering and clustering near together. That at this time they seek each other is unquestionable. I witnessed once, in front of my house at Nahant, a shoal of them, which was evidently in the act of spawning. Myriads of specimens had clustered together so closely that they formed an unbroken mass between which an

The Development of Scyphomedusse. 221

oar could not be thrust without hitting many at one blow. That they were actually spawning was ascertained by raising specimens from the water, when sperm was seen streaming freely from the appendages of the lower surface, and eggs flowing along the channel of their arms. It was about sunset and the closing night prevented me from ascertaining how long they remained together. The next day they were scattered by the wind, and a few days afterward immense numbers were found stranded upon the rocks and long sandbeach at Nahant." (1862, p. 76). A. Agassiz has also given expression to similar views on this point, though without definite facts other than the mere observation of the congregating of medusse at periodic times. (1865, p. 46).

Much more explicit is a recent reference of Conklin (1908, pp. 155-156), to the spawning of Linerges mercurius. In the case of this medusa there seems little doubt as to the facts given by Conklin and their bearing upon the matter under consideration. Something of a similar character is also known among Hydromedusae. The senior author has, in connection with the account of the development of Pennaria, made clear the intimate correlation of the spawning habits. He has also referred to the interesting occurrence of swarms of Rhegmatodes at certain times. But the latter have not apparently been associated directly with the breeding instinct, since they included specimens of all ages and conditions of maturity. He has also expressed serious doubt (1904, p. 26), of this interpretation as applicable to all cases. That the view of Agassiz cited above may find an occasional warrant need not be denied; but that it is general, or at slW frequent we do not believe. In many years of observation little evidence has been found to sustain the view. Furthermore, the fact that isolated specimens of both Aurelia and Cyanea have been taken, males and females bearing ripe gonads, the latter in various stages of maturity and with eggs in various stages of development, would further support it. Indeed, in the case of Cyanea, it is rather rare to find any considerable numbers together. The first collection included but three or four specimens, including both sexes. Again in the case of Aurelia, which we collected in large numbers about the first of August, there was no massing of numbers at any time.

222 Chas. W. and G. T. Hargitt.

Collections were made at all hours of the day in various localities, and while the sexes were seemingly in approximately equal numbers there was no evidence of '^ seeking each other," such as Agassiz asserts. Still a further fact of even greater significance remains to be noted, namely, that in both Aurelia and Cyanea spawning is not a single, or spasmodic process, but one continuing during several days. One finds on a given specimen embryos m all stages of growth, from blastulse to planulse, and eggs in all stages of cleavage, and at the same time the ovaries loaded with eggs in various stages of growth, from oogonia to ova in maturation. It is difficult to correlate this condition with the assumption of any sudden, single act of spawning.

Egg-laying — This feature has been referred to incidentally in a previous section. It only remains to call attention to a few points not already mentioned. It is possible to distinguish the males and females of Aurelia when sexually mature and bearing gonads: the male organs being milky-white, while in the female they are pale-pinkish or purplish, the eggs having these tints when viewed in any considerable mass. This is less marked in Cyanea, in which both sexes have whitish or cream-colored gonads.

So far as our observations go there is no definite time of day at which egg laying takes place. This is seemingly in sharp contrast with what is known in many other medusae. Conklin (loc. cit. p. 157), finds in the case of Linerges that this occurs about 8 a.m., and at no other period of the day." At this time he finds that for a short time "a. perfect epidemic of egg-laying takes place, after which no other eggs are laid till the following day." We have referred to a similar condition in Pennaria, though in this case it occurs in the evening, just about the twilight. The account of Professor Agassiz, previously cited, would indicate something similar in Aurelia. But this must be considered very doubtful. We have collected Aurelia at all hours of the day, and the senior author has kept females in the aquarium for days in succession, but in no case has there been apparent any such exhibition as would warrant the assumption that egg-laying in either of these medusae takes place at any definite time. We were the more particular upon this point in collecting eggs of Aurelia, since it was desirable to

The Development of Scyphomedusse. 223

obtain ova in earl}" maturation and cleavage. This it was not possible to do in any quantity, egg-laying appearing to be a more or less continuous process during all times of day, and for a number of days in succession. Another feature will make this more evident. In both Aurelia and Cj^anea the eggs are not definitely discharged from the female; but after dehiscense from the ovary into the gastric cavity, where apparently fertilization takes place, they are ' nursed' in pocket-like folds of the oral arms for some time or until they are well-developed planulse. Hence one finds on attempting to collect eggs that he gets all conditions of development at the same time, though in the immediate proximity of the mouth there will be a preponderance of cleavage stages and early blastulse. That egg-laying is not a single process, as shown above, is evident in that one finds specimens with the gonads in various stages of depletion. Furthermore, in a study of the histology of the gonads one finds eggs 'in all stages of growth, a fact incompatible with the assumption of the shedding of the entire crop at a single time.

The eggs, when they escape from the ovaries, have already gone through the process of maturation. This would seem to take place just prior to the rupture of the follicular membrane by the egg, as will be shown in a later connection.

Following the liberation of the sexual products, and in the case of the females the final escape of the larva), there seems to be a rather rapid decline in the vigor and activity of Aurelia and early death ensues. Many specimens may be found drifting along shore lines and in harbors, and, if there be wind or tide driving them shoreward, they become stranded and rapidly disintegrate. An examination of many such during the past summer showed that in most cases such specimens were dead or dying when they came ashore. The case of Cyanea seems somewhat different. This medusa seems to live for considerable time after passing the spawning period. As stated before, the breeding season is usually in spring. But it is not unusual to find many specimens during early and late summer swimming freely in the usual manner. It is, however, rare to find such specimens bearing genital products. In many specimens collected in Casco Bay during the past summer not one w^as found with gonads.

224 Chas. W. and G. T. Hargitt.


It was found by Kowalevsky (1873) that Cassiopea borbonica and Aurelia aurita went through regular cleavage, and a blastula with a relatively small cavity was formed. The entoderm was formed through a small invagination of the blastula wall, and was soon entirely separated from the ectoderm, a relatively larger cavity remaining between the two layers. Glaus (1878) stated that Ghrysaora passed its entire embryonic development within the ovary, emerging as a planula. The egg cells originated in the germinal epithelium and were covered by a follicle developed from this layer. The unequal cleavage, he believed, began while the eggs were still small and continued during the growing period, resulting in a blastula with shorter cells at one pole. From this region an ingrowth of cells took place and the entoderm was formed from these, the cleavage cavity being obliterated and the blastopore completely closed.

Haeckel (1881) studying Aurelia aurita, agreed with Glaus (Ghrysaora) that the first cleavage was not quite equal, and he referred to a differentiation of the smaller (animal) and larger (vegetal) blastomere, a difference believed to be retained for a considerable time. This inequality disappeared in later cleavage so that the blastula was composed of a large number of equal cells, enclosing a large cavity. The gastrula was formed by an invagination of one side of the blastula, the blastopore closing, but later the larval mouth ('nachmund') breaking through in the same place. As variations from the above processes he found that the cleavage might be quite unequal; that the invagination might be incomplete, the archenteron limited to about one third of the cleavage cavity, the remainder being filled with a jelly-like substance. When the gastrula was formed thus it usually transformed directly into an ephyra, skipping the planula, scyphostoma and strobila stages; or the blastopore of the completely invaginated gastrula remained open and the embryo settled down as an actinula with several tentacles, thus omitting the free-swimming planula stage. He also agreed with some earlier authors that the planula might bud or divide into several planulse each of which formed a scyphostoma.

The Development of Scyphomedusse. 225

Claus (1883) was the next to study Aurelia aurita. He called attention to the fact, known to early observers, that the eggs pass from the ovaries into the gastric cavity and from there into the folds of the manubrium where development takes place. Haeckel was incorrect in thinking the first cleavage unequal and especially in the differentiation of the two blastomeres, since, as Claus stated, the first cleavage furrow passing through the animal pole can not divide the egg into an animal and vegetal blastomere. The second cleavage Claus found to be meridional and the third equatorial. Up to this time cleavage was equal but thereafter might be unequal and form a blastula with shorter cells at one pole. As he found that the cleavage cavity, present as early as the 8-cell stage, increased only slightly and consequently remained small, he believed Haeckel was mistaken about the large cavity and wide invagination. Because of the small size of the invagination the cells were crowded together and resembled very much the appearance of the cell mass formed by ingression in Aequorea, but he w^as able to show that a real invagination had taken place. In Chrysaora Claus found essentially the same process of cleavage except for the smaller egg, the development within the ovary and the very large cleavage cavity. An ingrowth of cells from the thicker portion of the blastula wall formed a solid plug but with cells arranged in two rows, so that the process was quite close to an invagination. As the archenteron grew the ectoderm also extended, and for a long time there remained a space between the ectoderm and entoderm.

In Lucernaria Kowalevsky (1884) found a regular and equal cleavage, the first and second furrows meridional, the third equatorial. No cleavage cavity was formed, segmentation resulting in a solid mass of cells, an outer layer and an inner mass. The latter had its origin from a delamination of the early cleavage cells or, in some cases, from an actual immigration of entire cells. From the inner cells the entoderm was formed, the outer row being the ectoderm. Metschnikoff (1886) confirmed the absence of invagination in Lucernaria, and also described the results of a study of Nausithoe marginata and Pelagia noctiluca. In the former, although the third and fourth cleavages were unequal, the

226 Chas. W. and G. T. Hargitt.

cells of the blastula were so nearly equivalent in size that no polar differentiation was possible. The blastula which (unlike Aurelia aurita) had a very large cavity, took on a somewhat elongate form, the cells of one side became thicker and at this point an invagination occurred. Even during the invagination the cells began to be differentiated from the ectoderm cells and when the blastopore had closed there was a typical planula with two layers differentiated. Pelagia noctiluca differed mainly in the wider extent of the invagination and in the absence of an early differentiation of the entoderm cells. Also during invagination the ectoderm grew so that the two layers came to be further apart than at first. The invagination in Pelagia was confirmed much later by Goette (1893). Metschnikoff thus confirmed the earlier workers on the formation of the entoderm by invagination, and he believed this process to be a wide-spread one in the Acraspedota [Scyphomedusse] .

This general belief in the wide extent of invagination among the Scyphomedusse was vigorously attacked by Goette (1887), who claimed that Aurelia at least, and probably in other Scyphomedusse, gastrulation had been erroneously described. This was due, in his opinion, partly to the fact that sections has not been used in the study of the gastrulation. Furthermore since a coeloblastula was the starting point of the gastrulation and a coelogastrula the end result, it seemed to be necessary and was natural to assume that in the stages between, an invagination had taken place. But such an invagination does not occur in Aurelia" either as described by Haeckel or by Glaus. A sterrogastrula is formed, " . . and in Aurelia repeating exactly this process, the gastrulation occurs through a cell-inwandering." The cells which took part in the ingression came from the region of the blastula made up of shorter cells, and were either entire cells, or portions of cells set free by a process of delamination. The cleavage cavity thus became filled with a mass of cells, the entoderm, which by splitting apart assumed a position in a single layer. The ectoderm and entoderm then fused at one point and a prostoma was formed: "archenteron and prostoma therefore arise not through invagination, but through the hollowing out of a massive entoderm and

The Development of Scyphomedusae. 227

a breaking through of this cavity to the outside" (p. 5). Before the prostoma closes the embryo is exactly similar in appearance to an invaginate gastrula. Although he could not follow the entire process in Cotylorhiza, he found conditions which led him to believe that here also an ingresson of cells was the process leading to gastrulation. This view was opposed to Claus and in 1891 this investigator gave the results of further observations upon Cotylorhiza, Aurelia and Chrysaora. In Cotylorhiza he found no ingression, but a true invagination. Aurelia showed some ingression of cells which he believed did not take part in entoderm formation; rather an almost typical invagination was the process as he had found earlier (1883). In Chrysaora he confirmed his earlier results.

As a result of his own work, and from a summary of that of others, Hamann (1890) held to Goette's views that polar ingression was the more common process of entoderm formation in Scyphomedusse. His own observations upon Aurelia aurita, Chrysaora and Cyanea capillata led him to this conclusion, and from the papers of others he believed that a real invagination occurred only in Nausithoe and Pelagia, while at least ten other species of Scyphomedusse showed a polar ingression. McMurrich's (1891) observations upon Cyanea artica seemed to confirm this, though he found the ingression to be multipolar. Although the end result was a structure like an invaginate gastrula, invagination had taken no part in the process, according to McMurrich.

Once more new evidence of invagination was brought forward by Smith (1891) working upon Aurelia flavidula. He found, indeed, that there was an ingression of cells, which might begin long before invagination started. This ingression resembled that found by Goette in A. aurita, but it was not of constant occurrence; indeed the majority of Smith's preparations did not show any ingression, though invagination occurred in all. He found further that only three or four cells took part in the ingression, the nuclei of these always broke up into small particles and later the cells degenerated as Claus also found. Sometimes they persisted till gastrulation was completed, but did not take part in the entoderm formation, some of them were even forced through the entoderm

228 Chas. W. and G. T. Hargitt.

into the coelenteron where degeneration occurred. Hence Smith conchides that ingression plays no part in the entoderm formation of Aureha flavidula.

A comparative study of several species of Aureha and of Cyanea arctica was made by Hyde (1894) who found that the cleavage might be regular or irregular, equal or unequal. The first and second cleavage furrows were meridional, the third equatorial. After the third division the cells were commonly smaller at one pole. In all species a blastula was formed, the cavity usually appearing first in eight or sixteen-cell stage. In A. marginalis a real delamination of the blastula cells occurred and was multipolar in extent, in this resembling Lucernaria (Kowalevsky), the inner cells formed an irregular layer and finally the prostoma broke through. In A. flavidula an ingression of entire cells, or a delamination, took place from different parts of the blastula wall and the cells thus free in the cavity assembled at one pole. This pole bent in to form a small funnel-shaped opening, the free cells grouped themselves about this in a layer, and the gastrula was formed. This is somewhat similar to the condition in A. aurita (Goette), the entoderm being formed from a small portion of the inbent wall of the blastula and from the cells which had migrated to this pole. A second method of gastrulation was found resembling that described by Glaus (A. aurita) and Smith (A. flavidula), viz., an invagination, though a few cells might migrate from the wall and join with the invaginated cells. Hyde found that one of the characteristics of the cells which wandered into the cleavage cavity was that their nuclei were usually broken into small chromatin particles which were scattered through the cell, a condition noted earlier by Smith. In Cyanea arctica an invagination took place, but there were added to the invaginated cells some free cells arising by delamination from the blastula wall of the invagination pole. No migration of cells occurred in Gyanea. Thus in all three species Hyde finds a blastula with differentiated poles and in the cleavage cavity, often, a coagulated liquid and yolk grains. In the process of entoderm formation delamination occurred alone or in conjunction with other processes : in A. marginalis delamination alone; in Gyanea delamination limited to a

The Development of Scyphomedusse. 229

few cells at the pole where the invagination occurs, invagination being the chief process. In A. flavidula some blastula cells delaminate and some entire cells migrate into the cavity and these join a small invagination; or a large invagination occurs and this is only sometimes helped out by a few delaminated or migrated cells.

Hein's (1900) work upon Aurelia aurita gave results differing from those of Goette, Smith, Hyde, and others, in that he found it impossible to distinguish animal and vegetal poles because of the similarity of the cells of the blastula. He agrees with Glaus and Smith that from different regions of the blastula wall a few cells may migrate inward, but these never take part in entoderm formation, degenerating sooner or later. He found also an occasional migration of entoderm cells into the coelenteron where they degenerated. An invagination, with a rapid division of the invaginated cells, led to the formation of the entoderm, the blastopore persisting as a very fine canal between the archenteron and the outside.

Goette (1900), in a short response to Hein's work, upheld the work of Hyde in all particulars, and though he acknowledged that invagination was the method of gastrulation in some cases, he still maintained the view that ingression plays a more prominent and active part among Scyphomedusse. He discards entirely the view of Smith and Hein that the immigrated cells degenerate and take no part in entoderm formation.

C. W. Hargitt (1902a, 1902b) found that in Cyanea arctica the early cleavage stages were passed while within the gastric cavity or in the folds of the manubrium. Cleavage was total and regular, a typical blastula formed which by invagination gave rise to the gastrula.

Continuing his investigations upon Scyphomedusa, Hein in 1903 presented the results of work upon Cotylorhiza tuberculata. He found (as in Aurelia aurita) that a few cells ixdgrated from the blastula wall into the blastocoel, but there they always degenerated. Gastrulation was by invagination, as he found in Aurelia aurita, but in Cotylorhiza the blastopore soon closed while in Aurelia it remained to form the prostoma. In neither species was

230 Chas. W. and G. T. Hargitt.

he able to find any proof for Goette's assumption that the immigrated cells took an active part in entoderm formation; hence he believes Goette to be incorrect in this point. Concerning these same cells Hein refers to the statements made by Hyde that, the nuclei were usually broken into small fragments and scattered through the cell, and to the observations of Smith that no such cell ever had an intact nucleus. Hein concludes that such fragmented nuclei . . legen wohl einen Zerfall der Zellen zum mindesten nahe."

More recently Conklin (1908) has studied Linerges mercurius. The eggs are deposited about 8 a.m. in masses surrounded by jelly. Polar bodies form typically, the sperm enters at the vegetative pole and moves to the animal pole where it fuses with the egg nucleus. Cleavage begins at the animal pole and at the end of the first cleavage there is a small space between the two cells which is the first indication of the cleavage cavity. The second cleavage is meridional, the third equatorial and both begin in the center of the egg and pass outward. The animal pole becomes the ectodermal pole, the vegetal the entodermal. Gastrulation is usually by invagination, though sometimes an ingression of cells from the vegetal pole fills the cleavage cavity, and the archenteron only later forms by a splitting apart of the cells. The end result is the same in the two cases, and this Conklin believes is an indication of the close relationship of unipolar ingression and invagination.

From this review it will be seen that the chief controversy with regard to the early development of the Scyphomedusse has centered in the method of gastrulation. The differences in cleavage have been found rather insignificant and the results of gastrulation have in every case been the formation of a typical two layered planula. The main reason for the strict adherence to a belief in one or another single mode of gastrulation by some of the disputants appears to rest upon the belief that one mode must be more primitive than another; that there would be different phylogenetic relations indicated if this, rather than that, method were more common.

As early as 1881, Haeckel showed that there were marked varia

The Development of Scyphomedusse. 231

tions in the gastrulation of Aurelia aurita; and the later work of Hyde upon three species of Scyphomedusse showed other variations in the gastrulation even within a single species (A. flavidula). Hence there seems no reason further to insist upon the activity of any one, or any single process in entoderm formation. Nor does there appear to be any occasion to hold any longer to the view that differences in the gastrulation process have any necessary significance in phylogeny. Conklin has given expression to this view when he says (p. 163), " . . the form of gastrulation is of no fundamental or general significance, but that it depends upon individual or environmental conditions."

The observations described in the following sections are not intended to add anything essentially new to the controversy regarding gastrulation, but to record the facts brought out in a study of Cyanea arctica and Aurelia flavidula, the development of the former species never having been fully worked out. It may be said, however, that the results of this study and a careful comparison of the earlier papers has led to the conclusion that invagination is probably a more general and dominant method of entoderm formation in the Scyphomedusse than was thought by Goette, Hamann and others.

Cyanea Arctica

The development of Cyanea up to the formation of the planula takes place within the folds of the mouth lobes as stated in an earlier paper (1902 b) . Cleavage stages are passed through rather rapidly so that only the later stages are found in most medusse. Specimens obtained in early spring 1908, however, gave material which makes possible the study of the early development, though not the oogenesis and fertilization. In no case were there found in the mouth lobes more than an occasional egg which had not already begun to segment ; hence the formation of polar bodies and fertilization must occur either within the gastric cavity, or before the eggs leave the ovary. The conditions in Aurelia, which are described later, suggest the probability that the formation of the polar bodies occurs at about the time of the escape of the eggs from


232 Chas. W. and G. T. Hargitt.

the ovary, and that fertUization takes place a little later in the gastric cavity.

The material obtained from Cyanea does not permit the determination of the cytologic details of the maturation process nor of fertilization. But it was found that two polar bodies were formed one of which sometimes divided again. The polar bodies are of considerable size, relatively larger than polar bodies in the Hydromedusse. The chromatin is usually enclosed within a membrane, but is collected into several more or less distinct masses, not forming a reticulum. The polar bodies remain attached to the eggs for varying periods, being occasionally found in blastulse, though they commonly disappeared soon after the beginning of cleavage, probably through the rupture of the delicate egg membrane.


The first division of the egg is meridional, the cleavage plane in some cases (figs. 1, 4) being almost co-incident with the plane of the polar axis, while in other cases (fig. 11) there was considerable divergence between these two planes. There result two cells which are nearly equal in size (figs. 1, 3, 4), though rarely exactly so, and in a good many cases there is a marked inequality (fig. 2) one blastomere being perhaps only about half the size of the other. This irregularity is also quite marked in the cleavage of Aurelia. It can be followed without difficulty in the living egg, and one finds upon looking over a series that a considerable number exhibit this feature, not only in the size of the blastomeres, but in their arrangement also.

When the first division is completed there is often present a small space between the blastomeres (fig. 12). This can be seen in entire eggs as well as in sections, and while not present in all eggs it persists, when present, to become a part of the large cleavage cavity. This early appearance of the cleavage cavity, not described by earlier workers, Conklin (1908) found to be also characteristic of Linerges mercurius.^ He found among the organized subit may be doubted whether any such significance attaches to the cleavage cavity, as was earlier assumed, or whether the space referred to in the two cell stage can rightly be so regarded. — C. W. H.

The Development of Scyphomedusse. 233

stances an apparently more liquid substance, containing few yolk spherules, in the center of the egg. This he believes is the precursor of the cleavage cavity." There was no evidence of such a substance in Cyanea eggs, but in Aurelia the center of the egg in a few cases showed what appeared to be a similar differentiation. It is also clear that in both Cyanea and Aurelia a liquid substance is present in the early cleavage cavity. It may be rather doubtful whether this comes from a substance already differentiated in the unfertilized egg, surely not in all cases.

In eggs of most animals the second division is meridonal and other workers have generally described the same thing in the Scyphomedusse. So it was found in Cyanea (figs. 6, 11) that such might be the case. In view of the variations in the first cleavage it was not surprising to find instances (figs. 4, 13) where the second cleavage was equatorial, and this was not rare. This condition, of course, is often found in eggs which are erratic and irregular in cleavage as some of the Hydromedusse, Pennaria in particular. It was possible to follow the cleavage in the living Aurelia egg and to observe that in some cases the second division was. equatorial. When the first cleavage in Cyanea had been unequal a very common sequence was for the large blastomere to divide before the smaller giving a three-cell stage (fig. 5), a condition earlier noted in Cotylorhiza (Goette 1887) and Aurelia (Hyde 1894). The three cells were bf approximately equal size and thereafter the cleavage was fairly regular.

After the second division, cleavage is more or less regular, but the cells are usually unequal in size. There appears to be no very definite sequence of division, so far as could be determined, nor was there any marked synchronism, since there were found 3, 5, 7, 8, 10, 12, etc., cell stages, as Hyde earlier noticed. The impression obtained from the study of many eggs was rather that of a certain independence of each blastomere in its division. In the 8-cell stage (figs. 7, 14) the cleavage cavity, which first appeared in the 2-cell stage (fig. 12) was larger, and this increase in size continued through the later stages until in the blastula there was a large cavity surrounded by a single layer of cells which were smaller at one pole (figs. 10, 16). The cells were often so crowded

234 Chas. W. and G. T. Hargitt.

together that some were pushed toward the outside, not reachmg the cleavage cavity, and some toward the inside, not reaching the outer surface. This gave to the blastula something of the appearance of a double layer of cells. The cytoplasm was so filled with numerous closely packed yolk granules that cell boundaries were indistinct; the limits and number of cells must then often be estimated from the number of nuclei. As nearly as could be estimated about 400 cells are present in the blastula just before gastrulation. Figure 16 is a section of an early blastula with rather large cells ; these by further division become much more numerous and smaller.

As pointed out by the senior author in a previous paper (1902 b, p. 555) gastrulation in Cyanea may be easily observed in the living egg to involve a very obvious invagination. This is just as evident in Aurelia. One has only to remove some of the eggs to a watch glass where the various phases may be followed from the beginning to the closure of the blastopore. The entire process is compassed in a comparatively brief time.


Sections of Cyanea show that the cells of the blastula are longer at one pole (figs. 15, 16) ; whether this is the animal or the vegetal pole could not be determined on account of 'the absence of the polar bodies at this stage. The first indication of gastrulation is the flattening of a portion of one side of the blastula, usually the thinner side as Hyde (1894) noted, followed by an invagination. Often at the first the mouth of the invagination is broad and open, but it soon narrows until the blastopore is almost closed (fig. 17). The closure may be completed before the gastrulation is finished (figs. 17, 18). When the invagination begins the coelenteron may be broad, though shallow; as the blastopore closes the coelenteron becomes smaller and may be almost entirely absent (fig. -18). Later this cavity again becomes more evident as is shown in the figure of the cross section (fig. 19), the invaginated cells more nearly filling the cleavage cavity. Still later these infolded cells, which are the primitive entoderm cells, come to lie closely against

The Development of Scyphomedusse. 235

the outer or ectoderm cells, entirely obliterating the cleavage cavity.

While more recent investigations have tended to discount the older views as to the importance attaching to the methods of formation of the primary germ layers, it will not be without some interest to briefly refer to some phases of the process. In addition to the invagination of cells there may be some immigration, though this is not common and may take place before or after invagination. Such cells appear to take no part in entoderm formation, and while their ultimate fate was not determined, they probably degenerate as was found by Glaus (1891), Smith (1891) and Hein (1900, 1903). Some sections seemed to show that a delamination of the cells of the blastula might occur, the radially placed spindle of figure 15 clearly foreshadowing such a process. This is not at all a common condition and so far as could be determined the delaminated cells did not take part in the formation of the entoderm.

The nuclear spindles in cleavage stages are quite similar to regular spindles in other forms, having very clear, though delicate, spindle fibres with asters at either pole ; in no case was it possible to demonstrate the presence of any centrosphere or centrosome. The chromosomes in cleavage spindles are well separated, but extremely small. In no instance was there found an amitotic division during the cleavage.

One of the most characteristic of nuclear phenomena in the cleavage stages of Cynea was found in the resting" nucleus. In the majority of eggs, in early cleavage stages, the nuclei were composed of several vesicles (figs. 12-14, 20-24). Usually there were two of these vesicles almost i'dentical in appearance and of equal size; other nuclei were single or composed of three to five vesicles, sometimes quite unequal in size (figs. 22, 24). These vesicles were found rather commonly in the cells of all eggs from the 2-cell stage to a blastula containing 24 to 48 cells, though not all the cells in any one egg contained the several vesicles. In tne blastula just before invagination, and in the gastrula, they were less abundant and often entirely lacking, especially in the gastrula. Conklin (1908) in Linerges, sometimes found two equal chromo

236 Chas. W. and G. T. Hargitt.

somal vesicles in the telophase of the first division and believes that they represent the gonomeres of Hacker (1902), i. e., paternal and maternal nuclear constituents which have remained distinct and separate. Such a view is attractive, but the conditions in coelenterates other than Linerges seem not to bear out this interpretation. In Cyanea the presence of more than two vesicles, and of vesicles of unequal size, is rather opposed to this view. We are more strongly opposed to this suggested interpretation, because of facts presented in another paper (G. T. Hargitt, 1909) that some of the Hydromedusse are even less regular and constant in this feature. For example in Tubularia crocea it was found that in the cleavage stages, up to what represents a blastula, such double nuclei, and indeed any chromosomal vesicles, were absent, while in the cells of the blastula and in the developing ectoderm and entoderm cells double nuclei were common. Thus in the earlier stages where the distinctness of sperm and egg chromosomes should have been more evident there was a total absence of this condition in Tubularia. As is well known also, Dublin (1905) found that nuclei in Pedicellina which appeared double in somatic and germ cells were not to be considered as representing gonomeres. These several points appear to offer an objection which it is difficult to overcome. Since in many animals there are often formed many chromosomal vesicles in the telophase of division and these later unite into a single vesicle, w^hy may not these vesicles wherever found be better considered as simply stages in the reorganization of the nucleus, sometimes passed through and sometimes not, at least until it can be demonstrated in each case that the paternal and maternal nuclear constituents do actually remain distinct and separate. In Cyanea, Aurelia and certain Hydromedusse at least there seems to be no question of the absence of any evidence of gonomery. Chromosomal vesicles often form, and may delay a long time before uniting into a single vesicle or may even fail to so unite — perhaps as a result of rapidly succeeding division — but there is no evidence in this to warrant the assumption of the distinctness of paternal and maternal constituents.

The Development of Scyphomedusse. 237



Sections of the ovary of breeding medusae showed that young germ cells were still quite abundant, as well as quite a considerable number of older eggs, though the majority of the latter had been discharged. The eggs in the ovary secure their nourishment through the germinal epithelium to which they remain attached until mature ; no absorption of other cells takes place.

In Tubularia crocea it was found (G. T. Hargitt, 1909) that the stages immediately following the last oogonial division were of great interest for it appeared probable that a synapsis occurred at that time. Attention given to similar stages in Aurelia gave the following results. The last oogonial division, which is mitotic, gave rise to small oocytes in which the chromatin arrangement is not easily made out. There appears to be a spireme more or less massed together and concentrated toward one side of the nucleus and perhaps arranged in loops; no instance was found, however, in which this could be certainl}^ determined. In older oocytes (fig, 25) the chromatin was in a thread for only a short time, but all trace of a polar arrangement (if any existed) had been lost. The spireme disappears soon after growth begins and does not again reappear, in the eggs within the ovary, nor does a synapsis occur during the growth of the egg, for the stages of this period are sufficiently abundant to make it highly improbable that such a condition, if present, would be overlooked. One may only conclude then that if a synapsis occurs in Aurelia it probably takes place in the oocyte just before growth begins.

The changes in the chromatin of the germinal vesicle during the growth period involve, (1) the disappearance of the spireme, already noted, and the formation of a reticulum; and (2) the condensation of the chromatin near the end of the period and the formation of the chromosomes. The reticulum in the early growth period (fig. 25) is wide meshed and the chromatin is mostly assembled into a few large, rather dense, flocculent masses. In older

238 Chas. W. and G. T. Hargitt.

eggs (fig. 26) the reticulum is more complex, fine-meshed and granular, and the chromatin is in a large number of granular, less dense masses. The diffusion of the chromatin does not go beyond this condition, a marked contrast to what happens in Hydromedusae. In certain of the latter it has been found that characteristically the chromatin becomes so finely divided and so diffused that a stage is reached in which the nucleus appears to be without chromatin, though it is still present and later becomes more evident. While the reticulum in Aurelia will select an acid dye, the larger, denser masses always stain with the basic dye; hence the germinal vesicle never has quite the cytoplasm-like appearance that it does in nearly mature eggs of Hydromedusse, Near the end of the growth period the granular m? sses of chromatin undergo a process of condensation, leading to the assumption of a more definite shape (fig. 27), straight or twisted strands, loops and rings, considerably smaller than the original chromatin masses. During the progress of these changes the nuclear membrane has become wrinkled, the egg has reached its maximum size, the nucleolus is degenerating; it is therefore clear that the time for the formation of polar bodies is near at hand and the chromatin changes described are the beginning of chromosome formation. In the eggs of many animals the chromatin rings and loops become chromosomes arranged in tetrads, in Aurelia these are apparently absent in the maturation spindles (fig. 28, 29).


The first stages in the formation of the maturation spindle were not found, unless figure 26 is one such stage. Here a single small aster centres in two granules (centrosomes) which have, perhaps, just divided. If this be the beginning of the spindle it is certainly precocious, for the egg is not full grown and the chromatin and nucleolus are in an earlier stage than in figure 27 ; but in the latter and other similar eggs there is no indication of asters or centrosomes. Whatever the genesis of the spindle, it is at first tangentially placed, as shown in the figure of the incomplete spindle, (fig. 28). Figure 29, a polar view of the first maturation

The Development of Scyphomedusse. 239

spindle, shows a later stage with all the chromosomes at the equator. The chromosomes are smaller and more numerous than in figure 28, giving evidence that a splitting has taken place. In fig. 28 there are about 9-10 chromosomes, in fig. 29, 18-20 and in cleavage cells at least 18-20, so it appears that before the maturation spindle has formed the conjugation of chromosomes has occurred. In the ovaries examined five eggs were found which showed the first maturation spindle forming, but none showed the second spindle and none the polar bodies. These facts, then, make it clear that the polar bodies begin to form just before or at about the time the eggs leave the ovary. In all probability this process is completed and fertilization takes place while the eggs are in the gastric cavity.

A comparison of figures 28 and 29 with figures 27 shows how small a portion of the germinal vesicle takes part in the formation of the chromosomes of the maturation spindle, a point already noted in Coelenterates (Conklin 1908, Small wood 1909, G. T. Hargitt 1909). What has become of the rest of the nuclear contents? The chromatin in the maturation spindle (figs. 28, 29) is manifestly less than that in the germinal vesicle (figs. 25, 27), and no additional chromatin particles are found near the spindles. Many eggs show spherical bodies in the cytoplasm close to the nuclear membrane (fig. 25-27) sometimes with a vacuole about them, and the question arises : is this chromatin which has migrated from the nucleus before the rupture of the membrane? Such an interpretation has been given to apparently similar bodies in Hydromedusse eggs by C. W. Hargitt (1904 a, 1904 b, 1906) and Smallwood (1909) . These extra-nuclear bodies of Aurelia stain as intensely in the iron hematoxylin as do the chromosomes, but in hematoxylin-eosin they take the red or acid dye. This is evidence against their chromatic character, unles<* one assume that in the cytoplasm they change their staining reaction, a not improbable view. The conclusion must be reached that the greater part, at least, of the achromatic substance and superfiu^as chromatin at the time of the breaking of the membrane of the germinal vesicle escape into the cytoplasm and are rapidly absorbed or mixed with it. The relatively enormous size attained by the germinal vesicle

240 Chas. W. and G. T. Hargitt.

and the great amount of chromatin present, which does not take part in the formation of the definitive chromosomes, are points of considerable interest and significance, but are probably incapable of complete explanation in the present state of our knowledge. But may not a partial explanation possibly rest on the ground of a physiological need during the growth period, as earlier indicated by Wilson (1900, p. 128) when metabolism is extremely active, in the elaborating of reserve food material which shall furnish the energy for the following cell divisions and for growth? When the growth period is completed and the reserve food formed a considerable portion of the nuclear substance, now unnecessary, is cast aside, just how is of little importance.

The history of the nucleolus during the growth period contains some points of interest. This body in young oocytes stains red in hematoxylin-eosin ; in growing eggs sometimes red, sometimes blue or purple, even in eggs of the same size and of apparently similar age. Its composition is thus somewhat uncertain and its changes do not take place at any fixed or definite phase of the nuclear cycle. One or more small vacuoles are often present and each of these usually contains a small spot or granule which always stains red, regardless of the reaction of the rest of the nucleolus.

Near the end of the growth period the nucleolus becomes nearly transparent, but usually shows a deeply staining cap upon one side (fig. 27) which stains sometimes with acid sometimes with basic dyes. It appears as if substances were leaving the nucleolus during this period. Figure 27 suggests a connection between the nucleolar cap and the chromosomes as though nucleolar matter was passing into the chromosomes. This is not necessarily true, for the masses of chromatin are often present only in that side of the nucleus opposite the nucleolus and apparently with no connection between, though the reticulum undoubtedly furnishes an indirect pathway along which an interchange of material might take place.

A description of the cytoplasm at different stages will indicate what changes occur. In the young oocj^te (fig. 25) the cytoplasm

The Development of Scyphomedusse. 241

is composed of a dense mass of fine grains uniformly arranged, and staining blue (hematoxjdin-eosin) . The nearly mature eggs (figs. 27-29) have an alveolar cytoplasm, the walls of the large alveoli being made up of granules quite similar in size and staining reactions to the grains present in the young oocytes. Within each, alveolus is a more or less spherical body, not completely filling the vacuole, though of varying size, which always stains with the eosin. These are the yolk bodies and they are so abundant and so large that they give to the egg substance a red color with the stain used, the finer granules being hardly noticeable with low magnification.


Although the cleavage of Aurelia was not followed in detail, it appears to differ little if any, from that of Cyanea, and a similar blastula is formed. Hyde (1894) found that the gastrulation process was somewhat different in material from Maine and in that from the Massachusetts coast. If the differences be correlated with the environmental conditions of the regions, our material (from Maine) should resemble that portion of Hyde's which came from a similar locality. As a matter of fact, however, the gastrulation proves to resemble that of her Massachusetts material, and confirms the results of Smith (1891) and our results on Cyanea, all from the more southern material.

The presence of cells free in the cleavage cavity was very rare and from many different stages of gastrulation it is evident that invagination is the chief, if not the only process. Fig. 30 shows an invagination just beginning, only a small portion of the blastula wall being involved. The invagination extends inward, assuming a condition similar to that shown in fig. 31 ; this extension of the infolded layer of cells being due in part to cell division and in part to a change in the shape of the cells. The closure of the blastopore, which soon takes place, begins inside and proceeds outward (fig. 33).

242 Chas. W. and G. T. Hargitt,

Germ Layers

One can find other sections of gastrulse which appear to show an ingression of cells (fig. 32), but in all our preparations this is due only to orientation, for, if the gastrula be cut only slightly oblique, some cells of the entoderm are cut more or less transversely and give the impression of a mass of cells. Only when the plane of the section passes through the long axis of the invagination can the true condition of affairs be found. Whether the cells sometimes .found free in the cleavage cavity do actually take part in entoderm formation may possibly be answered by the conditions show in fig. 34. Here is a condition similar, perhaps, to those found by Hyde where an immigrated cell was joining the invaginated cells. It is questionable, however, whether this cell is actuallytakingpartinthe entoderm formation forthe invagination in this embryo is nearly completed and the entoderm cells arranged in a compact row except for the indentation caused by the cell %' . A similar cell, 'a\ is present in the coelenteron; this together with the position of '6,' and particularly the condition of the layer of entoderm cells near this cell, suggest the probability that cell '6' is being forced, or is migrating, through the entoderm layer, into the coelenteron. Such cells in the coelenteron degenerate. Further evidence touching the fate of the cells free in the cleavage cavity is found in the condition of their nucleus. In agreement with Smith (1891), Hyde (1894), and Hein (1900) it was found that in Aurelia these cells rarely (in our preparations, as in Smiths's, never) showed an intact nucleus, but there were only deeply staining granules present which may have been bits of a fragmented or degenerating nucleus. The presence of only scattered chromatin grains in the cells is pretty good evidence of the degeneration of those cells as Smith and Hein pointed out, though Goette (1900) and Hyde (1894) believe it does not warrant this interpretation. The entire absence of any sign of nucleus or even of chromatin fragments, which is true of some of these cells, suggests either a further degeneration of the cells or, more probably, that they are only cytoplasmic masses pinched off from some cell and hence doomed to degeneration. It is possible that in

The Development of Scyphomedusse. 243

some cases they only represent an aggregation of yolk granules and the coagulated liquid often found in the cle.avage cavity, and are not cellular.

From the observations on Cyanea and Aurelia it is clear that at least in some cases there is no sign of an ingression of cells, or delamination, leading to entoderm formation, while in all cases invagination is the chief, if not the only process.

It was mentioned earlier, and is shown in figs. 33, 34, that the blastopore begins its closure inside and proceeds outward. The result is the complete separation of the entoderm layer before the lips of the blastopore have fused in the ectoderm. Since the gastrulation has been by invagination the primitive entoderm cells are already in a definite layer, and further changes in the two layers are only differentiations of the layers already present. The entoderm layer sometimes obliterates the cleavage cavity and lies closely against the ectoderm at a very early period even before gastrulation is completed (fig. 31) . More often, however, the cleavage cavity is not filled till much later, and even after the embryo has begun to elongate into the planula there may be some space between the two layers (fig. 35, 36). When this elongation begins the cells of the ectoderm are long and slender and the nuclei are at the extreme outer ends (fig. 35), the cytoplasm being mostly limited to a very thin layer at the outer ends of the cells. The rest of the cell is packed with yolk bodies. There are some shorter cells which are limited to the deeper part of the layer. The entoderm cells are elongate, but relatively broad and rather few in number. Continued division results in the formation of long, very narrow, ectoderm cells (fig. 36) and, since the yolk is largely used up, the cytoplasm is more uniformly distributed and the nuclei are in the centre of the cells. The entoderm is still filled with yolk. The condition shown in this figure is one often met with and is evidence of the more rapid division of the ectoderm cells, thus pulling away from the entoderm layer. In some cases the two layers develop equally fast and remain closely joined at all times. From this condition, shown in fig. 36, the completed planula (fig. 37) is easily derived. The ectoderm uses up all its yolk, the cells, now completely and uniformly filled with cytoplasm,

244 Chas. W. and G. T. Hargitt.

have assumed their definite shape and relationship and the nettling cells have formed from the more deeply lying ectoderm cells. The entoderm, continuing its differentiation, has come to lie closely against the ectoderm again. The cells do not become so long and narrow as the ectoderm and the cytoplasm is more vacuolated than in the ectoderm cells. The yolk is almost used up; the last portion remaining is limited to the entoderm cells of the anterior end of the planula. At the posterior end (lower in the figure) the ectoderm is already differentiating into what will become the point of attachment of the planula. The entire surface is covered with cilia, not represented in the figure.

The formation of the definitive ectoderm and entoderm is, then, only a differentiation in size, shape, etc., the position of the individual cells remaining relatively the same. There is never a solid mass of cells in the embryo such as one finds in the early planula of the Hydromedusae, and the coelenteron is present from the first as the remains of the old archenteron of the gastrula.

Reference to the historical section will show that the difference in the results of this study and those of other investigators upon Scyphomedusse are chiefly differences in detail. The irregularity and inequality of cleavage shown by Glaus (1883),Haeckel (1881), Metschnikoff (1886), Hyde (1894) and others to be variable is here confirmed and shown to be even more variable. In some Hydromedusae irregularity is the rule. However in these Scyphomedusse the end result is the same whether irregularity or regularity is the chief character of the cleavage, and the variations and differences are then of only very minor importance. It has also been pointed out by many workers that considerable variation occurs in the process of gastrulation, but here again the end result is the same, viz., a typical planula, so that the differences are evidently of little significance. The methods of attaining the same result, though different, may only be indications of the individual differences of the eggs or of the environment as Gonklin (1908) pointed out. The controversies over these unessential points only tend to show that attempts to bring these and other processes into exact and constant agreement with some assumed ' law ' are futile, and often only indications of ignorance or bias.

The Development of Scyphomedusse. 245

Later Development The Planula

No attempt will be made to give technical details on phases of later development. There are yet problems involved in some of these which merit careful attention, and we much regret that our material does not include that necessary to such a study. It seems, however, that some brief review of a few points may be of value.

Referring to this phase Professor Agassiz has expressed views which seem more or less open to question. For example, concerning the time and circumstances associated with spawning he says, it might be supposed that the great destruction of these animals by the autumnal gales would put an end to the development of the eggs of the stranded specimens, but this is not necessarily the case. On the contrary, I believe .... that the coincidence of their spawning with the stormy season of the year is a provision to bring them into proper conditions for their future development and growth. Thrown among the rocks, upon the seaweeds, they become entangled and break up; but: by the time they are in pieces the eggs, which have been accumulating in the little pouches formed by the folds of the margin of the arm, have reached their planula state, and are ready to swim about independent as animals

as soon as they are cast off As with the returning tide

such specimens are set afloat again, it is evident that their brood may frequently make its escape into the water and undergo their

normal development after having been for a time ashore

The young soon become attached to rocks, dead shells, or seaweeds, and assume their polyp-like condition. . . . . The succession of fine days, along our shores during the month of October following the equinoxial gales, is the season during which the planulse, set free by the decomposition of their parents, float about in search of a resting place." It is hardly necessary to point out the teleologic bias which vitiates this account. And it is only necessary to point out that while it might be plausible if only Aurelia were concerned, what shall be said of Cyanea or others whose spawning season is April instead of August?

246 Chas. W. and G. T. Hargitt.

In the earlier account by the senior author (op. cit.), it was shown that the planulse might have variable life histories. It should be noted that these observations were made under the artificial conditions of the laboratory, and hence the variations might be due to the more or less artificial conditions. The same may also be said as to the scyphistoma.

Encystment. — This is a condition often common where development is limited to the laboratory. Attention was called to this in the earlier account. Figures 38-40 show the aspect of young polyps just emerging from the cysts, which in these cases become floats, by means of which the polyps may be borne for some time. Whether such a condition ever occurs in nature we have no means of knowing, but so far as recalled it has not been made a matter of record. All the observations point to the conclusion that the phenomena associated with encystment are expressions of adaptation due to unfavorable conditions of environment; and this may serve to reconcile certain more or less conflicting accounts of earher observers, more especially those of McMurrich (1891), and Hyde (1894).

The Scyphistoma

Concerning this phase the earlier work of L. Agassiz was perhaps the best of his entire account. Many of his admirable figs, (cf. op. cit. pi. xi), of both Aurelia and Cyanea would illustrate our own results quite fully.

The account here given relates almost exclusively to Cyanea, though scyphistomai of Aurelia have been kept under observation at several times during the progress of these studies. As in the life history of the planula, so in that of the polyp there is much variability. In a small proportion of specimens there was a metamorphosis into the ephyra within a period of about twenty days after the planula attached itself. In the larger number the period was much greater than this, thirty to forty days under average aquarium conditions, while in some cases there was no transformation even at the end of two months. In the case of Aurelia the polyp life is apparently much greater, usually several

The Development of Scyphomeclusse.


months or throughout the entire winter season. The senior author has taken polyps of Aurelia undergoing strobilation and giving birth to ephyrse during the month of April in a locality where during the former summer he had kept some of the same brood of polyps under constant observation for a period of more than a month without finding any sign of metamorphosis. This

Sketch (camera) of colony in watch-glass, showing various phases from planula; to scyphistoma.

is in confirmation of the observations of Agassiz (op. cit. p. 77), who reports strobilation as occurring in the month of February and March.

The polyps feed readily upon various minute organisms, such as larvae of echinoderms, copepods, etc., and in the aquarium have been found to turn cannibal and devour planula?, instances of which we have found several times, in a few cases having wit JOUHNAL OF MORPHOLOGY — VOI.. 21, NO. 2.

248 Chas. W. and G. T. Hargitt.

nessed the entire operation. As compared with AureUa the polyps of Cyanea are relatively small; they are whitish in color, making a most beautiful sight when viewed together in a watch glass by reflected light against a black background, (cf. text fig. 1,).

Stolonization. — This is much less common than in Aurelia, though not rare. In fig. 42 is shown such a case. Stolons may arise from the body or base of the polyp as shown in the figure, and may be very delicate thread-like structures, or often somewhat massive, and these may sub-branch, thus giving rise to little colonies of polyps formed by this mode. The budding of polyps directly from the body of a parent was not observed in Cyanea, but is not uncommon in Aurelia.

Strobilation.- — This feature is rather inconspicuous in Cyanea, owing to the relatively small size of the polyp, and the strobilse are relatively few in number, often but a single one arising from a given polyp at a given time. In other cases the polyp becomes polystrobilous, from three to five ephyrse being set free in early succession. In figs. 44-46 are shown phases of strobilation as drawn from nature by Mr. H. B. Bigelow, for which kindness it is a pleasure to express thanks.

It remains to mention another feature, namely, that concerning the origin and development of the tentacles. These arise by a process of budding from the margin of the peristome, and are usually four in number, constituting the primary set. In many cases, however, only two tentacles appeared at first and on opposite sides of the mouth; later two others would arise in the appropriate intermediate positions. In a few cases the polyp seemed organized in a trimerous fashion, three primary tentacles arising about a triangular mouth, to be followed later by a second set of three tentacles at intermediate positions, rendering the specimen hexamerous. The average number of tentacles in the polyp of Cyanea is sixteen, though this number is not definite, as many were to be found having twenty or more. In a few instances bifurcated tentacles were found, a not unusual condition among coelenterates.

The Development of Scyphomedussr. 249

The Ephyra

Touching this phase occasion will be taken to mention chiefly one feature, that relating to metamorphosis from the strobila. One of the first general signs of this change is that of the apparent atrophy of the scyphistomal tentacles. Concerning this feature Agassiz (op. cit.), says at this time the wreath of tentacles which crowns these bodies is cast off, and during the fair days of that season, in the month of March or early April, the saucer-like disks of the strobila begin to separate." As a matter of fact the tentacles are not cast off" at all, but are resorbed, as is well known. Various accounts have been given as to just how this takes place. According to Bigelow (1900), it is by a process of degeneration. According to Friedemann (1902), it is by a complication of modes, namely, a strangulation of the base of the tentacle, partly through the crumpling and atrophy of the tissues, with the cooperation of the phagocyte cdls. "Die Riickbildung der Tentakeln erfolgt theils nach vorhergehender Einschntirung an der Basis und nachfolgendem Abwerfen, theils durch Schrumpfung und Atrophie des Gewebes mit Hilfe von Phagocytaren Zellen."

These accounts do not seem to be confirmed, except in a limited degree, in the case of Cyanea. There is not apparent at any time the degenerative aspects described by Bigelow. Nor has the basal constriction referred to by Friedemann been observed. That something of phagocytosis may be involved seems altogether probable, though definite evidence of the actual operation of such phagocytic cells has not been observed. In one case actually followed from beginning to end it was found that some twentyfour hours were involved in the process. As is therefore evident, it proceeds slowly, a fact further suggesting the operation of resorption, due in part perhaps to phagocytosis, and perhaps in part to the direct influence of contiguous tissues. The latter would seem to be the more important and active of the two. That there is no marked evidence of distinctively degenerative processes involved may be inferred from the fact that for some time after resorption is under way, indeed till far advanced, the tent

250 Chas. W. and G. T. Hargitt.

acles retain their irritability, and active response, contracting after stimulus qu'te as usual, though less vigorously or promptly. This is more evident as resorption approached completion. This would seem to be what might naturally be expected. It seems, therefore, that it may be said that the reduction or atrophy of the tentacles is due primarily to direct resorption unaccompanied by distinct evidences of degenerative changes. There was evidence to suggest that in the process of tentacle resorption.those of the primary series, i.e., the perradial and interradial, were first involved, followed by those of the later series, but at irregular intervals. However, there was much variation in this respect, each tentacle behaving more or less in an independent or individual manner.

Except for certain theoretical considerations we might dismiss the subject at this point. As is well known, it has long been a rather current and general assumption that in Scyphozoa the rhopalia, or so-called sense organs, are metamorphosed tentacles. So wide spread is this view that one can hardly consult any of the current text-books of zoology in which it is not asserted without qualification or doubt. As an example the following citation from Hertwig's Lehrbuch, (Kingsley's translation, p. 246), may be given as a fair illustration. "Instead of a nerve ring there are eight nerve centers connected with the sensory pedicels. Each of these pedicels is a modified tentacle with an entodermal axis and an outer layer of ectoderm."

In connection with experiments on regeneration in Scyphozoa (1904), the senior author had occasion to express serious doubts as to the validity of this assumption. Later study and research has tended to confirm the doubt, and has strongly impelled the conclusion that there is no genetic relation whatsoever between these organs. Critical study of the actual process of tentacle resorption shows it to be purely physiological, quite as much so as that of such processes generally. It would be the height of absurdity to suggest that the urostyle of the frog's skeleton might be a metamorphosed polywog tail ; but hardly more so than the one here under review !

The Development of Scyphomedusae. 251

But aside from physiological objections, abundant facts of morphology go to discredit the assumption. Not to suggest such a priori reasons as that out of the large number of scyphistomal tentacles only eight, or in some cases four, should be thus modified, it may be worth while to call attention to the more significant fact that medusae are well known whose development is direct, and consequently have no scyphistomal tentacles from which rhopalia might be developed. Nor does it suflice to say that in such cases heredity may have established the condition. But other facts are significant. In the case of polydisc strobila only the primary ephyra as a rule has had the requisite tentacle primordium for such transformation; others of the series should be found to lack sensory bodies, if the assumption be vahd. But of course this is not the case. Furthermore, as already intimated, during the experiments on regeneration referred to there was not the slightest evidence that any tentacular primordium was necessary, or in the slightest degree concerned in the process. As is well known, it is not rare in such experiments to find appearing some heteromorphic organ arising in the course, such as the occurrence of a tentacle instead of an eye (crustacea) . Not the slightest evidence of the sort was found in the case under consideration. As may be noted (op. cit.), regeneration proceeded directly from the entoderm of the marginal pockets, or rarely from other portions of the margin. It was also practicable to trace step by step the histogeny of the organs from start to finish.

This view is supported also by other observers. Bigelow (1900), pointed out that in Cassiopea the rhopalia arose prior to the atrophy of the tentacles; and Friedemann (1902), p. 264, says explicitly that the sense bodies are not transformed tentacles. These observations go to confirm the earlier view of Goette (1887), that the sense organ can not be considered as homplogous witJi the tentacle of the scyphistoma.

We seem to have here another illustration of the baleful consequences of uncritical subservience to theory. It may be doubted if in the original hypothesis any attempt of a critical character was made to work out the primary genesis of the organs in question. To the time of the above mentioned experiments the writer has

252 Chas. W. and G. T. Hargitt.

to confess that it had not occurred to him to question the vaUdity of the assumption, and in the present case submits such facts and inferences as have seemed to bear more or less directly upon the problem, and seem to warrant the strictures expressed.


Agassiz, A. North American Acalephae. III. Cat. Mus. Comp. Zo6l.no. 2, 1865

Agassiz, L. Contributions to the natural histor;^ of the United States of Amer1862 ica. vol. 4. Second monograph, part 3. Discophorse.

BiGELOW, R. P. Cassiopea xamachana. Mem. Bost. Soc. Nat. Hist., vol. 5, p. 1900 229.

Claus, C. Studien iiber Polypen und Quallen der Adria. I Acalephen. Denk1878 schr. d. Kais. Akad. Wissensch. Wien, Math-naturw. Classe, vol. 38,

pp. 1-64, PL 1-11.

1883 Untersuchungen iiber die Organisation und Entwicklung der Medusen. Prag und Leipzig, 1883. 96 pp., pi. 1-20.

1891 tJber die Entwicklung des Scyphostoma von Cotylorhiza, Aurelia und Chrysaora. Arbeit. Zool. Inst. Wien, vol. 9, pp. 85-128, pi. 1-3.

CoNKLiN, E. G. The habits and early development of Linerges mercurius. 1908 Papers Tortugas Lab. vol. 2, Carnegie Inst. Pub. no. 103, pp. 153 170, pi. 1-8.

Dublin, L. I. On the nucleoli in the somatic and germ cells of Pedicellina 1905 americana. Biol. Bull., vol. 8, pp. 347-364.

Friedemann, Otto. Post-embryonale Entwick. Aurelia. Zeits. f. wiss. Zool.,

1902 Bd. 71, p. 241, etc.

GoETTE, A. Abhandlungen zur Entwicklungsgeschichte der Thiere. IV. Entwick1887 lungsgeschichte der Aurelia aurita und Cotylorhiza tuberculata.

Hamburg und Leipzig, 1887, 79 pp., 26 figs., 9 plates.

1893 Vergleichende Entwicklungsgeschichte von Pelagia noctiluca P6r, Zeitschr.f. loiss. Zool., vol. 55, pp. 645-695, pi. 28-31.

1900 Wie man Entwicklungsgeschichte schreibt. Zool. Anz., vol. 23, pp. 559-565.

Hacker, V. tJber das Schicksal der elterlichen und grosselterlichen Kernan 1903 telle. Jena. Zeitschr., vol. 37, pp. 297-400, pi. 17-20.

Haeckel, E. Metagenesis und Hypogenesis von Aurelia aurita. Jena. 1881., 36 pp. 1881 2 pi.

The Development of Scyphomedusae. 253

Ha&iann, O. tJber die Entstehung der Keimblatter. Internal. Monatschr. f. Anat. 1890 u. Physiol, vol. 7, pp. 255-267, 295-311, pi. 12.

Hargitt. C. W. a — Notes on Cyanea arctica. Science N. S., vol. 15, p. 571.

1902 6 — -Notes on the Coelenterate Fauna of Woods Hole. Amer. Natur.,

vol. 36, pp. 548-560.

1904 a — Notes on the Hydromedusa) from the Bay of Naples. Mittheil. Zool. Stat. Neap, vel, ol. 16, pp. 553-585, pi. 21-22.

1904 b — The early development of 'Pennaria tiarella McCr. Arch.f.Ent.Mech., vol. 18, pp. 453-488, pi.

1904 £ — The Medusa? of the Woods Hole region. Bull. Bureau of Fisheries, vol. 24.

1906 The organization and early development of Clava leptostyla Ag. Biol. Bull., vol. 10, pp. 207-232, ul. 9.

Hargitt, G. T. Maturation, fertilization and segmentation of Pennaria tiarella 1909 (Ayres) and of Tubularia crocea (Ag). Bull. Mus. Camp. Zool.,

Harvard Coll., vol. 53, pp. 161-212, pi. 1-9.

Hein, W. Untersuchungen liber die Entwicklung von Aurelia aurita. Zeitschr. f. 1900 uriss., Zool., vol. 67, pp. 401-438, pi. 24-25.

1903 Untersuchungen liber die Entwicklung von Cotylorhiza tuberculata.

Zeitschr. f. iviss. Zool, ol. v73, pp. 302-320, pi. 20-21.

Hyde, IdaH. Entwicklungsgeschichte einiger Scyphomedusen. Zeitschr. f. wiss. 1894 Zool., vol. 58, pp. 531-565., pi. 32-37.

Kowalevsky, a. Development of Coelenterates (Russian) Mem. Roy. Sac. 1873 Friends of Nat. Sci., etc., Moscow, vol. 2, p. 10.

1884 Entwicklungsgeschichte der Lucernaria. Zool. Anz., Vol. 7, p. 712717.

McMuRRicH, J. P. The development of Cyanea arctica. Amer. Natur, vol. 25, 1891 pp. 287-289.

Metschnikoff, E. Embryologische Studien an Medusen. 1886, Wien, 159 pp. 1886 11 pi.

Smallwood, W. M. a reexamination of the cytology of Hydractinia and Pen1909 naria. Biol. Bull., vol. 17, pp. 209-240, pi. 1-4.

Smith, Fr.\nk. Gastrulation of Aurelia flavidula. Bull. Mus. Comp. Zool., Har1891 vard Coll., vol., 22 No. 2, pp. 115-124, PI. 1-2.

Wilson, E. B. The cell in development and inheritance. Second edition, New 1900 York.

Accepted by The Wlstar Institute of Anatomy and Biology, March28, 1910. Printed Augusts, 1910.

254 Chas. W. and G. T. Hargitt.


Drawings of figures 1-37 have been made with the aid of the Abb6 camera lucida. The magnification indicated in each case is the original magnification; the figures have been reduced to f the original size.

Figs. 1-10. Cyanea arctica, drawings of entire eggs. X 360.

Fig. 1. Two-cell stage, showing polar bodies. Cells slightly unequal in size.

Fig. 2. Two-cell stage, cells unequal.

Fig. 3. Two-cell stage, cells equal.

Fig. 4. Two-cell stage, with polar bodies; the nuclei have divided the second time and the ensuing division would have been equatorial.

Fig. 5. Three-cell stage.

Fig. 6. Four-cell stage, polar view.

Fig. 7. Eight-cell stage, showing cleavage cavity.

Fig. 8. Sixteen-cell stage, all cells outlined.

Fig. 9. Twenty-four-cell stage.

Fig. 10. Blastula near the end of cleavage, at least 200 cells present.

Figs. 11-19. Cyanea arctica, from sections. X 715.

Fig. 11. Two-cell stage, showing polar bodies; the spindle shows that the second division would have been meridional.

Fig. 12. Two-cell stage with 'resting' nucleus multi-vesicular. The beginning of the cleavage cavity is shown.

Fig. 13. Four-cell stage. Three of the cells only are shown and multi-vesicular 'resting' nuclei. Polar bodies present, the second division has been equatorial.

Fig. 14. Eight-cell stage, with cleavage cavity and multi-vesicular nuclei.

Fig. 15. A blastula of 24 cells.

Fig. 16. An older blastula of about 300-400 cells.

Figs. 17-18. Longitudinal sections of young gastrulse.

The Development of Scyphomedusse.


256 Chas. W. and G. T. Hargitt.

Fig. 19. Cross section of an older gastrula showing the cleavage cavity and ccelenteron.

Fig. 20. Cyanea arctica. Unequal double nucleus from 4-cell stage. X 1340.

Fig. 21. Cyanea arctica. Bilobed nucleus with vesicles equal, from 2-cell stage. X 1340.

Figs. 22-24. Cyanea arctica. Multi-vesicular 'resting' nuclei from cleavage stages. X 1600.

Figs. 25-29. Aurelia flavidula. Sections from ovarian eggs. X 1900.

Fig. 25. Germinal vesicle from an egg about ^ grown. Two small oocytes also shown, at beginning of growth; chromatin in a spireme.

Fig. 26. Egg approaching maturity, though not fully grown. An aster and two centrosomes present. Chromatin in masses scattered through the reticulum.

Fig. 27. End of growth, the chromatin condensing to form chromosomes.

Fig. 28. First maturation spindle forming, chromosomes not all in the spindle (some chromosomes in another section).

Fig. 29. Polar view of completed first maturation spindle which is still tangential. A splitting of chromosomes appears to have taken place. (All chromosomes present in this section).

The Development of Scyphomedusse.


258 Chcas. W. and G. T. Hargitt.

Figs. 30-37. Aurelia flavidula. Sections of developing eggs. X 715.

Fig. 30. Very early stage of gastrulation. Yolk bodies present in all cells. In the cleavage cavity is a coagulated liquid.

Fig. 31. The invagination is nearly completed. The cells of the invagination are few in number and nearly cubical; some of these cells dividing.

Fig. 32. An oblique section giving the appearance of a cell ingression. This is only apparent, for such an ingression does not occur.

Fig. 33. Invagination completed and the blastopore closing, the entoderm layer already being separated.

The Development of Scyphomedusse.


260 Chas. W. and G. T. Hargitt.

Fig. 34. Invagination completed, the entoderm completed and the blastopore closing in the ectoderm layer, 'a' a cell in the coelenteron, 'b' a cell in the cleavage cavity which is apparently migrating into the coelenteron. Drawing made from 2 sections.

Fig. 35. Cross section of a very early planula, showing coelenteron and cleavage cavity.

Fig. 36. Longitudinal section of a developing planula. The ectoderm cells have nearly their definite shape and size and in their rapid growth have pulled away from the more slowly growing entoderm.

Fig. 37. Completed planula, anterior end uppermost. The differences between the ectoderm and entoderm are well shown. The yolk is used up except for a few granules at anterior end. The black bodies are nematocysts. At posterior end the entoderm is differentiating to form a point of attachment for the planula.

The Development of Scyphomedusae.


262 Chas. W. and G. T. Hargitt

Fig. 38. Figures a and h show encysted condition, and polyp just emerging from cyst, which serves as a float.

Figs. 39-40. Floating polyps at slightly varying stages of growth, and still floating on portion of cyst.

Fig. 41. Young polyp just emerged from cyst.

Fig. 42. Stolonization, a young budded polyp at left.

Fig. 43. Scyphistoma about at full development.

Figs. 44-46. Phases of strobilation, in the last the disks quite deeply divided.

Figs. 47-48. Late phases of inetamorphism, in the last the young ephyra ready to be liberated.






b 38








1 // /////

■///// //


17' . d

42 ^^

I. \






\ 46









Journal of MoRPHOLOGr.— Vol. 21, No. 2.



Director of the Pathological Laboratory of the Massachusetts General Hospital Assistant Professor of Pathology, Harvard Medical School

With Two Colored Double Plates

The theories that have been proposed regarding the nature and origin of the blood platelets, before the publication of my papers on this subject in 1906, may be grouped and briefly discussed under the following headings.

1. The blood platelets are fragments of the leukocytes.

The principal objection to this view is that the blood platelets when observed and stained in various well-known ways present such a characteristic structure and appearances as to make such an origin very improbable if not inconceivable.

2. The blood platelets are derived from the extruded nuclei of the red blood corpuscles.

The principal objection to this view is that the extruded and degenerated nuclei of the red blood corpuscles never show any similarity in appearance to, nor any transitions toward, blood platelets in preparations in which both of these elements are present and characteristically stained. I have been able to follow all stages in the fate of these extruded nuclei up to their final dissolving in the plasma in preparations of embryonic blood and of the blood of animals which had been intravenously injected with the hemolytic toxins, ricin and saponin. In these preparations in which the extruded nuclei of the red blood corpuscles are abundant I have in no instance seen a suggestion of the formation of a characteristic blood platelet from the extruded nucleus.


264 James Homer Wright.

3. The blood platelets are derived from certain parts or constituents of the red blood cell other than the extruded nucleus.

Various theories are grouped under this heading. The fundamental objection to all of them is that the most approved methods of preparation fail to give satisfactory evidence that the red blood cell at any period in its development contains a body which suggests a blood platelet and do not show any transitions between red blood corpuscles or their fragments and the blood platelets. Occasionally in smear preparations a blood platelet may be seen lying upon a red cell and this appearance has been interpreted as showing that the blood platelet is first contained within a red blood corpuscle, then later is extruded. In sections, in which the blood platelets are characteristically stained, I have never seen a blood platelet within a red blood corpuscle. The view that some, if not all of the blood platelets, are pinched-ofT portions or fragments of the red cells which have undergone certain changes whereby they present the characteristic staining and structure, which are brought out by the modern methods of preparation, is a pure assumption of fact to explain the obvious differences between fragments of red blood corpuscles and blood platelets. Fragments of red blood corpuscles, or so-called blood platelets containing hemoglobin, practically do not occur in the normal blood.

4. The blood platelets are a definite and independent kind of blood cell.

This is open to the objection that the blood platelet has no nucleus. The central granular portion brought out by proper staining methods cannot be regarded as a nucleus as has been maintained because it lacks the definite structure of a nucleus and the characteristic affinity for nuclear dyes.

5. The blood platelets are not independent cells or cell fragments but are of the nature of albuminous precipitates.

Against this view may be urged the characteristic appearances and structure of the blood platelets as brought out by various methods of preparation and the fact, confirmed by personal observation, that the blood platelets under certain conditions show continuous and amoeba-like change in form and outline.

• Histogenesis of the Blood Platelets. 265

6. The blood 'platelets are cells out of which the red blood corpuscles develop.

That they have nothing to do with the development of red blood corpuscles is shown by the absence of transitions between them and young red blood corpuscles in preparations from bone marrow appropriately stained, and also by well established facts in the development of the red blood corpuscles.

In brief, after much study of the subject and for the reasons given, as well as for many others, I am compelled to believe that all these theories of the origin and nature of the blood platelets are erroneous and untenable.

As one of the results of an extensive study of the blood and blood forming organs of various animals I have become convinced that the blood platelets are detached portions of the cytoplasm of those giant cells of the blood forming organs which have been named megakaryocytes by W. H. Howell to distinguish them from the multinucleated giant cells of the bone marrow, the so-called osteoclasts or polykaryocytes.

This view of the origin and nature of the blood platelets is based upon the following : —

Facts and Observations

By means of a special staining method, devised by me, which gives the so-called Romanowsky polychrome staining, I have been enabled to stain characteristically the blood platelets in sections of fixed tissues and organs so that they may be positively recognized and may be clearly distinguished from other histological elements. "*

The description of this method of staining will be found at the end of this paper.

I have studied especially sections of the bone marrow and spleen of the cat and young kitten. Similar material from man, mouse, dog, rabbit, guinea pig, white rat, and opossum has also been studied. The appearances observed in the preparations from the blood forming organs of man and of these animals indicate that the blood platelets originate from the megakaryocytes in all of them.

266 James Homer Wright.

In sections of the blood forming organs, and of blood vessels containing blood, stained by this special method, the blood platelets present the following appearances and characteristics. They consist of a hyaline blue staining substance in which are imbedded closely set, minute, red to purple staining granules. In deeply stained preparations this hyaline ground substance may not be apparent. Their well known disc shape is sometimes preserved so that they may appear as short rod-like bodies when viewed in certain positions. Their margins may be smooth or show irregular, small projections, of the hyaline ground substance. Elongated forms, sometimes several times as long as broad, occasionally occur. These have been described by F. Weidenreich in smear preparations. The red to purple staining granules may be aggregated in a more or less sharply outlined mass in the central part of the platelet so as to suggest a nucleus surrounded by a hyaline cytoplasm. In some platelets a clear, unstained, more or less sharply outlined vacuole-like area may be seen in the midst of the granules. This has also been noted in smear preparations by G. Schmauch and by F. Weidenreich.

The structure of the blood platelets, and especially the peculiar color taken l?y the granules within them, are very characteristic and sharply distinguish them from the other elements of the blood. The red corpuscles and granular precipitates caused by certain fixing agents in the plasma stain pink or green depending on the fixative used and the intensity of the staining. The nuclei of the leukocytes stain blue and the various kinds of granules in the cytoplasm of the leukocytes, according to the species of the animal, may or may not be characteristically stained as in smear preparations stained by modern blood staining methods.

Blood platelets, with the staining and other appearances above described, are found in sections of vessels doubly ligated during life in numbers corresponding to the numbers of the blood platelets in fresh blood. Moreover, in sections of material properly fixed by formaldehyde, in which no granular precipitate is produced, all bodies which are of the shape and size of blood platelets stain characteristically. For these reasons it is believed that all of

Histogenesis of the Blood Platelets. 267

the blood platelets are stained by this method and that the blood platelets are all of the same nature or kind.

Fixation in methyl alcohol or in corrosive sublimate produces in the plasma a more or less abundant precipitate in the form of granules of a size and shape closely resembling the blood platelets and which cannot be distinguished from blood platelets in preparations stained by the usual methods. This granular precipitate has doubtless been confounded with blood platelets by observers who have found an apparent increase of blood platelets in doublyligated vessels and in stagnant blood, or apparent blood platelet thrombus formation after cauterization of the vessel wall, because they did not use a specific staining method for the blood platelets in their experiments.

In sections of the blood forming organs the blood platelets may be very small in number because the blood has flowed out of them, taking with it the blood platelets, and because, with the stopping of the circulation, the platelets become irregularly distributed throughout the vessels. They are very numerous in the spleen, which is readily accounted for by the consideration that the structure of that organ favors the accumulation of various kinds of cells and blood corpuscles within it and prevents their escape from small pieces cut from it.

The giant cells or megakaryocytes in sections of the bloodforming organs present the following peculiarities which are of importance for the subject of this paper.

In the cytoplasm of the megakaryocytes or the giant cells are imbedded more or less numerous red to purple staining granules identical in appearance and staining with the granules in the blood platelets. Also in the cytoplasm small vacuole-like unstained areas may be seen like those in the blood platelets. In preparation fixed by methyl alcohol the granules may be arranged in more or less definite parallel rows or Hues coursing in various directions.

The cytoplasm of a minority of the megakaryocytes is prolonged into pseudopod-like processes of varying size, shape and number. These processes, which will hereafter be referred to as pseudopods, often occur as bud-like projections of the size of a blood platelet or as strands, of a width corresponding to that of a platelet, which

268 " James Homer Wright.

may attain a length greater than that of the diameter of an oil immersion field. Nearly all of the cytoplasm of some cells may be formed into pseudopods projecting in various directions.

The pseudopods commonly are seen to project into more or less well defined blood channels. This is especially clearly shown in the spleen of the young kitten where larger and smaller pseudopods are frequently seen projecting into veins through small openings in the vessel wall. A whole megakaryocyte in active pseudopod formation has been found in a blood vessel in the spleen.

The amount of cytoplasm associated with a megakaryocyte nucleus varies and such nuclei showing the usual signs of degeneration and senility with little or no cytoplasm about them are frequent. This is due to the well known fact that the megakaryocytes lose their cytoplasm. Thus pseudopods not connected with megakaryocytes occur not rarely.

The characteristic granules are most numerous and most closely crowded together in the cytoplasm of the larger megakaryocytes, of those with pseudopods, and of those in process of losing their cytoplasm and in detached pseudopods. The marginal or peripheral portion of the cell body is usually free from granules, is hyaline, stained blue and is sharply demarcated from the granulecontaining cytoplasm. The border of the cell may be smooth or show rounded or irregular projections of the hyaline cytoplasm. In the pseudopods the granules may be abundant throughout the cytoplasm or they may occupy only the axial or mesial portions leaving a sharply demarcated narrow marginal zone of hyaline blue staining cytoplasm, the borders of which may be smooth or show small projections of varying shape, just as does the hyaline blue stained marginal portion of the blood platelets. Thus some of the smaller pseudopods are identical in appearance with the elongated forms of the blood platelets in everything except that they are in continuity with the cytoplasm of the megakaryocyte.

In some megakaryocytes or pseudopods one or more small groups of the granules may be seen more or less definitely separated by a narrow zone of hyaline cytoplasm from the rest and arranged in one or more round or oval, more or less sharply outlined masses which are identical in appearance and staining reaction with the

Histogenesis of the Blood Platelets. 269

masses of granules in the blood platelets and may also contain vacuole-like unstained areas like those in the blood platelets. Such a small mass of granules is commonly found in a small budlike pseudopod and the appearance is thus produced of a platelet, the hyaline ground substance of which is continuous with the hyaline cytoplasm of the megakaryocyte or a larger pseudopod. The granular material in some slender pseudopods may be more or less completely segmented into these rounded masses and such a pseudopod may present the appearance of being composed of a chain of blood platelets united by the continuity of their hyaline ground substance, which in turn, is continuous with the hyaline cytoplasm of the megakaryocyte. The same appearance may be shown by undoubted pseudopods not connected with megakaryocytes and lying in blood channels, and also in the cytoplasm of cells in process of losing their cytoplasm.

The identity in appearance of the small portions of cytoplasm containing the separate groups of granules, and of the small pseudopods, with the various forms of blood platelets in every respect except that they are in continuity with the megakaryocyte, and the occurrence of such separate groups of granules in giant cells in process of losing their cytoplasm, have led me to the conclusion that such small portions of giant cell cytoplasm, and also the small pseudopods, by separation from the cell, become the blood platelets.

Bodies identical in appearance with blood platelets are commonly found near pseudopods, but only rarely are aggregations of free blood platelet-like bodies found associated with a cell in active pseudopod formation. The infrequency of the occurrence of such aggregations of platelets is easily explained by the consideration that by reason of the projection of the pseudopods into blood channels, and because of the close relations of the giant cells to the blood stream in the marrow, the platelets are usually swept into the circulation as soon as separated from the megakaryocytes or their pseudopods. That the megakaryocytes and their fragments have ready access to the blood stream is shown by the occurrence of detached pseudopods in the blood channels and by the well-known fact, first clearly pointed out by L. Aschoff, of

270 James Homer Wright.

the lodgement of them and of their more or less naked nuclei in the blood vessels of the lung.

The appearances observed are summarized and interpreted as follows :

All of the blood platelets are detached portions or fragments of the cytoplasm of the megakaryocytes, which are in such relation to the blood channels in the marrow that detached portions of their cytoplasm are quickly carried by the blood current into the circulation. The breaking up of the cytoplasm into the platelets occurs only in cells which have reached a certain stage of growth and development, and is probably rapidly completed when once begun. It takes place in various ways but usually by the pinching off of small rounded projections or pseudopods from the cell body or from larger pseudopods, or by the segmentation of slender pseudopods, or by the pinching off of longer or shorter pseudopods which may or may not undergo segmentation later. All or most of the cytoplasm of the giant cell is given off to the blood stream and the nucleus degenerates. The more or less naked nucleus is often carried by the blood stream to the lungs where it lodges in the capillaries. Before the separation of a platelet takes place the red to purple staining granules in that portion of the cytoplasm which is to form the platelet are separated from the rest by a zone of hyaline cytoplasm and arranged in a more or less sharply outlined, rounded or oval mass. The line of cleavage is through this zone of hyaline cytoplasm and this sharply outlined mass of granules becomes the central granular mass of the blood platelet which has been regarded by some observers as a nucleus.

These observations and conclusions concerning the origin and nature of the blood platelets have been confirmed by C. H. Bunting for the rabbit. On the other hand, H. Schridde, who also devised a method of staining the granules in the cytoplasm of the megakaryocytes, could not confirm them for the blood platelets of man, because the blood platelets in his preparations did not show the characteristically staining granules. This must be due to a defect in his staining method, for my method clearly

Histogenesis of the Blood Platelets. 271

brings out the same appearances in the megakaryocytes and blood platelets of man as in those of animals.

This conception of the histogenesis of the blood platelets derives additional support from the following considerations:

1. The observation by me, with the aid of Deetjen's method, of protoplasmic movements of identical character both in the hyaline marginal zone of the megakaryocytes and in the hyaline marginal zone of the blood platelets. These movements have been described by H. Deetjen and others for the blood platelets. I have seen the hyaline marginal zone of the megakaryocytes and of the blood platelets constantly changing its outline, sending out and withdrawing short processes of various shapes. This so-called amoeboid movement of the blood platelets is not surprising, because it is known that detached fragments of living protoplasm may exhibit movement.

In this connection I may state that I have seen a few megakaryocytes change their form very markedly, sending out and withdrawing pseudopods, such as are seen in the sections. This seems to show that the presence of pseudopods and protoplasmic prolongations of megakaryocytes in blood vessels, as I have seen in the sections, is not a passive act, due to local conditions of pressure in the tissue, but is a manifestiaton of vital activity. Amoeboid activity on the part of the megakaryocytes was suspected by J. Arnold, and has been affirmed by M. Askanazy.

2. A comparison of the numbers of blood platelets per cubic millimeter of blood in certain diseases, as estimated by various observers, with the histological findings in the bone marrow in the same diseases, suggests a relationship between the blood platelets and the megakaryocytes. Thus in pernicious anemia and lymphatic leukemia the blood has been found to contain abnormally few platelets, while the marrow in typical cases of these diseases as far as can be inferred from the reports in medical literature and from my own observations, undergoes profound changes in the character of its cellular constituents with marked diminution in the number of the megakaryocytes. On the other hand, in post-hemorrhagic and secondary anemia the blood platelets are increased in number and there is also increase in the

272 James Homer Wright.

amount of red marrow with consequen^t increase in the total number of megakaryocytes in the body. In so-called myelogenous leukemia the blood platelets are also increased in number, and in the cellular accumulations of this disease megakaryocytes do not seem to be an uncommon finding, although but little attention has been paid to them by pathologists. In view of the enormous increase of the marrow cells in this disease it must be obvious that the presence among them of a relatively small proportion of megakaryocytes means a great absolute increase in the number of such cells in the body.

Furthermore, C. H. Bunting has recently shown experimentally on the rabbit that synchronous with or preceding the appearance of an increased number of platelets in the blood stream the megakaryocytes are increased in number.

3. It would seem that the blood platelets do not appear in the embryo before the appearance of the forerunners of the megakaryocytes. Thus in an embryo guinea pig of about 4.5 mm. in length, I have found, free in the blood vessels, cells of about the size of the nucleated red blood corpuscles which have the characteristic staining of the megakaryocyte and differ from it only in being much smaller in size. A study of other embryos shows all grades of transition between this circulatory cell and the typical megakaryocyte. These small megakaryocytes may be seen in the blood vessels in the sections breaking up into typical blood platelets just as do the fully developed cells. (Fig. 17.) On the other hand in a smaller embryo guinea pig I have not found either the small megakaryocytes or blood platelets in the blood.

It is of interest to note in this connection that some at least of these forerunners of the megakaryocytes seem to be formed by a transformation of endothelial cells of blood vessels, because I have seen one of them apparently forming a part of the endothelium of a blood vessel in the yolk sac of a guinea pig embryo. This cell is pictured in Fig. 16.

4. According to my own and others' observations, bodies that are undoubtedly and obviously blood platelets are found only in the blood of mammals, and mammals are the only creatures that have megakaryocytes in the blood-forming organs. I have found

Histogenesis of the Blood Platelets. 273

undoubted, characteristically staining blood platelets in the blood of all of a considerable variety of mammals including the elephant, kangaroo, opposum and camel, and I have found megakaryocytes in the blood-forming organs of all mammals including the opossum, which I have examined under satisfactory conditions. The so-called spindle cells or thrombocytes of birds, amphibia, reptiles and fishes have been held by some writers to be the morphological equivalents of blood platelets, but my studies of the blood and blood-forming organs of these vertebrates have not led me to accept this view. I would offer the hypothesis that these peculiar corpuscles are rather the homologues of the megakaryocytes than of the detached fragments of their cytoplasm or the blood platelets, and that these two kinds of cell have been differentiated from one and the same type of cell which circulated in the blood of extinct vertebrates.

This hypothesis is based upon the following considerations: FiEST. As I have already pointed out, the forerunners of the megakaryocytes are at first circulatory cells in the blood of the embryo guinea pig. This fact points to the megakaryocyte as representing a circulatory cell in the ancestry of the mammals.

Second. The spindle cells or thrombocytes of certain amphibian blood have a cytoplasm which stains in the same way as does that of the megakaryocyte; namely, showing a granular red to violet staining endoplasm and a hyaline blue staining ectoplasm, (see Figs. 18 to 21) . Furthermore these cells in Batrachoceps attenuatus regularly lose their cytoplasm by a pinching off process and the portions thus detached appear as independent corpuscular elements of the blood with great likeness to blood platelets, for they have the same form and outline as blood platelets and the same red to violet staining central portion with vacuole-like spaces in it and the same hyaline blue marginal portion with the irregular or jagged edge. Figs. 18 to 21 show some of the various forms in which these cells appear and some of the platelet-like detached portions of their cytoplasm as well as some of the phases of the process of detachment. The likeness of these detached portions of the cytoplasm of these cells to blood platelets was first pointed out by G. Eisen.

274 James Homer Wright.

The Method

The material should be obtained immediately after death or taken from the living animal.

For fixation methyl alcohol, formaldehyde, or a saturated solution of mercuric chloride in a 0.9 per cent solution of sodium chloride, may be used. Methyl alcohol is not now recommended for fixation. Formaldehyde should not be allowed to act longer than forty-eight hours. The method is not applicable to material fixed in Zenker's fluid.

The tissue is dehydrated by alcohol followed by acetone, cleared in thick oil of cedar followed by xylol, and imbedded in paraffin.

The sections should not be more than 4j« in thickness.

Crystals of corrosive sublimate in the sections are to be removed by treatment with Gram's solution of iodine and alcohol.

The sections are stained while affixed to the slide by Meyer's glycerine-albumin mixture.

The staining fluid and the mode of its preparation are described below.

The staining, clearing and mounting is carried out as follows :

1. Equal parts of the staining fluid and distilled water are mixed in a small wine glass and immediately poured on to the slide. The measuring is conveniently done by means of a small pipette provided with a rubber bulb. At least 2 cc. of the freshly diluted staining fluid are thus spread out over the slide, which should be supported upon some object in such a way as to prevent the fluid from running off. The spreading out of the fluid in a layer is important because it facilitates the evaporation of the alcohol whereby the staining elements slowly precipitate out of solution and, while doing so, stain the tissue elements. This precipitate appears as a yellowish metallic scum which slowly forms on the surface of the mixture. The diluted staining fluid is allowed to act for about ten minutes when the preparation is immediately washed in water. The exact time required for the best results has to be determined for each batch of the staining fluid. The proper staining of the preparation may be judged by examining it by a

Histogenesis of the Blood Platelets. 275

yellowish artificial light under a low magnifying power after pouring back the diluted staining fluid into the wine glass. The stain is to be regarded as sufficiently intense and the staining process stopped by washing the preparation in water when the cytoplasm of the giant cells has acquired a bright red color and the fibrils of the reticulum begin to take on a red color also. If the staining is found not sufficiently intense the diluted staining fluid is poured back on the preparation and allowed to act longer. Over-staining and the formation of a black red granular precipitate on the preparation occur if the diluted staining fluid is allowed to act longer than a certain time.

2. Dehydrate in pure acetone.

On account of the great volatility of acetone some care is necessary to prevent the drying of the preparation, which should be avoided.

3. Clear in pure oil of turpentine.

4. Mount in a thick solution of colophonium in pure oil of turpentine.

Before mounting the preparation the superfluous turpentine should be carefully removed because this reagent rapidly takes up water from the air and thus may cause the clouding of the preparation or the fading of the stain.

The solution of colophonium is made by saturating a quantity of turpentine with powdered colophonium and keeping the filtered solution in the paraffin embedding oven until it has evaporated to the required consistence.

The use of acetone for dehydrating and of oil of turpentine for clearing and mounting is an important feature of the method, for these do not destroy the characteristic staining of the granules in the giant cells and platelets as do other similar reagents that I have tested.

The staining fluid is composed of a mixture of 3 parts of a solution of modified or polychromatized methylene blue and 10 parts of a 0.2 solution of eosin, w.g." (Gruebler) in pure methyl alcohol. It is permanent if kept in a well-stoppered bottle so that evaporation is prevented.

276 James Homer Wright.

The solution of methylene blue is prepared as follows: One gram of methylene blue, B. X. (Gruebler) is dissolved as thoroughy as possible in 100 cc. of a 0.5 per cent aqueous solution of sodium bi-carbonate in an Ehrlenmeyer flask. The flask and its contents are then placed in an ordinary steam sterilizer and kept at 100°C. for one hour and a half, counting the time after the steaming has become vigorous. When cool, the mixture is filtered and the filtrate is the modified blue solution. It must be of a well-marked purple color when viewed in a thin layer by the yellow transmitted light of an ordinary incandescent electric bulb. This color appears only after cooling.

It is important that the quantities mentioned should be accurately weighed or measured. An excess of eosin delays the appearance of the scum on the surface of the diluted staining fluid and the time required for staining will be longer than ten minutes. On the other hand, an excess of the modified blue component hastens the appearance of the scum and the staining may in ten minutes cause over staining and the granular precipitate to form on the preparation.

The preparations should be viewed by the light from an incandescent electric bulb which has a yellowish tint. This brings out more strongly the characteristic color of the granules in the megakaryocytes and in the blood platelets.

My thanks are due to Prof. S. H. Gage of Cornell University and to Prof. C. S. Minot, Dr. F. T. Lewis and Dr. J. L. Bremer of the Harvard Medical School for material for study. I am especially indebted to Dr. J. W. Dewis of Boston for a collection of preparations of the blood of various animals, the study of which first awakened my interest in the subject of the histogenesis of the blood platelets and was the starting point of the work upon which this paper is based. To Dr. Oscar Richardson, Assistant Pathologist, I am under many obligations for relieving me of much of the routine work of the Laboratory during the progress of this work.

Accepted by The Wistar Instltuteof Anatomy and Biology, April 28, 1910. Printed August 3, 1910.

Histogenesis of the Blood Platelets. 277


Arnold, J. Virchow's Archiv., Bd. 144, S. 411.

AscHOFF, L. Virchow's Archiv., Bd. 134, S. 11.

AsKANAZY, M. Miinch. Med. Wochenschr., 51 Jahrg. S. 1945. 1904

Bunting, C. H. Jour, of Exper. Med., vol. 11, p. 541. 1909.

Deetjen, H. Virchow's Archiv., Bd. 164, S. 239.

EisEN, G. Proc. California Acad. Sci., ser. 3, Zool. vol. 1.


Helber, E. Deutsch. Arch. f. Klin. Med., Bd. 81, S. 316.

Howell, W. H. Jour, of Morph., vol. 4, p. 177. 1891

Pratt, J. H. Johns Hopkins' Hos. Bull., vol. 16, p. 201. 1905

ScHMAucH, G. Virchow's Archiv., Bd. 156, S. 201. ScHRiDDE, H. Anat. Heft., 1 Abt. 99 Heft. (33 Bd., H. I.) Weidenreich, F. Verhandl. d. Anat. Gesellsch., Rostock 1.M.1.-5, Juni, S. 152. 1906

Wright, J. H. Boston Med. and Surg. Jour., vol. 154, p. 643. 1906 Virchow's Archiv. Bd. 186, S. 55.


The water color drawings were made by the author with the aid of a camera lucida. They are all drawn at the same magnification, except fig. 6, which ia drawn with a 3mm. Zeiss apochromatic oil-immersion objective, while the others are drawn with a 2 mm. objective of the same kind. The eye-piece used in all cases was compensating ocular 6. With the exception of figs. 18 to 21 all the figures were drawn from sections either of the bone marrow or spleen of the cat or kitten.

Fig. 1. Megakaryocyte in position of amoeboid activity with its protoplasm projecting far into the lumen of a blood vessel in the form of pseudpods. Blood platelets free and in process of segmentation from the pseudopods. Vacuoles in cytoplasm and in platelets.

Figs. 2-4. Megakaryocytes which have lost nearly all of their cytoplasm and present more or less degenerate nuclei. Platelets are shown free and in process of budding off from the cytoplasm. Vacuoles in cytoplasm and platelets.

Fig. 5. Megakaryocyte showing excessive development of pseudopods wdth blood platelets developing from them.

Fig. 6. Megakaryocyte with its cytoplasm prolonged into a blood channel in the form of a very long pseudopod. The continuity of the different portions of

278 James Homer Wright.

the pseudopod was shown by serial sections. At the right lower corner of the drawing are shown three blood platelets and a lymphocyte. The other small cells in the vessel are leukocytes.

Fig. 7. Megakaryocyte in the act of protruding a pseudopod through the wall of a blood vessel into its lumen. At the extremity of the pseudopod three platelets in process of development. Further down in the blood vessel four free platelets. Vacuoles in cytoplasm and in some platelets.

Fig. 8. Detached pseudopods projecting into a blood vessel and in process of segmentation into platelets. Free platelets also shown. Vacuoles in cytoplasm and in platelets.

Fig. 9. A mass of cytoplasm of a megakaryocyte with platelets budding off from it. Free platelets and different forms of white blood corpuscle also shown.

Fig. 10. Detached pseudopods and blood platelets. One of the pseudopods segmenting into blood platelets.

Fig. 11. Same as fig. 10, except that one of the larger granular red masses is in continuity with a giant cell not obvious in the section.

Fig. 12. Megakaryocyte protruding a pseudopod into a blood vessel through an opening in its wall, At the tip of the pseudopod two or three platelets in process of development and two or three developed platelets. The arrangement of the granules in rows is shown.

Fig. 13. Megakaryocyte showing a platelet in process of pinching off from a pseudopod protruding through the wall of a small blood channel.

Fig. 14. Megakaryocyte extending two pseudopods into a blood vessel through openings in its wall. A blood platelet pinching off from one of them. Two blood platelets free. Vacuoles in the cytoplasm of the giant cell are shown.

Fig. 15. Megakaryocyte with a blood platelet in process of budding off into a small blood channel. Two other blood platelets in the lower part of the figure.

Fig. 16. One of the forerunners of the megakaryocytes observed in the blood of a young guinea pig embryo. This cell is apparently in process of development out of an endothelial cell of a blood vessel. It still forms part of the endothelium of the wall of the blood vessel and it clearly is a transformed endothelial cell.

Fig. 17. One of the forerunners of the megakaryocytes in the blood of an early guinea pig embryo in process of breaking up to form blood platelets. Free blood platelets also are shown.

Figs. 18-21. Spindle cells or thrombocytes of Batrachuceps aUenuatus and free blood platelet-like corpuscles. Smear preparations stained by Wright's blood stain. Various phases are shown in the process of pinching off portions of the cytoplasm of the thrombocytes to form blood platelet-like corpuscles. The thrombocyte in fig. 18 has lost nearly all its cytoplasm by the process of pinching off of platelet-like fragments. The yellow body in one of the cells in fig. 21 is a fragment of a red blood corpuscle that has been taken into itself by the cell. Note the vacuoles in the platelet-like body in fig. 19 and in the pseudopods in fig. 21.













• A •




The Journal of Morphology.— Vol. 21 , No. 2.


The jorRXAL of MoRPHOLor.v— Vo1.21,No.2.


« <






N. M. STEVENS Bryn Mawr College, Bryn Mawr, Pa.

With One Hundred and Two Figures


In March, 1905, it was my privilege to avail myself of the opportunity offered by the newly opened winter quarters of the Marine Biological Laboratory at Woods Hole, Mass., to make a further study of the method of egg-laying in Sagitta elegans. A recent paper ('05) had described, as fully as was possible from fixed material alone, the later stages in the ripening of the ovum and its entrance into the oviduct which is temporarily opened up between the sperm-duct and the median oviduct wall. Questions arose as to the relation of the two ducts at the point of opening to the exterior, and also as to the activity or passivity of the ovum in its passage from the ovary to the reproductive pore.

Hertwig ('80) in Die Chsetognathen," (pp. 52-54 and PL 4, Fig. 13), describes and figures the sperm-duct as the oviduct, and the oviduct wall as the 'Keimlager.' He observed hving spermatozoa in the 'oviduct' but never in the ovary. Finding no opening from the ovary to the duct, he concluded that the eggs when ripe must enter the duct at its posterior end, and that the anterior portion of the observed duct must serve merely as a 'Samentasche.' This duct had been previously described by Krohn ('53) and by Leuckhart and Pagenstecher ('58) as a 'Samentasche' Kerferstein ('62) also saw spermatozoa in this canal, but nevertheless regarded the whole of it as an oviduct, and was of the opinion that there must be an anterior opening from the ovary into the oviduct. Grassi ('83) called the duct a sperm JOURNAL OF MORPHOLOGT — VOL. 21, NO. 2

280 N. M. Stevens.

oviduct, but did not discover by what method the eggs escaped from the ovary.

In two brief papers ('03, '05) I have shown that the duct described by previous investigators of Sagitta, is a sperm-duct (Samentasche), or sperm-receptacle, lying within the oviduct which is closed except when ripe eggs are being discharged. ('03, Figs. 1-3; '05, Figs. 1, 11, 12). The ' Keimlager ' of Hertwig, which forms the wall of the oviduct, I had suspected was not germinal epithelium at all and therefore not a part of the ovary proper, or only an accessory structure to be classed with the spermduct and the endothelial membrane covering the ovary, and not with the germ cells. This was one of the points which led to further investigation of the reproductive system of Sagitta.

Observations on Living Material

Through the kindness of Mr. George M. Gray, curator, I was able to sample the Sagitta material at intervals during March, 1905, and thus to time my arrival at Woods Hole when the laying season was just beginning.

The tim.e of day when Sagitta elegans discharges its eggs is not as definite as in the case of Sagitta bipunctata. The latter lays about sundown, while the former has been observed to discharge its eggs at various times between 11:00 a. m. and 6:00 p. m., and there is no reason for thinking that the eggs may not be laid at any other tim.e in the twenty-four hours.

In Sagitta collections brought in at about 9 :00 a. m. all stages of egg-ripening were found in different individuals. The nuclear membrane disappears from 15 to 30 minutes before the egg begins to push its way into the oviduct. That the egg does actively push its way between the oviduct wall and the sperm-duct, and by its own contractions or by shifting of material within the egg-membrane, make its way down the oviduct to the reproductive opening, I have no doubt, after observing the process in many individuals.

Fig. A is a freehand sketch' from a living Sagitta under a low power of the compound microscope, showing five eggs entering

Further Studies on Reproduction in Sagitta.



Fig. A. Freehand sketch of eggs entering the oviduct. " B. Egg entering anterior end of a closed oviduct. " C. a, h, c. Successive form changes of an egg moving down the oviduct.

282 N. M. Stevens.

one oviduct, as observed at 10:00 a. m. At 10:10 the egg a was wholly within the oviduct, at 10:15 c was in, and 11:00 all five were in, and a had moved down to x. Two eggs in the other ovary were still partly outside of the oviduct. In every case observed each egg made its way through its own opening into the oviduct, forcing apart the sperm-duct and the anterior wall of the oviduct. Fig. B is a sketch showing a ripe egg at the anterior end of an ovary, observed at 11:30 a. m., forcing an entrance into an otherwise wholly closed oviduct. The spermduct (s), containing live spermatozoa lies above the ripe ovum (o) and the younger oocytes.

In some cases a number of eggs — as many as nine — were crowded into one oviduct, occupying its whole length; in others a single egg was seen to pass down the whole length of the oviduct alone. In the latter case the egg was often drawn out in sausage shape so that it extended half the length of the ovary or more. One especially favorable egg was watched from 4:00 p. m. until it was extruded at 5:25 p. m. When first observed, it appeared as in Fig. Qa. Sketches were made at intervals of about 45 seconds, and though there was constant change in the form of the egg, the changes showed a regular rhythm, running from a through h and various intermediate forms to c, in from 2\ to 3 minutes, and then repeating the series of form changes. With each a-phase the egg made a slight advance down the oviduct. There was no visible muscular movement of the body corresponding to the various form changes observed in the egg. In Fig. C a an air (?) space is shown in the oviduct at each end of the egg. Such a space is always observable in the case of a single egg in the oviduct, in both living specimens and in sections. The first muscular movement observed occurred when the advance air-bubble reached the reproductive pore. The pore suddenly opened wide, the airbubble, covered by a thin film, passed out and broke, and the egg followed, the opening closing behind it (Fig. D a and 6). Both opening and closing had the appearance of a reflex. When several eggs were passing, crowded together, down the oviduct, the pore was only partly closed and then opened wide for another egg. Several times the sphincter closed too soon and cut an egg in two.

Further Studios on Reproduction in Sagitta. 283

Fig. D. a, b. Opening of tlic exterior .sphincter of the oviduct, and escape of the egg.

Fig. E. Spermatozoa escaf)ing from the; si)erm-duc,t uinh'r pressure on the cover-glass.

Fig. F. Longitudinal section of_ovary, oviduct, and sj>erni-duct.


N. M. Stevens.

This was very likely due to the abnormally slow passage of the egg under laboratory conditions.

Both living material and sections show that the sperm-duct and oviduct are entirely separate for their whole length, each having its own opening to the exterior. By exerting pressure on the cover-glass, a stream of sperm was forced out of the mouth of the sperm-duct when the oviduct, including its sphincter-like pore, was entirely closed. This is shown in Fig. E, an optical

Fig, G. Longitudinal section of an egg in the oviduct.

section of the sperm-duct with its external opening, and the walls of the oviduct adhering to the sperm-duct as they normally appear except when eggs are being discharged. In Fig. F, a camera drawing from a longitudinal section of Sagitta elegans, the same conditions are shown. The sperm-duct (sd) is here much enlarged but empty, and the oviduct (od) wall shows a characteristic fold which opens, in part at least, when an egg passes. In Fig. G, the sperm-duct contains spermatozoa and the oviduct is being opened up by an advancing egg (o), which pushes apart the spermduct and the anterior wall of the oviduct. An air-space (a) appears here as in the living material.

Further Studies on Reproduction in Sagitta. 285

These observations make plain the relation of the two ducts to one another and the activity of the egg in reaching the exterior when ripe. Study of a very complete series of sections of Sagitta from the beginning of post-larval life to maturity seemed to be necessary to settle satisfactorily the origin and development of these ducts. I had no opportunity to secure such material until the summer of 1909, when, through the courtesy of the directors of the marine laboratories at Pt. Erin, Isle of Man, and at Helgoland, I obtained an abundance of Sagitta bipunctata in various stages. I am especially indebted to Mr. Chadwick, curator at Pt. Erin station, for assistance in collecting and fixing material, as well as for many other courtesies. This material was studied while I was enjoying the privilege of working in the Zoologisches Institut at Wiirzburg, Germany. Later, eggs, embryos, and young Sagittas, 7| days old, were studied at Naples.

Origin and Development of the Ducts

Hertwig's well-known figures show the four primary germ cells free in the coelomic cavity of the gastrula, later embryo, and recently hatched young Sagitta. In the 10-day old Sagitta he figures the two pairs of primary germ cells attached to the body-wall, and each cell covered by a layer of endothelium. In my 7§-day larvae of Sagitta infiata, I find the germ cells not yet covered by endothelium, although the endothelial cells in some cases appear to be creeping up over them. Figs. 1 and 2, are longitudinal optical sections through the germ cells of whole mounts stained with borax carmine, and Fig. 3 is a cross section through one of the two primary oogonia (7^ days) . My later stages are all Sagitta bipunctata. The youngest, obtained from plankton at Naples, have from 4 to 7 cells in each ovary, and show in a single section (Fig. 4) two or three oogonia lying against the body wall and covered by a layer of endothelium. In ovaries containing from 16 to 20 oogonia the antrum has begun to form below the germ cells. Figs. 5 and 6 are sections of the two ovaries of the same individual, showing the same conditions as in Fig. 4, — a group of oogonia covered by a layer of endothelium, which is continuous with the

286 N. M. Stevens.

lining of the body wall. Just below the oogonia, a group of cells (Fig. 7) with a small lumen was found ; this is evidently the beginning of the antrum, of which the oviduct is a cephalad continuation. The section below shows a strand of cells running from the body wall to the transverse membrane which separates the two body cavities. Figs. 8 to 11 were taken from a somewhat older ovary containing 35 to 40 oogonia. Fig. 8 is the most posterior section of the ovary proper, showing a single oogonium surrounded by cells which will later give rise to the antrum and ducts. The section below (Fig. 9) shows more of the antrvim cells in the form of a fold or outgrowth of the inner or mesodermal layer of the body wall. (Cell outlines were not distinct.) The next section below contained the remainder of these cells, a thin strand extending to the transverse membrane. Fig. 10, the third section above Fig. 8 and Fig. 11, the tenth above Fig. 10, show nothing more than Figs. 5 and 6; i.e., no ducts yet formed. If, however, we take a considerably later stage where the growing oocytes ,can be distinguished from the oogonia (Fig. 12), we find in addition to the endothelial membrane, two other layers of cells between the body wall and the germ cells. These layers extend around the sides of the ovary forming a 2-layered crescent (od). There is as yet no lumen and no sperm-duct.

Fig. 13 shows the next stage, in which a different kind of tissue (sd) has come in between the two layers of the lateral wings of the future oviduct. This tissue evidently must either be proliferated from the layers which surround it or be formed by migration from those layers, as there is no other possible source. No cell outlines can be distinguished, the tissue appearing to be a syncitium of a fibrous nature with scattered nuclei. In this stage mitosis is frequent in the oviduct walls, but rare in the new tissue between them. Fig. 14 is a somewhat older ovary where the oviduct walls are completely separated in the middle by the meeting of the new tissue, which first appeared in the wings. In Fig. 15 we have a section of the lower or caudad end of the ovary, showing the oviduct walls continuous with the as yet closed antrum.. The reproductive pore is formed, but the cells have not separated to form the antrum chamber and external opening.

Further Studies on Reproduction in Sagitta. 287

The sperm-duct tissue has not appeared in the antrum region. By comparing this figure with Fig. F, one can easily see the relation of the embryonic layers to the adult oviduct and spermduct. In one case only, shown in Fig. 16, have I ever seen at this stage the sperm-duct tissue separated from the median oviduct wall, leaving a lumen corresponding to the space through which the ripe eggs pass to the exterior.

Somewhat later the whole oviduct with the eontained spermduct tissue contracts laterally, as shown in Fig. 17, changing the flat layer of sperm-duct tissue into an approximately cylindrical rod of syncitial tissue with nuclei rather irregularly scattered and vacuoles here and there (v). Figs. 18 and 19 show the appearance of an irregular lumen, and Fig. 20 the presence of spermatozoa in the more regular lumen of an older duct. In Fig. 21 we have a longitudinal, but slightly oblique, section through the reproductive opening and antrum of an adult Sagitta, showing the relation of the two ducts and their openings when the sperm-duct is slightly open and the oviduct closed. In Fig. 22 is shown a transverse section of an ovary of Sagitta bipunctata with a ripe egg in the oviduct, to the outer wall of which the sperm-duct (sd) remains attached. The walls of the oviduct are much stretched, but the lateral wings are not opened as one might expect them to be.

Thus we have traced the development of the oviduct from its origin in a fold of mesoderm at the base of the ovary, and the formation within it of a sperm-duct or ' Samentasche ' with a separate opening from that of the oviduct. These ducts are without doubt purely accessory mesodermal structures, and have no direct connection with the germ track of Sagitta. In adult ovaries the young oogonia and youngest oocytes are sometimes crowded against the median wall of the oviduct, and are even pressed in between the cells in such a manner that they appear to be a part of it, but I am convinced that this is only a secondary appearance and that there is always a portion of the oviduct wall between the germ cells and the sperm-duct.

The oviduct wall furnishes two accessory fertilization cells to each ovum, and the lateral wings m.ay be largely reserve cells to make good the loss from this source. The central ends of the cells of

288 ^ N. M. Stevens.

the oviduct wall stain as though they contained supporting fibers or possibly muscle fibers, though in watching the passage of eggs down the oviduct I have been unable to detect any muscular activity in the oviduct walls. I have thought that these fibers might be of an elastic nature, but have not tested their staining reaction with this point in view. They stain deeply with ironhsematoxylin, but so does any particularly dense portion of a cell. The stretching to which the walls of the oviduct are subjected when eggs are passing through it naturally suggests that the cells may have developed material of an elastic or supporting nature to meet this strain.

Maturation and Segmentation of Eggs in the Ovary

While making the above observations on egg-laying in Sagitta elegans, it was noticed that in several individuals asters were present in eggs which were either free in the ovary or were in the oviduct. The first cases noted were in material which had been kept in the laboratory over night, but later the same phenomenon was observed in individuals fresh from, the sea. Several such specimens were fixed and sectioned. Eggs were found free in the ovary in all stages of maturation and segmentation up to an 8-16-cell stage. Fig. 23 shows the first polar body and the second polar spindle of such an egg, and Fig. 24 the uniting male and female pronuclei. Figs. 25 and 26 are the first and second segmentation spindles in metakinesis and anaphase respectively. These figures make it plain that development here is that of fertilized eggs and not a case of parthenogenesis. Fig. 27 is a section of an 8-cell stage from an ovary in which several segmenting eggs were found. At present I have no satisfactory explanation to offer for this premature development of Sagitta eggs. Nothing of the kind was seen, either in material from Naples or in that from Woods Hole in 1904, the only approach to it being the cases of two polar spindles in eggs of one individual described in "Further Studies on the Ovogenesis of Sagitta" ('05, p. 246, Figs. 11, 15, 16, and 17). Whether these segmenting eggs could ever be laid seems doubtful, for the moment of breaking away from the oviduct

Further Studies on Reproduction in Sagitta. 289

wall seems to be the opportunity to open a way to the exterior. Failure to make their way into the oviduct may be the whole explanation of the conditions observed. The eggs are already fertilized and therefore ready to go on developing unless some added stimulus from the sea-water were necessary to cause them to complete their maturation and begin to segment. That this is not the case is evident; after the egg is fertilized and has broken away, maturation and the early segmentation stages go on whether the egg is laid at the normal time or is retained in the ovary. Most of these eggs had decidedly irregular outlines suggesting amoeboid movements and supporting my conclusion that the eggs make their way into and down the oviduct by their own activity.

Elpatiewsky's 'Besondere Korper'

The appearance of Elpatiewsky's ('09) paper on "Die Urgeschlechtszellenbildung bei Sagitta" invests these segmenting eggs with a new value. Elpatiewsky finds in the development of normally laid eggs a stainable body which, from the first segmentation on, marks the germ track ; and, possibly, is in the sixth division the determining factor in the separation of the germ plasm into primary oogonia and primary spermatocytes. This 'besondere Korper,' as Elpatiewsky designates it, is first seen near the vegetal pole of the egg when the two pronuclei are in the center of the egg, and the first cleavage plane passes a little to one side of it. Although Elpatiewsky states that he first finds this homogeneous stainable body in the stage where the two pronuclei are uniting, my first thought on reading the paper was that this body must have some relation to the accessory fertilization cell which I had described ('05) as degenerating after the egg breaks away from the oviduct wall. This description was based on the degenerate appearance of the cell in eggs nearly ready to break away, and on the fact that I had been able to find no trace of it in the eggs after they entered the oviduct. Indeed, I suspected that in some cases the accessory cell might be actually pulled out of the egg.

290 N. M. Stevens.

I had no difficulty in finding Elpatiewsky's 'besondere Korper' in the free eggs of Sagitta elegans, provided that they had reached the stage when the two pronuclei were in the center of the egg, as in Fig. 24. Fig. 28 shows this body from the same egg as Fig. 24, and Fig. 29 a and h is another similar stage, while Figs. 30 and 31 are sections from 2-cell and 16-cell stages. It was also found in all stages between those of Figs. 29 and 31, but when I cam.e to look for it in earlier stages, of which I had an abundance, both of Sagitta elegans and of Sagitta bipunctata, no connection between the 'besondere Korper' and the accessory fertilization cell could be traced. Figs. 32 a and 6 show an egg in which the accessory cell {a) is evidently degenerating, and the nucleus of the egg is breaking down preparatory to the first maturation mitosis; Figs. 33 a and h another egg, still attached, with the first maturation spindle in prophase. In the latter case the nucleus of the accessory cell looked quite normal, while in the former the whole cell stained dark and a somewhat lighter portion in the center of the cell was the only indication of the disappearing nucleus. In general it may be said that at this stage the accessory cell looks shrunken and dark, and the nucleus has either degenerated or is on the point of doing so. Fig. 34 shows an egg in which the two accessory cells were both much shrunken when the first maturation spindle was in metaphase. In such a stage as Fig. 32 a the accessory cell is about the same size as the ' besondere Korper' (Fig. 29) , and both stain deeply. The accessory cell, however, has a dis/ tinct cell boundary and the 'besondere Korper' quite a different structure, being composed of a non-staining homogeneous matrix filled with deeply staining granules. Fig. 35 shows one case where this body was not spherical but irregular in outline, consisting of somewhat scattered masses, at a stage when the first segmentation spindle was in metaphase.

In maturation stages between Figs. 32, 33, 34, and Fig. 29, I find absolutely no trace of either accessory fertilization cell or 'besondere Korper.' This is in entire argeement with the conclusion of Elpatiewsky, and is based on examination of many preparations of Sagitta elegans (1904 material) in which eggs containing the first maturation spindle in metaphase were passing down the ovi

Further Studies on Reproduction in Sagitta. 291

duct, and also on many preparations of the same species containing eggs developing free in the ovary (1905 material). Several preparations of Sagitta bipunctata with eggs in the oviduct were examined with the same result.

In one egg of Sagitta elegans, containing the second maturation spindle, I did find, near the vegetal pole, a spot of denser cytoplasm (Fig. 36 a and h) which, however, was not at all stained with ironhsematoxylin. This may have been an early stage of the 'besondere Korper,' but could, I think, hardly be a transition stage between the degenerating accessory cell and the 'besondere Korper,' since degenerating nuclei nearly always stain dark.

It also seemed possible that the granules of chromatin-like material extruded from the nucleus in preparation for maturation of the egg ('03, PI. 1, Fig. 2 6), a phenomenon which is even more marked in Sagitta elegans (Fig. 37), might have some part in the formation of the 'besondere Korper,' but apparently these granules all disappear completely before this peculiar body becomes visible. At the stage shown in the two figures referred to, small black granules are scattered all through the nucleoplasm and a quantity of larger granules are found outside the membrane between the nucleus and the point of attachment of the egg. If either these granules or the accessory fertilization cell take part in the formation of the 'besondere Korper,' it must be after an intermediate stage in which the material does not stain.

In a recent paper, Buchner ('10) traces Elpatiewsky's 'besondere Korper' to the degenerate nucleus of my 'accessory' cell, but curiously enough he describes the recently laid eggs as containing "1) das Degenerat der Strangzellen, 2) das Spermium, zwar schon mit einem Strahlenhof umgeben, aber noch nicht zum mannlichen Vorkern aufgequollen, 3) die Telophase der ersten Reifenteilung des Eikerns," and then he refers to his Fig. 5, which is not the stage which he describes but the one where both Elpatiewsky and I first find the 'besondere Korper.' (Compare Buchner Fig. 5, Elpatiewsky Fig. 3, Stevens Figs. 24, 28, 29).

The mit Eisenhsematoxylin sich tief schwarzendes Netz" to which Buchner refers and upon the importance of which he lays considerable stress in connection with his hypothesis of the chromi

292 N. M. Stevens.

dial nature of the 'besondere Korper' and of the nucleolus-hke bodies of the oogonia and oocytes, I have spoken of in describing the development of the oviduct. I find no such connection with the 'accessory' cell as he figures on page 435, and a careful review of all my preparations of material from four species has only served to confirm my previous account of the appearance and function of the two epithelial cells which connect each oocyte with the sperm-duct.

The Accessory Fertilization Cells

Fig. 38 shows an exceptionally good section of the accessory or connecting epithelial cells in Sagitta bipunctata. The fertilization canal, which Buchner says he has not been able to find, is shown, and the spermatozoon (s) is partly in the second cell. It is only occasionally that one finds the spermatozoon in the cells, but when found, it is not to be mistaken for any such connecting fiber as Buchner figures. In the broad portion of the canal there is a large, granular non-staining spindle-shaped body (6) which may, I think, be some substance attractive to the spermatozoa. This I first saw in living eggs while attempting to observe the entrance of a spermatozoon into the fertilization canal. In the living eggs both connecting cells can be distinguished, though their nuclei are invisible and the canal extending through both cells is plainly seen, as is also the spindle-shaped body in the widest part of the canal. Fig. 39 is a camera sketch from such a living egg of Sagitta inflata, with the nucleus visible but near the periphery of the egg, indicating the approach of maturation. Many spermatozoa were around the mouth of the canal, but I have never been able to detect one in it, possibly because I have used too late a stage for observation.

Fig. 40 shows another case where a part of a spermatozoon is seen in a section of the canal in the inner accessory cell. Figs. 41a and h show the same cells from an egg of Sagitta minima. Here the fertilization tube (c) is coiled in the outer cell. This I also observed in the living egg. Fig. 42 is a section through these cells in Sagitta elegans showing several portions of the more or less

Further Studies on Reproduction in Sagitta. 293

winding fertilization canal (c). The eggs of Sagitta elegans have a much thinner membrane than those of Sagitta bipunctata. In this figure one sees the usual conditions in the region of attachment of an egg which is nearly ripe, — the ends of the surrounding cells of the oviduct wall stain more deeply, and one also finds the stain showing between cells, simulating fibers. The nuclei of two cells adjoining the outer connecting cell have perfectly black nuclei, probably indicating degeneration. In Fig. 43 is shown the place where an egg has broken away without making its way into the oviduct. The cells surrounding the torn connecting cell (a) stain almost perfectly black on a well differentiated slide. It is also true that when an egg is passing from the ovary into the oviduct the cells on the border of the opening through which the egg is passing stain deeply, indicating some change connected with the ripening of the egg. In Sagitta inflata the nuclei of the connecting cells stain darker, and it is more difficult to say how early degeneration begins. Fig. 44 shows the two connecting or accessory cells of a comparatively young egg and Fig. 45 of an older egg. In both figures sections of the fertilization tube (c) are seen.

The 'Besondeke Korper' in the Primary Germ Cells

My material does not include the stage in which the ' besondere Korper ' divides in the sixth segmentation mitosis, which gives rise to the two primary germ cells, but I find the remains of this body in many sections of young gastrulse. In Sagitta bipunctata they are very often conspicuous in a stage where the germ cells are still buried in the gastrula wall (Fig. 46 k). In this particular case these bodies were as deeply stained as in the earlier stages. When the two germ cells are in mitosis, I find only less deeply staining fragments (Fig. 47 a and 6, k) . Likewise in the 4-germ cell stage I find some fragments (Fig. 48). Corresponding bodies are found in Sagitta inflata and Sagitta elegans, but in both they are less conspicuous. Fig. 49 is an early 2-germ cell stage of Sagitta elegans, showing one gray mass in cell a, none in 6 ; in the next section there were two more such masses in a, but none in h, indi

294 N. M. Stevens.

eating that the fragments disappear somewhat earlier than in Sagitta bipunctata. Figs, 50 and 51 show the remains (k) of the 'besondere Korper' in Sagitta inflata; here it is granular and yellowish, not taking the hsematoxylin. My sections all indicate, as Elpatiewsky says, that division of the 'besondere Korper' is unequal, a larger portion going to one of the first two germ cells than to the other.

As to whether this body is a special device for determining sex in a hermaphrodite organism, I do not think we can safely express an opinion without more evidence drawn from the embryology of other hermaphrodite forms. The special object of my renewed


Fig. H. Freehand sketches showing relative positions of the two pairs of primary germ cells.

study of the germ cells of Sagitta at Naples last year was to see whether the division of the first two primary germ cells to form two primary oogonia and two primary spermatogonia is in any way a visibly differential mitosis. I have not been able to detect anything of that nature. Before Elpatiewsky's paper appeared, I had noted the fact that the two primary germ cells never divided synchronously. One finds one cell in prophase, the other in metaphase; one in metaphase, the other in anaphase; one in anaphase, the other in telophase, etc. Having also noticed that the arrangement of the four germ cells in a row is a secondary matter, I had suspected that the differential division might be the earlier one. When the two cells are in mitosis, the two spindles in metaphase often stand nearly at right angles, and one finds the four cells soon after this mitosis in various positions (Fig. H a, h, and c). In the older gastrulse, after the entoblast folds are formed, the four cells usually form a nearly straight row, as in Fig. 51. Fig. 52 shows the four germ nuclei in an egg of Sagitta in

Further Studies on Reproduction in Sagitta. 295

fiata, as one looks down into the gastrula cavity. Fragments of the 'besondere Korper' are found only in the two cells with the smaller nuclei which are the products of the later mitosis. In the stage shown in Fig. 51, 1 am unable to detect any constant difference in either cytoplasm or nuclei of the cells, and even in young ovaries and testes consisting of a single cell (Figs. 1 to 3) or containing a number of oogonia or spermatogonia, the cells are so much alike than one must depend upon other anatomical features to determine whether one is looking at a section of a young ovary or a young testis.

Intranuclear Granules and Network

In my- 1903 paper I referred briefly to certain large black granules found on the inside of the nuclear membrane of young oocytes, and also to the reticular network" conspicuous in the nuclear membrane of all older oocytes up to the time when the egg membrane forms. While I was working on the development of the sperm-duct and oviduct in Professor's Boveri's laboratory, he suggested that it might be interesting to investigate the origin of this network in view of its possible relation to the mitochondria recently described in the cytoplasm of many kinds of cells. At the time I could not trace it back farther than the young oocytes, perhaps a little younger than those shown in Figs. 4 and 6, PI. 1, '03, and I thought it desirable to examine more carefully than I had previously done the embryonic and larval stages of the germ cells, with this question in mind. Sections of those, however (Figs. 1 to 3, 46 to 52) , throw very little light on the point. The ero.bryonic germ cells show some granules, but nothing really comparable to those of the young oocytes. The oogonia of young ovaries show several rather large granules, or chromatin-nucleoli, against the nuclear membrane (Figs. 8 and 10). Dividing oogonia also show these nucleoli outside of the spindle in both metaphase (Fig. 53) and anaphase (Figs. 54 and 55). In the daughter nuclei no such nucleoli are found, and it therefore seems probable that these chromatin-nucleoU consist of material thrown out by the chromosomes in the interval between two mitoses, and dissolved


290 N. M. Stevens.

in the cytoplasm during or after each mitosis. In Fig. 56, 6 is a cell in prophase and a two recent products of mitosis. The resting nuclei of oogonia often show as many granules in tangential section as in Fig. 57 or 58, but it is in the young oocytes at the very beginning of the growth stage that one finds relatively large masses of this chromatin-like material inside the nuclear membrane. It most often lies at the two ends of an elongated nucleus, and several of the chromosomes are closely associated with it (Fig. 59). In Fig. 60 one of these masses has begun to divide. Figs. 61 and 62 are nuclei of slightly older oocytes with the large masses broken up into smaller ones, and the volume of these intranuclear granules is also apparently increased. Fig. 63 is a tangential section of a somewhat later stage showing the beginning of the network which replaces the masses and granules in older oocytes. These granules, as well as the reticular network, stain like the chromosomes with iron-hsematoxylin, borax carmine, saffranin, and Benda's stain for mitochondria. Figs. 64 and 65 show the characteristic network which lines the nuclear membrane of older oocytes of Sagitta bipunctata. The pattern consists of two parts, the network and the irregular figures in the openings. These two parts vary in prominence in different species. In Sagitta bipunctata the meshwork is more prominent, and in older eggs the central figures disappear first. Figs. 66 a and b are tangential and median sections of the same nucleus. Figs. 67 a and b similar sections of the nucleus of an older egg, showing the breaking up and gradual absorption of the network as the egg approaches maturity. When the chromosomes have been reduced to the maturation size, the stainable network has practically all disappeared from the nuclear membrane (Fig. 68). In Sagitta minima, one finds in young oocytes stellate or amoebalike figures (Figs. 69 and 70) , which later unite to form a network (Fig. 71), which may or may not have rather indistinct figures in the spaces (Fig. 72). In Sagitta infiata I usually find such a pattern as in Fig. 73, and Fig. 74 from an older egg. In these three species the network is more prominent than the contained figures. In Sagitta elegans it is the reverse — the central figures

Further Studies on Reproduction in Sagitta. 297

are heavier and more lasting than the surrounding network (Fig. 75). Fig. 76 is from an older egg, where the netw^ork has nearly all disappeared and the central figures are spread out thinner. In Sagitta decipiens (Fig. 77 and 78) we find the irregular figures without the network.

As the above described intranuclear network is evidently derived from the large chromatin-like masses and granules of the younger oocytes, and these masses and granules are either developed under the influence of the chromosomes, or, more probably, are material extruded from the chromosomes of the earlygrowth stage, I think we must regard it as comparable to the chromatin-like material frequently given off from the chromosomes in early growth stages of both spermatocytes and oocytes (see Boring, '07, Figs. 62-67, and King, '08, Figs. 26-30), and not to the mitochondria which is found outside the nucleus, usually in the form of fibers, and which has not been satisfactorily traced to a nuclear origin. Ordinarily when one sees such material thrown out by the chromosomes or spireme, one is inclined to regard it as waste material, but the case of Sagitta, where the material forms such a conspicuous pattern on the nuclear membrane during the greater part of the growth stage of the oocyte, strengthens the growing opinion that such material may have a specific function in connection with the growth-process of the egg. So far as I can see, there is no evidence whatever that either the chromatin-like masses seen against the nuclear membrane in spermatogonia, oogonia, and young oocytes, or the elaborate pattern of older oocytes, can possibly be derived from Elpatiewsky's ' besondere Korper ' or from the degenerate inner connecting, or fertilization cell of the ovum, as Buchner ('10) has suggested.

Synapsis and Spermatogenesis

In regard to synapsis, I have no new evidence to add to that given in my two earlier papers. I have found nothing opposed to the conclusion that in Sagitta we have a case of parasynapsis in the oocyte and telosynapsis in the spermatocyte. Synizesis and bouquet stages (Figs. 79-81) are not uncommon among the

298 N. M. Stevens.

young oocytes, but it is very difficult to find the synapsis stages figured in my '05 paper, PL 16, Figs. 18-25. The chromosomes are extremely small and favorable stages are rarely met with.

In my first paper on Sagitta ('03) I briefly described the principal stages in the spermatogenesis of Sagitta bipunctata, showing the reduced number of chromosomes to be nine, formed by telosynapsis. Mention was made of the fact that one or two large nucleoli were often seen near one or both poles of the first maturation spindle. In a later paper ('05) on the spermatogenesis of insects, some additional figures (PL 7, Figs. 226-241), based on further study of the body spoken of as a nucleolus in the earlier paper, were given. This was of especial interest at that time on account of its possible homology with the heterochromosomes of insects.

After further study of the element in question in Sagitta elegans, new material of Sagitta bipunctata from Pt. Erin and Helgoland, and considerable work on Sagitta bipunctata, S. minima, and S. inflata, in aceto-carmine preparations, at Naples last year, I am convinced that the elements figured as x in the spermatogonia, and growth stages of the spermatocytes (PL 7, Figs. 226-233) are nucleoli comparable to those described above in oogonia, oocytes and spermatogonia, while the elements so designated (x) in the maturation mitoses (Figs. 235-241) are products of premature division of one of nine bivalent chromosomes. I find some individuals in which the nine chromosomes all behave alike in all or in most cases; while in other individuals, one divides prematurely and appears at or near the two poles of nearly every spindle in both first and second spermatocytes. In the first material, which I prepared at Naples and studied at Wiirzburg in 1902, I found only a very few cases of this phenomenon; so few that I did not publish any figures. The figures in the 1905 paper were taken from material obtained from Naples the same year, but used at the tim.e only for the study of the maturation stages of the ova. Sagitta elegans showed the sam.e peculiarity, and the appearance of daughter chromosomes at each pole of the maturation spindle in metaphase was much more frequent. In the Pt. Erin material the difference between individuals in this

Further Studies on Reproduction in Sagitta. 299

respect was very striking. In some specimens all of the chromosomes were in the equatorial plate without exception; in others the polar chromosomes were seen in every spindle ; in still others both conditions were found in the same cyst. This variation was also. found in the aceto-carmine preparations of the three Naples species. The number of chromosomes in all of the species studied is the same, nine for the reduced number. The individual chromosomes are much more distinct in aceto-carmine preparations, and I think there can now be no doubt as to the count.

Figs. 82 and 83 are first spermatocyte equatorial plates of Sagitta bipunctata from spindles in which all of the chromosomes were in the equatorial plates; Fig. 84 from a spindle where eight were in the equatorial plate and one daughter chromosome (.Ti and X2) at each pole. All of the spindles from this individual showed the polar chromosomes. Some of these spindles also showed the tetrad nature of the chromosomes (Fig. 85). Fig. 86 is a prophase showing the premature division of one chromosome (x). Sagitta inflata proved to be the best material for studying the spermatocytes with aceto-carmine. Fig. 87 is an equatorial plate containing the nine chromosomes, and Fig. 88 a side view of a spindle with the two daughter chromosomes (x) near the poles of the spindle and in different positions, so that they appear somewhat unequal in size. In spindles which do not show the prematurely divided chromosome, one often sees a larger bent chromosome (Fig. 89), which apparently is the erratic one, and may or may not divide precociously. It will also be noticed that this chromosome is attached to the spindle fibers in a different way from the others. Fig. 90 shows the polar chromosomes in such a position that they look equal, and in Fig. 91 they are already dividing precociously a second time. Fig. 92 is a second spermatocyte spindle showing the polar chromosomes. Fig. I, a to h, shows the various forms which the bivalents may assume in first spermatocyte prophases and metaphases. The dumb-bell form (a) is by far the most common; h I have seen only once; while the other figures are not infrequent in aceto-carmine preparations. Fig. 93 also shows in an oblique view of a daughter plate the splitting of the chromosomes for the second maturation mitosis.

300 N. M. Stevens.

I have as yet been able to find no satisfactory explanation of the fact that in the four species in which the spermatogenesis has been examined, one of the nine chromosomes behaves so peculiarly, sometimes dividing synchronously with the others, sometimes precociously in both maturation mitoses. One naturally wonders if this chromosome is in any way homologous with the heterochromosomes of insects, whose distribution is closely connected with the determination of sex. Of course in a hermaphrodite organism, there is no question about determination of sex, as the two sexes are combined in one individual, and all fertilized eggs must contain both male and female sex determiners, or at least must be able to produce both male and female germ cells.

abed ®fgh


Fig. I. Forms which the bivalent chromosomes assume in prophases and metaphases of the first spermatocyte.

So far as the chromosomes are concerned, there is no evidence of any such reducing mitosis as might give oogonia, oocytes and ova purely female, and spermatogonia, spermatocytes, and spermatozoa purely male. If sex determiners are present in the germ cells, both must be present in both oocytes and spermatocytes and reduction must give both eggs and sperm of two kinds with reference to sex. Selective fertilization would then be necessary to give eggs containing both determiners, and capable of producing hermaphrodite organisms. It may then be possible that the 'besondere Korper' discovered by Elpatiewsky is a mechanism for determining dominance of the sex determiners in the male and female germ cells; i.e., determining that one pair of primary germ cells shall give rise to ovaries, the other pair to testes. In support of this suggestion it is important to determine whether anything comparable to this 'besondere Korper' is present in the

Further Studies on Reproduction in Sagitta. 301

segmenting eggs of other hermaphrodite organisms, and in either male or female germ cells of those insects in which an equal pair of heterochromosomes has been described in the male germ cells.

It is my intention to investigate further some of the points referred to in this paper as soon as there is opportunity to work again on fresh material.

I desire in this place to express my appreciation of the opportunity afforded me to collect material at Pt. Erin, Helgoland and Naples, and of the privileges and courtesies which I enjoyed at the Zoologisches Institut, Wiirzburg, and. at the Stazione Zoologica, Naples.


1. Eggs of Sagitta, when ripe, apparently make their way into a previously closed oviduct and move down the oviduct to the reproductive pore by their own activity.

2. The ovary proper and the ducts are entirely distinct structures as to their origin. Each ovary develops from one of the four primary germ cells, while the antrum and oviduct develop from a fold or outgrowth of the mesodermal layer of the body wall below the ovary, and the cells which form the sperm-duct or ' spermentasche ' originate from the oviduct wall by migration or delamination.

3. It is suggested that the deeply stained network that can often be seen in the median part of the oviduct wall next to the sperm-duct may be of an elastic nature.

4. In Sagitta elegans eggs free in the ovary have been found in all stages between that of the metaphase of the first maturation mitosis and a 16-cell stage. No satisfactory explanation of this phenomena was apparent.

5. Elpatiewsky's 'besondere Korper' was found in all of these free eggs, beginning with the stage in which the two pronuclei were in the center of the egg.

302 N. M. Stevens.

6. No connection between this 'besondere Korper' and the secondary accessory fertilization cell could be traced.

7. The two accessory cells with the fertilization canal are found in all of the species studied (5), and have been observed in living specimens.

8. The author confirms Elpatiewsky's conclusions as to the 'besondere Korper' in the primary germ cells, and also as to the probability that the division which produces two germ cells is the differential mitosis.

9. The granules and network found on the inside of the nuclear membrane of immature oocytes appear not to be mitochondria, but material derived from the chrom.osomes of the very young oocytes. Whether this is waste material or material which functions in connection with the growth of the oocytes is an open question.

10. All of the species examined have nine chromosomes in the spermatocytes, one of which sometimes behaves like an equal heterochromosome bivalent. The precocious division of this chrom.osome is very striking in some individuals and entirely absent in others.

Accepted by The Wistar Institute of Anatomy and Biology, May 6, 1910. Printed August 4, 1910.

Further Studies on Reproduction in Sagitta. 303


Boring, A. M. A study of the spermatogenesis of twenty-two species of the mem1907 bracidse, Jassidae, CercopidseandFulgoridse. Journ. Exp. Zool. i.

BucHNER, P. Keimbahn und Ovogenese von Sagitta. Anat. Am. 35. 1910

Elpatiewsky, W. Die Urgeschlechtszellenbildung bei Sagitta. Anat. Am. 35 1909

Grassi, B. I. 1 Chaetognathi. Fauna Flora Neapel. Mongr. 5. 1883

Hertwig, O. Die Chsetognathen. Jena. 1880

Keferstein. Untersuchungen iiber niedere Seethiere. Zeit. wiss. Zool. 12. 1862

King, H. D. The oogenesis of Bufo lentiginosus. Journ. Morph. 19. 1908

Krohn, a. Nachtragliche Bemerkungen iiber den Bau der Gattung Sagitta. 1853 Archiv.f. Naturges. Jahrg. 19. Bd. 1.

LeuckhartundPagenstecher. Untersuchungen iiber niedere Seethiere. Archiv 1858 /. Anat. Physiol, u. unss. Med. Berlin.

Stevens, N. M. On the ovogenesis and spermatogenesis of Sagitta bipunctata. 1903 Zool. Jahrb. 18.

1905 Further Studies on the ovogenesis of Sagitta. Zool. Jahrb. 21.

1905 Studies in spermatogenesis with especial reference to the "accessory chromosome." Carnegie Inst. Pub. 36.

304 N. M. Stevens.


(Figures reduced I from the camera drawings)

Figs. 1 and 2. Longitudinal optical section of male (0 and female (o) primary germ cells in Sagitta inflata, 7| days. 2mm-8.

Fig. 3. Cross-section of 7| day Sagitta through one of the primary oogonia (o). 1.5 mm.-12. Fig. 4. Section through young ovary containing 6-7 oogonia. 2 mm.-8.

Figs. 5 and 6. Sections through two ovaries of the same individual containing 16-20 oogonia. 2 mm. 6.

Fig. 7. Section next below the ovary of Fig. 6, showing beginning of antrum and oviduct (od). 2 mm.-6.

Figs. 8 and 11. Sections of same ovary. 2 mm.-6.

Fig. 8. Posterior section of ovary proper showing only one oogonium (o).

Fig. 9. Next section posterior to Fig. 8, showing the fold or outgrowth of mesoderm which will give rise to antrum and oviduct.

Fig. 10 Fifth section anterior to Fig. 8.

Fig. 11. Tenth section anterior to Fig. 10.

Fig. 12. Cross-section of older ovary, showing the oviduct (od) as a 2-layered crescent-shaped structure. 2 mm. -4.

Fig. 13. One-half section of older ovary showing sperm-duct tissue (sd) in one wing of the oviduct (od). 2 mm. -4.

Further Studies en Reproduction in Sagitta.


7^. :^.6./;Uj^ 6

1 3

306 N. M. Stevens.

Fig. 14. Older ovary, showing sperm-duct tissue (sd). 2 mm. -4.

Fig. 15. Longitudinal section through the antrum region before the external opening (p) has become functional. Same stage as Fig. 14. 2 mm. -4.

Fig. 16. Crescent-shaped oviduct, showing an opening (od) between the spermduct tissue {sd) and the median wall. 2mni.-4.

Fig. 17. Oviduct and sperm-duct contracted laterally. 2 mm. -4.

Figs. 18-20. Sperm-duct, showing development of lumen (1). 2 mm. -4.

Further Studies on Reproduction in Sagitta.



^. ^. S. U^.


308 N. M. Stevens.

Fig. 21. Longitudinal section of adult, showing sperm-duct {sd) slightly open to the exterior. 2 mm. -4.

Fig. 22. Cross-section of adult ovary, showing an egg (o) in the oviduct (od)' B. and L. ^-A.

Fig. 23. First polar body and second polar spindle of an egg free in the ovary2 mm.-12.

Fig. 24. Two pronuclei of egg free in the ovary. D-4.

Fig. 25. First segmentation spindle of free egg. 2 mm. -4.

Fig. 26. Second seg. spindle of free egg. 2 mm.-4.

Fig. 27. Section of egg in 8-cell stage, free in the ovary. D-4.

Fig. 28. 'Besondere Korper' of same egg as Fig. 24. 1.5 mm. -4.

I i:riher Stu- .;l:'. on Reproduction in Sagitta.


^ T^Jf .^i/!

310 N. M. Stevens.

Figs. 29 a and b. Free egg containing 'besondere Korper' (iv) and pronuclei (n). 4 mm. -2.

Fig. 30. Two-cell stage of egg free in the ovary, showing the 'besondere Korper.' 4 mm.-2.

Fig. 31 . Section of 16-cell stage, free in the ovary. 4 mm. -2.

Figs. 32 a and b. Sections of attached egg, showing degenerating accessory cell (a) and nucleus (n) breaking down for maturation. 4 mm. -2.

Figs. 33 a and b. A slightly later stage, egg still attached but mat. spindle formed. 4 mm. -2.

Figs. 34 a and b. Slightly later stage. 4 mm. -2.

Fig. 35. Exceptional case where the 'besondere Korper' (A') is irregular in form. 1.5-6.

Figs. 36 a and b. Second mat. spindle (a) and spot of denser cytoplasm at vegetal pole {b) of free egg. 1.5 mm. -2.

Further Studies on Reproduction in Sagitta. 311

3 4b 36a



3 6b


312 N. M. Stevens.

Fig. 37. Nucleus of egg near maturity, showing cast-out granules of chromatinlike material.

Fig. 38. Section through the two accessory fertilization cells (Oi an), showing granular body (b) and spermatozoon (s) in the canal. 1.5 mm. -4.

Fig. 39. Optical section of accessory cells in living egg of Sagitta inflata, showing canal (c) and granular body (6). 2 mm.-4.

Fig. 40. Section showing spermatozoon in second accessory cell. 1.5 mm. -4.

Figs. 41 a and b. Two sections through the accessory cells of an egg of S. minima. 1 . 5 mm.-4.

Fig. 42. Similar section from S. elegans. 1.5 mm.-4.

Fig. 43. Sections from S. elegans, showing degeneration where an egg has broken away without entering the oviduct. 1 . 5 mm.-4.

Fig. 44. Section of the accessory cells of a young egg of S. inflata. 1.5 mm. -4.

Fig. 45. Similar section of an older egg. 1 mm. -4.

Further Studies on Reproduction in Sagitta. 313


314 N. M. Stevens.

Fig. 46. Two primary germ cells of S. bipunctata still in the gastrula wall, showing the 'besondere Korper' (K). 1.5 mm.-6.

Fig. 47. Two sections of the two primary germ cells of S. bip. in mitosis, showing 'bes. Korper' (K). 1.5 mm. -6.

Fig. 48. Section of the four primary germ cells of S. bip., showing 'bes. Korper' (K) in one cell. 1.5 mm. -6.

Fig. 49. Two primary germ cells of S. el. preparing for mitosis. 'Bes. Korper. ' in delayed cell. 1.5 mm. -4.

Fig. 50. Two sections of first two primary germ cells of S. infl., showing 'bes. Korper.' 2 mm.-4.

Fig. 51. Four primary germ cells of S. infi. 1.5 mm. -6.

Fig. 52. Four primary germ cells of S. infl., early stage, as seen in gastrula cavity. 2 mm. -4.

Fig. 53. Oogonium in metaphase, S. bip. 2 mm. -12. Figs. 54 and 55. Oogonia in anaphase. 2 mm.-12.

Fig. 56 a. Two young oogonia without granules, and b an oogonium in prophase with granules. 1.5 mm. -6.

Further Studies on Reproduction in Sagitta.







316 N. M. Stevens.

Figs. 57 and 58. Resting nuclei of oogonia, showing granules under the nuclear membrane. 2 mm. -12.

Fig. 59. Young oocyte, showing masses of chromatin-like material at the two ends of the nucleus. 1.5 mm.-12.

Figs. 60-62. Later stages with several chromatin-like masses inside the nuclear membrane. 1.5 mm. -12.

Fig. 63. Young oocyte, showing beginning of network. 2 mm. -4.

Figs. 64 and 65. Tangential sections of nuclei of S. bip. showing characteri- tic network. 1.5 mm. -4.

Fig. 66 a and b. Tangential and median sections of nuclei of half-grown oocytes of S. bip. 2 mm. -4.

Fig. 67 a and b. Similar sections of older oocytes, network disappearing. 2 mm. 4.

Fig. 68. Section of nucleus of nearly mature oocyte of S. bip. No granules or network. 2mm. -4.

Figs. 69 and 70. Sections of nuclei of S. min. showing amoeboid figures inside the nuclear membrane. 1.5 mm. -4.

Figs. 71 and 72. Tangential sections of nucleus of older oocytes of S. min. 1.5 mm. -4.

Figs. 73 and 74. Tangential sections of nuclei of oocytes of S. inflata. 1.5 mm.-4.

Further Studies on Reproduction in Sagitta.
















318 N. M. Stevens.

Figs. 75 and 76. Tangential sections of nuclear membrane and network of oocytes of S. el. 1.5 mm. -4.

Figs. 77 and 78. Similar sections from S. decipiens. 1.5 mm.-4.

Figs. 79 and 80. Synizesis stages of young oocyte of S. bip. 1.5 mm. -12.

Fig. 81. Bouquet stage of young oocyte of S. bip. 1.5 mm. -12.

Figs. 82 and 83. First spermatocyte metaphase, S. bip., 9 chromosomes in the eq. plate. 2 mm.-12Xl|.

Fig. 84. First spermatocyte metaphase, 8 chromosomes in the equatorial plate, one (x) precociously divided. Mag. as above.

Fig. 85. First- spermatocyte metaphase showing tetrads. Same mag.

Fig. 86. First spermatocyte prophase showing x divided. Same mag.

Fig. 87. First spermatocyte metaphase of S. inflata, 9 chromosomes in the eq. plate.

Fig. 88. First spermatocyte metaphase of S. infl., one chromosome precociously divided.

Fig. 89. First spermatocyte metaphase, one chromosome (x) larger, bent, and peculiarly attached to the spindle fibers.

Fig. 90. First spermatocyte metaphase showing tetrads and precociously divided chromosomes.

Fig. 91. First spermatocyte metaphase, x dividing again.

Fig. 92. Second spermatocyte prophase, showing x divided.

Fig. 93. First spermatocyte anaphase, oblique view of daughter plate showing longitudinal splitting of chromosomes. 2 mm. -12X2.

Further Studies on Reproduction in Sagitta.



7 6


















\ \ il



^/. 7^. S. cUJ^,




8 9






From the Department of Comparative Anatomy, Harvard Medica' School



Introduction 321

I. Description - 324


Embrj'os of 11 mm. in length 324

Embryos of S mm. in length 332

Embryos of 5 mm. in length 335

Embryos of 3.5 mm. in length 336

Larvae of 15 mm 338

Larvae of 19 mm 339

Adult Opsanus 342

II. Discussion 344

Morphological divisions of the fore-brain roof 344

Epiphysis 347

Posterior commissure 349

Superior commissure 350

Post-velar arch 351

Velum transversum 352

Paraphysis 353

III. Conclusions 354

Bibliography 356

Reference letters 359


In 1901 Minot published an account of the morphology of the pineal region in which it was shown that the characteristic structures included therein had their beginnings in constant well-defined divisions of the fore-brain roof. These subdivisions constitute a succession of arches and intervening projections into the ventricle, and in Acanthias are recognizable in embryos of 11.5 mm.


322 Robert J. Terry.

in length, a stage in which the diencephalon and telencephalon are demarcated. The names proposed for them, taken in order from before backward, are:

1. Paraphysal arch 4. Superior commissure

2. Velum tr&nsversum 5. Epiphysis

3. Post-velar arch * 6. Posterior commissure.

According to the author, The homologues of all these parts, exist probably in all vertebrates."

Other questions suggested and discussed in the paper concern the distinction between the paraphysis and paraphyseal arch, the genesis of the choroid plexus, the significance of the great difference of development of the post- velar arch among animals, the relations of the superior commissure and the position of the posterior commissure with reference to the diencephalon and mesencephalon.

Following the direction of research given in Minot's paper, Dexter ('02) and Warren ('05), in investigations of the avian, amphibian and reptilian brains, have supported the view of the general occurrence of the primitive subdivisions and also have noted especially the distinction between the paraphyseal arch and organ.

The present study ^ was undertaken with the object of extending the same line of inquiry to the teleostean brain. In view of the differences in the mode of formation of the medullary tube in the elasmobranchs and teleosts, misgivings were felt at the start that the early form of the fore-brain roof would necessitate an interpretation of the value of its parts in terms of the elasmobranch type rather than present parts easily identified and directly comparable with the divisions of the shark's brain. To some extent this was true, but the early appearance of the epiphysis, posterior commissure and velum was sufficient to make certain the interpretation of the remaining regions. The method of study adopted

^ This investigation was made during the year 1906-7 while the writer held an Austin Teaching Fellowship in the Department of Histology and Embryology in the Harvard Medical School. For the opportunities there afforded for anatomical research and especially for the encouragement received from the director of the department, Professor Minot, he is deeply giateful.

The Pineal Region in Teleosts.





Fig. 1. Reconstruction of the pineal region of an Opsanus embryo of 11 mm. H.E.C., series no. 121. X 167. A part of the right wall of the ventricle has been removed and the paraphysis has been sectioned in the median plane.

has also helped in the identification of the regions in the smallest embryos. As the starting point, there was selected from the series an embryo in which all the structures pecuUar to the pineal region could be identified beyond question. Each structure was then traced back through smaller and smaller embryos to the stage in which it was first appairent.

The material basis for most of the observations recorded is Opsanus (Batrachus) tau, the toad-fish, a teleost represented in

324 Robert J. Terry.

the Harvard Embryological Collection by an extensive series of sections of embryonic and larval stages. In addition, the brains of specimens half and fully grown were obtained and prepared for dissection and microscopical study. Other fishes included in the research are Salvelinus, Fmidulus, Ameiurus, Lepidosteus, Amia, Acanthias and Petromyzon.

I Description

Opsanus Embryos of 11 mm. in Length

Epiphysis. In embryos of this stage the epiphysis is located in the middle of a depression intervening dorsally between the midbrain and telencephalon (fig. 1, ES) . Its direction is nearly dorsad, corresponding in this respect with the posterior epiphysis of Salmo embryos of 7mm. Hill ('94) found the posterior epiphysis in larval salmon of 13 mm. directed forward, so that it presented dorsal and ventral surfaces. The end of the organ lay close to the epidermis, whereas in Opsanus it reaches a level half-way between the diencephalic roof and the ectoderm. Differentiation of the outward form has already begun and sagittal sections (fig. 2, E) show an ovoid end-vesicle surmounting a cylindrical stalk. The latter expands at its base where it joins the diencephalic roof, and presents a slight curve in the sagittal plane in adaptation to the superior commissure lying just anterior. The epiphysis of Opsanus at the present stage, like that of Clupea (Holt '91), is a solid structure. These teleosts differ, therefore, from the embryos of Salmo in whi ch, as shown by Hill, the epiphysis is hollow. Sections of the solid epiphysis give evidence, however, of a difference in the the structure of the peripheral and central regions, the former appearing deeply, the latter but lightly stained. Moreover the protoplastoi of the peripheral coat is nucleated, whereas that of the central region is non-nucleated. ^ This sort of structure is

2 There are no cell boundaries in the protoplasm of the epiphysis at this stage; the organ has a syncytial structure and the nuclei are all located in the peripheral parts, forming a more or less even layer.

The Pineal Region in Teleosts.


Fig. 2. Median section of the pineal region of an Opsanus embryo of 11mm. H.E.C., series no. 122, section no. 118. X 130.

V E S.

Fig. 3. Transverse section of the pineal region of an Opsanus embryo of 11 mm. H.E.C., series no. 121, section no. 99. X 97.

326 Robert J. Terry.

to be seen in both the end-vesicle and stalk, the clear central portion of the latter extending quite to the base and to the margin of the ventricle. In later stages a cavity forms, taking the place of the region of lightly-staining protoplasm, and the peripheral nucleated layer then becomes a surrounding wall. The thickness of the peripheral layer varies somewhat in different parts of the vesicle, but there is no marked contrast in this respect between the anterior and posterior regions of the layer. In transverse (fig. 3, ES) and in frontal sections, the epiphysis is seen to have a median position. The point of union of the stalk with the roof of the third ventricle is also midway between the intermediate tubercles; that is to say, there is no approximation of one or the other of the tubercles toward the middle line. Regarding the presence of a second epiphyseal organ, the following observation was made. In the series of transverse sections no. 121, a small rounded body was found to the left of the median line anterior to the epiphysis and surmounting the superior commissure (fig. 4, A). In section no. 101 (fig. 5, A) continuity was traced between this body and the roof of the third ventricle anterior to the commissure. Structurally it consists of a peripheral nucleated stratum surrounding a central clear area of protoplasm.

The depression on the dorsal aspect of the brain between the mesencephalon and telencephalon is filled with a loose mesenchymal network. This tissue surrounds the epiphysis and connects the end-vesicle with the membranous roof of the cranium by a dense broad band. Hill states that the end of the epiphysis in Salmo of 13 mm. projects into a mass of undifferentiated tissue lying between it and the epidermis. Blood vessels are present at the sides and at the back of the epiphyseal stalk. One of them is constant and traverses the mesenchyma in the median plane. The pigmentation of the tissues about the epiphysis described by Cattie ('82) appears in the later stages of development of Opsanus. Differentiation of the epidermis over the pineal region does not occur except in connection with the formation of the pit organs of the lateral line system.

Superior Commissure. Fig. 2, taken from a median sagittal section of the pineal region of an Opsanus embryo of 11 mm.,

The Pineal Region in Teleosts.


Fig. 4. Section no. 102 of the same series as fig. 3. X 97.

Fig. 5. Section no. 101 of the same series as fig. 3. X 97.

328 Robert J. Terry.

shows the superior commissure (S) lying immediately anterior to and in contact with the base of the epiphysis. It occupies a position in the outer part of the diencephalic roof, resting ventrally upon a thin ependymal layer (Comp. Acanthias, 22 mm., Minot, '01). The component fibers, non-medullated at this period, pass over the surface of the post-velar arch to the intermediate tubercles.

Posterior Commissure. The posterior commissure is located in that fold of the brain-roof which is recognized as the boundary between the mid-brain and fore-brain. Sagittal sections show it to be folded transversely. The two layers resulting from this disposition of the commissure are continuous at their ventral edges. They extend from side to side and lie, the one dorso-caudad of the other (fig. 2, PC). Rabl-Ruckhard ('82), Haller ('98), Mayser ('82), and other writers, have noted these two divisions in teleosts. Haller has described the commissure in Salmo as composed of a dorsal and a ventral part, the former belonging exclusively to the lobi optici and carrying fibers from lobe to lobe, the latter made up of mixed fibers of the longitudinal tracts. In the toad-fish, a septum of mesenchyma, continuous with the same tissue around the epiphysis and mid-brain, separates the two layers for a considerable distance. The ventral or anterior stratum, an even layer of fibers, is separated from the third ventricle byathick ependyma. It reaches forward as far as the epiphysis, occupying that region of the diencephalic roof called the Schaltstuck or pars intercalaris (Burckhardt) . Some of the fibers pass around the base of the epiphyseal stalk. The dorsal or posterior stratum is related to the wall of the mid-brain. Its thickness varies inversely with that of the underlying epithelium in such manner that, whereas the ectal surface of the stratum is even, the deep surface is irregular. When followed in the dorsal direction, this stratum merges with the. ectoglia layer of the mid-brain roof.

Velum transversum. (Fig. 1, V). This structure, projecting from the roof of the brain into the ventricle some distance in front of the superior commissure, is naturally divisible into two parts. One of these is a low, broad, transverse fold of the roof, the other a large ovoid body, hanging from the middle of the fold by a short pedicle (fig. 2, V). In the embryos of teleosts, ganoids, and elas

The Pineal Region in Teleosts.


mobranchs, the velum consists of a simple reduplication of the epithelium of the brain-roof. Such a velum is found in Opsanus at an earlier period of development and persists as the transverse fold of the present stage (fig. 6, VL). The median lobe, although constituting the greater part of the velum at this time, is a secondary modification of the middle part of the fold. The latter, followed anteriorly, goes over into the paraphysis, posteriorly into the post- velar arch. On either side, a prominent sagittal ridge of the fore-brain roof can be seen extending backward within the


Fig. 6. Section no. 97 of the same series as fig. 3. X 97.

ventricle to the intermediate tubercle (figs. 3, 5). It marks the lateral extent both of the velum and paraphysis. Regarding the contour of the median lobe, one finds the surface broken by fissures of greater or less depth, dividing the whole mass into lobules. The part of the lobe lying behind the level of the pedicle is greater than that anterior. It will be seen by referring to fig. 2, that the lobe projects some distance caudad beneath the postvelar arch. The velum, even at this early stage, affords evidence of being adapted to a secretory function. Sections of the lobe show a peripheral thick epithelial tunic, thrown into folds and sup

330 Robert J. Terry.

ported by mesenchyma rich in blood-vessels. The latter, continuous with vascular channels in the vicinity of the epiphysis, postvelar arch and paraphysis, are close under the velar epithelium. No difference was observed in the thickness of the epithelial coat of the anterior and posterior aspects of the velum as has been described in Acanthias by Minot ('01) and in Acipenser by Kupffer


Post-velar Arch (Fig. 2, PV). The name post-velar arch" was given by Minot ('01) to the curve in the brain roof which lies between the velum and the epiphyseal anlage. This is the part of the diencephalic roof called Zirbelpolster by Burckhardt and the region of the post-paraphysis of Sorensen ('94). Out of this region is developed in the ganoids and some of the bony fishes the evagination called by Goronowitsch ('88) the dorsal sac. In Ospanus, the arch does not expand into a sac, but on the contrary diminishes in extent and finally disappears. In the stage under discussion, the post- velar arch is a small dome-shaped evagination of the diencephalic roof extending sagittally between the superior commissure and velum and going over laterally into the intermediate tubercles. At no time is the evagination so extensive as it is in Salmo where, as Hill ('94) states (p. 242) it bears some resemblance in form to the epiphysis." The arch in Opsanus rises a little higher than the base of the epiphysis but not so high as the paraphyseal fold. Its simple cavity can be followed forward to the pedicle of the velum, on either side of which it becomes continuous with a short blind recess (fig. 6). In transverse sections through the posterior part of the velum, these two small recesses appear as lacunae surrounded by the velar and ependymal epithelium. In structure the post-velar arch consists of a layer of rather thick ependyma. Anteriorly, this continues witli slight increase in the height of its constituent cells into the velar epithelium. Posteriorly, it changes .abruptly to a very thin layer at the level of the superior commissure. The latter lies upon the epithelium of the caudal part of the arch, over which it extends in a lateral direction as far as the intermediate tubercles. Folds of the epithelium are not present, nor are there any diverticula such as Herrick ('91) has found in the walls of the dorsal sac of

The Pineal Region in Teleosts. 331

Lepidosteus. In a word, there is no differentiation of this part of the diencephahc roof; it resembles the simple post-velar arch, found by Minot in Acanthias embryos. The vascularity of the dorsal sac and of the region around it was observed by Balfour ('77) and has been described by subsequent investigators. Blood sinuses of considerable size are present in Opsanus in the mesenchyma dorsad of the post-velar arch, and are connected with the vessels of the epiphysis and velum.

Paraphysis. (Fig. 2, P). Gaupp's ('97) and Studnicka's ('05) reviews of the extensive literature of the pineal region contain few references to the paraphj^sis in teleosts. The question of its presence in the class has received little attention. Three years after Selenka's ('90) discovery of the organ, Burckhardt found a paraphyseal rudiment in the trout. Later, Studnicka ('95) described a paraphysis in two teleosts, Lophius and Anguilla. In adult Lophius the paraphysis appears as an evagination of the brain wall in front of the velum. Its occurrence is not constant. In young Anguilla the organ is a relatively large, thinwalled sac connected by a narrow base with the brain. In 1905 the same investigator described the paraphysis in two other bony fishes, Cepola rubescens, in which it appears as a conical sac in front of the velum, and Belone acus, where a rudimentary paraphysis is indicated by uneveness of the lamina supraneuroporica. The paraphysis of Opsanus is a simple transverse fold of the roof of the telencephalon just anterior to the velum (fig. 1, P). Its walls are composed of an epithelium differing markedly in its greater thickness and staining properties from the tela of the fore-brain. In sagittal sections the paraphysis appears tentshaped, with an anterior oblique and a posterior perpendicular wall going over into the velar fold. Of these two walls the anterior is so newhat thicker. Followed in the lateral direction the the organ is clearly defined to the same extent as the velar fold, that is, as far as the sagittal ridge of the fore-brain roof, described on p. 329. The simple cavity of the paraphysis presents no diverticula and communicates freely with the telencephalic ventricle.


Robert J. Terry.

Opsanus Embryos 8 mm. in Length

Epiphysis. The dorso-caudal direction of the -epiphysis at this stage brings its posterior surface into contact with the midbrain (fig. 7, E). The organ is relatively shorter now and in form




Fig. 7. Median section of the pineal region of an Opsanus embryo of 8 mm. H.E C, series no. 116, section no. 78. X 116.8.

Fig. 8. Transverse section of an Opsanus embryo of 8 mm. H.E.C., series no. 115, section no. 36. X 116.8.

Fig. 9. Transverse section of the pineal region of a Salvelinus embryo of 10 mm. H.E.C., series no. 455, section no. 42. X 84.

somewhat oval. Its structure is essentially as in the 11 mm. embryo and calls for no particular description. Just to the left of the base of the epiphysis there can be seen in transverse sections of the diencephalic roof an oval, compacted mass of nuclei pro

The Pineal Region in Teleosts. 333

jecting above the surface. The center of the mass consists of clear protoplasm, the whole structure bearing a close resemblance •to the epiphysis of the 3.5 mm. embryo (fig. 8, A). This small bud is evidently the beginning of the anterior rudimentary epiphysis noted in the 11 mm. embryo. The examination of Salvelinus embryos, resulted in, the discovery of two epiphyses as was anticipated in view of their presence in Salmo. They are represented in transverse section in fig. 9. It will be seen that these bodies are of unequal size, that the larger one is located in the median plane and the smaller one to the left. A cavity is present in the larger epiphysis.

Hill and other investigators have remarked upon the absence of mesodermal tissue between the epiphysis and ectoderm in very young teleost embryos. This condition was observed in the present study in the embryos^of Salvelinus (fig. 10), Fundulus, and also Amia (fig. 11). It is not so however in Opsanus embryos, for there is always present between the brain and the epidermis a continuous layer of mesenchyma. Within this tissue the end of the epiphj^sis is to be seen (fig. 7, E). It is worthy of note that, whereas the end vesicle of this organ is pressed closely against the epidermis in the above named fishes, it is farther removed in Opsanus. A large blood sinus, lying between the superior commissure and the mesenchymal layer, is in contact with the anterior surface of the epiphysis.

Superior Commissure. It is at this stage of development that the superior commissure first appears as a clearly defined bundle of fibers. Its position relative to the epiphysis is the same as in the 11 mm. embryo. Sagittal sections show that the fibers are incompletely separated by a range of nuclei into dorsal and ventral groups.^ In fig. 7, it will be observed that the commissure lies upon the ependyma of the post-velar arch.

Posterior Commissure. Excepting that the fibers of the commissure are spread to a relatively greater extent upon the epi ' Transverse sections reveal an intermingling of certain of the ependymal cella with fibers of the commissure, a condition interesting in connection with Mrs. Gage's ('95) observation that, in Diemyctyius, processes of the endymal cells traverse the commissure.


Robert J. Terry.

^ P A

Fig. 10. Median section of the pineal region of a Salvelinus embryo of 10 mm. H.E.C., series no. 458, section no. 60. X 84.

Fig. 11. Median section of the pineal region of an Amia embryo of 10 mm. H.E.C., series no. 12, section no. 91. X 84.

Fig. 12. Median section of the fore-brain of an Opsanus embryo of 5 mm. H.E.C., series no. 110, section no. 55. X 115.

physeal stalk, the conditions described for the 11 mm. stage obtain in the specimens under discussion.

Post-velar Arch. Compared with the 11 mm. embryo, the postvelar arch is relatively larger. The evagination is more pronounced and there is presented in consequence a considerable posterior surface over which the superior commissure is spread. This relation is very striking in Amia (fig. 11, S).

The Pineal Region in Teleosts. 335

Velum Transversum. The transverse fold and median lobe are both present, the extent and relations of the former being the same as in the 11 mm. embryo. Differentiation in the form of the middle lobe has not begim; its surface is smooth and it is connected directly with the transverse fold.

Paraphysis. There is no paraphyseal evagination at this stage. In its place the brain-roof forms a low arch or dome, sharply defined laterally and anteriorly. This paraphyseal arch rises somewhat above the general level of the fore-brain tela, with which it contrasts in its thicker and more deeply staining epithelium. Caudally the arch goes over into the transverse fold of the velum.

Opsanus Embryos 5 mm. in Length

Epiphysis. Instead of the more or less conical, projecting epiphysis observed in the preceding stages, there is now present a simple, arch-like evagination of the diencephalic roof (fig. 12, EA). The middle of this arch is differentiated both in form and structure from the rest (fig. 12, E). It is raised slightly above the general level and presents in sections the same structural divisions, peripheral and central, as were found in the epiphysis of the older embryos. The arrangement of the nuclei in a superficial layer is interesting in connection with Eycleshymer's ('92) observation of the migration of nuclei in the epiphyseal evagination of Amblystoma embryos and of a somewhat similar phenomenon in the optic vesicles. The epiphyseal arch occupies the median plane, its summit lying within the mesenchyma which now fills the interval between the brain-roof and the epidermis. The latter presents no special features in this region. A second epiphj^seal outgrowth was not observed and so it appears that this smaller bud is a later development than the principal epiphyseal organ. In Acanthias, a projection into the ventricle, seen in sagittal sections, marks the site of the future superior commissure and limits the epiphyseal arch anteriorly. Since this projection is not well defined in Opsanus, the arch goes over without sharp limit into the post-velar region.

336 Robert J. Terry.

Superior Commissure. This tract is not present.

Posterior Commissure. In an embryo of 6.5 mm. the posterior commissure appears in sagittal sections (fig. 13, PC) as a large clear area in the ectoglia of the brain-roof, limited within the deep fold between the fore- and mid-brain. Throughout this area a network of fine processes, continuous with the ependyma, can be seen. The dorsal surface of the commissure presents an indentation, the beginning of the division into the two parts seen in later stages. This indentation is found in Amia (fig. 11, P.C) 'and in the trout as represented in Kapffer's ('06) fig. 150. This investigator says, (p. 131): Sowohl die Commissura anterior wie die posterior lassen bei der Forelle anfanglich zwei scharf geschiedene Portionen erkennen." In Opsanus the anterior part reaches the epiphysis and lies in contact with its posterior surface.

In embryos of 5 mm. the commissure is stretched out in the superficial or ectoglia layer of the brain-roof, caudad of the epiphyseal arch (fig. 12, P C). Haller ('98) has seen a similar disposition in selachians and remarks that it is transitory in the teleosts. The extended posterior commissure has been observed in other forms, as, for example, Ammocoetes (See Kupffer, '06, fig. 47), and it has been found in the present investigation in Fundulus embryos of 7 mm. The fore-brain in front of the epiphysis presents in sagittal sections an angular bend dividing it into a longer posterior and a shorter anterior segment (fig. 12). Behind the bend a large blood sinus lies in contact with the roof which at this point projects into the ventricle. This is the beginning of the velum transversum. The epithelial roof in front of the angle goes over anteriorly by a slight curve into the thicker terminal wall of the ventricle. This curve is to be compared with Minot's paraphyseal arch.

Opsanus Embryos S.5 mm. in Length

Sagittal sections show the brain roof considerably thickened in the epiphyseal region where a slight folding represents the beginning of the epiphyseal arch (fig. 14, EA). A short distance

The Pineal Region in Teleosts.


p c

Fig. 13. Median section of the pineal region of Opsanus. Length 6.5 mm. H.E.C., series no. 113, section no. 67. X 400.

Fig. 14. Median section of the pineal region of an embryo Opsanus. Length 3.5 mm. H.E.C., series no. 107, section no. 50. X 500.

Fig. 15. Pineal region of an Opsanus embryo of 15 mm. Median section. H.E.C., series no. 1183, section no. 145. X 84.


338 Robert J. Terry.

caudad, the site of the posterior commissure is indicated by a slight projection of the roof into the ventricle and the presence of a small elongate area of clear protoplasm (Comp. Hill, '99). Between the epiphyseal arch and the site of the posterior commissure, there appears a segment of the brain-roof not seen in the older embryos as a separate region. Its relations identify it as the pars intercalaris (fig. 14, PI). The segment of the roof anterior to the epiphyseal arch is thicker than that which is posterior; also it is relatively and absolutely thicker than the corresponding part in older stages. It presents, as in the 5 mm. embryo, a flexure in the sagittal plane, but exhibits no trace of the subdivisions evident at that stage.

Opsanus Larvae 15 mm., in Length


Epiphysis. In fig. 15, taken from a section a little to one side of the median plane, the epiphysis is seen to be inclined well forward over the paraphyseal region of the fore-brain. The epidermis is now folded to form a deep, wide groove extending transversely between the lines of the supraorbital pit organg'(fig. 15C,). Although standing in a position posterior to the level of groove, the epiphysis is inclined forward so that its axis, if prolonged, would meet the groove. This topographical relation is the only one which was observed between the two organs. The small epiphyseal bud, noted in younger stages, is not present. A mesenchymal layer, in which the bony cranial roof is to form, now covers the great dorsal fontanelle of the chondrocranium. Over a considerable area of this layer are attached strands of connective tissue which radiate from the end of the epiphysis.

Superior Commissure. Owing to the disappearance of the postvelar arch, the superior commissure now lies between the velum and the epiphyseal stalk.

Posterior Commissure. The two divisions are even more sharply limited toward each other than is the case in the earlier stages. The mesenchymal septum is a well defined fold in a membrane which, followed posteriorly, covers the mid-brain, and anteriorly joins with the connective tissue over the diencephalon. The

The Pineal Region in Teleosts. 339

anterior part of the commissure rests upon the intercalated division of the diencephalic roof, reaching forward to the base of the epiphysis. The posterior division forms a superficial fiber layer of the mid-brain in this region.

Velum Transversum. The middle lobe of the velum is relatively larger than in the preceding stage. In its growth backward it has invaded the region of the post- velar arch and has come therefore to lie below the epiphyseal stalk and the intercalated part of the diencephalon. The lobules comprising it are numerous and each includes a blood sinus whose walls are closely related with the epithelial covering of the velum.

Post-velar Arch. The reduction of this region, already referred to, goes hand in hand with the backward growth of the velum. The latter appears to have taken up and included the epithelium of the arch. That this process takes place was however not proved, for the epithelium of the arch presents no characters by which it can be distinguished from that of the pedicle and base of the velun: .

Paraphysis. The thick epithelial coat of the velum passes over into the roof of the telencephalon for a short distance, giving place to the flat epithelium of the tela. Where this change occurs the roof is elevated into a slight but conspicuous transverse fold lying just beneath the end of the epiphysis. This rudimentary paraphysis presents a simple structure, consisting, as in the preceding stages, of a rather high epithelium, resting upon a thin stratum of connective tissue which contains but few vessels.

Opsanus Larvae of 19 mm. in Length

Epiphysis. The forward inclination of the epiphysis is more marked than in the preceding stage. The elongated stalk, bent over the superior commissure, is continued into the now much enlarged end-vesicle. In the latter a cavity is to be seen for the first time. This space, occupying the region which in earlier stages was characterized by the presence of clear, non-nucleated protoplasm, is traversed by fine fibrillae continuous with the surrounding walls (fig. 16). The conclusion that these fibrils


Robert J. Terry

from a protoplasmic syncytium, derived from the central clear protoplasm and continuous with the walls of the epiphysis rests on the following evidence. The fibrils are first seen at the time of the appearance of the central cavity, not after it is formed. The cavity of the stalk is completed after that of the vesicle and a syncytium of fibrils is seen as the central axis of protoplasm disappears in the process of cavity formation. In staining reactions and in structure the fibrils agree with the central protoplasmic mass, except at the periphery where they resemble, in

Fig. 16. Pineal region in median section. Opsanus of 19 mm. H.E.C., series no. 1188, section no. 203. X 84.

these respects, the surrounding walls. There is, therefore, no line of demarcation between the net -work and the protoplasm of the walls. The latter have undergone no differentiation at this time; there are as yet no cell boundaries. Bone has appeared in the cranial roof and to its under side are fixed the connective tissue bundles that radiate from the tip of the epiphysis. A commissural canal has replaced the transverse groove of the integument, seen in the 15 mm. larva, lying somewhat further in advance of the end vesicle of the epiphysis than did the groove (fig. 16 C).

The Pineal Region in Teleosts.


Fig. 17. Dorsal aspect of the brain of adult Opsanus. X 7.

Superior Coinmissure. This remains as in the preceding stage.

Posterior Commissure. The fibers of the commissure are now separated into bundles by deUcate septa continuous with the underlying endyma.

Velum Transversum. The pedicle of the median lobe now extends backward to the level of the superior commissure. As a consequence of the backward growth of the velum the dorsal part of the diencephalic cavity has been reduced to a mere cleft extending transversely between the intermediate tubercles. The epithelium of the lobules is everywhere elevated into tufts supported by vascular connective tissue.

342 Robert J. Terry.

Paraphysis. Velar epithelium extends forward toward theparaphyseal region where there is to be seen a slight transverse fold of the brain-roof. This fold, which appears to be the remains of the paraphysis, forms the anterior end of a median longitudinal groove of the tela, evident in transverse sections. The groove is the beginning of a deep median invagination which in the adult toad-fish separates two diverticula of the caudal end of the telencephalic ventricle.

Adult Opsanus

Epiphysis. The fully developed epiphysis presents a form not

uncommon among the teleosts (fig. 17). It consists of an oval

end-vesicle terminating a long slender stalk, the whole structure

being directed cephalad in the median plane and suspended in

the meninges between the fore-brain tela and the cranial roof.

The stalk, which is slightly fusiform, measures 6 mm. in length

and 0.08 mm. in its greatest diameter. The end-vesicle measures

0.6 mm. in length and 0.38 mm. in greatest breadth. There is no

angle between the stalk and vesicle but a gentle curve extends

throughout the length of the organ. The cavity, present in both

stalk and vesicle, does not communicate with the ventricle. It is

traversed by protoplasmic processes forming a wide mesh-work

from wall to wall. An artery and vein are associated with the

epiphysis throughout its whole extent. In the connective tissue

along these vessels and around the distal half of. the organ black

pigment is present in considerable amount. The commissural

canal of the lateral line system is now far anterior to the end of

the epiphysis and there seems to be no further relation between

these organs. A parietal foramen or fossa of the osseous cranial

roof is not present.

Superior Commissure. There is no change in the structure and relations of this commissure from what was last observed.

Posterior Commissure. The two divisions of the posterior commissure are still recognizable; the intervening connective tissue septum now is less distinct. The anterior division contains some longitudinal fibers which extend in a thin layer around the base of the epiphyseal stalk and on to the intermediate tubercles.

The Pineal Region in Teleosts.


Fig. 18. Transverse section through the pineal region of an adult Opsanus. W.U.C, series no. 13, section no. 517. X 21.

Velum Tnansversum. The appearance of the median lobe of the velum is much the same as that of the 19 mm. larva. The covering epithelium consists of a single layer of long club-shaped cells, grouped into prominent tufts. The larger free end of the cell is often irregular, fringed or lacerated, sometimes rounded and regular. Neither cilia nor cuticulae are present, but long shreds of some substance extend from the ends of the cells into the ventricle there to become continuous with a coagulum which is always found close about the velum. In the coagulum are rounded bodies and masses which are stained like the clubbed ends of the cells. The nucleus is located near the free end of the cell and, between it and the base, granules are sometimes to be seen which stain deeply. The epithehum rests upon a rather thick reticular membrane, separating it from numerous underlying blood-vessels, i

344 Robert J. Terry.

Post-velar Arch. This structure can no longer be said to exist, its place being occupied by the velum.

Paraphysis. There is no trace of a paraphyseal differentiation of the fore-brain tela. The latter, in its posterior part, has, however, been changed from the simple dome-form of the larval stages by the appearance of bilateral diverticula. In fig. 18, it will be seen that the ventricle is continuous from side to side, but that the tela is depressed in the mid-line to form a broad, shallow groove opposite the stalk of the epiphysis. Fig. 19, which is taken from a section at the level of the velar pedicle, shows a median partition between the two wide diverticula of the forebrain ventricle. This septum extends from the bottom of the groove, noticed in the previous figure, and contains, besides some large blood-vessels going to and away from the velum, the stalk of the epiphysis.

II Discussion

Morphological Divisions of the Fore-brain Roof

Burckhardt ('94, a & b) recognized in types of all vertebrate classes the presence and constant relations of the following structures :

Paraphysis Epiphysis

Velum Transversum Pars Intercalaris

Zirbelpolster Posterior Commissure. Superior Commissure

The forecast of these structures in the embryonic brain has been described by Minot ('01) who, as mentioned in the beginning of this paper, found that the pineal region of Acanthias at an early stage presented six constant divisions. The divisions, in the form of arch-like evaginations and alternating depressions into the ventricle, were named by this author according to their subsequent differentiation. Further observations upon the pineal region of embryos of other animals led to the belief that they were fundamental and that homologous parts might be found in all vertebrates.

The Pineal Region in Teleosts.



Fig. 19. Section no. 542 of the same series as in figure 18. X 21.

Fig. 20. Median section of the pineal region of Ameiurus. Length 10 mm. H.E.C., series no. 388, constructed from sections nos. 89-92. X 290.

346 Robert J. Terry.

Dexter ('02),' in his study of the development of the paraphysis in the fowl, has identified and figured subdivisions of the forebrain roof, comparable with those described by Minot ; and Warren ('05), who has found the arches in Necturus, has shown a very striking resemblance between the embryonic pineal regions of this amphibian and Acanthias.

In regard to teleosts it is probable that Minot's subdivisions of the embryonic fore-brain are present throughout the class. In Kupffer's ('06) figure of a trout embryo of 53 days, the epiphysis is represented as an elongate evagination, but the other subdivisions of the the roof are shown to have the form of arches and intervening folds. A post-velar arch and the invaginations of the velum and the posterior commissure are shown in Hill's ('94) figure of Salmo fontinalis of 42 days. Opsanus embryos present the six snjbdivisions and also a pars intercalaris. The intercalated part first appears as a distinct segment in embryos of 3.5 mm. lying between the posterior commissure and the epiphyseal arch. As the result of forward growth of the commissure over the intercalated part, the latter as such disappears; that is to say, it no longer remains a segment interposed between the posterior commissure and the epiphysis. It can be recognized, however, in all later stages in the stretch of ependyma underlying the anterior division of the posterior commissure. The fundamental divisions appear less clearly defined than they do in Acanthias and, moreover, they are not all evident at the same time as is the case in the dog-fish. In Opsanus, the epiphyseal arch is present in the smallest embryo studied as are also the posterior commissure and intercalated part; all three can be seen during a brief period (embryos of 3.5 to 5 mm.). By the time the posterior commissure has grown over the pars intercalaris (embryo of 6.5 mm.), the velar invagination is seen. The epiphyseal arch disappears with the formation of the main epiphysis, the post-velar and paraphyseal arches and the superior commissure presenting themselves at this time (embryos of 8 mm.).

As to the relation which these divisions bear to the neuromeres, no direct evidence was obtained in the present study. Kupffer ('06) has identified the region of Burckhardt's Zirbelpolster (Minot's

The Pineal Region in Teleosts. 347

post-velar arch) as the median dorsal part of his parencephalic segment which in turn, he derives from the second neuromere. The same author finds the intercalated part to be the roof of his synencephalic segment, derived from the third neuromere. Regarding the relation of the epiphysis to these segments, Kupffer ('06) says: ^'Es tritt namlich die Commissura posterior, die man als dorsale Grenzmarke zwischen dem Vorder und Mittelhirne festzuhalten hat, nicht zwischen den Segmenten p and se auf — hier entsteht die Epiphyse" (p. 175). The segments J9 and se are the parencephalic and synencephalic segments. The paraphysis is the product of the telencephalic segment derived from the first neuromere, and the velum marks the dorsal boundary between the telencephalic and parencephalic segments. According to Johnston ('05) the second neuromere gives rise to the optic vesicles; from its narrow dorsal part is formed the velum. The epiphysis, according to this author, belongs to the third neuromere.


The two epiphyseal outgrowths of Opsanus differ in their early form and relations from those of Salmo, Coregonus and the other teleosts which Hill ('91, '94) studied. In the first place they are not true evaginations but solid outgrowths of the brain-roof. As in the case of the solid epiphysis of Clupea, a cavity traversed by fibers is later formed in the main organ. Holt ('91) regarded the fibers as a coagulum and found no eye-like structure in the epiphysis, but Studnicka ('05), who looks upon the two walls of the pineal vesicle of Peftomyzon as retina and pellucida, hints at a comparison of these syncytial nets with the remains of a corpus vitreum of the parietal organ. The network in Opsanus is derived, as already shown, from the lightly staining protoplasm occupying the axis of the epiphysis, and it therefore cannot be considered a coagulum of some possible secretion of the walls of the organ. A secretion discharged into the cavity of the epiphysis of Opsanus would have no outlet. While the organ is moderately vascular it does not conform in structure with any of the ductless glands. On the other hand there is little evidence

348 • Robert J. Terry.

in support of the theory of the epiphysis of this teleost being an ocular organ, either rudimentary or degenerate. Whatever may be the significance of the syncytial network, its formation in Opsanus goes hand in hand with the development of the epiphyseal cavity, a. process analogous with that which produces the cavities of the central nervous system in the teleosts.

As to the fundamental question of the independence of origin of the two epiphyseal vesicles, the evidence afforded by Hill's material is not convincing. In Opsanus the two outgrowths are entirely separate and there is no question of one of them being developed from the other. The smaller bud appears later than the definitive epiphysis and arises from the diencephalic roof. It lies at first to the left and a little in advance of the main organ but secondarily comes into connection with the superior commissure and the post- velar arch. This forward migration recalls the shifting of the anterior vesicle of Amia and Lacerta.

The end of the epiphysis, from the time it is first seen in Opsanus, is closely related with the overlying tissues. In the smallest embryo the cranial mesenchyma extends between the epiphysis and ectoderm. In larger embryos and in larvae, strands of this tissue and finally connective tissue fibers proceed from the epiphysis to the roof of the cranium. There is no evidence to show that the function of these connecting strands is anything more than a passive one in fixing the epiphysis, but the early appearance of a bond between the end-vesicle and the overlying parts is suggestive of some other relation. Dean '95) has expressed the opinion that the epiphysis of fishes is connected with the innervation of the sensory canals of the head, at the same time opposing the theory of its relation to a median eye. In the present study attention was specially directed to a search for evidence in support in this view, but there was nothing observed which pointed to a relation between the epiphysis and the lateral line system, beyond the fact that the former is at one time directed toward the supraorbital commissural canal and is approximated rather closely to it.

The forward inclination of the epiphysis, occurring at the time when the roof of the cranium is first laid out in the mesenchyma.

The Pineal Region in Teleosts. 349

points to the influence of Cranial development on the position of this organ.

The statement made by Goette ('75) that the pineal body in Bombinator arises from a bridge which connects the brain and ectoderm has repeatedly been reaffirmed and denied. Van Wijhe ('83) and Hoffmann ('84) have presented evidence in support of the connection; Mihalkovics ('77), Balfour ('85), and later investigators assert that the epidermal bridge has no existence. Locy ('93) claims that the beginning of the epiphysis in the shark can be seen in the medullary plate. No recent investigator has found a continuity between the epiphysis and ectoderm after the formation of the medullary tube, although a close relationship has often been observed between the end of the pineal body and the outer germ layer. This condition and the absence of any intervening mesoderm led Hoffman to believe that the epiphyseal anlage was laid out before mesodermal formation had commenced. In Opsanus of 3.5 mm., when the epiphysis is just discernible, a mesenchymal stratum stands between it and the ectoderm and there is no evidence of continuity between these parts.

Posterior Commissure

It was shown in the descriptive part of this paper that the posterior commissure, arising in the ectoglia layer of the brain-roof, is, at first, located posteriorly to a pars intercalaris, that it spreads forward over this region and finally becomes included within the fold that intervenes between the mid-brain and diencephalon. Moreover, it was found that the commissure in the older embryos, in all stages, and in the adult fish, presents two distinct divisions, anterior and posterior, separated by a connective tissue septum. This mode of development has been observed in other bony fishes, to which reference has been made above (p. 336), and its division into strata has been described and represented in figures of the teleostean brain. It appears, therefore, that a type of posterior commissure appears among bony fishes, characterized by the presence -of two strata of fibers separated by a partition. The commissure lies neither altogether in the wall

350 Robert J. Terry.

of the mid-brain nor in the diencephalon, but is so situated that its posterior layer stands in connection with the former, while its anterior stratum is spread out in the roof of the latter caudad to the epiphysis. In form and position, therefore, it differs from the corresponding tract of the elasmobranchs, in which fishes it has been shown to be a fiber bundle associated wholly with the midbrain (Ehlers, 78; Edinger, '99; Minot, '01). With this difference in the commissures of the two classes of fishes is correlated the difference presented by the pars intercalaris, which is extensive in the teleosts, small or absent in the elasmobranchs.

Regarding a pineal nerve, the evidence was insufficient to warrant the statement that such a structure is present in Opsanus. There is that close relationship between the posterior commissure and the epiphysis which has been described by Edinger ('99) in Scyllium and sturgeon, by Kupffer ('93, '06) in the trout, and shown by Dean ('96) in his figure of Amia (Comp. fig. 11). In the smaller embryos of Opsanus the posterior commissure extends further upon the base of the epiphysis than it does in the adult, recalling the relation of a tractus pinealis. Many fibers were followed to the base of the epiphyseal stalk but their terminal relations were not discovered.

Superior Commissure

The observation made by Osborn ('84) on the position of the superior commissure in front of the epiphysis has been many times confirmed and in recent years emphasized by Minot ('01) and Dexter ('02). " Cameron ('04), however, appears unwilling to concede that the commissure should be considered as closely related to the epiphysis. He states that in all vertebrates it is situated behind the root of the choroid plexus of the third ventricle. The definition of the position of the commissure given by Osborn, Minot and Dexter implies a topographical relation to a constant organ. The structural relationship between the commissure and the epiphysis in the ganoids described by Herrick ('91) and by Eycleshymer and Davis ('97) is another reason, beyond that of mere topography ,for associating the two parts. Yet, the position

The Pineal Region in Teleosts. 351


of this fiber bundle immediately in front of the epiphysis does not always obtain, as Cameron points out, and a definition of its relations applicable to all cases must wait until more observations have been made. One relation appears to be constant, namely, that the commissure is associated with the post-velar arch. This relationship is shown in median sagittal sections of the brain where it appears that the diencephalic roof in front of the epiphysis is composed of two strata, the cellular ependyma and the narrow fiber layer of the superior commissure. When these are followed in a direction away from the median plane they are found to pass over to the intermediate tubercles. Here all three of the fundamental layers of the brain wall, as recognized by His ('89) and Minot ('92, '03) can be seen, the Randschleier or ectoglia layer being formed by the fibers of the superior commissure. The latter may therefore be regarded as the ectoglia layer of the postvelar arch, small and limited to the posterior aspect of this region in Opsanus, but more extensive in Amia and Petromyzon. The origin and early disposition of the posterior commissure would warrant the same conclusion respecting its relation in the brain wall.

Post-velar Arch

The post-velar arch reaches its highest development in Opsanus when the embryos are about 8 mm. long, and subsequently disappears, probably by incorporation with the velum. The epithelium of the arch is like that of the median lobe of the velum and probably functions as a secreting surface. Hill ('94) and Leydig ('96) have found that this part of the brain-roof in teleosts is specially differentiated to form ridges and secondary folds of the epithelium resting upon a connective tissue foundation, in some cases vascular. Leydig states that in the trout the cells are higher toward the summit of the arch. The great expansion of the post-velar region in the ganoids is well known through the writings of Balfour, Huxley, Wiedersheim, Goronowitsch, Wilder and others in recent years. In Amia and Lepidosteus, Kingsbury ('97) found the dorsal sac, velum and metaplexus lined with an

352 Robert J. Terry.

ependymal epithelium which appears to consist of secreting cells, and Herrick ('91) describes the pouch of the diatela as possesed of a wall composed of a single row of cells with long cilia or flagella. It appears, therefore, that in these ganoids the epithelium of the arch is especially modified and lines a great evagination of the diencephalic roof. Regarding the elasmobranchs, Minot ('01)

says the post-velar arch remains small, hence the

velum seems to arise later very close to the mouth of the epiphysis." The connection of the smaller epiphyseal bud with the postvelar arch by forward shifting brings up the question of its possible relationship with those large outgrowths of this region which have been observed by Schauinsland (SeeKupffer, '06) inCallorhynchus,


by Gierse ('04) in Cyclothone, and by Handrick (See Studnicka, ,'05) in Argyropelecus. Commenting on the latter, Studnicka, ('05) says: Der Fall ist sehr wichtig, da er zeigt, dass der de norma breitere Dorsalsack unter Umstanden sich in ein enges schlauchformiges Gebilde verwandeln kann. Dieselbe Erscheinung kann man bekanntlich auch beiderParaphysebeobachten; auch diese tritt einmal als ein enger Schlauch, ein anderes Mai wieder in der Gestalt eines breiten Sackes (Paraphysealbogen — Sedgwick Minot) auf."

Velurn Transversum

The velum transversum of the teleosts, according to the current descriptions, consists of a simple transverse fold of the fore-brain roof, having smooth surfaces and a free ventral margin. RablRuckhard ('82) regards the velum as the starting point, phylogenetically, of the choroid plexus, although he found no differentiation in this direction in the bony fishes. In them this organ seems to be less advanced in its development than it is among the selachians, where it has been shown that projections are formed on either side which are regarded by Minot ('01) as the anlages of the choroid plexuses of the lateral ventricles. In Acanthias a superficial coat appears upon the ependyma of the anterior surface of the velum, the nature of which is uncertain, but as

The Pineal Region in Teleosts. 353

Minot says ('01, p. 91), . . . suggests . . . the formation of secretory spherules." Gentes ('08) regards the -velum of torpedo as a true choroid plexus. The enormous development of the velum of Opsanus is a striking peculiarit}^ of the brain of this bony fish. The more important structural characters which have been described on p. 342, seen in all the adult and larval specimens, are those belonging to the true choroid plexus and establish the velum of the toad-fish as such an organ.

Among the teleosts possessing a rudimentary velum, Belone acus has been cited by Studnicka ('05) . In the brain of the Ameiurus embryo, shown in fig. 20, the velum is rudimentary and there is a further resemblance to Belone in the peculiar form of the epiphysis. This consists of a large flattened end-vesicle supported upon a rather slender and tortuous stalk. As to the significance of the primitive velum the view recently expressed by Johnston ('09) is interesting, namely, that the velar invagination begins early on account of the withdrawal of material from the alar plate to form the optic vesicle." With such a relationship between these structures, some variation in the development of the optic vesicle might be expected in those animals where the velum is rudimentary or absent.


The occurrence of a paraphysis in a number of teleosts, as discovered by Burckhardt and Studnicka, lends support to the interpretation that has been given to the fold of the fore-brain roof in Opsanus. The higher epithelium of the fold differentiates this rudimentary organ from the tela anterior to it, and its position, immediately in front of the velum, corresponds with the location of the paraphysis in all forms in which it has been observed. This relation of paraphysis and velum or choroid plexus is responsible for the identity of the former organ remaining hidden for so long a time. In Opsanus the paraphyseal fold is clearly no part of the velum for it appears after the latter has been formed, differs from it in structure and has only a brief existence.

The part of the embryonic fore-brain roof which is to give rise


354 Robert J. Terry.

to the paraphysis has usually the form of a dome. Since Minot ('01) named it the paraphyseal arch and demonstrated that the paraphysis arises from it, its presence has been recognized in all the vertebrate classes. (Comp. Dexter, '02; Warren,* '05; Kerr, '03; Johnston, '09).'

Ill Conclusions

1. A. The six morphological divisions of the fore-brain roof recognized by Minot are present in Opsanus, and probably also in Salmo, Salvelinus and Amia.

B. These divisions, in the form of arches and alternating projections into the ventricle, are not all present at the same time in Opsanus as in Acanthias.

2. A. A pars intercalaris is to be seen as an independent segment of the brain-roof between the posterior commissure and the epiphysis in embryos of Opsanus 3.5 mm. long.

B. In the adult it remains as a thick stretch of ependyma, supporting the anterior stratum of the posterior commissure.

3. A. There are two epiphyses connected with the brain of the toad-fish, one of them being a mere rudiment.

B. The main epiphysis lies in the mid-line and develops a stalk and end-vesicle.

C. The rudimentary organ makes its appearance some time after the definitive epiphysis is differentiated, lies at first to the left and a little in advance of it and subsequently migrates forward into the region of the post-velar arch.

D. The origin of these organs is entirely independent, the one from the other.

E. Both epiphyses are originally solid outgrowths, the main organ springing from the epiphyseal arch, the rudiment from the diencephalic roof behind the superior commissure after the disappearance of the epiphyseal arch.

F. The cavity which develops secondarily in the end-vesicle

Warren, John. On the paraphysis and pineal region in Lacerta and Chrysemis marginata. Assoc. Am. Anat. 25th Session. Boston, December 30, 1909.

The Pineal Region in Teleosts. 355

and stalk of the main epiphysis includes a weak meshwork of protoplasmic processes continuous with the surrounding walls.

G. Continuity between the epiphysis and ectoderm was not observed; the fibers that extend between the end- vesicle and the overlying parts are mesenchymal in origin.

H. There is no parietal foramen and no differentiation of the epidermis of the epiphyseal region in Opsanus.

I. A nerve connection between the epiphysis and lateral line system does not obtain.

J. A pineal nerve was not discovered.

K. There are two epiphyses in Salvelinus, the chief organ being median in position, the subordinate outgrowth to the left of the former.

4. A. The posterior commissure of Opsanus has its origin in the ectoglia of the brain-roof.

B. In teleosts the commissure is divided into two parts, the one associated with the mid-brain, the other with the intercalated part of the diencephalon.

5. A. The superior commissure lies, in the toad-fish, immediately anterior to the base of the epiphysis.

B. It arises in the ectoglia of the diencephalic roof, retains the relation of an ectoglia layer in the brain-wall of the adult and may be regarded as an incomplete ectoglia stratum of the post-velar arch.

6. A. The post- velar arch attains its maximum extent in the embryos of Opsanus and early begins to diminish and finally disappears.

B. Its place is taken by the velum in its backward growth.

C. Its epithelium probably becomes incorporated with the velum.

7. A. In the development of the velum of Opsanus, a transverse fold and a median lobe are formed, the latter differentiating as a choroid plexus of the fore-brain ventricle.

B. The velum of Amieurus embryos is rudimentary.

8. A rudimentary paraphyseal organ is developed from the paraphyseal arch, appearing later than the epiphysis and disappearing during the early larval life of the toad-fish.

356 Robert J. Terry.


Balfotjb. The development of the elasmobranch fishes. Journ. Anat. and Phys.,

1877. vol. 2.

1885. Works. Edited by Foster and Sedgwick. Lond.

Bukckhardt. Zur vergleichenden Anatomie des Vorderhirns bei Fischen. Anat. 1893. Anz., vol. 9.

1894a. Die Homologien des Zwischenhirndaches und ihre Bedeutung fiir die Morphologie des Hirns bei niederen Vertebraten. Anat. Anz., Jahrg. 9.

1894b. Der Bauplan des Wirbeltiergehirns. Morph. Arbeit., vol. 4.

Cameron. On the presence and significance of the superior commissure through1904. out the vertebrates. Journ. Anat. and Phys., vol. 38.

Cattie. Recherches sur le glande pineale (epiphysis cerebri) des plagiostomes, 1882. des ganoides et des teleostiens. Arch, de Biol., vol. 3.

Dean. Fishes living, and fossil. New York. 1895.

1896. The larval development of Amia clava. Zool. Jahrb., vol. 9.

Dexter. The development of the paraphysis in the common fowl. Am. Jour. 1902. Anat., vol. 2.

Edinger. Das Zwischenhirn. Abhandl. der Senckenb. naturforsch. Gesellsch.,

1892. vol. 18.

1899. The anatomy of the central nervous system of man and of vertebrates in general. Trans, by Hall. Philadelphia. Ehlers. Die Epiphyse am Gehirn der Plagiostomen. Zeitsch. f. wiss. Zool.,

1878. vol. 30, supplement.

Eycleshymer. Paraphysis and epiphysis in Amblystoma. Anat. Anz., vol. 7. 1892.

Eycleshymer and Davis. Epiphysis and paraphysis in Amia. Jour. Comp.

1897. Neurl. Psych., vol. 7.

Gage, S. P. The brain of Diemyctilus viridescens. Wilder Quarter Century Book.

1893. Ithaca.

Gaupp. Zirbel, Parietalorgan und Paraphysis. Ergeb. der Anat. und Entwickl.,

1897. vol. 7.

Gentes. Sur le d^veloppement de la glande inf undibulaire et des plexus choroides

1908. dorsaux chez la torpille. Comtes rend. soc. biol. Paris.

GiERSE. Untersuchungen liber das Gehirn und die Kopfnerven von Cyclothone

1904. acclidens. Morph. Jahrb., vol. 32.

GoETTE. Entwicklungsgeschichte der Unke. Leipzig. 1875.

Goronowitsch. Das Gehirn und die Cranialnerven von Acipenser ruthenus. 1888. Morph. Jahrb., vol. 13.

The Pineal Region in Teleosts. 357

Haller. Bau des Wirbeltiergehirns. I. Salmo und Scyllium. Morph. Jahrb.,

1898. vol. 26.

Herrick. Topography and histology of the brain of certain ganoid fishes. Joiir.

1891. Comp. Neurl. Psych., vol. 1.

Hill. The development of the epiphysis in Coregonus albus. Jour. Morph.,

1891. vol.3.

1894. The epiphysis in teleosts and Amia. Jour. Morph., vol. 9.

1899. The developmental history of the primary segments of the verte brate head. Zool. Jahrb., Aht.f. Anat. u. Ontog., vol. 13. His. Die Neuroblasten. Abhandl. d. math.-phys. Classe d. kotiigl. sdch. Gesellsch.

1889. d. Wissensch., vol. 15.

Hoffmann. Zur Ontogenie der Knochenfische. Arch. f. mikros. Anat., vol. 23.

1884. Holt. Observations upon the development of the teleostean brain with especial

1891. reference to that of Clupea harengus. Zool. Jahrb., Abt. f. Anat. u.

Ontog., vol. 4. Huxlet. On Ceradotus Forsteri. Proc. Zool. Soc. London.

1876. Johnston. The morphology of the vertebrate head from the view-point of the

1905. functional divisions of the nervous system. Jour. Camp. Neurl. Psych., vol. 15.

1906. The nervous system of vertebrates. Philadelphia.

1909. The morphology of the fore-brain vesicle in vertebrates. Journ. Comp. Neurl. Psych., vol. 19. Kerr. The development of Lepidosiren paradoxa. Part 3. Quart. Journ.

1903 Micros. Sci., vol. 46.

Kingsbury. Encephalic evaginations in ganoids. Jour. Comp. Neurl. Psych.,

1897. vol.7.

KuPFFER, VON. DieEntwicklung des Kopfes von Acipenser sturio. Studien zur

1893. vergleichenden Entwicklungsgeschichte des Kopfes der Kranioten.

Heft 1. Munchen.

1906. Die Morphogenie des Centralnervensystems. Hertwig's Hand buch der vergleichenden u. experimentellen Entwicklungslehre der Wirbeltiere. vol. 2, part 3. Leydig. Zur Kenntnis der Zirbel u. Parietalorgane. Abhandl. Senckenb. natur 1896. forsch. Gesellsch., vol. 19.

LocY. The derivation of the pineal eye. Anat. Am., vol. 9.

1893. Mayser. Vergleichende anatomische Studien iiber das Gehirn der Knochenfische.

1882. Zeitsch. f. wiss. Zoolog., vol. 36.

MiHALKOvics, VON. Entwicklungsgeschichte des Gehirns. Leipzig.

1877. MiNOT. Human embryology. New York.



Robert J. Terry.

MiNOT. 1901. On the morphology of the pineal region, based upon its development in Acanthias. A7JI. Jour. Anat., vol. 1.

1903. A laboratory text-book of embryology. Philadelphia. OsBORN. Preliminary observations upon the brain of Menopoma. Proc. Acad.

1884. Nat. Sci. Philadelphia.

Rabl-Rxjckhard. Zur Deutung und Entwickelung des Gehirns der Knochen 1882. fische. Arch. f. Anat. u. Physiol.

Selenka. Das Stirnorgan der Wirbeltiere. Biolog. Centralbl., vol. 10.

1890 SoRENSEN. A comparative study of the epiphysis and roof of the diencephalon.

1894. Jour. Comp. Neurl. Psych., vol. 4.

Studnicka. Zur Anatomie der sogenannten Paraphyse des Wiibeltieregehirnes. '

1895. Sitzungsber. der konig. bohm. Gesellsch. der Wissensch. in Prag. 1900. Untersuchungen iiber das Ependym der nervosen Zentralorgane.

Anat. Hefte, vol. 15. 1905. Die Parietalorgane. Oppel's Lehrbuch der vergleichenden mikrosko pischen Anatomie der Wirbeltiere. vol. 5, Jena. Van Wijhe. Ueber Mesodermsegmente und die Entwicklung der Nerven des

1883. Selachierkopfes. Verhandl. konigkl. Acad. v. Wettensch. vol. 22. Warren. The development of the paraphysis and the pineal region in Necturus

1905. maculatus. Am. Jour. Anat., vol. 5.

Ueber das Parietalorgan der Saurier. Anat. Am. Jahrg. 1.

Adapted from the German


1907 . Comparative anatomy of vertebrates

by W. N. Parker. London. Wilder. The dipnoan brain. Am. Nat., vol. 21. 1887. 1896. Dorsal sac, aulix and diencephalic fiexture

Psych., vol. 6.

Jour. Comp. Neurl.



anterior epiphysis



commissural canal of the lateral


line system






diverticulum of the telence


phalic ventricle






epiphyseal arch



epiphyseal stalk



epiphyseal end-vesicle


intermediate tubercle






paraphyseal arch

posterior commissure

pars intercalaris

post-velar arch

superior commissure


telencephalic roof

velum transversum

median lobe of velum transversum

transverse fold of velum transversum.


H. H. NEWMAN and J. THOMAS PATTERSON From the Zoloogical Laboratory, University of Texas



I. Introduction 360

A . Review of the literature 360

B. Material and methods 363

C. Purpose and scope of the present paper 364

II. The female genitalia 365

III. Number, arrangement and sex of embryos 367

A. Number of embryos 367

B. Airangement of embryos 368

C. Sex of embryos 370

IV. The early embryology 371

A. The earliest stages of Fernandez 371

B. The primitive streak stage 374

C. The five to seven somite stage 380

V. History of the placenta 384

VI. History of the amnion 393

VII. History of the allantois and the umbilicus 396

VIII. Pairing of the embryos 397

IX. Conditions in vesicles containing five foetuses 401

X. The question of identity of embryos 405

XL Specific polyembryony and the determination of sex 406

XII. Summary of evidence for specific polyembryony 409

Bibliography '*-^^

1 Contribution from the Zoological Laboratory of the University of Texas. No. 105.

360 H. H. Newman and J. T. Patterson.

I. Introduction A. Review of the Literature

It is not our present purpose to attempt any comprehensive review of the literature dealing with the development of the Edentata, nor even of that treating especially of the armadillos. It seems advisable, rather, to limit our survey to those contributions, a knowledge of which is essential to an understanding of the problem of specific polyembryony.

That certain species of armadillos bring forth at a birth young all of one sex has been known for over a century. According to Azara,2 a writer of the eighteenth century, the natives of Paraguay and of the Argentine Republic knew that this was true for the Mulita (Tatu hybridum). Any observant hunter, who had been fortunate enough to capture a litter or two of young animals in a burrow with the mother, might readily have noted such a unique state of affairs, for the sexes are easily distinguishable.

In the latter part of the nineteenth century Herman von Jhering, ('85 and '86), met with similar statements on the part of the natives of Brazil and was sufficiently interested to attempt a scientific confirmation of what had been until then merely an interesting piece of folklore. Two pregnant females came under his observation, the uterus of each of which contained eight male foetuses, all in exactly the same stage of development. Each foetus was described as having its own separate amnion; but all were surrounded by a common chorion.

These conditions were interpreted in a subsequent paper by the same author as indicating the origin of the several embryos from a single fertilized egg, and it was further assumed from the facts in hand that the splitting of the original single germ into separate embryonic primordia occurred at some period after fertilization. Von Jhering apparently saw nothing more fundamental in this situation than the discovery of a new type of animal reproduction to which he gave the name temnogenesis." Its bearings on the problems of sex determination and of heredity

^ Referred to by von Jhering.

Development of the Nine-Banded Armadillo. 361

were not appreciated. To him however belongs the credit of having discovered specific polyembiyony in the Mulita.

No attempt was made to secure evidence, either internal or external, of the validity of von Jhering's suggestion until Rosner, ('01), took up the subject in connection with his studies of human monochorial twins. On the basis of a histological examination of the ovaries of one pregnant female of the South American ninebanded armadillo he attempted completely to discredit the idea that the several embryos of a litter arise from a single fertilized ovum. Since his observations strike at the very foundations of the question of polyembryony in the armadillos it seems necessary to review his work in some detail.

The genitalia of two pregnant females were sent to him by von Jhering, and an examination showed that the ovaries of only one specimen were sufficiently well preserved to admit of histological examination. Sections of the other pair of ovaries showed that a large percentage of follicles contained more than one egg. There were in all 52 large follicles : 1 1 with 2 eggs, 7 with 3, 2 with 4, 1 with 5, and 1 with 7. The two largest follicles contained four eggs, exactly the number necessary to produce the four embryos habitually brought forth in a litter of this species. Since the youngest follicles never contained more than one egg the conditions seen in the older ones must have resulted from secondary fusions of adjacent follicular walls, which subsequently disappeared in such a way as to form a common cavity. The author's figures are evidently accurate representations of actual observations and are calculated to convince the reader. Especially striking is the figure of a reconstruction of a series of sections through a large pluriovular follicle in which each of the eggs has its own thick coating of discus proligerus cells.

Rosner believes that the observed condition of four embryos surrounded by a common chorion is to be explained by the following sequence of events: four adjacent follicles fuse in such a way that four eggs are thrown into a single cavity ; on the rupture of this compound follicle the four eggs are discharged simultaneously, descend the fallopian tube held together in a mass by means of their discus proligerus cells, become fertilized, undergo

362 H. H. Newman and J. T. Patterson.

cleavage and come to a common point of attachment in the uterus; subsequently the contiguous walls of the four blastocysts atrophy and a single vesicular chorion is produced.

Were Rosner's observations a record of the normal conditions in the armadillo ovary the question of specific polyembryony would assume an aspect entirely different from that suggested by von Jhering, and we would need to seek no further for an explanation of the observed conditions. The observation that all the embryos in a litter are of the same sex was summarily dealt with by Rosner who considered it as interesting but in no way connected with the presence of a common chorion. Fortunately however there is now every reason to believe that Rosner's material was pathological or otherwise exceptional, for no subsequent investigator has been able to find in the armadillo ovary conditions such as he described.

Cuenot, ('03), while engaged in the study of the problem of the determination of sex, examined the ovaries of one pregnant and of one virgin female of the species investigated by Rosner. In the ovaries of the pregnant specimen there occurred only one follicle of the pluriovular tyipe and this contained only two small, rather abnormal ova. Out of 119 follicles in the ovaries of the virgin female however three contained two or three eggs, but none was found with the number requisite to give rise to the number of young habituall.y born in a litter.

Until quite recently no further progress was made toward the solution of the problem. In 1909, however, there appeared ahnost simultaneously and quite independently, two contributions to the subject, one by Fernandez, ('09), on the Mulita (Tatu hybridum), and the other a preliminary report by the present writers, ('09), on the Nortn American armadillo (T. novemcinctum) . The two species evidently agree very closely in many of the more fundamental details of development but differ sufficiently to make it both interesting and valuable, from the comparative standpoint, to have the developmental history of both species worked out in the fullest detail.

Fernandez presents somewhat detailed descriptions of seven rather early embryonic stages and enters upon a brief discussion

Development of the Nine-Banded Armadillo. 363

of some of the more important questions involved. He was especially fortunate in securing in a good state of preservation two very young embryonic vesicles in which the demarkation of the several embryonic primordia had not yet manifested itself. For the equivalent of this stage we have looked in vain and hence, for the present at any rate, are compelled to rely on Fernandez's description for an explanation of our own earliest stages. Since it is necessary constantly to refer to Fernandez's work in the body of the text no further comment of an introductory character is needed here.

At this point it becomes necessary to refer to our own preliminary report in order to correct the description of fig. 3 in that paper. The specimen there figured was presented to us with the statement that it was intact in every respect, except tha: the uterus and the contained vesicle had befen slit open along the mid-ventral line. On the basis of this statement, together with a study of the external features, we reconstructed the vesicle in situ. Our subsequent investigations of fresh specimens has led us to suspect that what we took to be a young vesicle was in reality only the villous portion of a somewhat later stage.

B. Material and Methods

During the past two years we have had the opportunity of examining 137 females of the native armadillo, together with a considerable number of males. During the breeding season hunters employed to collect material for us covered a wide range of territory in south-central Texas. These men were frequently obliged to haul the living animals through rough country for distances of fifty miles or more in order to reach an express ofSce whence they could be shipped to our laboratories. As a rule a number of days elapsed between the capture of the animals and their arrival in Austin. This delay would serve in part to explain our ill success in securing the earliest embryonic stages. In order to obtain a complete series we believe it will be necessary either to breed the animals in captivity or to accompany the hunters on their expeditions so as to lose no time in examining freshly

364 H. H. Newman and J. T. Patterson.

fertilized females. Although we fully expect to secure the earliest stages in the course of time it seems inadvisable for us to postpone the publication of the results thus far obtained, results sufficiently clean cut in themselves to form the basis of a self-consistent and fairly well rounded embryological account.

At present we have in our possession seventy embryonic vesicles comprising a close series of stages ranging from the primitive streak stage to birth.

Little need be said about the methods employed. To each animal that reached the laboratory was given a number and a page in a ledger where all facts that might be of interest were recorded. In case the carcase was to be thrown away complete records of all data that might be useful in the future were kept. The ovaries of the majority of the females were fixed in the standard cytological fluids. Every^art taken from a given specimen was numbered accordingly. Much of the data thus gathered proved useful during the course of the work and we have no doubt that all of it will ultimately serve to throw light on future investigations.

C. Pur-pose and Scope of the Present Paper

In this our second contribution to the developmental history of the armadillos the main purpose in view is to establish the fact of specific polyembryony and thus to clear the way for future investigation. A more or less tentative explanation of its causes and of the conditions and relations that result from it is hazarded on the strength of the evidence now in hand, which is internal in contradistinction to that derived from an examination of the ovaries and testes, no detailed discussion of which is attempted at present.

Although the question of polyembryony is the central problem it is impossible to treat of it as an isolated phenomenon for the reason that many curious developmental processes are intimately associa ted with i t . The history of the amnion and of the placenta, for example, would be indecipherable apart from the fact of polyembryony, and the inter-relationships of the embryos admit of a rational explanation on no other basis. The associated phenome

Development of the Nine-Banded Armadillo. 365

non of germ layer inversion is also (indissolubly) bound up with polyembryony and in turn involves many peculiar and interesting relations.

Any adequate treatment of the principal problem will therefore necessitate the presentation of a somewhat complex arraj^ of facts whose combined verdict will, we trust, establish our main contention.

Except in the case of the two earliest stages described no attempt is made to present a detailed account of the organogeny of the species. No doubt such a study would reveal many facts of interest to the specialist in mammalian embryologj'-, but would serve only to cloud the main issue with obscuring details.

II. The Female Genitalia

The uterus is simple and not unlike that of the primates in form. In the non-pregnant condition it varies somewhat in size and shape according to the previous history of the individual. In old females that have produced a number of litters the organ though non-pregnant may be distended to several times its normal size, often leading the observer into the vain hope of finding the earliest stages. The uterus of the virgin adult presents a less modified condition and will furnish a basis for the accompanying detailed description.

The average dimensions of the non-pregnant uterus are as follows: 13 mm. from the tip of the fundus to the junction of the cervix with the vagina, 15 mm. between the points of entrance of the two fallopian tubes, and 10 mm. deepdorso-ventrally. Viewed from the dorsal aspect the uterus appears to be broadly kiteshaped (fig. 7) with the posterior angle blending into the vagina. The fallopian tubes are approximately straight where they enter the uterus, but near the ovaries are strongly convoluted, each ending in a hood-shaped fimbriated infundibulum, which, with the aid of a posteriorly directed flap of the broad ligament, covers a large part of the ovary and thus renders the escape of the ovum into the body-cavity well-nigh impossible. The points

366 H. H. Newman and J. T. Patterson.

of entrance of the fallopian tubes are about equadistant from the tip of the fundus and the vagina, thus rendering the cavity of the uterine body much larger as compared with that of the cervix than is the case in the human uterus, where the tubes enter practically at the distal end of the organ.

The ovaries are kidney-shaped having the convex side directed anteriorly, with reference to the axis of the animal. In virgin females the two ovaries are approximately equal in size, but in individuals that are or have been recently pregnant there is always a considerable difference in the size of the two ovaries. The larger one may be two or three times as large as the smaller, and this greater size is invariably due to the presence of a single enormous corpus luteum, the actual bulk of which may be much greater than that of the remaining ovarian tissue. There are found notinfrequently smaller bodies (resembling in histological appearance the large corpus luteum) which are crowded to one end of the ovary and suggest by their shrunken and irregular form that they are either relics of a previous pregnancy or simply the lutea of ova which were never fertilized. It may be stated without hesitation however that there is never more than one large and prominent corpus luteum in the ovaries of a pregnant female.

The mucosa of the uterus is undoubtedly deciduate in character, as may be seen in the illustration of a section taken from a series cut through a pregnant uterus and its contents (fig. 1). Even at the comparatively early period represented it can readily be seen that the mucosa is separated from the outer layers of the uterus by a lymph space of considerable magnitude.

Since the young embryonic vesicle always gains attachment to the mucosa near the tip of the fundus it is not a difficult matter to orient it with reference to the uterine axis. It will be found convenient to refer to the fundus and cervix ends of the vesicle, the former being the original attached and the latter the original free end. The axis of each embryo is also related to that of the uterus, in that its anterior extremity is directed towards the cervix end of the vesicle, except in advanced conditions when the length of the umbilical cord occasionally permits an embryo to reverse its position within its amniotic sac.

Development of the Nine-Banded Armadillo. 367

The pregnant uterus assumes a variety of shapes in different individuals. At approximately the same period of pregnancy it may be either elongated or comparatively broad, either blunt or pointed at one or both ends, and either simple or clearly bilobed dorso-ventrally at the fundus end (figs. 42 and 43). These various forms are not due to the position or arrangement of the foetuses, which in this respect are practically constant, but probably to individual variation influenced by the previous functional history of the organ.

III. Number, Arrangement and Sex of the


A. Number of Embryos

In sixty-five out of seventy cases there were four normal embryos in a vesicle. It may be assumed then that four is typical for the species. Three atypical conditions occurred which may be listed as follows:

1. Vesicles containing five normal embryos (three cases, nos.

28, 91, 108).

2. Vesicle containing three normal embryos each measuring 15 mm. and one decidedly abnormal embryo 7 mm. in length (no. 57). No doubt this vesicle was destined to produce a threeembryo litter.

3. A case of twins (no. 137 ). These were born in captivity. A very careful examination of the uterus and intestines of the mother convinced us that there were no other young born. This may have been a case somewhat like the preceding except that two embryos degenerated instead of one.

There appear not infrequently in otherwise normal embryonic vesicles small amniotic sacs that usually contain the more or less completely degenerated remains of what may once have been extra embryos. In one case (no. 108), a vesicle with five normal embryos, such a sac appeared, which, if truly the representative of an extra embryo, would furnish an example of a six

368 H. H. Newman and J. T. Patterson.

embryo vesicle. In another case (no. 17), which is peculiar in several other respects, there occurred a small empty amniotic sac fused firmly to the wall of the Trager and connected with the amniotic sac of a normal embryo by means of an amniotic canal similar to those of the other embryos. In still another case (no. 9) a fairly large sac in the Trager region was connected by means of a poi'fect amniotic canal with that of a normal embryo (fig. 44) . There is little doubt but that these sacs represent the remnants of supernumerary embryos and as such are the equivalent of those described by von Jhering and Fernandez. It is interesting to note in this connection that Tatu novemcinctum shows a stronger tendency toward stability in the number of foetuses in a litter than does T. hybridum. There is evident, however, in the latter species, a tendency to produce eight young in a litter, just twice the number typical for our species. The numbers of individuals in a litter ranges, however, from seven to twelve.

B. Arrangement of Embryos

In order to clear the way for the description of the early embryonic conditions it should provisionally be pointed out that the four embryos of this species are arranged in pairs, one pair to each lateral half of the uterus. The upper embryo of the left hand pair usually occupies the dorsal amniotic quadrant and is therefore referred to as the dorsal embryo " (no. Ill) . The lower embryo of the left hand side occupies the left lateral amniotic quadrant and is referred to as the left lateral" embryo (no. IV). The lower embryo of the right hand pair occupies the ventral amniotic quadrant and is the "ventral embryo" (no. I), while its mate, occupying the right lateral quadrant is spoken of as the "right lateral" embryo (no. II). Nos. I and II constitute the right hand pair and nos. Ill and IV the left.

The orientation of the vesicle in the uterus and the arrangement of the four embryos with reference to the vesicle and to one another is rather precise, so that a plane running from the mid-dorsal to the mid-ventral line of the uterus would divide

Development of the Nine-Banded Armadillo. 369

y. s. w.

Fig. 1. Outline camera drawing of a transverse section through a pregnant uterus measuring about 15 mm. long by 14 mm. wide. Line D-V is drawn from the points lying at the middle of the dorsal and ventral sides of the vesicle. It divides the section of the vesicle into halves. Embryos I and II lie in the left hand half, and III and IV in the right hand half, a.a., line of attachment of the amnion to the vesicle; e.;;., a small extra chorionic vesicle, which is not fused with the larger one; i.l., intestinal loop; l.s., lymph sinus between the wall of the vesicle and the uterine mucosa, um. X 9.

the two pairs of embryos and their placental areas from each other. There may be a secondary shifting of the positions of the various amniotic sacs, so that in the definitive condition one may find the upper embryo of the right hand pair occupying the doi'sal position, which in the great majority of cases is occupied by the upper left hand embrj^o. Such a shifting might easily occur at any time before the walls of the various amnia fuse firmly with the chorion, a process that does not occur until a late period of gestation. Previous to this time each amnion is attached to the chorion only along a meridional line, an attachment that would permit the whole sac to swing almost as readily to one side as to


370 H. H. Newman and J. T. Patterson.

the other. Reference to fig. 1 will show that the amnion of embryo II, especially after the amnia have increased considerably in size, might readil}^ overlap the line D-V, so its embryo would occupy the dorsal amniotic quadrant. The same shifting might equally well occur on the ventral side. Such shiftings might take place however without affecting in any way the point of the embryonic attachment, which is immediately adjacent to the original amniotic attachment (fig. 1, a. a.). Such departures from the typical arrangement of embrj^os in the vesicle are rather rare, and are not to be considered as of prime importance, for they in no way affect the pairing of embryos, a relationship depending on the point of attachment of the latter which is equivalent to their point of origin. The significance of this arrangement is discussed in a subsequent chapter.

C Sex of Embryos

In thirty-eight embryonic vesicles the foetuses are sufficiently advanced to permit of the accurate determination of their sex. There is no exception to the rule that all embryos in a vesicle are of the same sex.

Although the armadillo hunters claim that males are considerably more numerous than females we find no inequality of sexes in the sets of embryos in our collection, exactly half of which are male and half female. In the small collection of nine advanced sets of mjilita embryos Fernandez found that six were female and three male. On this basis he proceeds to discuss the significance of the apparent disproportion of sexes in the species. No doubt a larger collection of embryonic sets would have shown no such disproportion, for in our earlier survey of the subject of sex distribution we found a much larger proportion of males.

Development of the Nine-Banded Armadillo. 371

IV. The Eaely Embryology

In. the development of the nine-banded armadillo we find that striking peculiarity, met with in the rodents, of germ-layer inversion. In the case of the armadillo the inversion is intimately bound up with the formation of the four embryos, and without it the mechanics of specific polyembryony, as found here, would be inexplicable. The possession of a common amnion by the embryos at an early stage could only occur as a sequence to inversion, and strongly suggests that the embryos are the product of a single fertilized egg.

In the present description of Tatu novemcinctum we shall begin with the primitive streak stage, and leave out of account the younger embryos (except for a brief reference to the work of Fernandez) until we shall have secured a series covering that important period. In dealing with the following stages considerable emphasis is placed upon the embryological details, and especially upon the relations existing between the embryos. This is done because these stages furnish the strongest internal evidence for polyembryony that has been brought forward.

A . The Earliest Stages of Fernandez

It will be necessary to refer to the work of Fernandez, especially to the part in which he describes his youngest two stages; because they hold the key not only to the morphology of the older embryos of Tatu hybridum, but also, we believe, to that of the stage of T. novemcinctum which we are about to consider.

Fernandez secured two specimens of his earliest stage, and the one he describes in detail was cut longitudinally into twentythree sections (10 microns thick). It was found attached to the mucus membrane at the bottom of a fold at the fundus end of the uterus.

Fernandez correctly interprets the condition presented in this early stage as one having been brought about through the process of germ-layer inversion, and compares the vesicle to corre

372 H. H. Newman and J. T. Patterson.

spending stages of the rat and the mouse, described respectively by Selenka, '84, fig. 29, Taf. XIV., and Mehssinos, '07, figs. 38 and 39 Taf. XXXIV. He thus finds the vesicle composed of three sacs lying one within the other: the innermost one is the ^ectoderm, the middle the entoderm, and the outer the trophoblast (hinfalligen Ectoderm), which at the proximal or attached end of the vesicle is differentiating into the Trager. The similarity between the vesicle of Fernandez and those figm-ed by Melissinos (his figs. 38 and 39) is particularly striking, though, as he points out, there are several differences. In the first place, the mesoderm is not yet formed and the so-called Trager cavity scarcely can be regarded as homologous with that of the mouse. In the second place, the parietal layer of the yolk-sac entoderm is not complete, but is wanting in the distal portion of the trophoblast. If, however, we may be allowed to make a suggestion based on a study of his photograph (fig. 6, Taf. XIX), what appear to be scattering cells lying along the inner surface of the distal trophoblast might well be interpreted as representing the remains of the parietal layer of the yolk-sac. This would make this early stage of the Mulita very closely resemble the corresponding stages of several other forms, as illustrated in the figures of such investigators as Selenka ('84), Robinson ('92), Jenkinson ('00), and Mellissinos ('07).

The most interesting portion of this young vesicle of the Mulita is the inner sac, for it is the primordium out of which the ectoderm of the several embryos later differentiates. Fernandez points out the significant fact that it gives no indication of being a multiple structure, such as one would expect to see if the vesicle were the product of the fusion of several eggs.

The second stage of Fernandez is decidedly more advanced than the preceding, and was found lying loose in the fundus end of the uterus. In the preserved condition it measured 3 mm. long by 2.3-2.5 mm. wide. The general condition of the germ layers in this vesicle is made clear in the slightly modified copy of his second text-figure (fig. 2). The figure, which is a diagram of a median longitudinal section passing through two embryos, is shaped like a horse shoe. The entire convex anterior and lateral

Development of the Nine-Banded Armadillo. 373

Fig. 2. A diagrammatic longitudinal section of an early stage of the Mulita. ex.c, extraembryonic body cavity; en., entoderm; am.c, amniotic cavity of the embryo; am.c.c, beginning of the amniotic connecting canal; c.am.c, cavity of the common amnion; ms., mesoderm; ??i. p., medullary plate; ir.c, Trager cavity; tr.e., Trager epithelium; (slightly modified after Fernandez).

margins represent the entoderm of the inverted yolk-sac, while the concave posterior margin is covered with Trager epithelium. Between these two regions occurs a narrow zone where the vesicle was attached to the uterine wall (marked X) .

The Trager cavity {ir. c.) is situated in the concave space roofed over by the Trager epithelium. While in some respects this cavity is comparable to that of the rodents, yet for the most part any such comparison would appear to be strained. The difficulty standing in the way of pointing out any true homologies, however,

374 H. H. Newman and J. T. Patterson.

must be attributed to the incompleteness of the history of these early stages — a fact which Fernandez freely admits.

Within the limits of the vesicle there are two distinct cavities : one the general cavity of the vesicle (ex. c), and the other the common amniotic cavity (c. am. c). The former is lined throughout with mesoderm, and the latter with ectoderm.

The embryos, w^hich are in the medullary plate stage, lie in pocket-like diverticula from the lateral margins of the floor of the common amnion; and each embryo is connected with the latter by a short tube, which is the beginning of the amniotic connecting canal. The common amnion, together with its accompanying embryos, is the product of the inner ectodermal sac of the earlier stage. It is not at all easy to explain fully the manner in which the various structures presented in this vesicle develop out of the primordia of the preceding vesicle, although the history of several of them is self evident. To go from this to the succeeding stage is, however, an easy step, and we shall therefore pass directly to it as exemplified in our youngest vesicle of Tatu novemcinctum.

B. The Primitive Streak Stage

We were fortunate in being able to secure from the uterus the entire embryonic vesicle in practically a perfect state of preservation. The opportunity was thus afforded not only to make a detailed stud}^ of the relations existing between the different embryos but also to obtain a drawing of the vesicle as a semi transparent object (fig. 12). In the preserved condition it measured 7 mm. wide by 9 mm. long. It is slightly flattened dorso-ventrally but in general outline is shaped like an inverted balloon, with two lateral horn-like projections which fit into the openings of the fallopian tubes. These horns persist for a considerable time and are of great service in aiding one to maintain the correct orientation of the vesicle during its early developm.ent.

The surface of the vesicle presents two distinct regions, the lower of which fits into the fundus end of the uterus and is recognized as the Triiger. It is therefore covered by Trager epithe

Development of the Nine-Banded Armadillo. 375

Hum. At the extreme lower end there is a small cap-like area where the primitive attachment knots or cord's of the Triiger epithelium are beginning to disappear. The other region occupies the upper two-thirds of the vesicle and differs from the preceding both in its greater transparency and in the complete absence of a trophoblast. This region is the yolk-sac of the inverted type, and consequently is covered with the entoderm. It is rather indistinctly divided into two portions: (1) the central zone occupied by the embryos and their vascular areas, and (2) the cap-like upper third in which the almost complete transparency is obstructed by the presence of the common amnion and its connecting canals.

Two of the embryos lie on the upper side (corresponding to the ventral side of the uterus) and two on the lower side of the vesicle. Each embryo is connected with the Trager region by a rather broad band, the belly-stalk, and is surrounded by an amnion. Since there is an inversion of germ layers, the embryos when viewed from the outside of the vesicle are seen from their ventral aspects; hence, the posterior portion of each amnion is invisible except as seen through the semi-transparent embryo. Anteriorly, however, the lateral margins of the amnia are clearly distinguishable and are seen to pass forward as the tube-like, amniotic, connecting canals. These lie on the inner or mesodermal surface of the yolk-sac, to which they are loosely attached, and in passing forward they converge and finally enter the common amnion. They do not communicate with this by four distinct openings, but by two, for just before reaching it, the canals belonging to the dorsal and left lateral embryos unite to form a single tube, as do also those belonging to the ventral and right lateral embryos. As will be pointed out in another section, this fusion of the canals is an indication of the pairing of the embryos since the union in each case is between individuals of a pair.

The common amnion at this stage is a comparatively small vesicle lying at the extreme cervix end of the vesicle. The manner in which this condition has been evolved from that seen in the second stage of Fernandez is not difficult to figure out. On the one hand, the cavity of the embryonic vesicle has undergone

376 H. H. Newman and J. T. Patterson.

an enormous extension, due in part to the natural growth of the vesicle and in part to the modification in the shape of the Trager wall, which has changed from concave to convex; on the other hand, the common amnion not only has failed to keep pace with this rapid expansion of the embryonic vesicle, but has actually ceased to grow at all, and is destined soon to degenerate and disappear. In the rapid growth of the embryonic vesicle the embryos gradually have been drawn away from the common amnion, and consequently their connections with it have been pulled out into the long, slender, tube-like canals.

The embryo viewed from the dorsal side shows the exact relations existing between it and the amnion (fig. 13). In general outline the embryo is slipper-shaped and throughout the greater part of its length the amnion conforms to this contour. Both anteriorly and posteriorly the amnion narrows down rapidly — in the former direction to produce the amniotic canal {am. c. c.) and in the latter to form the posterior amniotic process {p. am. .p), which ends blindly above the Trager. The level at which the amnion becomes narrower than the belly-stalk varies in different embryos. In the embryo in question it cuts in some distance posterior to the mouth of the allantois, but in other cases it may cut in at a level somewhat anterior to this point.

The entire embryo, from the anterior end of the medullary plate to the posterior tip of the amnion, measures 3.5 mm., but the embryo proper is only 2.5 mm. long. Running through the central part of the medullary plate is the elongated primitive streak, in which is a well developed primitive groove with a faintly defined primitive pit at its anterior end. The primitive streak is exactly 1 mm. long, and has at its anterior end a distinct head process measuring 0.28 mm.

The outline of the allantois is seen through the embryo, and begins a short distance back of the posterior end of the primitive streak and extends through the mesoderm of the belly-stalk, finally ending some distance anterior to the tip of the amnion. Fernandez does not describe the development of the allantois in the Mulita, and this stage is, of course, too far advanced to give any clue to the exact nature of its origin.

Development of the Nine-Banded Armadillo. 377

Lateral to the embryo is seen the beginning of the yolk-sac or vitelline circulation. At this time the blood islands are well developed and incipient blood vessels are represented by a network of anastomosing cords of mesoderm. About midway between any two contiguous embryos there is a band-like area extending from the Triiger to the upper limit of the area vasculosa. The band represents the region where the boundaries of the vas-cular areas of adjacent embryos come together, and thus corresponds to the sinus terminalis of other forms, except that it is double in composition. At the anterior margin of the vascular area of each embryo the sinus terminalis tends to form the arc of a circle, a tendenc}^ which, if not inhibited by the crowding of four embryos, would result in the production of a circular sinus exactly as in other forms. As a result of this retardation by crowding the anterior margin of the vascular zone of the four embryos is in the form of a series of scallops.

For an appreciation of the condition of the germ layers it is necessary to turn to a study of representative sections. In the most typical of these, such as that taken through the primitive pit, the neural portion of the ectoderm is thick and has the general appearance of that of corresponding stages of other forms (fig. 19). The outer ends of the section curve decidedly upward, especially the one on the right, but for the most part this is due to the fact that the embryo conforms to the general curvature of vesicle. At the ends of the section the medullary plate turns upward to form the amniotic ectoderm, which is composed o^ a single layer of cells.

In the central part of the section the entoderm is composed of rather flattened cells, which, however, remain distinct from the overlying mesoderm. Beyond the limits of the primitive streak it becomes thicker and its cells are cuboidal in shape. It must be kept in mind that the entoderm actually forms the outer surface of this region of the vesicle; for the trophoblast has practically disappeared and there are found only a few of its cells scattered here and there along the outer surface of the entodermal layer.

378 H. H. Newman and J. T. Patterson.

The mesoderm is arising from the primitive streak region in the characteristic manner, and laterally it thins out and, at the point where the ectoderm turns up to give rise to the amnion, divides into two layers, one following closely the amniotic ectoderm and the other the yolk-sac entoderm.

Through the middle of the head process (fig. 18 h. p.) the entoderm at the center of the section is barely distinguishable from the mesoderm, and in many places the union of these two layers is very intimate. This must be looked upon however as a condition which is in all probability secondary. In the region of the head process proper the mesoderm cells are closely packed together, but are entirely separate from the neural plate.

Anterior to the head process the mesoderm rapidly thins out practically to a single layer of cells and is easily distinguishable from the entoderm (fig. 17).

Anterior to this section the mesoderm passes into a thickened region of the entoderm, which obviously has nothing to do with the mesoderm, but owes its existence to a proliferation of entoderm cells (fig. 16, p. p. h.). It was not detected in the whole mount preparations of the embryos, but its extent is easily determined by a study of sections. The thickening runs through the first five sections beginning with the anterior tip of the embryonic shield, and its width is equal to its length, and it therefore forms a circular plate about 45 microns in diameter. In every respect this circular spot corresponds to the "protochordal plate" of Hubrecht, ('08), who has laid especial emphasis upon it as a region where the entoderm is clearly a source of mesoderm formation. Whatever may be one's conviction regarding Professor Hubrecht's interpretation one can at least be certain that the thickening is purely of entodermal origin in this species. Our series is here too incomplete to permit of tracing out the history of the protochordal plate, and thus to see whether its definitive condition is simply that of mesoderm formation, or whether it contributes to the formation of the fore-gut or the oral plate.

It should be stated here that the protochordal plate at the stage under discussion thins out to a single layer towards its margin,

Development of the Nine-Banded Armadillo. 379

where it gradually passes into the surrounding entoderm. In many places the mesoderm cells are beginning to migrate in between the plate and the ectoderm, and especially is this true in the more anterior sections (fig. 21). Tn this section, which shows six of the mesodermal cells, the anterior limit of the protochordal plate is represented. A very short distance in front of this the sections pass through the amniotic canal (fig. 20), which is seen to be composed of two layers, a rather thick inner ectodermal layer, and a thin outer mesodermal layer. In some places the canal is loosely connected with the underlying mesoderm of the yolk-sac, but for the most part it merely lies in contact with the latter.

In sections lying posterior to the primitive pit there is nothing of especial note until we come to the region where the allantoic tube takes its origin. The mouth of the allantois is in the form of a deep groove traversing the ventral side of the anterior end of the belly-stalk (fig. 22, al) . This is lined with an especially thick entoderm and gradually fades out anteriorly, but posteriorly suddenly narrows down to form the tube. The mesoderm of the belly-stalk appears to extend laterally to form the two wing-like processes, which are to be interpreted as representing cross section of the belly-stalk bands (6. b.). Externally these are covered with an epithelium, but within are composed of a loose mesodermal tissue in which run the umbilical blood vessels together with their accompanying sinuses. In section the posterior amniotic process is triangular in shape, and is not much more than half the width of the belly-stalk.

In sections taken through the posterior end of the embryo (fig. 23) the allantois is reduced to a slender tube, having a small lumen. The amnion is here triangular in cross section with the lower angle coming in close proximity to the allantoic entoderm. The mesoderm has much the same shape as in the preceding figure, but may be divided rather indistinctly into two portions : (1) the allantoic mesoderm which surrounds the entodermal tube, and has the cells compactly arranged; (2) the more distal wings or belly-stalk, bands through which the blood vessels run.

The semidiagrammatic longitudinal section of the primitive streak stage is shown in fig. 24, and in connection with what has

380 H. H. Newman and J. T. Patterson.

been said above concerning the transverse sections, this may be studied with profit. The entoderm in this section can be traced from the protochordal plate back along the entire length of the embryo. Throughout the greater part of its length it is composed of flattened cells, but near the posterior end of the primitive streak these cells become cuboidal, and in the region of the mouth of the allantoic tube (al) take on a columnar appearance. Posterior to the allantoic opening the yolk-sac passes back and ends abruptly at the margin of the Trager epithelium {tr. e.).

While the median section does not show the lateral bellystalk bands which form the main connections between the embryo and the Trager, it does, however, bring out with clearness the union between these two as seen at the extreme tip of the embryo. This connection {ms. co.) is simply a backward and downward continuation of the allantoic mesoderm, which passes over into the general mesodermal lining of the Trager region.

C. The Five to Seve7i Somite Stage

The general relations existing between the various parts of the embryonic vesicle in this stage closely resemble those of the primitive streak stage, but the vesicle is almost twice as large, measuring 15 mm. long by 14 mm. wide (fig. 14). Owing to this increase the horns are not only relatively but actually shorter than in the preceding stage. The Trager has undergone marked differentiation and shows a tendency to overgrow the yolk-sac region. The common amnion with its canals presents the same general features as before.

The most interesting changes have occurred in connection with the development of the embryos, and it is to these that we would direct attention. In the first place emphasis should be placed upon the fact that the embryos are not equally differentiated, for the dorsal and left lateral have each, five pairs of primitive segments while the ventral and right lateral embryos have seven. In other words, the individuals of the same pair are in the same stage of development.

Development of the Nine-Banded Armadillo. 381

In the five somite embryo (fig. 30) the neural folds have not yet coalesced to form the brain vesicle, and consequently the neural groove is open throughout its entire length. The posterior ends of the neural folds embrace the much reduced primitive streak. The embryos are bounded laterally by an area pellucida, which is rapidly being invaded by the blood cords.

In sharp contrast to this embryo is the individual from the other pair showing seven somites (fig. 31), and unless one were from the first aware that they were members of the same set of embryos, one would not so classify them. There are really only six and onehalf somites in this embryo, for the most anterior or cephalic pair is connected witn the head mesoderm and is somewhat smaller than the succeeding pairs (fig. 15). There is a slight indication of an eighth pair being cut off from the anterior end of the unsegmented paraxial mesoblast.

The amnion has undergone several marked changes, chief among which are (1) its enlargement in the cephalic region of the embryo and (2) its reduction in width at the level of the distal part of the belly-stalk. In this stage the neural folds have risen up and coalesced to form a portion of the neural tube. The point where the fusion first occurs is at the level of the mid-brain region, and from this place it progresses both backwards and forwards. The anterior progress of the union, however, takes place rather slowly and the final closing on the under side of the fore-brain to form the neuropore does not occur until a period much later than this.

At the posterior end of the diverging folds the reduced primitive streak is seen as a broad plate, which in the mid-ventral region is slightly concave, and by transmitted light appears to be decidedly thicker than the lateral portions. The notochord is seen to arise from the anterior end of the primitive streak and to extend forward between the folds. At the point of origin of the notochord the primitive streak is unusually thick, forming a distinct primitive knot, just back of which is the suggestion of a primitive pit. At the posterior end of the primitive streak the entodermal allantois is faintly visible. It extends backward lying beneath the floor of the posterior amniotic process, and falls far short of reaching the tip of the latter.

382 H. H. Newman and J. T. Patterson.

The belly-stalk'now shows a tendency to form into two bands at the proximal or attached end. Each band later carries an umbilical artery and vein from the placental disc to the embryo, that is, they form the attachment of the mnbilical cord to the wall of the vesicle. The anterior margins of the bands are turned up to form scroll-like structures beyond which the scale-like villi of the Trager are beginning to extend out over the yolk-sac (fig. 15 s. v.).

There is yet to be considered the yolk sac circulation. This consists of a net work of anastomosing mesodermal cords, which in section are seen to be (fbmposed of a central mass of incipient blood cells, surrounded on the upper side by an attenuated layer of mesoderm and on the lower b}^ the entoderm (fig. 8, b. c). These cords do not become hollowed out even at a much later period than this. Indeed it is doubtful whether they ever become functional blood vessels.

In considering the details of structure we shall confine our accounts to a brief description of a series of transverse sections of the five somite embryo, and to the median longitudinal section of a seven somite embryo.

In the region of the neural fold the neural groove has become greatly deepened to form the first rudiment of the brain vesicle (fig. 26, n. g.), and the lateral margins of the medullary plate have become tucked in beneath, thus forming a bay on each side that is at once recognized as the lateral extensions of the headfold ( h. /.) . In consequence of this folding the extreme lateral portions of the amniotic cavity have had the marginal parts of the medullary plate withdrawn from them, with the result that the walls of the amnion have more or less collapsed, obliterating the cavity. In all probability the obliteration is an artifact, due to the rupture of the amniotic canals and the consequent escape of the amniotic fluid.

In the central region the entoderm has undergone a transformation to produce the notochord ( n. ch.) which consists of a row of columnar cells. Already the entoderm shows signs of beginning to grow beneath the notochord, so that this structure will soon be cut off from the archenteron. • The primordia of the pharyn

Development of the Nine-Banded Armadillo. 383

geal pouches ( ph. p.) are seen as bays of entoderm lying on each side of the neural tube.

The mesoderm in this region is in two rather distinct forms; the outer portion is epithelial in character and conforms to the general contour of the entire surface of the section; and the other part is composed of mesenchyme and lies to each side of the imperfectly formed brain vesicle, and consists of scattering stellate


The medullary plate gradually grows narrower as one passes backward until the region of the somites is reached, where its width is about one- third that of the entfre embryo. The margins of the entoderm have almost grown together beneath the notochord. The mesoblastic somites are partlj^ constricted off from the lateral plates, which are undergoing the process of splitting into the somatic and splanchnic layers, between which is the weakly developed coelome.

In the region of the proximal part of the allantois (fig. 28) the belly-stalk bands are very much folded, having their outer margins turned up to form the scrolls that were noted in fig. 15. The umbilical blood vessels in the bands are well organized and are lined with an endothelium. The only other structure worthy of special mention is the posterior amniotic process which is reduced to a small flat tube.

The final section of this series to be considered here is one taken through the posterior end of the amnion (fig. 29). The amnion and median posterior portions of the belly-stalk bands are connected by a rather slender stalk with the Trager (ms. co.). The exact nature of the Trager will be considered in another section, and it remains here merely to point out that the original primitive knots are being rapidly transformed into villi.

The longitudinal section of the seven somite embryo (fig. 25) should be compared with that of the primitive streak stage, in order to bring out the most significant changes occuring in development. The notochord lies exposed throughout the greater part of its length, but at each end it is covered beneath with the entoderm. At the posterior end, where the notochord is covered over, the entoderm is seen to turn back on itself for a short dis

384 H. H. Newman and J. T. Patterson.

tance (fig. 25, en'). This is doubtless only an expression of the same process noted in the study of cross section, in which it was seen that the entoderm was growing in beneath the notochord. The primitive streak has become greatly reduced, due to its transformation into the embryo. The final change to which we would call attention is seen in the great reduction in the length of the allantoic entoderm {al). It is now not more than one-half of its former length, and is soon destined completely to disappear.

V. History of the Placenta

Certain isolated stages in the development of the placenta have been described for at least three species of armadillo.

Kolliker (76), Milne-Edwards (78), and Duges (79-'80), successively described the placental conditions seen in rather advanced vesicles of the South American nine-banded armadillo. Of these accounts that of Milne-Edwards appears to be themost detailed. The embryonic vesicle is described as being a pearshaped body covered with a chorion, the proximal and distal parts of which were thin and membranous, while the middle part formed a thick, vascular, four-scalloped ring, composed of four fused placentae.

A stage similar to that just cited was recently described in somewhat greater detail bj^ the present writers, ('09), and illustrated with two diagrammatic figures. This description of the North American variety of the species seems to agree closely with that of the South American variety as given by the authors just referred to. No doubt we have essentially the same species on both continents.

The only other reference to the placentation of Tatu novemcinctum is that of Lane ('09), who described in some detail the afterbirth of a specimen sent to him from central Texas.

A more comprehensive account of placental conditions is found for Tatu hybridum. Von Ihering states with reference to an advanced stage of placentation, that there is a zonary placenta which has nothing in common with that of the carnivora, but must be considered as a "placenta annularis composita. Each of the

Development of the Nine-Banded Armadillo. 385

eight discoid placentae is pressed against the margins of the two contiguous ones so that the whole set forms a ring or zone encircling the vesicle at right angles to the long axis of the uterus.

The most detailed account of the armadillo placenta yet published is that of Fernandez, who describes several important early stages of this structure in connection with his account of the early development of the Mulita.

Chapman ('01), gives a detailed description of the after-birth of a single specimen of Dasypus sexcinctus. Excellent figures of all structures involved accompany the text. As seen from the foetal side the placenta appears to be truly discoidal in form, but on the maternal side the distribution of the villi is decidedly different from that usually found on that type of placenta. The markedly arborescent villi are arranged in a broad, somewhat lobose ring around the margin of the disc, leaving the centre of the latter free of villi, a condition strongly reminding one of a much earlier stage in the development of the placenta of Tatu novemcinctum, when the original saucer-shaped Trager has begun to produce villi along the free overgrowing margin, but has a comparatively non-villous central area. The forked connection of the umbilicus with the placenta is almost identical with that found in our species. In view of these striking similarities in the placental details of the two species one is led to conjecture that the conditions found in six-banded armadillo closely approximate the ancestral conditions of the more highly specialized armadillos, of which Tatu hybridum seems to be the most pronounced example and T. novemcinctum the next.

In view of the fact that there has yet appeared no complete and consecutive account of the history of the placenta of any species of armadillo it seems worth while to devote a special chapter to a description of the conditions seen in our species.

For the earliest condition it will be necessary once more to call attention to the youngest embryonic vesicle of Fernandez. Here we find surrounding the true embryonic layers the trophoblast, which is attached to the uterine mucosa by means of a thickened disc or plug of trophoblast tissue, called the Trager. This attachment disc is to be considered as the primary placenta. As the


386 H. H. Newman and J. T. Patterson.

vesicle develops the Trager assumes a saucer-shaped form, as seen in vesicles 10 and 18 (figs. 12 and 14).

It will have been noted that, owing to the inversion of germ layers, the whole yolk-sac region of the vesicle is covered externally with entoderm, and that the trophoblast layer of this region, which in species with a diffuse placenta ultimately forms the outer lining of the villi, has practically disappeared. In the Trager region, however, the original trophoblastic epithelium persists in a somewhat modified form. This region of the vesicle consists of an inner layer of mesoderm, at this time rather thin and free of blood vessels, and an outer trophoblastic layer of true epithelial character, from the surface of which protrude branching and anastomosing cords of trophoblast tissue, which give to the Trager a characteristic rough or ridged appearance (fig. 12). These cords of cells appear to function at first as adhesive pads in that they no doubt serve to give the vesicle a firmer grip upon the uterine wall.

In the primitive streak stage these Trager cords, when examined histologically, show themselves to be composed of solid masses of cells with lar^e nuclei and deeply staining cytoplasm, surrounded by a rather flattened layer of epithelium continuous with that covering the general surface of the Trager. Mitotic figures are of frequent occurrence among the cord cells, showing rapid cell proliferation. In some respects the appearance of the tissue suggest a glandular function, and it may well be that from it a secretion is given off which subsequently facilitates the penetration of the villi into the uterine mucosa. That these cords of cells are of trophoblastic origin seems certain, for themeso' derm, the only other layer in this region of the vesicle, is a thin membrane entirely separate from the trophoblast, which at this period it has not begun to invade. The Trager cords then must be formed by a process of rapid local cell proliferation which causes masses to protrude from the surface and frequently to overgrow it to such an extent that they appear to be almost completely constricted off (fig. 9).

Taking the primitive streak stage as the last phase of the primitive placentation, we may notethac the Trager occupies roughly

Development of the Nine-Banded Armadillo. 387

one-third of the area of the embryonic vesicle (the remainder consisting of the yolk-sac region), that the embryos are attached to the Trager by paired bands of mesoderm, equivalent to the belly-stalk of the primates, and that the central area of the Trager is freer from thickenings than the periphery.

The function of the Trager or primary placenta appears to be not so much nutritive as merely adhesive, since there are at this time no blood-vessels in it by means of which nutriment might be conducted to the embryos. It is highly probable that whatever nutriment reaches the embryos comes to them by a process of osmosis through the thin wall of the yolk-sac region of the vesicle.

The formation of the secondary placenta occurs entirely within the confines of the Trager and involves at tne beginning practically its whole area. ' A very instructive stage in the development of the placenta is seen in vesicle 18, (figs. 14 and 15). Here the Trager epitnelium has been pushed out into short scaly villi, which show a tendency to overlap one anotner as well as the margin of the yolk sac region. These protuberances have been in vaded by a stroma-like mesenchyme, which has arisen from the original thin mesodermal epitheliam lining both Trager and yolksac regions of the vesicle. The free ends of the scale-like villi are tipped with masses of solid gland-like tissue derived by the breaking up of the branching cords of earlier stages into numerous knots which are carried out to the extremities of the individual villi. Although the general Trager epithelium which surrounds the villi has persisted in the form of a rather thick syncytial layer the knots are bare of covering except for the presence of an extremely thin layer of much flattened and scattered cells. The knot cells therefore are in a position to come into most intimate contact with the uterine tissues and probably serve as organs of penetration, softening the maternal tissues by means of a secretion and forcing open a path for the villi, in much the same way as the diamond tips of drills cut away the harder materials and open up a path for the shaft. These Trager knots forming the tips of the villi appear to persist throughout almost the entire foetal life in a form practically identical with that just described.

388 H. H. Newman and J. T. Patterson.

The tip of one of the branches of an arborescent villus is shown in fig. 11. The terminal knot of cells is seen to be practically, naked, while farther down in the villus are shown blood vessels containing nucleated blood cells.

Although the formation of villi occurs at first over almost the entire area of the Trager, somewhat more advanced stages clearly show the beginning of a tendency for them to become restricted into four distinct patches near the boundary line between the Trager and yolk-sa*c and around the umbilicus of each embryo. The villi of other regions cease to grow and remain short, as in fig. 3, even flattening down into small rounded prominences which probably serve no nutritive function. Small patches of these flattened villi are scattered over the central area of the Trager as well as between the newly formed placental discs of the various embryos.

During this period the Trager area of the vesicle has been growing more rapidly than the yolk-sac region, the boundary between the two remaining at all times definitely marked. In fig. 3 is shown semidiagrammatically the conditions in vesicle 11 in which four discoid placentae are clearly marked off from the surrounding areas of scattering flat villi. At this stage the placentation is obviously discoid for each embryo.

In vesicle 14, (fig. 4) a decided change is in evidence. The four formerly quite separate discs have undergone a considerable increase in diameter and have come into very intimate contact along contiguous margins. This fusion is more complete between the placentae of embryos I and II and between III and IV than between II and III or I and IV. The significance of tiais is discussed later. A further change is seen in that the villous margin of the Trager region has overgrown the yolk-sac region (not fusing at this time with the latter) and has extended the placental area of the vesicle along the sides of the cervix cavity as far as the os uteri. Judging by the size and abundance of the arborescent villi in this placental annex it seems obvious that it plays the jDrincipal nutritive role at this period. One might compare this overgrowing fringe of branching villi to the cricoid placenta of Dasypus sexcinctus.

Development of the Nine-Banded Armadillo.


a. V.


Fig. 3. A semi-diagrammatic representation of a vesicle seen from the dorsal side. II, III, and IV are the placental discs of the embryos so numbered. Note that those belonging to the paired embryos III and IV are closer together than II and III. f.v., flattened villi of the Trager; h., horn of the yolk sac. X 2.

Fig. 4. A semi-diagrammatic drawing of the dorsal view of a vesicle slightly older than that seen in fig. 3. This shows the fusion of the placental discs I, II, III, and IV into a zone. Note that the fusion between the discs of III and IV (of paired embryos) is more intimate than between II and III. In the cervix region of the vesicle the dorsal part of the overgrowing placental ring, p.r., has been removed to show the smooth yolk-sac lying within (y.*.). The ring was fused with the wall of the cervix at "z". The dotted line lying j ust above the discs represents the line along which the upper part of the ring was cut. X 2.

The yolk-sac region of the vesicle is from this period on cut off from all contact with the uterine wall except at the mouth of the uterus where a small circular area remains uncovered by any outer layer. This condition persists until birth except that the overgrowing ring of arborescent villi undergoes a gradual degeneration, as the placental discs increase in functional prominence until the long, branched villi become mere flattened prominences, which serve only to slightly roughen the membraneous area at the cervix end of the vesicle.

390 ■ H. H. Newman and J. T. Patterson.

The fundus end of the vesicle is still villous to some extent, but the villi are so small and scattered as to interfere only slightly with the transparency of the membrane. One can readily view the embryos in situ through this end. Subsequently the villi of this region disappear entirely with the exception of occasional small tufts that might readily be overlooked. In several vesicles (nos. 116 and 117) this region was seen to be four-lobed owing to the presence of two thickened bands of tissue crossing each other at right angles (figs. 37). These may indicate a demarkation of the several embryonic primordia earlier than that seen in the differ entiation of the embryos themselves.

Stages intermediate between that shown in fig. 4 and the definitive condition can best be shown by a series of photographs.

Fig. 34 shows a somewhat older vesicle, in which the area at the fundus end is seen to be smooth and almost free of villi. The lobing of the composite zonary placenta is only slightly marked.

In fig. 35 is shown the cervix end of a stage slightly more advanced than the preceding one. The heavy coating of arborescent villi is seen to cover the entire cervix end of the vesicle with the exception of the small area that lies across the mouth of the uterus.

The dorsal surface of another vesicle, approximately of the same age as the last, is seen in fig. 36. The vesicle is attached to the shrunken cervix of the uterus. Here is evidenced the tendency on the part of the composite zonary placenta to divide into two double lateral discs. The deep notch occurs between the placental areas of embryos II and III. The small lobe (d. b.) is destined to persist as a bridge between the two lateral discs.

Two farther steps in the development of the definitive placenta are seen in figs. 38 and 39. The vesicle has grown to be several times the size of that shown in fig. 34. Coincident with this great increase in surface the villi in the composite zone have increased ,in functional importance while those that previously overgrew the yolk-sac region of the vesicle have degenerated, leaving a membraneous area at the cervix pole, which in time becomes as large or even larger than that at the other end of the vesicle.

Development of the Nine-Banded Armadillo. 391

In fig. 40 is seen a condition slightly more advanced than that described in detail in our preliminary paper. There is now at each pole of the oval vesicle a star-shaped clear area, with a broad, deeply notched placental zone between, which still shows distinct signs of its origin from four discoid placentae. The notches are more deeply cut along the dorsal and ventral lines than along the lateral, where the placentae of the paired embryos I and II and likewise III and IV are so intimately fused as barely to show the points of union.

Shortly after the condition just described the placenta takes on what appears to be approximately the definitive condition. The tendency to form two well defined lateral discs is carried still farther, but in no case have we observed the complete separation of the two placental areas. As a rule the bridge between the two main discs is narrower on the dorsal side than on the ventral, but its narrowness is compensated for by the presence of a heavier coating of villi and by that of rather large placental blood-vessels which serve to connect one main disc with the other. It seems to be almost invariably the case that the division into the two double lateral discs strikes only approximately along the boundary lines of the original discoid areas, for colored injections forced into the placental vessels of individual foetuses run across the narrow placental bridges and invade more or less extensive and clearly marked villous areas of the other main disc. Such a condition is well shown in fig. 41.

The umbilical cords which may be from 18 to 20 centimeters long are attached rather near the fundus margin of the placentae except in rare cases where five foetuses occur and involve the crowding of one or more unbilical cords away from the margin.

Although a litter of young armadillos was born in the laboratory we were not fortunate enough to secure the after-birth and therefore cannot describe this final stage of the placenta. A comparison of the size and degree of development'^of the new-born young with the oldest foetuses in our possession convinces us that the conditions just described stand as definitive. Yet Lane in his reconstruction of the after-birth of the single individual under observation fails to find connecting bridges between the main discs. He

392 H. H. Newman and J. T. Patterson.

may have observed a rare case in which the Hne of separation into lateral discs passes exactly between the placental areas of the two dorsal and the two ventral embryos. Moreover we find no such clearly marked non-villous areas at the two poles as he describes. The smooth area at the cervix end is in all of our specimens very small and circular in outline, while that at the fundus end is only vaguely outlined and frequently shows patches of fiat villi.

Any attempt to classify a placenta with the above history meets with grave difficulties, as one might conjecture from the multiplicity of terms applied to it by different writers. Kolliker in his original description of the conditions of the embryonic membranes of T. novemcinctum refers to the placenta as discoidal and deciduate. Milne-Edwards considers it to be compound zonary in structure. Beddard describes it as dome-shaped and deciduate; while Lane suggests the term zono-discoidalis indistincta," subdividing Strahl's class ^^zono-discoidalis" into two varieties, ^'distincta and ^'indistincta."

Somewhat similar placental conditions, as found inT. hybridum, are designated by von Jhering as indications of a placenta annularis composita." Chapman's use of the term deciduate cricoid" appears to be apt for the placenta of the six-banded armadillo.

Of all these terms the one that appeals most strongly as descriptive of a certain rather persistent phase in the development of this multiformed structure is that used by von Thering, placenta annularis composita," but one must not forget that at first it is simply discoidal, then cricoid, then tetra-discoidal, later annularis composita, and finally incompletely doubly discoidal.

If animals are to be classified according to the form of their placentae, a method of classification that is fortunately falling into disrepute, it would be very difficult to classify the nine-banded armadillo, unless we arbitrarily decide to select some particular developmental phase of the placenta as a criterion for classification. In such cases one would be led to chose either the primary or the definitive condition and would thus call the placenta either "simply discoidal" or "incompletely doubly discoidal." Other terms scarcely find a rational basis.

Development of the Nine-Banded Armadillo. 393

The conjecture that the compound placenta of T. novemcinctum has been derived without any fusion of four embryonic vesicles from a condition similar to that described bj^ Chapman for Dasypus sexcinctus, is very tempting in view of the evident close relationship of the two species and the striking resemblance that exists between them in the details of the placenta, umbilicus and other structures. This if true would furnish one of the most cogent proofs of polyembryony, since we find in the more highly specialized species a quadruple placenta, which at a rather early period closely resembles the definitive placenta of a more primitive species that gives birth to single young or to twins. ^

VI. History of the Amnion

From Fernandez's description of his earliest stage it is clear that the common amniotic cavity is at first the hollow of the ectodermic vesicle, which, through the inversion of germ layers, has come to lie within an envelope of entoderm. Regional differentiation .of this ectodermic vesicle produces the ectodermal portions of the embryonic primordia, which are at first contained within a single vesicular amniotic cavity. Subsequently the individual embryos sink into pockets in the floor of the common amnion, which has evidently become fused to the walls of the yolk-sac at the cervix pole of the embryonic vesicle. The posterior end of each embryo has become fixed by means of the primordium of the belly-stalk to the margin of the Trager, and consequently, as the yolk sac gradually increases in size, the embryos are drawn away from the common amnion, retaining connection with it only by means of slender tubes, the amniotic connecting canals (figs. 12 and 14). It has been shown that each pair of embryos withdraws from the

^ We are informed by Mr. Robert D. Carson, superintendent of the Philadelphia Zoological Garden, that a female six-banded armadillo in captivity gave birth to:

1. A single male, on May 10, 1901.

2. Twin males, on April 6, 1902.

3. Twins (male and female), on July 19, 1902.

394 • H. H. Newman and J. T. Patterson.

common amnion into a single pocket and leaves for a short distance a single connecting canal. Later each member of these pairs loses its connection with its partner and acquires its own canal. This secondary separation of the pairs produces a forking of each of the original two connecting canals, a condition that persists for a long time.

After the embryos have left the common amnion the latter probably becomes functionless and ceases to grow. Fortunately however it persists with all of its connections through a considerable developmental period, furnishing evidences of polyembryony and of embryonic pairing. In fig. 44 it is shown still typical in form with its connecting canals entire but with their lumens interrupted with plugs of tissue. The regions between the plugs have become distended through local secretion of amniotic fluid, so that the canals as a whole present a decidedly moniliform appearance. In fig. 45 a somewhat more advanced stage of degeneration in these structures is seen. The common amnion can no longer be recognized but the canals are still clearly defined. Each of these shows a number of pronounced bead-like swellings, one of which may represent the remains of the common amnion. These canals may persist until stages as advanced as that shown in fig. 33, but are seldom to be detected in later stages.

The posterior amniotic processes, which in early stages were seen to be closely associated with the development of the allantois, do not persist in so marked a form in our species as in the Mulita. Only in rare cases does one see any traces of these structures at a period later than the five to seven somite condition (fig. 15). In vesicle 17, however, one of the embryonic amnia is connected by means of an amniotic canal with a sac as large or larger than the common amnion but lying at the opposite pole of the vesicle. This condition is no doubt exceptional and may be accounted for on the supposition that the posterior amniotic process of one of the embryos, on account of its unusual length, protruded far down into the Trager region, came into contact with and united with it, and subsequently swelled into an amniotic sac at the point where "its terminal bulb fused with the Trager wall.

Development of the Nine-Banded Armadillo. 395

Another exceptional condition is that seen in fig. 46, where branching from a typical amniotic canal of one of the embryos, is an accessory c^nal running to an empty amniotic sac at the center of the Trager. Such a condition is doubtless due, as was stated in another place, to the presence of the remains of a degenerated fifth embryo. Teratological amniotic structures similar to those just described were observed in a number of other cases. In most instances there seems to be no doubt that they represent the retarded or degenerate remains of supernumerary embryos. The frequent occurrence of similar rudimentary embryos in Tatu hybridum and in our own species seems in itself a strong piece of internal evidence of specific polyembryony, for, on the basis of the origin of the several embryos from separate eggs, it would be difficult to understand why some should develop into complete embryos, and others, in the same vesicle and under practically identical conditions, should meet with so little success.

After the closure of the lumens of the various amniotic canals all communication between the four or more amnia is cut off; and henceforth each embryo has its own separate amnion in as true a sense as in those mammals that produce several entirely independent young. The developmental history of these envelopes is moreover in no important way different from that of other mammals except that in late stages a gradual fusion occurs, first of all with the wall of the chorionic vesicle and later with one another, where, through the pressure of growth their walls have come into contact.

Various representative stages in the later history of the amnion are seen in the photographs herewith presented. In fig. 44 the amnia may be seen to lie rather closely applied to the bodies of the embryos. In fig. 33 the cavities of the individual amnia have increased greatly in size and the sacs have assumed an ovoid form with the narrower end directed toward the cervix pole of the vesicle. In fig. 34, an external view of the fundus end of a somewhat older vesicle, the amnia are seen pressed against the membraneous area of the Trager, producing at points of contact an added transparency, reminding one of windows through which the embryos can clearly be viewed.

396 H. H. Newman and J. T. Patterson.

Even after the embryos have reached a length of 4 cm. the amniotic sacs are still quite free from one another, but a little later they begin to fuse along contiguous surfaces. Not until about a month before birth however do they become inseparably bound together. After the fusion is complete the amnia occupy the entire cavity of the vesicle and divide it into (normally) four quadrants of equal size, each running from pole to pole. This nearly definitive condition was described in detail in our preliminary account and needs no further attention here. In fig. 46 the edges of the amniotic partitions separating adjacent embryos may be seen at a." The umbilical cords are always attached just to the left of the partitions.

VII. History of the Allantois and the Umbilicus

The early history of the allantois was shown to be very intimately bound up with that of the belly-stalk or primitive umbilicus This intimate connection persists as long as the allantois retains a distinguishable structure. In stages of the degree of advancement shown in vesicle 17 and 11 (figs. 1 and 44) the entodermal allantois is seen as a slender cord of cells more or less closely fused with the umbilicus and showing here and there traces of a former lumen. The outlines of the mesodermal allantois, however, are no longer distinguishable from the tissues of the belly-stalk. The allantois of the armadillos seems then to be entirely vestigeal in later stages of development.

The umbilicus arises directly from the primitive belly-stalk, which was shown in the description of vesicles 10 and 18 to consist of paired flat bands of mesoderm uniting the posterior end of the embryo to the margin of the Trager or primitive placenta. That the mesodermal allantois contributes some tissue to the definitive umbilicus has already been intimated, but at no time do allantoic blood vessels function. The placental circulation is carried on exclusively by the umbilical vessels, paired arteries and veins. Each artery arises along the inner margin of a belly-stalk band, while each vein forms in the scroll-like outer margin. In later stages the two bands fuse at a short distance from the vesicle

Development of the Nine-Banded Armadillo. 397

and continue to the bodj^ of the embryo as a single somewhat flattened cord. The forked connection between the cord and the vesicle is maintained as a characteristic feature of the placentation. In the definitive condition the umbilicus measures from 18 to 20 cm. in length and about 1 cm. in greatest diameter. The veins are longer than the arteries and take an open spiral course along the flattened edges of the cord.

VIII. Pairing of the Embryos

In our preliminary paper attention was called, in treating of the nearly complete identity of the four embryos, to indications of a still closer resemblance between the individuals of the right and left hand pairs. In attempting to derive the four embryos from the blastomeres of the four-cell stage the following suggestion was offered: "This possible interpretation receives a striking confirmation in the fact that the four embryos can be arranged into two pairs, the individuals of which approach almost complete identity; and these identicals are not only adjacent to each other but are also attached to placental discs that are closelj^ united. If all four embryos are derived from a single egg, this is exactly what we should expect to find; for surely the individuals derived from one of the blastomeres of the two-cell stage ought to be more nearly similar to each other than to the individuals of the other blastomere."

The subsequent acquisition of a large amount of additional data has served only to strengthen our conviction concerning this strong tendency toward pairing among the four embryos: a tendency that expresses itself in the method of separation of the embryos from the common amnion ; in the fusion of the four discoid placental areas into two double lateral discs; in the different rates of development seen in the embrj^os of a single vesicle ; and in the closer resemblance, as a rule, between the paired embryos of one double placental disc than between the embryos in general.

The forked arrangement of the amniotic canals, as was pointed out in connection with vesicles 10 and 18, shows that the embryos retreat from the common amnion in pairs and that only when at

398 H. H. Newman and J. T. Patterson.

some distance from the latter do the individuals of a pair sever their intimate connection and acquire separate amnia. Subsequently these embryos show their pairing in their mode of attachment to the definitive placental discs, embryos I and II being attached to the right hand disc and III and IV to the left.

Fernandez calls attention in the case of the Mulita to the exact identity in stage of development among the embryos of a set. That this is not always the case in our species is well brought out by a comparison of figs. 30 and 31, two embryos from vesicle 18.

Fig. 30 represents embryo III, and IV was identical with it. Fig. 31 was taken from embryo II but would serve equally well as a figure of I. The difference in degree of development between the two pairs is well marked not only in the number of somites (5 in III and IV and 7 in I and II), but in the conditions in the head region and in other parts.

It is not likely that a difference in rate of development between the two pairs is of common occurrence, but the clear case of it just presented seems worth recording not only on account of its rarity but because it serves to emphasize the tendency of the individuals of a pair to be alike, but somewhat different from the equally identical opposite pair.

Although of very common occurrence the pairing of embryos on the basis of resemblances in the total number of scutes in the nine bands of armor, is not without exception. In many cases the pairing is so marked as to be startling, as for example in one case where I and II each has 555 plates and III and IV each has 548; or in another case where I and II have respectively 551 and 552 and III and IV have respectively 560 and 559. In many other cases the pairing is obvious but not so clean cut.

There are on the other hand two cases where there was a close resemblance between three embryos, but one was strikingjy different, as for example where II, III, IV have respectively 544, 545, 543 and I has 549; or again where I, II, III have respectively 562, 565, 564 and IV has 573. Finally two cases occurred in which, if any pairing at all exists it appears to be between I and III and between II and IV, as for example where I and III have respectively 544 and 546 while II and IV have 550 and 548.

Development of the Nine-Banded Armadillo. 399

On the whole however, in spite of these exceptions, the general rule holds good, that the closest resemblances occurs between paired embryos.

In this connection it should be mentioned that even where there is exact resemblance between the individuals of a pair in the total number of scutes in the nine bands of armor, there is no perfect correspondence with respect to individual rows. The resemblance in total numbers of scutes is however, a matter of more importance than the exact manner of their arrangement into rows, which is a secondary process. Each primary scute is the equivalent of a well defined hair group arid these groups, as can be seen in other regions of the body, are quite definite units, although subject to more or less shifting before reaching their final arrangement into rows. Tn a subsequent paper we expect to make a special study of variation and heredity in the elements of the armor and shall in this place refrain from any more detailed reference to the subject.

Another source of data, however, which furnishes striking evidence of pairing is seen in connection with a fairly common tendency for regional fusion of adjacent bands of armor, or for the occurrence of interrupted and of incomplete bands in definite regions. Such atypical conditions occur in from three to four per cent of all cases, a fact that we have established from an examination of considerably over a thousand shells. This comparative rarity of occurrence, while it renders the collection of data on pairing and identity difficult, gives to such data an added value, in that chance resemblances are very unlikely to occur.

Only four cases of strikingly atypical armor arrangements have so far been discovered in the collection of foetuses now in our possession. In one case in embryos I and II there occurred a remarkably atypical scute arrangement in the first band of armor, while III and IV were quite normal. In a second case I and II showed a slight fusion between the first two rows at the right hand margin, while III and IV showed a much more eKtensive fusion in exactly the same region. The pairing in this case was only a matter of degree of fusion, but there was a decided difference in extent of the region of fusion in the two pairs. In a third

400 H. H. Newman and J. T. Patterson.

case III and IV exhibit almost precisely the same atypical condition, a short interruption in the first band a little to the left of the median line; II has an interruption in the same band, involving considerably more than half of the total length of the band, while I, although appearing to be perfectly normal, seems to have carried the tendency toward the suppression of a band to the extreme in that the whole band is lacking. In a fourth case one of the four embryos shows a short fusion between the first two rows on the left hand side, while the other three are perfectly normal.

Three out of four cases, then, furnish strong evidence of pairing, while the fourth case, which is after all atypical only to a minimum extent, affords an exception, whose weight can scarcely be sufficient to discredit the evidence of the other cases.

Although the pairing of embryos is not always perfectly obvious the cumulative evidence in favor of its general occurrence is convincing and must have some fundamental significance, an understanding of which is undoubtedly closely bound up with the earlj^ developmental mechanics as we shall attempt to show\

It has occurred to us that the division of the four-scalloped placental band into right and left lateral discs migh t be dependent upon the fact that the blood supply of the uterus comes from the paired ovarian blood vessels that enter the uterus laterally. It is true that the paired embryos, with very few exceptions are located on the same side of the uterus, but that the pairing is in any way caasally related to the fact of their location near the entrance of a single maternal blood vessel is highly improbable, because the maternal blood does not reach the embryos.

It has also been suggested that the close resemblance between the individuals of a pair might be due to admixture of foetal blood, but we have demonstrated by the use of colored injections that the placental area of each embryo is sharply circumscribed and that no blood passes from one embryo to another. A common blood environment then cannot be held accountable for the near approach to identity seen in the pairs. Moreover it has been shown that long before there was any sign of the definitive placentation, and hence before there was any circulation of blood, pairing of embryos was evident in the relationship of the amniotic

Development of the Nine-Banded Armadillo. 401

connecting canals and in one case, in the degree of development of the embryos.

These observations force us to the conviction that the orientation of the vesicle in the uterus and the pairing of the embryos are expressions of the cleavage polarity and symmetry of the ovum. The cell products of the first two blastomeres would occupy the right and left halves of the early blastocyst and the daughter cells derived from the first two blastomeres would normally hold their relative positions as quadrants of such a blastocyst, so that, although it may not be possible to note any definite demarkation of embryonic primordia until a much later stage, they may be well defined from the first. When however pairing seems to exist between diagonally opposed embryos it might conceivably be due to a shifting of blastomeres in the four-cell stage, which could readily occur in such loose cell aggregates as prevail in early mammalian cleavage stages. A shifting upwards of two diagonally placed blastomeres and a consequent shifting downward of the other two would bring about a recombination of blastomeres into two new pairs without interfering with the hereditary tendencies of the individual units. Such an appeal to the imagination of the reader would scarcely be justified were it not the logical outcome of a failure to explain the conditions on any other basis. We are much inclined, in spite of Fernandez' failure to note any indication of a demarkation of separate embryonic areas in his earliest vesicles, to believe that such areas exist from the beginning and express themselves as separate primordia only on the differentiation of embryos. This view is in direct opposition to that of Fernandez who holds that up to the time when the separate embryos are distinguishable, the vesicle is a single embryo.

IX. Conditions in Vesicles Containing Five Foetuses

Out of a total of seventy embryonic vesicles there occurred three in which there were five foetuses. In all of these the sex could be determined and, curiously enough, they were all males. Whether or not this condition is universal could not be determined. If how^ever it should prove that all five-embryo sets are miales it


402 H. H. Newman and J. T. Patterson.

would mean that sex is determined by certain conditions in the egg. With only three cases in hand a discussion of the matter would be unprofitable.

In two cases out of three it was possible to enumerate the scutes in the nine bands of armor and on that basis to determine the varying degrees of resemblance among the embryos.

The occurrence of five embryos involves a decided asymmetry of the placental and amniotic elements and an atypical arrangement of the embryos. In each case the condition of two main lateral discs was maintained, but one of these discs, the one to which three embj:'yos were attached, was considerably larger than the other. An examination of the larger disc shows that in each case it is composed of only two, not three, primary discs. One of the primary discs, on the side where three embryos are attached, is twice the normal size and to it are attached in S3nmmetrical fashion the umbilical cords of two embryos. Apparently there is no regularity about the position of the double disc. In one case the double disc is ventral, and in the other two right lateral in position. Believing that the two embryos attached to a single primary disc are the equivalent of one typical embryo, we shall give them the same number, as for example, I and I'.

The following conditions are found in vesicle 91, the relative positions of the embryos being indicated in the diagram of the placenta, represented as cut open along the narrow dorsal bridge and laid out flat (fig. 5) . The number of scutes in the nine bands of armor are indicated on the figure. It will be noted that there is distinct pairing on the normal side of the vesicle, between embryos III and IV; that the resemblance between the two embryos on the large disc (I and I') is equally close; but that there is a wide difference bewteen these two embryos and the single embryo on the same side (no. II).

In vesicle 108 somewhat similar conditions exist, but the vesicle is laid open along the ventral bridge (fig. 6). Embryos II and II', having a common primary placental disc, are identical in the number of scutes but widely different from embryo I, which is attached to the other primary disc on the same side of the vesicle. Embryos III and IV are quite different from those on the other side, but are fairly similar to each other.

Development of the Nine-Banded Armadillo.


Fig. 5. Diagram of the placenta of vesicle no. 91, showing the placentation and the numbers of scutes in the nine bands of each embryo. Cut open along the dorsal notch.

Fig. 6. The same scheme for vesicle 108. Cut open along ventral notch.

404 H. H. Newman and J. T. Patterson.

In the case of the third five-embryo vesicle a satisfactory enumeration of the scutes was not found possible, but the position of he large disc was the same as in 108.

In all three cases the anmia of the three embryos occurring on the same side are irregularly arranged. Instead of occupying whole quadrants of the subspherical vesicle the amnion of one or more embryos is forced away from one end and crowded past the opposite end, thus causing the amniotic partitions to run diagonally across the placental discs instead of taking a meridional course from pole to pole as in typical cases. The relative positions of the embryos is of course correspondingly irregular so that one is immediately struck by it when the vesicle is first exposed to view.

The high degree of mal-adjustment seen in these vesicles would seem to indicate that the occurrence of more than four embryos is the expression of a coenogenetic tendency to carry polyembryony a step farther by a doubling of the present typical number of embryos. In the Mulita this condition has been attained and there exists a strong tendency to double again, as seen in the frequency of vesicles containing nine or more embryos. It appears probable to us in view of the occurrence of one case of twins in our collection, that in T. novemcinctum specific polyembryony had its origin in the acquisition of a habit of producing identical twins in a fashion similar to that seen in other mammals, that the inversion of germ layers made it easy for this tendency to express itself still more fully in the habitual production of four embryos. The production of more than four embryos in our species seems to involve so great a disturbance of a very accurate adjustment of embryos and embryonic membranes that it seems highly improbable that a larger number of embryos will ever become typical.

It would be interesting to find out whether there is in T. hybridum anytedency of the embryos to arrange themselves into two groups corresponding' to the right and left sides of the vesicle. A study of Fernandez' photographs (figs. 1 and 2) would seem to indicate that such is the case. It is hoped that this matter will receive some attention and that the degree of resemblances among the embryos of the various sets will be determined.

Development of the Nine-Banded Armadillo. 405

X. The Question op Identity of Embryos

In the case of identical or monochorial twins the question of close resemblance has been much discussed and the impression seems to prevail that the individuals of a pair show such marked similarity in their finer details of structure as to be practically identical.

In our earlier contribution to this subject we were inclined to look for the resemblances between the embryos of a litter and to understimate the value of the points of difference. Now however that we feel that the question of specific polyembryony has been established, the differences among embryos interest us more than the resemblances, because they indicate a rather marked degree of versatility in the hereditary possibilities of a single fertilized germ cell.

The only point of unfailing identity among the individuals of a litter is that of sex. In 38 cases where the sex was definitely determined there was no exception to the rule that all embryos in a vesicle are of the same sex.

So far as dimensional differences go there is again practical identity, although in a few cases there seemed to be a slight difference in the size of the two pairs. In comparing one individual with another we were forced to admit that they differed only in the minutest details, such as the number of scutes in the armor. A comparison on this basis is just about as searching as would be a comparison of the number of feathers in a given feather tract of two birds, or of the hairs in a given hair area of two mammals. We have for the present limited our comparison to the total number of large scutes (with corresponding underlying bony plates), in the nine moveable bands of armor. The extreme range of variability in the total number of these plates (in all of the individuals so far examined) is rather wide, running from 511 to 620, a range of 109. In a number of cases the individuals of a litter exhibit a range of only five or six scutes, but as a rule the range is wider, averaging in all cases studied twelve, or less than oneninth of the total range of our sample of the species. Whether or not this represents a closer esemblance than exists between

406 H. H. Newman and J. T. Patterson.

the individuals in a litter of rats or other mammals cannot at present be determined.

Although the difference between the two pairs of a litter may on the average be rather marked, that between the individuals of a pair seldom exceeds three scutes and averages in all cases observed less than three, while cases of absolute identity in the total number of scutes is of frequent occurrence.

It will be remembered also that in our discussion of pairing a considerable mass of evidence was adduced to show that even in atypical scute arrangements a high degree of identity existed between pairs, while in most cases the pairs differed greatly from each other. All of these observations go to show that the identity between the individuals of a pair is a very real thing but that the there is nothing approaching true identity' between the pairs. The condition may well be described as a case of double identical twins.

XI. Specific Polyembeyony and the Determination

OF Sex

The first clue to the existence of polyembryony in the armadillos was furnished by the discovery that all of the individuals of a litter are of the same sex. This together with his observation of a common chorion, led von Jhering to surmise that all of the embryos of a vesicle arise from a single fertilized egg. That this flash of insight foreshadowed the discovery of a truth has been sufficiently demonstrated, we believe, by Fernandez for Tatu hybridum and by us for T. novemcinctum.

Identity of sex then is in some way closely bound up with the phenomenon of polyembryony. Presumably all of the individuals of a litter are of the same sex because they have been derived from a single fertilized ovum ; but this presumption involves the corollary that sex is determined in the germ before any demarkation of embryonic rudiments has occurred. The only alternative is that similarity of environmental conditions during the developmental period has the effect of producing offspring all of the same sex, an alternative with no factual basis, as is shown by the

Development of the Nine-Banded Armadillo. 407

following observations : that at a very early period each embryo is surrounded by its own amnion; that a little later each draws maternal nutriment from a separate area of the uterine wall ; and that there is no admixture of foetal blood. We are therefore driven to the conclusion that sex is determined before there occurs any splitting of the single germ into separate embiyonic primordia.

Opinions differ as to the exact period at which this splitting takes place. Fernandez maintains, on the basis of his studies of the early blastocyst of the mulita, that there is no trace of polyem.bryonj'^ until after the two primary germ layers have been laid down. What he probably means is that previous to this time there is no visible demarcation of the germ layers into isolated blastodermic areas. That the real separation of embryonic rudiments occurs at a much earlier period, even during the early cleavage stages (in our species at the four-cell stage), seems probable in view of the discovery of pairing among the embryos, a phenomenon for which no other explanation offers itself; and by the observations of Marchal, ('04), and Silvestri ('06), on the parasitic hymenoptera, where each embryo in a set takes its rise from a single cell of a rather advanced cleavage Stage.

It seems highly probable then that the tissues involved in each of the four quadrants of an embryonic vesicle, whether or not they may show a demarcation, do really arise as the lineal descendants of one of the first four blastomeres. In this sense the four embryos are delimited at the four cell stage. It is hardly to be expected that any demarcation would be visible before the beginning of the period when the separate embryonic shields are differentiated.

The question as to the exact period of separation of the several embryonic rudiments is one that cannot at present be definitely settled. Even if one should be fortunate enough to obtain the early cleavage stages it is improbable that he would be able to observe any essential departure from the usual plan of mammalian cleavage, for a blastomere of the four cell stage would have the same appearance whether it were destined to produce a whole or only a quarter of an embryo.

408 H. H. Newman and J. T. Patterson.

It seems probable from our studies of the ovaries that the tendency to polyembryony is inherent in the unfertihzed egg, which is the seat of a developmental vigor somewhat more intense than that exhibited in the ova of other mammals. This extra expresses itself sometimes by parthenogenetic divisions and at other times in the formation of fairly regular morulae within the confines of the Graafian follicles. That polyembryony is simply a more normal expression of the same superabundant energy in the female germ cells seems highly probable, and we would offer this as a tentative explanation of the physiology of polyembryony, pending an exhaustive study of a large collection of ovaries.

Taking it for granted then that sex is determined in the undivided oosperm, the question naturally arises as to which of the two germinal elements is the sex determiner. Cytological examination of the ovaries reveals no dimorphism of the ova. They all have 32 chromosomes and are equally alike in other respects. The possibility that sex might depend on which of the two ovaries produced the egg that became fertilized as suggested by the work of Dawson ('09) . This writer maintains on observational grounds that in the human being the male producing ova come from the right and th^ female producing ova from the left ovary. The corpus luteum served to indicate which ovary functioned in any given pregnancy. In the armadillo we have an exceptional opportunity to put Dawson's theory to a test, for the corpus luteum of this species is a very prominent feature of the ovary that has functioned. A study of our data reveals the fact that the corpus luteum is found with almost equal frequency in right and left ovaries, which coincides with the exact equality of male and female litters. Unfortunately for the theory, however, there is no correlation between the sex of the embryos and the dextrality or sinistrality of the functional ovary. Out of twenty cases in which the right ovary contained the corpus luteum, the sex of the embryos was male in seven and female in thirteen; while out of thirteen cases in which the left ovary held the corpus luteum, the sex was male eight times and female five. Evidently then the position of the functional ovary has no determining influence on sex.

Development of the Nine-Banded Armadillo. 409

There is on the other hand excellent evidence that the male cell may act as a sex determiner. Studies of the spermatogenesis of our species show that the spermatogonial number of chromosomes is in all probability 31, one less than the oogonial. There is moreover in the reduction division a very definite and obvious odd chromosome, which precedes the other chromosomes to the pole of the spindle and serves to institute a dimorphism of the spermatids. That the odd chromosome is concerned with the determination of sex is as probable for the armadillo as for the insects and other forms in which it has been described. Both rest on the same observations. Since it is our intention to make a detailed study of the cytology of the germ cells in this species, it must suffice for the present to have indicated the sort of external evidence of polyembryony and of sex determination we have at our cdmmand.

The discovery of so clear a case of an accessory chromosome in a mammal is in itself worthy of mention, since it brings us* a few steps nearer to the discovery of the physiology of sex determination in man. In addition to the intrinsic value of this discovery, however, we are afforded another strong proof of specific polyembryony, in that it is highly improbable, on the basis of the origin of the embryos of a vesicle from several fertilized eggs that each of these eggs would be fertilized by the same kind of spermatozoon. Such a possibility could be realized only through the instrumentality of selective fertilization, the occurrence of of which has never been successfully demonstrated.

XII. Summary of Evidence for Specific Polyembryony

1. The uterus is simple, resembling that of the primates, which give birth typically to one offspring at a time.

2. There is never more than one large corpus luteum in the ovaries of a pregnant female.

3. In over 90 per cent of vesicles the number of normal embryos is four, a number that suggests their origin from the blastomeres of the four-cell stage. It is also unlikely that this number of ova would so often be given off at the same time.

410 H. H. Newman and J. T. Patterson.

4. The fact that all of the embryos of a set are invariably of the same sex strongly suggests their origin from a single fertilized egg.

5. The definite orientation of the embryos in the vesicle, and of the vesicle in the uterus, precludes the possibility of their origin from several eggs, even though these might conceivably be simultaneously given off from the ovary.

6. The inversion of germ layers presents a condition in both Tatu hybridum and in T. novemcinctum, which could not be attained by the union of several eggs to form a single vesicle. This is the strongest piece of evidence for specific polyembryony that has been advanced, and, to our minds, is practically conclusive.

7. The Trager or primitive placenta, common to all four embryos, is the morphological equivalent of that seen in the monembryonic vesicles of certain rodents.

8. The overgrowing fringe of arborescent villi seen in middle stages of gestation reminds one strongly of the cricoid placenta seen in the monembryonic vesicle of the six-banded armadillo, figured by Chapman.

9. The existence of partial or rudimentary embryos is evidence against the idea that the several embryos have been derived from separate eggs, for it is difficult to understand why some should develop perfectly, while others, under the same environmental conditions, should have so little success.

10. The pairing of embryos points to the origin of each pair from one of the first two blastomeres.

11. The presence of an accessory chromosome in the male germ cells suggests that the spermatozoon is the sex determiner. On this basis the fertilization of several eggs always by the same kind of spermatozoa seems highly improbable.

Development of the Nine-Banded Armadillo. 411


Bailey, Vernon. Biological Survey of Texas. North Amer. Fauna, no. 25.

1905. Beddard, F. E. Mammalia. Cambridge Natural History, \o\.\0.

1902. Chapman, H. C. Observations upon the placenta and young of Dasypus sexcinctus

1901. Proc. Acad. Nat. Sci. Philadelphia, pp. 1-4.

Ctjenot, L. L'ovaire des Tatous et I'origine des jinneaux. C. R. Soc. Biolog 1903. T. 60 , pp. 1391-1392.

Dawson, E. R. The causation of sex. London, H. K. Lewis Co. pp. 1-190.

1909. Duces. Annates des Sciences Naturelles, Sixieme Ser. Zool. 9. p.l.

1879. Fernandez, Miguel. Beitrage zur Embryologie der Giirteltiere, 1. Zur Keim 1909. blatterinversion und spezifischen Polyembryonie der Mulita (Ta tusia hybrida Desm.). Morpholog. Jahrb. Bd. 39, pp. 302-333. HuBRECHT, A. A. W. Early ontogenetic phenomena in mammals and their bear 1908. ing on our interpretation of phylogeny of the vertebrates. Q. J. M. S., vol. 53, pp. 1-181.

Jhering, H. von. Ueber die Fortpfianzung der Giirteltiere. Sitzungsberichte der

1885. konigl. preuss. Akademie der Wissenchaften. Heft 47, S. 105.

1886. Ueber Generationswechsel bei Saugetieren. Archiv f. Anatomie

und Physiologic, Physik. Abteilung., s. 442-450. Nachtrag zur Entwicklung von Praopus. Ebenda. s. 541-542. Jenkinson, J. W. A reinvestigation of the early stages of the development of

1900. the mouse, Q. J. M. S., vol. 43, pp. 61-82.

KoLLiKER, A. Lehrbuch der Entwicklungsgeschichte des Menschen. p. 362.

1876. Lane, H. H. Placentation of an armadillo. Science, N. S.Vol. 29, p. 715.

1909. ^

1909. Some observations on the habits and placentation of Tatu Novem cinctum. Bull. State Univ. of Oklahoma, no. 1. pp. 1-li.

1909. A suggested classification of edentates, idem, no. 2. pp. 21-27.

Marschal, p. Recherches sur la biologie et le developpement des Hynenopt^res

1904. parasites. I. La polyembryonie specifique ou germinogonie.

Arch. Zool. Exper., Series 4, vol. 2, pp. 257-335.

Mellissinos, Konst. Die Entwicklung des Eies der Maus. Archiv f. mikr,

1907. Anat. Bd. 70 pp. 587-628.

Milne-Edwards, A. Sur la conformation des placenta chez le Tamandua, 1872. Ann. des Sci. Nat., 15.

1878. Recherches sur les enveloppes fcetales du Tatou a neuf bandes.

Ann. Nat., Ser. 6. Zool. T. 8.

Newman H. H. and Patterson, J. T. A case of normal identical quadruplets

1909. in the nine-banded armadillo, and its bearing on the problems of

identical twins and of sex determination. Biol. Bull., vol. 17,

no. 3, pp. 181-187.


H. H. Newman and J. T. Patterson.

Robinson, Arthur. Observations upon the development of the segmentation

1892. cavity, the archenteron, the geiminal layers, and the amnion in

mammals. Q. J. M. S., vol. 43, pp. 369-456. RosNER, M. A. Sur la genese de la grossesse gemellaire monochoriale. Bull.

1901. Acad. Sc. de Cracovie.

Selenka, Emil. Die Blatterumkehrung im Ei der Nagethiere. Wiesbaden,

1884. pp. 67-99.

SiLVESTRi, FiLippo. Contribuzioni alia conoscenza biologica degli Imenotteri

1906. parassiti. I. Annali R. Scuola Sup. d' Agricoltwa. Portici. vol.

6. 15 Gennaio, 1906.


a.a., amniotic attachment to wall of

vesicle. at., allantois.

al.en., allantoic entoderm, al.ms., allantoic mesoderm. am., amnion. am.c, amniotic cavity. am.c.c, amniotic connecting canal. a.v., aborescent villi. b.b., belly-stalk bands. b.c, blood cords. b.s., belly-stalk. b.v., blood vessel. c, cervix of uterus. c.a., clear area of Trager. c.am.c, canal of the common amnion. c.e., canal enlargement. C.I., corpus luteum. CO., coelome. d.b., dorsal bridge. d.n., dorsal notch. ec, entoderm. en., entoderm. e.v., extra chorionic vesicle. ex.c, extra embryonic body cavity. f.g., fore-gut.

f.n.t., floor of the neural tube. f.t.. Fallopian tube. f.u., fundus end of uterus. f.v., flattened villi of Trager. h.f., head fold. h.ms., head mesoderm. h.p., head process. i., infundibulmn of the Fallopian tube.

i.l., intestinal loop. I.S., lymph sinus. ms., mesoderm.

ms.co., mesodermal connection. m.p., medullary plate. n.ch., notochord. n.g., neural groove. n.l.l., notch of the left lateral lobe. 0., ovary.

p. am., posterior amniotic process. p. am.c, posterior amniotic cavity. p.p.h., protochordal plate of Hubrecht. p.p., primitive pit. p.r., placental ring. p.s., primitive streak. s., somite.

s. am.c.c, supernumerary connecting canal of amnion. s.t., sinus terminalis. S.V., scale-like villi. t.m.p., tip of the medullary plate. tr., Trager. tr.c, Trager cavity. tr.c, Trager epithelium. tr.k., Trager knots. u.m., uterine mucosa. v., villi. vg., vagina. y.s., yolk-sac. y.s.en., yolk-sac entoderm. y.s.w., yolk-sac wall. I, II, III, and IV, refer respectively to the ventral, right lateral, dorsal, and left lateral embryos.

Development of the Nine-Banded Armadillo. 413




Fig. 7. The genitalia of an adult virgin female as seen from the dorsal aspect b.l., broad ligament; c, cervix of uterus; c.l., coprus luteum in left ovary; /. fundus end of uterus ;/.<., fallopian tube; 2., infundibulum; o., ovary. X 3.

Fig. 8. Portion of the yolk-sac wall in the region of the area vasculosa (see fig. 15). It shows three blood cords in section. These are made up of a central core of solidly packed cells, b.c, which are surrounded by the mesodermal epithelium, ms. X 215.

Fjg. 9. Cross section of the Trager of our youngest vesicle (fig. 12), showing three adjacent Trager cords or knots (tr.k) ; tr.e., Trager epithelium. X 265.


H. H. Newman and J. T. Patterson.


Fig. 10. Crosssectionof scale-like villi, f., of the vesicle in fig. 14. TheTrager knots are still covered with a thin epithelium. The epithelium of the villi has become a syncytium. The mesoderm has proliferated cells which have invaded the villi, but as yet blood formation has not taken place. X 265.

Fig. 11. The tip of a villus from a more advanced stage, showing, in addition to the features described in pi-eceding figure, the well developed blood vessels, b.v. X 265.

Development of the Nine-Banded Armadillo.


«=?i^- ,-..■.>, 1-// tvi . • T-^AXA c/



^feS;.-:i. --.■■■:-:■,■ ■•:f^ .-.J'- ^':<h^




Fig. 12. A detailed drawing of vesicle no. 10 as seen from the ventral side as a semi-transparent object. The embryos in white (I and IV) are on the upper side of the vesicle, and since there is an inversion of germ layers, these are seen from their ventral aspects. Embryos II and III are shaded, and lie on the far side of the vesicle. For a fuller description see text. X 9.


H. H. Newman and J. T. Patterson,

— am




liG. 13. A detailed drawing of embryo I (fig. 12)as seen from the dorsal aspect, that is, as viewed from the inside of the vesicle, am., amnion; a.mc.c, connecting canals of the amnion; j^.am., posterior amniotic process; b.s., belly-stalk; tr., Trager. X 25.

Development of the Nine-Banded Armadillo.



^"-^ c a


Fig. 14. Ventral view of vesicle No. 18, seen as a semitransparent object. The Trager is covered with scale like villi, which overlap the lower margin of the yolksac. This vesicle should be compared with that shown in fig. 12, in which the lettering is the same. X 5.



H. H. Newman and J. T. Patterson.




FiG. 15. A detailed drawing of embryo I (fig. 14) as seen from the dorsal side. A photograph of this embryo is shown in fig. 31. al., allantois; am. c.c, connecting canal of amnion; b.b., belly-stalk band, note that the band is much more distinct on the left side than on the right; p.s., primitive streak; s.v., scale-like villi; tr., Trager region, which shows the villi as seen from their under sides. For a fuller description see text. X 21.

Development of the Nine-Banded Armadillo.


m S



Note — Figs. 16-23 represent a series of transverse sections taken through various regions of an embryo from the same vesicle as that shown in fig. 13.

Fig. 16. A section taken through the anterior end of the medullary plate. The most important feature of this section is the thickening of the entoderm to form the "protochordal plate" of Hubrecht, p.p.h. X 130.

Fig. 17. A section taken through the medullary plates at a point lying half way between the fore end of the head process and the anterior tip of the embryo. The entoderm is distinct from the mesoderm, which is scarcely more than one cell thick. X 130


H. H. Newman and J. T. Patterson.



P'iG. 18. A section taken through the middle of the head process. In this region the entoderm is very intimately associated with the mesoderm, especially in the central part of the section. X 130.

P'iG. 19. A section taken through the primitive pit. It shows the primitive streak proliferating mesoderm in the characteristic manner. X 130.

Development of the Nine-Banded Armadillo. 421



Fig. 20. A section taken through the connecting canal, which is seen to be composed of two layers, ectoderm on the inside and mesoderm on the outside, and is loosely connected with the mesoderm of the yolk-sac wall. X 143.

Fig. 21. A section taken through the tip of the medullary plate. This is the first section that shows the anterior end of the protochordal plate of Hubrecht. Note that there are a few scattering mesoderm cells (?ws.) that have wandered in between the plate and the ectoderm. X 143.

Fig. 22. A section taken through the belly-stalk at the level of the mouth of the allantois {al. ) . The cavity of the posterior amnotic process {p.am.c. ) does not cover more than one-half the width of the section. The mesoderm of the belly-stalk extends laterally to form wing-like processes. These are the belly-stalk bands {b.b.) through which the umbilical blood vessels pass to the Trager. X 143.


H. H, Newman and J. T. Patterson.

Fig. 23. A section taken through the belly-stalk near the posterior tip of the allantric entoderm {al. en.). The mesoderm is indistinctly divided into two portions. (1) that forming the belly-stalk bands, and (2) that part immediate'y surrounding the allantoic tube — this may be called the allantoic mesoderm (al.ms.). The belly stalk is here separated from the wall of the yolk-sac by a space (ex.c ), which is only a part of the general extra-embryonic cavity. X 143.

Development of the Nine Banded Armadillo.
















-:3 ^

03 (p

CD ^

o - CO


O <u

to ' '

O (B

g ^ CO


S fH


^ -^

v- <B

o o

03 OJ

-^ "S

CO 03

r^ a.

CO t-i


a o



02 H

"3 CD

-►^ a 'Ec <^

o :=;

^ o3

03 >,

i 03 +2

.JL tu


^ -g

fe a o o















































Qi ^




^— '


. ^




d rt^


-2 a -i

■7^ Sh o3

O o <B

^ -O jd

'^ O „,

^ to Ji

- S "^

M _ •+^ 03 to 0)

bC -^

.S 0^

go CI, ^


.2 T3

-7^ -^


.2 =3 +^ . ^








"^ <X> u CO

° X

a .

o ^

1 « 

O -73 " -(J

2 o --^

_,^ 03 S3 tH

>> CO


to Qj d

QJ O rj

+j '^H :2

5 -G °


Sh O


M 03

S -°

o3 t-i

•" a



o to

^ bC

0-2 o 0)

03 ■+^

s .S ^ o



CD 'TT! CD o

t>. o3


CO <1

' • "^

03 ^-^ 03 ^










o3 O

o -fi O

-^ 2



5 s

-^ .2

03 O



Note, figs. 26-29 represent a series of transverse sections of a five-somite embryo.

26. A section through the region of the head-fold. The brain vesicle is in the process of formation, and the neural groove (n.g.) has become very deep. The notochord {n.ch.) is represented by a row of cells, and to each side of it the entoderm is bayed to form the pharyngeal pouches (ph.g.). X 68.

27. A section through the somite region. The somite shows a distinct cavity, and the coelomic cavity is forming. The entoderm is beginning to close in beneath the notochord. X 68.

28. A section through the proximal part of the allantoic tube. The bands of the belly-stalk have become much folded, and contain a number of umbilical blood vessels. The posterior amniotic process has become reduced to a very small tube. X 68.

29. A section through the posterior mesodermal connection (ms.co.) of the belly-stalk. The Trager shows the villi in the process of formation. X 30.





■'" ^ nc h. p-h.p



n.c h


en a !.e n. 3l.m s.






30. One of the five somite embryos (III) of vesicle No. 18 (see figs. 14 and 32). Note how tlie embryo is attached to the Trager by means of the belly-stalk (b.s.). The area vasculosa, like that of the chick, does not extend in to the embryo, but is separated from it by a clear space which corresponds to the area pellucida. On the right is seen the compound sinus terminalis (s.L) lying between the vascular areas of the two contiguous embryos. The posterior prolongation of the amnion is not clearly seen, but its extreme tip is indicated by the leader, p. am. X 16.

31. A seven somite embryo (I) of this same vesicle. For a description of this embryo see the detailed drawing shown in fig. 15. X 16.

32. The dorsal view of the vesicle reconstructed in detail in fig. 14. The cervix end is slightly torn and is turned under, consequently the common amnion and its canals are not shown in the photograph. The turning under of the torn piece also makes the vesicle appear shorter than it really is. At o.m. may be seen the scalelike villi beginning to overgrow the lower portion of the yolk-sac. X 2.15.

33. A vesicle cut open along the mid-ventral line and spread apart to show the pairing of the embryos. It will be noted that the embryos are arranged so that the right-hand pair (III and IV) is the mirrored image of the left-hand pair (I and II). At this stage the amnia are still distinct, and in shape are oval with the broad end directed toward the fundus. X i










34. A view of the fundus end of a vesicle which contained embryos measuring 31 mm. head rump length.* In the portion of the vesicle lying within the margin of the placenta are seen four window-like spots. These are the areas where the amnia come in contact with the wall of the vesicle. The fundus end is now practically free of villi. X |

35. A view of the cervix end of a vesicle in which the embryos measured 31 mm. The clear yolk-sac is seen through the opening in the rather thick placental overgrowth. The margin of this opening represents the place where the placenta is attached to the uterine mucosa at the cervix end of the uterus. X f

36. The dorsal view of a vesicle which is still attached to the cervix of the contracted uterus. This vesicle shows a distinct placental bridge ip.b.) connecting the lateral placentae, and also a number of blood vessels at the fundus end. Embryos 32 mm. in length. X f

37. A view of the fundus end of a vesicle which contained embryos measuring 33 mm. This view shows two points worthy of especial note: (1) the four-lobed appearance of the fundus membrane, due to constrictions occurring between the fundus areas of the individual embryos (seen more clearly before fixation) ; (2) the persistence of a few villi, which in the photograph appear as scattering black

, 5^pecks. X f .. ;38. A view of the ventral side of vesicle, with embryos measuring 36 mm. The cervix end of the yolk-sac is clearly visible, and blood vessels are seen at the fundus end. The placental bridge although present is not clearly brought out in the photograph. X §

39. A view of the vcitral side of a vesicle which contains embryos measuring 53 mm. The division of the zone-like placenta into right and left halves is clearly brought out. The fundus end of the vesicle is now practically free of both villi an 1 blood vessels, and the membranous area at the cervix is much larger than in the preceding figure. X f

^Unless otherwise stated, the length of the embryo will mean the head-rump measurement.










40. The dorsal view of a vesicle in a rather advanced stage of development. The embryos measure 155 mm. from tip to tip. The dorsal notch, d.n., although extending down to near the meddle of the vesicle, does not completely separate the lateral placental discs. X f

41. Dorsal view of a vesicle showing the difinitive condition of the placenta. The placenta is divided into two lateral discs, each of which is distinctly bilobed. The notch between the two lobes of the left lateral (on right) disc is clearly shown in the photograph {n.l.l.). The discs are united to each other both on the dorsal and ventral side by placental bridges, the one on the dorsal side {d.b.) being the narrower. The original arborescent villi at the cervix end have greatly degenerated, and have become reduced to flat, blunt knobs. The embryos in this vesicle are about 210 mm. from tip to tip. X |.

42. Right lateral view of a uterus showing a dorso-ventral bilobing. Embryos are 48 mm. long. X ^.

43. Ventral view of a pear-shaped uterus, which contained embryos measuring 52 mm. This and the preceding uterus show two of the several forms that have been observed. X |











44. A vesicle split open to show the internal relationships of the different parts. The amniotic connecting canals are seen to pass from the anterior ends of the amnia to the spot occupied by the common amnion. This vesicle also shows a supernumerary canal (s.am.c.c.) extending from a small vesicle in the Trager wall to the canal belonging to the lower, right-hand embryo. In the entire condition the vesicle measured 24 mm. wide by 29 mm. long, (see fig. 3 for a diagram of the placenta.) Very slightly enlarged.

45. A vesicle laid open in a manner similar to the preceding. At the distal end of each canal is shown a series of bead-like enlargements (c.e.). The origin of the canal from the anterior tip of the amnion is shown with especial clearness in the embryo lying nearest the foot of the plate. In the entire condition the vesicle measured 24 mm. wide and 30 mm. long. Very slightly enlarged.







46. A vesicle cut open along the mid-ventral line to show the relationship of the embryos to each other and to the wall of the vesicle. Each of the four amniotic partitions (a), which have been cut off close to the chorionic wall, lies just to the left of the umbilical cord. These are attached to the wall near the tips of the placental lobes at the fundus end. The left lateral placental disc is indistinctlyseen thiough the chorionic wall, and the notch separating it from the right lateral disc is marked with the larger "n", while that indicating its division into the two lobes is designated by the smaller "n," X 3.

47. A photograph of vesicle no. 108, which contained five embryos. This vesicle was cut open along the mid-ventral line. Embryos nos. I, II, and II, are attached to the large, right lateral placental disc, and embryos III and IV to the smaller, left lateral. (Seetextfor a fuller description and significance.) X |.











From Dartmouth College, Hanover, N . H.



Introduction 425

Historical review 426

Definition of terms 437

Material and methods 438

Division into stages 441

Description of embryos 442

General conclusions 477

Summary 479

Bibliography 481


The investigation of which this paper is an account was conducted in the Zoological Laboratory of Dartmouth College under the direction of Dr. William Patten. The method which Dr. Patten used in his studies of the development of the nervous system and sense organs of arthropods, an examination of external markings of specially prepared embryos of a very early stage, has been applied here to a study of these organs in Amblystoma. I wish to express my indebtedness to Dr. Patten for his suggestions upon my entering on the work and for his careful supervision throughout.

It is the purpose of this paper to offer a slight contribution to the solution of the problem of vertebrate cephalogenesis. The subject has been treated, however, in its narrower aspect strictly as a problem in amphibian embryology with such general references to wider questions as has been required to make clear the history of the work in this field of research.


426 Lei and Griggs.

Historical Review

Research in the field of amphibian embryology has already yielded results of considerable value. Neuromeres have been observed not only in the neural plate but also in the neural crests and the neural tube; the anlage of the lateral eyes has been traced back to pigment spots, and the early origin of the ear in relation to lateral line organs has been observed.

Before reviewing these topics, however, it will first be necessary to clear the way by considering two minor points, which have been the cause of considerable confusion, the closing of the blastopore and the formation of a series of grooves in front of the blastopore.

The closing of the blastopore has been made the subject of extensive research. It is agreed that the circular form by a more rapid inward lateral growth becomes an oval, then a narrow slit. The next change, however, is a matter of dispute.

Jordan ('93) finds that in the newt Diemyctylus the next step is a fusion of the walls of the slit for a very short distance either at the posterior end or at the anterior end or, as is more usual, at both ends at once. Next there is a fusion of the two walls at the center of the slit. Thus there are formed two temporary pores the distance between which is slightly less than the original diameter of the circular blastopore. The anterior of these two pores is the opening of the neurenteric canal and the posterior one becomes the anus.

Eycleshymer ('95) in his study of Ambly stoma and Rana palustris confirms Jordan's statement of the origin of the anus and of the neurenteric canal but he does not describe any preliminary fusing at the ends of the slit-like blastopore. He finds that there is occasionally "si pear-shaped opening with the smaller end anterior instead of the usual dumb-bell outline." He describes the position of the anus when first formed as lying just outside the neural crest.

Morgan ('97) finds that in the Urodeles the anus is formed as described by Jordan and Eycleshymer but in the Anura here is a concrescence o the slit-like blastopore posteriorly and

The Nervous System of Ambly stoma. 427

the anus appears later as a new structure at the point where the posterior end of the blastopore disappeared, a condition which had already been described by Robinson and Assheton as "a reopening of a temporarily closed blastopore." Morgan was the first to observe that the thickened blastoporic lip was bounded by a definite groove, which he saw, however, only in front of the blastopore as a sickle-shaped" depression. This thickened blastoporic lip is, according to Semon ('01), more prominent in Ceratodus where it forms a broad circular rim surrounding the slit-like blastopore, and this rim is bounded on the outer edge by a circular groove, a condition which, as will be shown, is very similar to that found in Amblystoma.

The closing of the blastopore leaves on the surface of the egg a narrow groove commonly called the primitive groove. This structure was confused by the earlier investigators with other grooves lying in front of it. Miss Johnson ('84) has recorded her observation of but one groove running over the surface of the egg for a distance equal to about three fourths of the length of the neural plate. She homologizes the groove with the primitive groove of higher vertebrates. Miss Johnson makes one statement which rather refutes her theory of the presence of but one groove when she says that " the front end of the primitive groove deepens into a distinct pit." It will be shown that in Amblystoma this pit is in reality a distinct groove which has an entirely different origin from the groove which arises from the closing of the blastopore.

Robinson and Assheton ('91) consider only the posterior portion of the long groove the true primitive groove, limiting the term to that portion of the groove which is formed by the lips of the blastopore. Jordan ('93) is more exact in his definitions. He holds that the first or primitive groove never quite equals the diameter of the original blastopore, and that it shows a fusion of the primary germ layers throughout its length, while the second groove shows no fusion of layers, although there is an apparent fusion" at the anterior end where Miss Johnson found an anterior pit.

428 Leland Griggs.

Eycleshymer ('95) in his observations on Amblystoma not only observed a distinction between neural and primitive grooves but he records a division of the neural groove into two parts. The anterior part of the neural groove usually appears first before the neural crests are clearly formed, and then "sl second groove is often observed lying between the posterior end of the neural groove and the blastopore." This groove he calls the dorsal groove." He notes that it is sometimes absent and again sometimes the neural groove appears to run forward as a continuation of the slit-like blastopore." He considers the dorsal groove as really "a part of the neural groove having arisen in precisely the same manner." His figures indicate in later stages an anterior pit, but no mention is made of it. Eycleshymer has thus gone farther than any other writer in analyzing the single primitive groove of Miss Johnson into three grooves, "the primitive groove," "the dorsal groove," and "the neural groove."

More recent writers have failed to notice these three grooves. Morgan ('97) who adopts the old terminology applies the term "primitive groove" to the whole series of grooves in the frog, but he makes the significant statement that in older embryos "the primitive groove is narrower." It will be shown in this paper that this apparent narrowing of the old groove is in reality the appearance of a new groove after the disappearance of the old one.

Semon ('01) insists that in Ceratodus there is undoubtedly a lengthening of the groove which arises from the closing of the blastopore. It must be admitted that this is what would be expected since the groove lies in the growing region of the embryo. The condition in Amblystoma seems to resemble very closely the structure which he has figured for Ceratodus. Earlier authors who claimed that the primitive groove was shorter than the diameter of the original blastopore were correct so far as the first appearance of the groove was concerned, but Semon has shown that they overlooked the fact of a later growth which can be demonstrated in Ceratodus only by sections which show the typical fusion of the germ layers extend

The Nervous System of Ambly stoma. 429

ing over a distance considerably greater than the diameter of the blastopore.

The natm'e of these grooves is not only in itself an interesting problem which is not yet wholly solved, but the relation of the grooves to the nervous system is also very important. It has quite generally been taken for granted that these grooves, one or all of them, are neural" grooves, but this relation has never been well established.

In turning now to the more important problem of the early segmentation of the neural plate we find considerable divergence of views. Kupffer ('93) describes the cephalic plate of Salamandra atra as being divided into eight segments or primary neuromeres. The cephalic plate or "Hirnplatte" has no definite boundary posteriorly but appears to coincide in extent with the later Archencephalon," which the author finds to be a rather vaguely defined vesicle comprising the region of the fore-brain, mid-brain and a part of the hind-brain. Froriep ('91 and '93) has made similar observations, although less complete on Salamandra maculosa and Triton. In the cephalic plate of the former he saw three or four segments, and in that of the latter he saw five, but in both cases there is an unsegmented tip which he thinks is large enough to represent three or four segments. Froriep, however, thinks this appearance of segmentation is merely due to the structure of the underlying mesoderm.

Later authors have made observations similar to those of Kupffer and Froriep, and, like them, have failed to agree on an interpretation. Eycleshymer ('95) finds in the neural plate of Necturus and of Amblystoma several large segments which, he considers, are caused by the mesoderm. Locy ('95) describes a few faint lines of division in the open neural plate of Amblystoma and four or five prominent divisions in the anterior end of the plate of Rana palustris, but since these divisions do not agree in number with the divisions of the neural plate which he considers to be true neuromeres he denies their metameric value. Hill ('00) has found in the solid anlage of the trout brain ten segments extending from the anterior tip of the plate to the

430 • Leland Griggs.

region of the anlage of the ear. Two of the dividing grooves appear deeper than the rest and so furnish convenient landmarks which no previous investigator had been able to find. The anterior of the two grooves proves to lie between the fore and the mid-brain, while the posterior one lies just back of the cerebellum. Thus Hill is able to show that there are three of these segments in the fore-brain, and two in the mid-brain. The general features of this segmentation were confirmed by the study of the chick of one somite although here the larger grooves were absent.

A comparison of the results obtained by these various authors is difficult, especially as to the number of segments, since Hill is the only one to find definite persisting landmarks. But the numerous observations as to the presence of some kind of division in the open neural plate is sufficient to warrant further research.

Turn ng now to the segmentation of the neural crests as they first appear at the sides of the neural plate we find that in Amblystoma embryos such a segmentation has been observed by Eycleshymer ('95) and Locy ('95). The former considers the appearance of segmentation merely an artificial scalloping due to the reagents used, because he finds variation among various embryos. The latter, however, regards the segments as true neuromeres, but he did little work on Amblystoma and that merely to confirm his work on Acanthias.

In Acanthias Locy ('95) finds a series of segments in the neural ridges which he is able to trace through the closing of the neural tube and the formation of the brain vesicles. The fore-brain contains three segments, the mid-brain two and the hind-brain nine. Hill ('00) has confirmed these observations in his study of the chick embryo. He finds that the grooves which separate the segments extend entirely across the neural plate. This is the only observation yet made showing that the segmentation of neural crests agrees with that of the open neural plate. The accuracy of this observation has been called in question by Kupffer ('03).

The Nervous System of Amblystoma. 431

The segmentation of the closed tube in the region of the medulla was observed by the early anatomists and embryologists (Balfour, '81). These segments were first carefully studied by Orr ('87) in his work on Anolis. He was the first to use the term neuromere, and his description of a typical neuromere has ever since served as a criterion of a neural segment in the closed neural tube. He says: "Each neuromere is separated from its neighbors by an external dorso-ventral constriction, and opposite this is an internal sharp dorso-ventral ridge — so that each neuromere {i.e., one lateral half of each) appears as a small arc of a circle. The constrictions are exactly on each side of the brain. The elongated cells are placed radially on the inner curved surface of the neuromere. The nuclei are generally nearer the outer surface and approach the inner surface only toward the apex of the ridge. On the line between the apex of the internal ridge and the pit of the external depression, the cells of adjoining neuromeres are crowded together, though the cells of one neuromere do not extend into another neuromere." He found six neuromeres in the hind-brain. The first and fifth have no nerve connection; the second is connected with the fifth nerve; the third to the sixth nerve; the fourth to the seventh and eighth nerves; the sixth to the ninth nerve. He considers that the mid-brain is a single neuromere and that the thalamencephalon comprises two neuromeres. He does not regard the secondary fore-brain as being a true neuromere. McClure ('90) was the first to claim that these neuromeres of the closed tube extended in a regular series to the anterior tip of the brain. He found that Amblystoma nad one less neuromere in the hind-brain than Orr and Hoffman had found in reptiles. He considered that the the difference was due to a fusion of two neuromeres. He found in the forebrain evidences of two neuromeres and possibly a portion of a third. Waters ('92) confirmed McClure's account of the absence of one neuromere in the hind-brain of Amblystoma, but in the teleost brain he found the whole number, in the hind-brain six, and assigned to the mid-brain two, and to the fore-brain three. Kupffer ('03) from the results obtained Dy McClure and Waters concludes that Amblystoma presents

432 Leland Griggs.

an exception to the usual number of neuromeres in ttie hindbrain.

Very extensive observations on the neuromeres of the closed tube have been made by Locy ('95) and Hill ('00). Their work has been of especial value owing to the fact that they have shown the direct correspondence between the segments of the neural tube and those of the neural crests before the tube is formed. By following carefully the history of the segments in the anterior end of the crests they observed in the anterior end of the neural tube a few transitory segments which had not been seen before. Locy found in the fore-brain of Acanthias three segments, in the mid-brain two and in the hind-brain nine, counting three in in the vagus region which are not included in the brain by earlier writers. Hill in his work on the trout and chick confirmed Locy's investigations. Johnston ('05) also claims to have confirmed these results in his work on Amblystoma which he has not yet published.

Thus we have presented to us by Locy, Hill and Johnston a fairly complete account of the neuromeres of the brain. According to this account there are eleven similar divisions in the anterior end of the neural plate which constitute a cephalic plate, not, however, marked off from the rest of the plate except by their subsequent history. The segments extend laterally across the neural crests and after the tube is formed the divisions extend completely around it. The segments in the hind-brain are those already investigated by earlier authors. They persist until the nerve relations can be fairly well determined. In the fore-brain and mid- brain, however, the segments are transitory. In the fore-brain they completely disappear before the differentiation of the secondary fore-brain and thalamencephalon.

Kupffer ('03, '05) has recently described a series of segments, his so-called secondary neuromeres, which he has shown to be quite generally present in vertebrate embryos. His fore-brain divisions telencephalon, parencephalon, synencephalon, although they agree in number with Locy's divisions, are evidently not the same, for Locy has shown that his divisions disappear before there is any trace of this later differentiation of Ihe fore-brain.

The Nervous System of Amblystoma. 433

Kupffer's first and second mesencephalic neuromeres, however, are evidently the same as Locy's. Kupffer, as has already been stated, considers these divisions described by Locy and himself as of secondary origin and not true primary neuromeres, maintaining that the true neuromeres are the large divisions in the open neural plate which he has described as being apparent in some forms of amphibians.

Neal ('98) is the severest critic of the theory of neuromeres as developed by Hill and Locy. After an investigation of Acanthias, the same form which Locy used, he disagreed with Locy both as to observation and interpretation. He found that the lobes on the opposite sides of the plate do not correspond in number or position, neither do they show any definite relations to the mesodermal somites," and there is no constancy in different individuals." He found these apparent segments transitory and was unable to find any relation between them and the later ventral segmentation of the neural tube. He pointed out another apparent inaccuracy in Locy's work in that "the line which separates the expanded cephalic plate from the region posterior to it marks the posterior boundary of the auditory invagination," instead of lying "just in front of the point where subsequently the vagus nerve begins" as Locy saw it. From this he concludes that Locy can not have traced his so-called neuromeres correctly into the later divisions of the brain.

Neal accepts Orr's criteria of a neuromere but he adds the following important point — " the best criteria are such as associate the supposed neuromeres metamerically with other structures known to be segmental." Judged by this standard he concludes that the fore-brain and the mid-brain represent each one neuromere, the former being associated with the anterior head cavity of Piatt and the sensory olfactory nerve, and the latter with Van Wijhe's first somite and the ophthalmicus profundus of the fifth nerve. Neal's method is undoubtedly very valuable in showing positive evidence in favor of the presence of true neuromeres in the hind-brain, but it is very difficult to apply so severe a test to the mid-brain and the fore-brain because it has been demonstrated so clearly that the nerve com

434 Leland Griggs.

ponents and the mesodermal somites of this region have undergone profound alterations. If, as the most recent investigation seems to show, the nervous system, appearing first, presents a simpler and more unaltered condition than the other two systems, then it may well serve as a basis for the study of segmentation of the head; and other organs should be shown to correspond to it rather than vice versa.

From this survey of the literature dealing with the segmentation of the nervous system of the Amphibia and allied forms it seems fair to conclude that the presence of neuromeres has been proved conclusively. As Hill ('00) says, There is substantial agreement among observers as to the six segments in the hind-brain." The problem of the segmentation of the fore-brain and mid-brain, however, may still safely be considered unsolved. While the evidence in favor of the segmentation of the neural crests and the open plate seems very strong, the difference of opinion as to the constancy, number and relations of the segments is so great that considerable more research will be necessary before we have a fully established theory of primary neuromeres.

In solving the problem of the morphology of the vertebrate head the subject of sense organs is second in importance to that of neuromeres. The origin of the olfactory organ, of the eyes both lateral and parietal, of the ear and the lateral line organs have all been made the subject of- careful investigation, particularly in the Amphibia and other groups among the lower vertebrates. "For the purposes of this paper only one of these topics will be considered in detail, that of the origin of the lateral eyes.

It was observed by Balfour ('81) that the anlage of the eye appeared before the closing of the neural canal. ■ He has quoted with approval a statement by Lankester that the original vertebrate must have been a transparent animal with an eye or pair of eyes within the brain."

Whitman ('89) observed the anlage of the eyes on the surface of the amphibian egg in the open neural plate stage. In Necturus before the neural crests have begun to move together he found that "the basis for the eye is already discernible as a cir

The Nervous System of Amblystoma. 435

cular area." Whitman has proposed a theory which to-day attracts more attention than Lankester's theory, although it will be seen that the two theories do not necessarily conflict. His hypothesis is as follows: The medullary plate of the vertebrate is undoubtedly an enormous extension of the ancestral invertebrate plate. Sense organs lying originally outside the neural plate have probably in consequence of this extension in width, been brought within the medullary area."

Eycleshymer ('94) continued Whitman's work on the origin of the eyes of the Amphibia. Of the three forms which he studied, Necturus, Amblystoma, Rana palustris, he found the last named the most satisfactory for his purpose on account of the deeper pigment of the eye spots and their greater histological differentiation. He studied very fully the histology of the pigmented areas, but he leaves it a little in doubt from his plates and descriptions whether those areas are actually on the neural plate or between the plate and the crests, a very important point in testing Whitman's theory that the ancestral invertebrate plate has widened out so far as to irfclude the region of the lateral eye. His conclusion in favor of the hypothesis that the vertebrate eye must have been located originally within the brain, and his statement in support of this theory that "this was precisely the case in Rana palustris," must be taken to mean that the original brain comprised both neural plate and neural crests.

In criticism of this theory of the relation of the eyes to the primitive brain it may be suggested that the neural plate proper represents the central nervous system of the ancestors of the vertebrates, and that the anlage of the vertebrate lateral eyes are located between the neural plate and the neural crests. If this be true it is probable that the medullary plate has not undergone such an enormous extension as Whitman supposed and that the ancestral vertebrate eyes were not located on or in the brain, as Lankester and Eycleshymer supposed, but on the margin of the neural axis, as in arthropods, and that by a folding over of the non-nervous sides and roof of the brain they became in included in the neural canal in a similar way to that described by Patten for the arachnids (Patten, '89). Such a view as this obviates the necessity of assuming that the nervous tissue in

436 Leland Griggs.

creased in bulk by appropriating other tissues. To prove the correctness of this view it would be necessary to show that the anlage of the eyes does not lie on the neural plate. If we turn to work outside the field of amphibian embryology for evidence we find that the neural plate is not very sharply marked off from the neural crests in those animals which have been studied most. Fororiep ('05) locates the '^Sehgrube" of Torpedo on the edge of the neural plate in the area called by His Flugelplatte," and Locy ('95, '97) locates the optic groove of Acanthias and the chick on the sides of the neural furrow as the walls begin to rise to form the canal. None of these forms, however, has a clearly marked groove between the neural plate and the neural crest and therefore, although these authors seem to believe that the eye spot is located on the anlage of the brain, it is clear that this interesting point still remains unproved.

Locy's work on Acanthias is especially interesting inasmuch as he was the first to find '^accessory optic vesicles." He observed that the first, or true optic groove, covered the space of three neuromeres and following this was a series of five or six accessory vesicles occupying the space of one neuromere each, although this relation to neuromeres was not always strictly observed. Locy has compared this arrangement with that of the same organs of the leech embryo described by Whitman ('89). In the process of development the accessory vesicles degenerate except the first pair which according to Locy form the pineal eye. A study of the chick confirmed the work on Acanthias.

This review of the literature dealing with the problem of the development of the nervous system of the Amphibia has revealed the need of further research along several lines. First, the series of early grooves which appear at about the same time as the neural plate should be carefully investigated and their relation to the nervous system determined. Then attempts should be made to trace Kupffer's primary neuromeres into the later divisions of the brain, and the exact position of the anlage of the eye should be determined and also its relation to numbered neuromeres.

The Nervous System of Amblystoma. 437

Definition of Terms

Since, as has been shown in the preceding review, authors have not agreed upon tferms and since many of the old terms have proved to be misleading it seems necessary to adopt a definite system of nomenclature. The terms used in this paper are as far as possible those of earlier writers. Where a choice between two old terms was made, or where a new term was chosen, it was done with the purpose of showing the position and relations of the structure involved without suggesting doubtful homologies.

In the first place the common term, primitive groove, has been avoided altogether. This word has been used in various senses by different authors, and furthermore it suggests a doubtful homology which it is not the purpose of this paper to discuss. The groove formed by the closing blastopore will be called the blastogroove. The two grooves lying in front of the blastogroove which have frequently been called the primitive groove will be called the anterior and posterior germinal depressions. Since these terms are perfectly colorless as regards homology they leave the way clear for an unprejudiced discussion of the meaning of the structures.

The term, neural groove, which has often been applied to one or all of these grooves just mentioned is not used in this connection, because the neural nature of these three early grooves has not been demonstrated. The true neural groove is a later structure appearing after the other grooves have begun to degenerate and persisting as the faint narrow groove lying in the bottom of the neural canal.

The term, neural plate, will be used in its usual sense. The division of this plate into well defined regions demands further definition. The terms, hirn-platte and cephalic plate, which have been loosely applied to the widened anterior end of the plate, will be discarded for the term procephalic lobes, denoting that part of the neural plate which lies in front of a distinct transverse furrow which will be called the transverse cephalic groove. The region behind this groove may be theoretically

438 Leland Griggs.

divided into metencephalic plate and spinal cord plate, although in Amblystoma there seems to be no trace of a dividing groove.

Connected with the neural plate are some less important structures which should be defined. The narrow groove bounding the neural plate will be called the peripheral groove. The low ridge running around the neural plate just outside the peripheral groove will be called by its usual name of neural crest.

In Amblystoma the neural plate undergoes at an early period a sharp downfolding. This will be called the infundibular depression.

The term neuromere is one that needs careful definition. In this paper it will be applied to the early divisions of the neural plate which have been called primary neuromeres. These divisions are considered as having true metameric value. The later divisions of the neural tube will be called neuromeres in so far as they can be shown to be identical with the divisions of the open neural plate. The use of such terms as primary and secondary neuromeres will be avoided because they are contradictions of the idea that a neuromere is a genuinely segmental structure.

To sum up, the old terms neural plate, neural crest and neuromere will be used substantially as they have been used before, while the term neural groove will be used in a new and restricted sense. The new terms, blastogroove, anterior germinal depression, posterior germinal depression, peripheral groove, transverse cephalic groove, procephalic lobes, infundibular depression are new terms applied to newly discovered structures or to newly discovered divisions in old structures.

Material and Methods

Amblystoma punctatum was the animal chosen for study, mainly on account of the large egg which when first laid is about two mm. in diameter. This is larger than any frog's egg and much larger than the egg of Amblystoma tigrinum but considerably smaller than the egg of Necturus. The anlage of the nervous system of Amblystoma punctatum, however, is

The Nervous System of Amblystoma. 439

nearly as large as that of Necturus and the markings on the neural plate are much more distinct.

The abundance of eggs also makes Amblystoma punctatum a good object for study, for, as Locy has already pointed out, in the study of neuromeres a very large supply of material is essential. In the vicinity of Hanover, N. H., the eggs are very common in the early part of April, being found in large bunches in all the small ponds and pools and even in the shallow ditches by the roadside. In the years 1903, 1904 and 1905, between three and four thousand eggs were collected for this study.

In fixing the embryos several fluids were used, chromic acid, picrosulphuric acid, Perenyi's fluid. All three of these fluids proved to be useful for various purposes. For the study of surface views chromic acid gave the best results. The eggs after being dissected out of the main mass of jelly were dropped into a 1 per cent solution. After remaining in this fluid for about half an hour they were freed from the outer envelope by the use of fine pointed forceps and needles and then transferred to a fresh solution of the same strength where they were allowed to remain for about four hours. The material was then thoroughly washed in running water for several hours and finally transferred to 70 per cent alcohol. The alcohol was changed as often as it became turbid. Before the eggs were ready for study they were immersed for a few minutes in a weak solution of eau-de-Javelle, which removed the inner membrane and the last traces of any albuminous precipitate. This treatment gave beautiful surface preparations. For dissection Perenyi's fluid was bei^ter. Aiter older embryos had been fixed in this fluid for several hours and then hardened in 70 per cent alcohol it was possible to dissect out the brain with fine needles and sweep it clean with a delicate brush or the brain could be dissected with the sense organs and the ganglia left attached to it. For paraffin sections Kleinenberg's picro-sulphuric acid was the most satisfactory for fixing embryos.

The division of the embryos into a series of stages proved to be a difficult task. A single set of eggs killed at one time contains several stages, making it necessary to examine the eggs

440 Leland Griggs.

one at a time. The embryos were first divided roughly into the conventional stages such as yolk plug stage, open neural plate stage. These stages, however, owing to the fact that the markings of the open neural plate are very transitory, had to be subdivided.

Some difficulty was encountered in thus subdividing the principal stages owing to a remarkable variation in the younger embryos. Some of the grooves and infoldings were so deep that abnormalities seemed at first sight to be very common. Further study of these variations showed that they were not promiscuous abnormalities but variations along certain well defined lines. Those structures which in a majority of embryos were faintly marked, and in a few embryos were not seen at all, were very plainly marked in others. The deeply furrowed specimens were, therefore, the most valuable for study when it could be shown, as it usually could without difficulty, that the deep furrows correspond in extent and position with the faint furrows in other eggs. Patten has already pointed out in his study of Limulus embryos the principle that in the process of recapitulation those embryos which vary from the normal average type may show best of all some features of the ancestral condition. This seems to be especially true in a study of neuromeres. Some embryos seem to show no neural segmentation in the early stages, the majority show it faintly, a few show it very distinctly, but in all cases the number of neuromeres is the same where they show at all and the size of the neuromeres is approximately the same. Hence we are obviously not dealing with a meaningless variation, but we are justified in making a careful study of the most clearly sculptured neural plates.

A second difficulty in the way of dividing the embryos into a series of stages is found in the fact that several distinct structures are forming on the surface of the egg at the same time and with a relative rapidity which varies with the different eggs. For example, if the eggs are assorted according to the closing of the blastopore the stages will not show a consecutive development of the neural plate, for among the eggs with a wide open


The Nervous System of Amblystoma. 441

circular blastopore there may be some that show the outline of the neural plate and others that show no trace of the plate. This difficulty becomes much greater when the several distinct parts of the neural plate are taken into consideration.

The following division into stages is based upon the development of the particular structure which is made the special' object of study in the general period under consideration. Thus the eggs before the appearance of the neural plate are divided into three stages; these stages depending upon the relative development of the germinal depressions, not upon the condition of the blastopore. After the appearance of the neural plate, however, the peripheral groove, the transverse cephalic groove, the neural crest and the neuromeres are successively the landmarks for division and it is necessary to ignore the germinal grooves, for they show many degrees of degeneration in each stage. After the closing of the neural canal these peculiar conditions are no longer found and assortment into stages becomes comparatively simple.

The method described above will be made clear by the following summary of stages.

Division into Stages

Stage 1. The posterior germinal depression has appeared in front of the blastopore. The blastoporic rim is bounded by a faint narrow groove.

Stage 2. The anterior germinal depression has appeared and the posterior depression shows signs of degeneration from behind forward.

Stage 3. The two germinal depressions appear very closely united. In a few eggs the blastogroove is now formed.

Stage 4. The area of the neural plate has become marked off by the peripheral groove. The condition of the germinal grooves and of the blastogroove varies.

Stage 5. The neural crests have appeared and there are indications of the transverse cephalic groove.

Stage 6. The procephalic lobes become divided into neuromeres.


442 . Leland Griggs.

Stage 7. The first neuromere folds down to form the infundibular depression.

Stage 8. The neural crests move together and fuse to form the neural tube.

Stage 9. The neural crests become segmented. The otic pit appears.

Description of Embryos

Stage 1 (Fig. 1, B). The posterior germinal depression (pgd fig. 1, B), the first structure to appear on the surface of the egg in front of the blastopore, is formed while the egg is still spherical before the yolk plug has been withdrawn. There are no indications of any growth in length of the depression, or of any origin at a constant point from which it grows either anteriorly or posteriorly, but the groove appears at the very outset as a long shallow depression of uniform length in different eggs.

In order to distinguish between this groove and the grooves which appear later it is important to note carefully its character. It extends from a point an appreciable distance in front of the blastopore to a point about half the distance across the upper hemisphere of the egg. It is nearly as wide as the diameter of the blastopore just before the latter changes from its circular to its oval form. Its depth varies owing to the presence of several shallow pits one of which is usually found at the anterior end forming a fairly well marked anterior limit to the depression.

It has already been mentioned that there is considerable difference of opinion in regard to the number of grooves which appear on the surface of the amphibian egg. Miss Johnson ('84), Schultze ('88) and Erlanger ('90) have maintained that there is but one groove, the primitive groove, formed by a concresence of the lips of the blastopore and then extending forward over the surface of the embryo. A study of these early stages of Amblystoma supports the position taken by the later authors, Robinson and Assheton ('91), Jordan ('93), and Eycleshymer ('95), that there are two grooves, the posterior one derived from the blastopore showing a fusion of cell layers and the anterior

The Nervous System of Amblystoma.


■ •■■■■'"■r.X

■-.- - *»'.X,





' ■■■' t^





"- .w::t.::z—iP




m c



Fig. 1. Surface views of the first four stages. A, B, C are views of entire embryos and D, E, F are outline drawings of neural plates. These and the remaining figures of this paper were drawn with the aid of the camera, a, anus; agd, anterior germinal depression; bg, blastogroove; bp, blastopore; br, blastoporic rim; id, infundibular depression; nee, neurenteric canal; pg, peripheral groove; pgd, posterior germinal depression; teg, transverse cephalic groove.

one showing no fusion. In Amblystoma the evidence for this view is still stronger ; for when the posterior germinal depression is formed the blastopore is still wide open, and furthermore there is an appreciable distanc'e between the posterior end of the depression and the anterior rim of the blastopore. It is not the purpose of this paper to enter into a further discussion of the morphology of the so-called the primitive groove. It is desired merely to emphasize the fact that the groove which is here called the posterior germinal depression arises independently of the blastopore.

444 Leland Griggs.

Stage 2 (Fig. 1, A; Plate I, fig. 1). The posterior germinal depression now shows some changes (pgd fig. 1, A). It is narrower, deeper and more clearly defined since the shallow pits have disappeared. The anterior end has not changed in position, but the posterior end has begun to widen out and disappear leaving a considerable distance between the closing blastopore and the posterior germinal depression.

A new groove, the anterior germinal depression (agd) a structure very constant in its nature and present in all eggs of the right stage, has now appeared in front of the posterior depression. In the photograph (Plate I, fig. 1) the depression is not so sharply defined as in the figure. It becomes more and more sharply defined in succeeding stages, however, as may be seen by examining the photographs of older embryos. It is evidently not one of the pits described in the preceding stage since it is deeper, narrower, and longer, and furthermore the pits of the posterior depression have all disappeared at this time. These facts together with its very important later history give sufficient ground for considering it a separate and distinct structure.

This anterior germinal depression is evidently the groove which Eycleshymer ('95) in describing the early development of Rana and Amblystoma has called the neural groove proper, which he found to be connected with the groove derived from the blastopore by a third groove which he called the dorsal groove, evidently identical with the posterior depression described above. Morgan ('97) for Rana has figured the posterior part of what he has named the primitive groove as forming first, and later the anterior part appearing as a continuation of the part first formed. Evidently he observed the same process as Eycleshymer has described more fully. Neither of these authors, however, has described in detail the character of these two grooves nor traced their later history.

The value of adopting the new terms anterior and posterior germinal depression must now be apparent. It has just been shown to be a mistake to call the posterior depression a primitive groove, if by that term is meant a groove derived from the blastopore. It is obviously just as much a mistake to call the

The Nervous System of Amblystoma. 445

anterior depression a neural groove since it forms before any well defined anlage of the nervous system appears and when its later history, as will be shown, shows no particular relation to the nervous system. The term germinal depression does not, by begging the question at the very outset, confuse the problem of the morphology of the various grooves, but leaves the field clear in which to follow the history of these important structures.

Stage 3 (Fig. 1, C. Plate I, fig. 2)). This stage shows the three early grooves present at one time. At the anterior end lies the anterior germinal depression, in the middle the posterior germinal depression, and at the posterior end of the series the blastogroove. These three grooves are distinguished from one another not only by order of appearance and position but also by nature of the grooves themselves as will be shown later by means of transverse sections.

The anterior germinal depression (agd, fig. 1, C) shows some changes when compared with the preceding stages. It is narrower, deeper and longer and is more closely united with the posterior depression. In fact in this stage it is more difficult than at any other time to distinguish between the two depressions. Yet this is possible in every embryo, the majority showing the distinguishing features much better than the one illustrated in Plate I, fig. 2. By reference to fig. 1, B and E, typical representatives of their respective stages, it is seen that although it may be difficult to locate the exact point of union of the two grooves, yet the grooves are readily distinguishable from each other by a difference in width and depth.

The posterior germinal depression shows further signs of degeneration by widening out and disappearing at the posterior end. Scattered shallow pits again appear in an irregular manner as in the first stage

The blastogroove is formed by the concresence of the lateral lips of the blastopore. This groove together with the neurenteric canal can be described better in connection with one of the older stages. It may be noticed here that the neurenteric canal (nee) forms the anterior limit of the blastogroove and that the

446 Iceland Griggs.

latter's length is not greater than the diameter of the circular blastopore.

Stage 4 (Fig. 1, D, E, F. Plate I, figs. 3, 4, 5) . From this time on the three grooves, the development of which has just been traced, undergo rapid modification. The posterior germinal groove shows unmistakable signs of degeneration in all eggs of this stage. In fig. 1, D, and in Plate I, fig. 3 there is no sign whatever of the original groove, a condition which is found in about one-third of the eggs. In a few eggs, although the groove has disappeared, there is a faint dark line marking its former position.

The only remaining portion of the anterior germinal depression (agd) in fig. 1, D, is a short deep groove or rather pit, and this deeper portion of the original groove may be seen in some eggs before the disappearance of the main part of the groove (agd, fig. 1, E). Miss Johnson ('84) seems to have been the first one to describe this "anterior pit" as she calls it but she failed to notice the series of germinal grooves, apparently considering the anterior pit as a part of one long primitive groove. Eycleshymer's ('85) figures of Necturus show a similar structure and he seems to have considered the pit a part of the "neural groove." His figures as well as those of Miss Johnson are representations of older embryos than those of this stage of Amblystoma now under consideration, and the single long groove evidently corresponds to the neural groove proper which will be described shortly. Hence their figures showing the relation of the pit to a single long groove are correct. The history of this pit, however, as will appear from an examination of the various stages shown in fig. 1, indicates that it is in reality the deeper and persisting part of the anterior germinal depression. A study of succeeding stages corroborates this view.

The neural plate in this stage is marked off by a peripheral groove (pg). This new groove does not take its origin from a given point from which it grows forward or backward but from the first it extends for a considerable distance along the sides of the plate and, although very faintly, across the anterior end of the plate. It cannot at its first appearance be traced around the posterior end of the plate. As the photograph and the

The Nervous System of Amblystoma. 447

figures illustrating this stage show, the changes which develop in this groove consist of a widening and deepening and a growth backward toward the blastopore.

Lying in the peripheral groove at a point opposite the posterior end of the anterior depression is a pit or short groove (teg, fig. 1, E, F), one on each side of the neural plate. This pit is the beginning of the broad transverse cephalic groove which later marks off the anterior end of the neural plate from the posterior end. It is visible in a majority of the eggs as soon as the peripheral groove appears but in a few instances as in fig. 1, D, no trace of it appears in this stage. Not onlj^ is the anterior end of the plate, the procephalic lobes as this region may be called, marked off by the groove but also, when the eggs are hardened in chromic acid, by a striking difference in color. The procephalic lobes are darker than the remainder of the plate, a distinction which remains until about the time of the closing of the neural tube. Thus at a very early stage color and a definite boundary mark out the region which proves to be the anlage of the brain. The transformation of this anlage, the procephalic lobes, will be the principal object of attention in the succeeding stages.

Stage 5 (Figs. 2, 3, Plate I, figs. 6, 7). This stage is characterized by the appearance of the neural crest (nc, fig. 2). The crest first appears as a pair of short longitudinal thickenings at the sides of the neural plate opposite the transverse cephalic groove (fig. 2, E). Each half of the crest grows rapidly backward until it reaches the blastoporic rim (br, fig. 2, A) which in a few eggs is still distinctly visible. After the disappearance of the blastoporic rim the two sections of the crest fuse behind the neural plate (fig. 2, B). At a slightly later period the crest is continued around the anterior end of the plate (fig. 2, C). This description of the origin of the neural crest agrees with the more general statement of Eycleshymer ('95) that in the Amphibia the neural bands arise independently" and differentiate in situ."

It has been quite generally assumed that the neural crest which surrounds the open neural plate is identical with the later


Leiand Griggs.

Fig. 2. Neural plate and neural crest just before the appearance of neuromeres, fifth stage, a, anus; agd, anterior germinal depression; ap, anterior pit; bg, blastogroove; br, blastoporic rim; id, infundibular depression; li, lateral infolding; n4, fourth neuromere; nc, neural crest; nee, neurenteric canal; ng, neural groove; pel, procephalic lobes; pep, postcephalic plate; pgd, posterior germinal depression; teg, transverse cephalic groove.

neural crest from which the dorsal ganglia of the spinal nerves arise. Almost conclusive evidence is found in the fact that the crests in the open plate and in the closed tube hold the same position, lying in both cases immediately adjacent to the edges of the plate or to their line of fusion in the tube. Johnston ('05) has recently shown by a series of sections that the rounded or cubical cells of the crest in the open plate stage can be easily followed into the anlage of the dorsal roots of the cerebral nerves. His sections are drawn from Amblystoma embryos. My obser

The Nervous System of Amblystoma. 449

vations agree entirely with his. It will be assumed then in any further consideration of this subject that the generally accepted position in regard to the identity of the neural crests at the various stages is correct.

The relation of the neural crests to the neural plate as pictured in fig. 2 at once suggests the hypothesis of the longitudinal zones of the brain which has been elaborated in detail by Johnston ('02, '05), but this particular theory of the neural crest is not supported by the conditions just described. From the fact that in Amphioxus the greater part of the nerve elements homologous with the neural crest lie inside the neural tube and that many of these elements are found there in fishes and even in Amphibia, Johnston draws the inference that the neural crest was originally a part of the central nervous system and that there "has been a progressive separation of material for the ganglia from the brain." It has just been shown, however, that the neural crest develops at a later period than the neural plate and that the two structures are separated by an appreciable distance. Furthermore it will presently be shown that the crest and plate are not segmented in the same way, a fact already noted by Locy ('95). The evidence from the embryology of Amblystoma then does not show a close relation between neural plate and neural crest but rather it seems to indicate that the plate represents the anlage of the central nervous system of the primitive vertebrate and that outlying structures must have a different homology. This position is further strengthened by a study of neuromeres and sense organs, as will presently appear. It will be shown to have an important bearing on the conception of the structure of the primitive vertebrate head.

Another important event which occurs in this stage is the formation of a new groove, the neural groove, in the region first occupied by the germinal depressions. This groove can readily be distinguished from the posterior germinal depression by the fact that the latter at this time has either entirely disappeared or is fast degenerating while the new groove is deeper and sharper than the old grooves and is much narrower than the posterior depression (ng, fig. 2, D). In Plate I, figs. 4, 5, 6, the posterior

450 Leland Griggs.

germinal depression is in various stages of degeneration, while in Plate I, fig. 1 1 the sharply defined neural groove is seen, being especially prominent in the posterior half of the neural plate. Photographs of succeeding stages, all except Plate I, figs. 8, 9^ show the neural groove. The anterior germinal depression persists as a long narrow pit near the anterior end of the neural groove but not extending exactly to the end of the groove (agd^ fig. 2, F). The neural groove is seen in but a small proportion of the eggs of this stage but after stage 7 is present in all the eggs and persists to the closing of the neural canal. It may properly be called a neural groove. While the two grooves called the germinal depressions appear before there is the slightest indication of a nervous system and begin to degenerate before the anlage of the nervous system is clearly defined, this third groove appears as the neural plate is forming and persists even in the fully formed neural canal. It is the only groove the history of which is closely identified with that of the nervous system.

The germinal depressions present in this stage show no new features. The anterior depression in the majority of eggs is short and deep and clearly marked (agd, fig. 2, A, C; Plate I, figs. 4, 5, 6), but in a few eggs it is impossible to make the distinction between the anterior depression and the posterior depression or neural groove. The posterior depression, as in stage 4, is either absent (fig. 2, A) or is very wide and shallow (fig. 2, B, C, E).

The transverse cephalic groove has now extended in a few eggs until it entirely cuts off the procephalic lobes (teg, fig. 2, A, Plate I, figs. 6, 7). This groove shows considerable variation in length and width as the figures show. Moreover, it may extend clear across one-half of the neural plate before it shows at all on the other half (teg, fig. 2, C; Plate I, fig. 6). In the latter case the embryo usually shows other signs of more rapid development on that side where the groove is present. The infundibular depression (id, fig. 2, C), for example, may show on one side only or in later stages the neuromeres may develop faster on one side than the other.

The Nervous System of Amblystoma. " 451

The infundibular depression in a majority of cases (id, fig. 2, C, F,) begins to form where the neural groove meets the peripheral groove, but when it appears before the formation of the neural groove it lies in front of the anterior germinal depression. In later stages the infundibular depression becomes a very important landmark on the procephalic lobes.

The blastopore, the transformation of which goes on more or less independently^ in point of time compared with the transformation of the anlage of the central nervous system, has now in a majority of the eggs given rise to three important structures, the blastogroove, the anus and the neurenteric canal. The circular blastopore becomes an oval (bp, fig. 1, A, B) then narrows into a slit (bg,fig. 1, C) and since the walls of the slit in reality touch each other the slit becomes a groove, the blastogroove. The length of the groove is approximately the diameter of the original blastopore. In some cases it may seem to be a little shorter, probably because there is a slight concrescence of the, walls of the groove or slit at one or both ends. This account agrees with the description given by Jordan ('95), Eycleshymer ('93) and Morgan ('97) for various Amphibia. The statement of these authors, however, that the anus in Salamandra is a persisting part of the blastopore needs confirmation. It is maintained by Morgan that in the Anura the anus is a new structure and in Amblystoma a careful study of this region leads to the same conclusion. At this stage a posterior extension to the blastogroove has appeared carrying the slit backward until it touches the groove which in earlier stages bounds the blastoporic lip (bg, fig. 2, B, fig. 2, B, a). This backward extension of the blastogroove is the anus. In other words the anus in Amblystoma forms directly behind the blastogroove and in connection with it, so that in appearance it is the posterior end of the groove, there being no distinct boundary between the two. This forms a condition between that of the frog, where the anus develops separate from the blastogroove, and that of the newt, where if Morgan's account is correct the anus is a persisting part of the blastopore. At the anterior end of the blastogroove there remains a small pore, the neurenteric canal (nee, fig. 2, B).

452 Leland Griggs.

A large part of the eggs as Eycleshymer has observed show no sign of this canal. His inference, however, that the canal is never formed in those eggs in which it is not seen does not follow necessarily, for like the transitory structures already described it may appear and disappear very quickly. Eycleshymer's statement could be proved only by following through the individual development of a single egg. In the absense of such evidence it is logical to conclude that the neurenteric canal, like the posterior germinal depression and the neuromeres, is of very general occurrence. The growth in length of the embryo which becomes apparent in this stage involves the question of whether or not the blastogroove grows in length. This has been a matter of dispute. Semon ('01) for Ceratodus has given perhaps the most detailed and careful description of a growth in length of the blastogroove. The earlier writers, Miss Johnson ('84) and Schultze ('88), have confused the blastogroove with other grooves and therefore their description of an elongating groove hardly affords any light on this problem. Jordan ('93) distinctly states that the blastogroove of the newt does not exceed in length the diameter of the blastopore. Morgan ('97) and Eycleshymer ('95) have not described any elongation in the various forms of Amphibia which they have studied. An elongation of the blastogroove in Amblystoma is evident after the disappearance of the neurenteric canal (bg, fig. 2, B) . This is a result of the general elongation of the region in which the groove lies. The neural plate, too, elongates rapidly in this region as can be seen by comparing drawings A and B in fig. 2, noting in each case the length of the procephalic lobes as compared with the entire length of the plate. Such an elongation carries back the posterior end of the blastogroove and the anus. There is, however, no growing forward of the blastogroove since its anterior end does not approach the transverse cephalic groove but rather each end is slowly receding from the transverse groove. Now that che history of all four of these grooves has been traced, a more detailed comparison may be made by means of transverse sections. Fig. 3, A shows the anterior germinal depression in its average condition as regards depth of groove, while fig.

The Nervous System of Amblystoma.


Fig. 3. Transverse sections of the open neural plate. E, F, G are highly magnified to show the cell structure, agd, anterior germinal depression; bg, blastogroove; ng, neural groove; pgd, posterior germinal depression.

3, B shows the posterior depression of the same embryo. The difference between the two grooves is a difference of depth. Fig. 3, C, a section of a slightly older embryo, shows the much more sharply defined neural groove as it appears in the posterior half of the neural plate. The walls of this groove form a more acute angle than the walls of the posterior depression and the depth of the groove is decidedly greater. The anterior end of the blastogroove, shown in fig. 3, D, has about the same depth as the neural grove but it will be noticed that the walls of the blastogroove form a sharp angle with the surface of the neural

4.54 Lei and Griggs.

plate, a condition very different from that shown in fig. 3, C. These differences, although not very great, go to support the position maintained in this paper that four different grooves appear in the median dorsal line of the embryo of Amblystoma.

Some account should also be given here of the histology of the grooves. It has been assumed by many writers that there is one primitive groove running over the dorsal surface of the embryo showing throughout its length a fusion of the germ layers. Miss Johnson ('84), however, has found that in the newt there is a fusion of germ layers only in the vicinity of the closed blastopore and that there is an apparent fusion" in the anterior pit" as sne calls the deeper portion of the anterior germinal depression. In Amblystoma the ectoderm and endoderm are clearly separated in the region of the neural groove and of the posterior depression out the condition in the anterior depression and in the b astogroove is different. At the deepest part of the anterior germinal depression the ectoderm is pressed down against the endoderm (fig. 3, E). The fusion, however, is not real but merely "apparent" or mechanical for there is a readily recognizable difference in shape and contents between the two types of cells, and there are no cells between the two layers intermediate in character. In the case of the blastogroove there is a real fusion of layers (fig. 3, F). At the posterior end of the groove the cells can be followed through from the ectoderm to the endoderm. There is no space between the two layers, and the cells lying midway between the two are intermediate in character. Toward the anterior end of the blastogroove, however, the anlage of the notocord has differentiated and the two germ layers have separated (fig. 3, G). A few small undifferentiated cells may still be seen in the bottom of the anterior end of the blastogroove resembling those at the posterior end.

The peripheral groove and neural crest in this stage are extending backward and meet around the posterio end of the neural plate (fig. 2, A, B) . This process clearly defines the posterior limit of the neural plate and it is seen that the anus lies outside the neural plate in the region of the neural crest (a, fig. 2, B).

The Nervous System of Amblystoma.




Fig. 4. Surface views of entire embryos to show the development of neuromeres, sixth stage, agd, anterior germinal depression; id, infundibular depression; li, lateral infolding; nl, n2, n3, n4, the four neuromeres of the procephalic lobes; nc, neural crest; ng, neural groove; pel, procephalic lobes; rs, retinal spot; teg, transverse cephalic groove.

Stage 6 (Fig. 4, 5; Plate I, figs. 8-11). The following description of the development of the neuromeres in the neural plate should perhaps be prefaced by a few statements showing the evidence upon which it is based. The drawings are all made from specimens killed and hardened in chromic acid but the number and arrangment of the neuromeres has been confirmed by a study of living eggs, as well as by eggs killed in other fluids and by longitudinal sections. Four is the largest number of neuromeres found in the procephalic lobes. Where there are ess than four there is always apparently room for the development of the full number

The first neuromere to appear lies just in front of the transverse cephalic groove (n4, fig 4, A). Its anterior boundary is formed like the transverse cephalic groove from a pair of pits which develop in the peripheral groove and which in a few eggs are visible in earlier stages (fig. 2, A). In this stage the groove is seen in various degrees of development. It is nearly complete in fig. 4, A and fully formed in fig. 5, A. The three remaining neuromeres appear in the same way but do not follow^ any definite order in their appearance. Fig. 4, A and B, represent

456 Leland Griggs.

typical eggs showing the first and last steps in the process, while some of the intermediate steps are shown in fig. 5. Plate I, figs 8 and 9 show the faint appearance of one neuromere. These two embryos are especially valuable because they show, although faintly, the anterior and posterior germinal depressions and hence it is possible to locate the neuromere exactly. It lies opposite the posterior end of the anterior depression as can be seen plainly in Plate I, fig. 9. Plate I, fig. 10 shows all four neuronieres. The first and most anterior one is the faintest and is rather difficult to see in a photograph. Plate I, fig. 11 shows the neuromeres as seen from the side. Their shaded posterior margins appear as dark bands, the first neuromere being very faintly marked off from the second. While it is very difficult to show the details in photographs they are offered as a general confirmation of the more exact history of the neuromeres as shown by drawings. This follows the convincing method used by Locy ('95) in his treatment of the subject of neuromeres in the embryo of Acanthias.

This brief history of the development of the important procephalic neuromeres should be supplemented by a description of some details of minor importance. The neuromeres are usually not all of equal size. The fourth neuromere, the first to appear in point of time, is often very prominent (n4, fig. 4, B, fig. 5, D) and the second, although never higher than the others is sometimes much wider (n2, fig. 4, B, fig. 5, D). The first neuromere is usually the most faintly marked of all. Variations from the typical pattern are common particularly in the direction of so called hemi-embryos" (fig. 5, B, E). Neal ('98) has used this fact of lack of correspondence between right and left sides as an argument against Locy's theory of neuromeres. Such a condition, at least in Amblystoma, seems to show rather that the right and left sides develop more or less independently. After the full development of the neurome:es there is an entire correspondence between the two sides as will be shown in the next stage. Very rarely there are indications of more than the usual number of neuromeres. Among the hundreds of eggs examined there were only two such specimens found one of

The Nervous System of Amblystoma.


Fig. 5. Neuromeres in various stages of development. F is a diagram showing all the features of the fully developed neural plate, a, anus; ad, anterior depression; agd, anterior germinal depression; ap, anterior pit; bg, blastogroove; id, infundibular depiession; li, lateral infolding; nl, n2, n3, n4, four neuromeres of the procephalic lobes; nc, neural crest; pel, procephalic lobes; pen, postcephalic neuromeres; pep, postcephalic plate; pg, peripheral groove; pgd, posterior germinal depression; rs, retinal spot; teg, transverse cephalic groove.

which is shown in fig. 4, C. The irregularity in the appearance of the neuromeres suggests abnormality especially since the cases are so very rare. It seems a fair conclusion that the variations mentioned in this paragraph in respect to size of neuromeres, degree of development of right and left sides, and number of neuromeres are only such as one might reasonably expect to find in working over material of this nature.


458 Leland Griggs

This group of four neuromeres constituting the procephalic lobes 's further marked off from the rest of the plate by its height (fig. 5, D; Plate I, figs. 11, 12), and also by its darker color as in previous stages. These distinctive characters make it convenient to apply the term 'Hagma" suggested by Lankester for a more or less isolated and independent group of segments. The aptness of this term will become more apparent as the development of the procephaic lobes is followed in the succeeding stages and when it is discussed in the conclusion.

It has already been noted in this paper that there are few observations of neuromeres present in the open neural plate. Kupffer ('93) found six or seven neuromeres in the expanded end of the plate of Salamandra atra. Froriep ('91, '93,) found in the same region of Triton five neuromeres and in Salamandra four, but he later denied their real metameric value. Locy ('95) found a few large divisions in the cephalic plate of Amblystoma and in the plate of Rana palustris which, however, he did not regard as true neuromeres. Eycleshymer ('95) found divisions in the neural plate of Amblystoma which he regarded as artifacts. Hill ('00) found in the anlage of the combined forebrain and mid-brain of the trout and the chick five neuromeres. Although these authors differ as to the number of segments that shall be assigned to the brain region of the neural plate and although they fail to agree as to the meaning of the segments, yet the evidence taken as a whole is strongly in favor of the presence of true neuromeres in the open neural plate. The cause of disagreement, particularly as to the number of segments, seems to be the lack of landmarks. The broad anterior end of the neural plate cannot be traced directly into a well defined region of the brain without some such boundary as the transverse cephalic groove. Kupffer admitted that he was unable to find such a landmark in Salamandra. Locy used the entire expanded portion of the plate as forming in a general way the anlage of the brain. Hill is the only author who has yet described anything corresponding to the transverse cephalic groove of Amblystoma. He found that in the solid anlage of the trout brain there were two deep '^dorsal grooves" one in

The Nervous System of Amblystoma. 459

front of the mid-brain and one behind the cerebellmn, and in the closed tube of the chick the constrictions in front of and behind the mid-brain are deeper than the others. The groove behind the cerebellum in the trout may possibly be homologous with the transverse cephalic groove of Amblystoma, but the peculiarities of the formation of the brain of the former make such comparisons of little value. These observations, then, while they furnish proof of the actual segmentation of the neural plate, emphazise particularly the great importance of the presence of such a land mark as the transverse cephalic groove of Amblystoma.

Scarcely less important than the neuromeres are the anlagen of the lateral eyes. These appear as a pair of oval pigmented depressions lying between the neural plate and the neural crest (rs, fig. 4, B, fig. 5, E). In Amblystoma, although the retinal spot is unquestionably present in a small proportion of the embryos, it is not nearly so prominent as in Necturus and Rana palustris. In the last named form it is very prominent in all eggs of the open neural plate stage, occupying exactly the same position as in Amblystoma between the neural plate and the


In the review of the literature on this subject it was shown that the first observers of this retinal spot Whitman, Eycleshymer, Locy, Hill were not specific as to its exact position in relation to the edge of the neural plate. They have described the spots as being located in a general way on the plate, not beside it. In Amblystoma the retinal spots are very clearly located just lateral to the neural plate in the peripheral groove. If the proposition that the neural plate represents the ancestral brain be granted, then the eyes were located not on or in the brain, as some authors have claimed, but just lateral to it in a position corresponding to that found in some of the higher invertebrates. A careful consideration of this question of the position of the retinal spots is important in the understanding of the history which follows in succeeding stages.

As the neuromeres form on the procephalic lobes the neural crests in this region become scalloped on their inner surfaces

460 Leland Griggs.

to correspond to the segments of the neural plate (fig. 4, B), Dut the crest on the upper surface and on the outer edge is smooth except in a very few cases where some of the grooves between the neuromeres are extended laterally clear across the crest. This description differs from that of Locy ('95) who has pictured for Amblystoma not only a few large neuromeres in the open neural plate but also a beaded appearance in the neural crests. This beaded appearance was not apparent in any of the embryos of this stage used in the preparation of this paper.

The neural crests in this stage begin to move in toward the median line. The change is seen first at the posterior end of the neural plate where the two crests meet just in front of the anus (fig. 5, E). The anus which is now dorsal in position sooq noves to a ventral position and loses all connection with the neural crests. In this respect Amblystoma seems to agree exactly with the condition which Morgan ('97) has described in detail for Rana. One point in addition should be noted. The infoldings which in the earlier stages extended for a short distance to the right and the left of the anus now fade away and disappear as can be seen by a comparison of A and E in fig. 5. Further changes in connection with the moving in of the neural crests will be described as they are pictured in connection with older stages.

The neural groove, the origin of which has already been traced, is easily recognized in all embryos (fig. 4, A, B). The posterior germinal depression with which the neural groove might be confused has now disappeared except in a very few eggs and the only remaining part of the anterior depression is the deep pit which has been shown in the preceding stage. The blastogroove disappears from view as the neural crests close over and in a majority of eggs it can no longer be distinguished from the neural groove (fig. 5, E). This condition of the various grooves persists until the neural canal is completely closed.

The open neural plate with its various structures, grooves, crests, neuromeres, etc., has now reached its maximum development. Fig. 5, F, is a diagram illustrating the different structures, all of which appear in every egg in the course of its develop

The Nervous System of Amblystoma. 461

ment although a single egg is rarely found which shows them clearly all at one time as they are shown in the diagram. The two halves of the plate for example often develop unequally and, moreover, the order of appearance and disappearance of the various structures varies greatly in different eggs, and again there are a few transitory structures which in some eggs may never be prominently developed. The confusion caused by these three factors has been carefully considered in the description of the preceding stages. It is evident that there are a few important general factors underlying the course of the development of the neural plate; the markings on the surface of the embryo take the form of grooves and ridges; these grooves and ridges run transversely and longitudinally marking off transverse and longitudinal zones. Such a conception of zones will be made clearer by the following review of the various stages in the development of the plate.

The embryonic area is first divided longitudinally by a series of grooves, the anterior germinal depression (agd, fig. 5, F), the posterior germinal depression (pgd), and the blastogroove (bg), all three of which develop in the median line of the embryo in the order named. Then to the right and left of the median line develop the low wide ridges of the neural plate which is bounded by the peripheral groove (pg). Outside of the peripheral groove another ridge, higher and narrower, develops to form the neural crest (nc). Later the first three grooves disappear, except the anterior portion of the anterior germinal depression, and their place is taken by the neural groove which is not shown in the diagram. The transverse zones appear after the longitudinal zones. First the groove which is called the transverse cephalic groove (teg) divides the procephalic lobes from the rest of the plate, then the fourth neuromere (n4) is formed, then the first three neuromeres appear, not following any regular order. In a few eggs of later stages neuromeres appear behind the first four of the procephalic lobes but their history cannot be traced in Amblystoma nor even their number and arrangement determined. In this list the blastopore, the retinal spot (rs), the lateral infolding (li) and the infundibular


Leland Griggs.

depression (id) have been omitted. The retinal spots may as Locy has claimed be a part of a longitudinal series of sensory patches but Amblystoma shows only the first pair unless the lateral infoldings have some such significance. In any case both the retinal spots and the lateral infoldings lie in the peripheral groove. The blastopore after it has been transformed into the blastogroove also takes its place among the longitudinal zones. The infundibular depression in its later development includes the whole of the first neuromere and so marks that transverse division to a certain extent as standing apart from the other neuromeres.

The following table gives concisely the arrangement of the structures mentioned above.

A. Longitudinal Divisions

1. posterior germinal depression. 1.

2. anterior germinal depression.

3. blastogroove, developing from the

blastopore. 2.

4. neural plate. 3.

5. peripheral groove, with lateral

infoldings and retinal spots.

6. neural crests. 4.

7. neural groove, which forms where

the two germinal depressions and the blastogroove wholly or in part disappear.

B. Transverse Divisions

procephalic lobes, a tagma marked off by the transverse cephalic groove.

fourth neuromere.

first three neuromeres, the first later folding down as the infundibular depression.

neuromeres of medulla and spinal cord, a region where tagmata cannot be observed in Amblystoma.

Stage 7 (Fig. 6, 7; plate I, figs. 12-15). It has already been shown that in front of the procephalic lobes the peripheral groove becomes dee )er and wider forming a very conspicuous depression, the infundibular depression (id, fig. 6, D). This grows deeper and wider in older embryos until finally the entire first neuromere is folded down and lies on the floor of the depression (nl, fig. 6, C) where it remains permanently at a level considerably lower than that of the other neuromeres.

The whole process takes place very quickly with the resuly that a very few embryos show intermediate stages. Nearly all are in the condition shown in plate I, figs. 12-15 where three very regular neuromeres are visible in the procephalic lobes^








Fig. 6. Neural plates of the seventh stage showing the downfolding of the first neuromere and the development of neuromeres behind the procephalic lobes, a, anus; agd, anterioi germinal depression; id, infundibular depression; nl, n2, n3, n4, neuromeres of the procephalic lobes; pen, postcephalic neuromeres; pep, postcephalic plate.

Fig. 7. Longitudinal sections cut a little to one side of the median line showing development of neuromeres and infundibular depression, id, infundibular depression; nl, n2, n3, n4, neuromeres of the procephalic lobes; np, neural plate; pen, postcephalic neuromeres; pep, postcephalic plate; teg, transverse cephalic groove.

4G4 Leland Griggs.

the remaining one being entirely hidden from view in a deep furrow at the anterior end of the neural plate. The sections illustrated in fig. 7 show the process of infolding. In section A there is not yet any appearance of the infundibular depression. The three succeeding sections show the origin and development of the depression and the flexure that is produced in the neural plate by the infolding. The neuromeres cannot be readily observed in sections A and B but in C and D the raised procephalic lobes are easily recognized and in the latter it is possible to identify the neuromeres particularly the fourth. The relation of the infundibular depression to the neuromeres will be shown again in the next stage.

This infolding involves also the anterior pit which represents the original anterior germinal depression (agd, fig. 6, C). The pit may be seen for a short time lying on the floor of the infundibular depression but it sooq disappears and it cannot be seen at all in embryos after the closing of the neural canal, even by a careful dissection of the brain to uncover this region. The infundibular depression itself, however, persists as a very important landmark.

Kupffer ('04) has given a very good account of this region of the embryonic brain, the hypencephalon, as he calls it, in a series of vertebrates. According to him it is bounded behind by a transverse ridge, evidently the anterior edge of the first neuromere as seen in Amblystoma, w^hich he calls the tuberculum posterior, but in front in the earlier stages there is no definite boundary. He observes that the infundibulum develops from the posterior ventral part of the hypencephalon while the optic chiasma forms in front of the infundibulum on the floor of the hypencephalon. It should be noted that he failed entirely to trace the neuromeres of the open neural plate into this stage. The determination of the exact relation of the first neuromere to this infundibular depression would certainly form an important contribution to the history of this region.

Behind the procephalic lobes there appears a new series of neuromeres, the postcephalic neuromeres (pen, fig. 6), which apparently develop in order from before backward. They

The Nervous System of Amblystoma.



Fig. S. Entire embryos showing the first steps in the closing of the neural canal, and the relation of the neuromeres to the developing cerebral vesicles, stage eight, c, cerebellar crest; e, eye;fb, fore-brain; id, infundibular depression; nl, n2, n3, n4, neuromeres of the procephalic lobe; ol, optic lobe; rs, retinal spot.

are usually indistinct and in at least half the eggs of this stage they cannot be seen at all. Fig. 6, B, illustrates the best specimen of the entire collection for this purpose, but other eggs vary both in number of the neuromeres and the appearance of the neuromeres. In examining the four embryos shown in fig. 6, it will be noticed that a few narrow neuromeres are present. While such narrow neuromeres appear only rarely, when they do appear they always alternate with the usual wide neuromeres and there are no neuromeres intermediate in size between the two forms.

It might be expected that these postcephalic neuromeres would be found in groups or tagmata like the procephalic neuromeres. Thus we can speak theoretically of a metencephalic tagma, the anlage of the medulla, and of a spinal cord tagma, the anlage of the spinal cord, but in Amblystoma the postcephalic neuromeres are too rudimentary and transitory to discover any such relations.

As the neural crest closes over the plate all signs of segmentation behind the procephalic lobes disappear. No landmarks are left that can be followed into the adult brain.

Stage 8 (Figs. 8, 9, 10, 11). In this stage the closing of the neural tube is completed. The crests meet first in front of the


Leland Griggs.


Fig. 9. Later stages in the closing of the neural canal and the development of the cerebral vesicles. C is a view of the inside of a halved neural tube, and D is a view from the side of an entire embryo, anp, anterior neuropore; c, cerebellar crest; e, eye; fb, fore-brain; nl, n2, n3, n4, neuromeres of the procephalie lobes; ol, optic lobe; rs, retinal spot; teg, transverse cephalic groove.

Fig. 10. Sections to show the structure of the walls of the neural canal. The diagram A shows the location of sections B and C. C is an older embryo than B. The diagram D shows the location of sections E, F and G. F is an older embryo than E, and G is older than F. e, eye; c, cerebellar crest; mbl, mb2, divisions in the optic lobes or mid-brain; ol, optic lobes.

The Nervous System of Amblystoma. 467

anus as described above, next they meet over the posterior part of the cephalic lobes and from these two points the process of closing goes rapidly on. The anterior end of the canal remains open for a brief period as the anterior neuropore (anp, fig. 9, B), but the neuropore does not lie at the extreme anterior end of the canal, for there is a slight concresence of the ridges in front of the neuropore leaving a groove which is descernible in all eggs of the right condition, and furthermore the blind end of the neural tube always extends an appreciable distance in front of the neuropore. These relations may be easily seen in the specimens shown in figs. 8 and 9.

This concresence of the fold in front of the neuropore is an important factor in determining the anterior limit of the neural axis. Kupffer and other authors have considered the anterior border of the neuropore as the anterior end of the brain but Johnston ('05) has more recently shown that this point owing to a concresence, such as has just been described, varies greatly in different types of vertebrates. Johnston's conclusion, however, that the anterior end of the brain from a morphological point of view lies at that point at which the brain plate meets the general ectoderm at the same time tiat it comes in contact with the anterior end of the endoderm" does not form a criterion that can be applied to the conditions found in Amblystoma, for the neural place does not come in contact with the general ectoderm since the neural crest intervenes (fig. 8, A) . If the open segmented neural plate be considered as the neural axis, or the anlage of the neural axis, then the anterior margin of the first neuromere constitutes the anterior limit of the nervous system, a point, however, which cannot be traced into the adult brain of Amblystoma although it is visible just before the closing of the neural tube. On the other hand, if the anterior border of the neural crest be taken as the point in question, on the ground that the neural crest forms an integral part of the central nervous system, at least in modern vertebrates, then the neural axis ends at a point slightly in front of the neuropore, a point which in Amblystoma can no more be determined with accuracy in adult stages than the anterior limit of the neural plate. Thus Amblystoma

468 Leland Griggs.

although it shows the weakness of the old theories does not present any exact point that may safely be taken as the anterior limit of the adult brain.

Before going further into a discussion of the neuromeres of this stage it will be necessary to describe the origin of the landmarks of the adult brain. As soon as the neural crests begin to close a constriction appears in the neural canal marking off the primary fore-brain vesicle (fb, fig. 8). Then a second constriction appears marking off the mid-brain vesicle (ol, fig. 8). When once these landmarks have been identified they may be readily traced back to embryos as young as those shown in fig. 8, but they cannot be seen in any younger specimens. This process of the development of the two primary brain vesicles becomes very clear by tracing the series backward in the following order — fig. 9 B, fig. 9 A, fig. 8 B, fig. 8 A. A third landmark ,the first division or segment of the hind-brain vesicle, appears in the neural crest (c, fig. 8.) In its later stages (c, fig. 9) it appears as the first hind-brain neuromere or cerebellar neuromere of the closed neural tube. The cerebellum arises from its roof.

The relation of the neuromeres to these three well recognized landmarks is clearly a matter of prime importance but its importance may be overestimated as compared with the relation of the whole group or tagma of procephalic neuromeres to the parts of the adult brain. This tagma has been shown to be older then the individual neuromeres and to be more prominent and regular in its appearance. The first problem is to determine the relation of the tagma as a whole to the primary brain vesicles; the relation of the individual neuromeres to these vesicles or to other later landmarks may prove to be a different and more difficult problem. It is possible that in Amblystoma in which the segmentation of the neural plate is somewhat rudmientary the original number of neuromeres in each tagma may be subsequently changed by fusion, or by temporary suppression. Be that as it may, it is better to consider first the relation of the procephalic lobes as a whole to the adult brain.

The posterior boundary to these lobes has been shown to be the great transverse cephalic groove, the first landmark to appear

The Nervous System of Amblystoma.


Fig. 11. A is a section through the eyes and optic lobes, B is a portion of the inner wall of the eye shown in A greatly enlarged to give the cell structure, C is a portion of the open neural plate seen in cross section showing the structure of the neural crest and peripheral groove, e, eye; nc, neural crest; ol, optic lobe; pg, peripheral groove.

on the surface of the neural plate. An examination of the surface views of the stage in which the primary brain vesicles are beginning to appear shows this groove lying just behind the cerebellar swelling of the neural crest (fig. 8). An embryo that has been split in halves shows this relation to better effect (fig. 9, C). Behind the fourth neuromere is the transverse cephalic groove (teg). Directly above the fourth neuromere is the cerebellar crest (c.) A careful examination of Plate 1, fig. 15, shows the cerebellar crest and its relation to the posterior boundary of the procephalic lobes exactly as in the figures just described. The

470 Leland Griggs.

evidence, therefore, seems fairly conclusive that the region of the procephalic lobes includes the fore-brain, mid-brain and that part of the hind-brain in which the anlage of the cerebellum is found.

The determination of the relation of the individual neuromeres to landmarks of the adult brain is a more difficult problem. It should be borne in mind that this is the last stage in which it is possible to identify the neuromeres with certainty since they very quickly disappear as the neural crests close over into the neural tube. The only trustworthy evidence then must come from the embryos of this stage. It will be best to consider the neuromeres separately beginning with the first or most anterior. This neuromere very clearly lies on the floor of the fore-brain vesicle (nl, fig. 8, A, fig. 9, C). A small portion of the second neuromere is bent down with the first until it may perhaps be considered to be involved with the infolding (n2, fig. 9, C). Johnston's ('05) statement that 'Hhe fore-brain is formed by a relatively enormous growth of the first neuromere especially of its dorsal part" seems to go considerably beyond the facts as they appear in Amblystoma. All chat can actually be observed here is that the first neuromere forms the floor of the infundibular depression wJiich at this stage constitutes the ventral portion of the primary fore-brain vesicle. That the cells of the first neuromere actually contribute to the formation of the sides and roof of the adult fore-brain may be very probable but other structures may very likely be involved also. The second and third neuromeres in the same way form the floor of the primary mid-brain vesicle (n2, n3, fig. 8A, fig. 9, C). The second neuromere as has been stated sometimes appears to extend slightly into the fore- brain region, but the third lies wholly in the midbrain. The fourth neuromere covers a region on the floor of the hind-brain vesicle lying directly below the cerebellar crests (n4, fig. 9, A, C). All these various relations nay be summed up by the statement that the first neuromere contributes to the formation of the fore-brain, the second and third to the formation of the mid-brain, and the fourth to the formation of the anterior part of the hind-brain. It should be borne in mind, however,

The Nervous System of Amblystoma. 471

that other structures outside of the neural plate may have a part in the formation of the brain.

The position taken by Patten is that the vertebrate brain is built up of various elements, some of which lie morphologically outside of the primitive neural axis. This hypothesis has been very suggestive. Some attempt has been made in this paper to test it by tracing as far back as possible the very beginning of Lhe various parts of the brain, and by discriminating very carefully between structures that belong primarily to the neural plate and those that do not belong to it. It has already been shown that the germinal depressions are probably not "neural" grooves. The history of the neural crest has shown that it also is evidently not an integral part of the neural plate since it makes its appearance only after the plate is well marked off, since it is separated from the plate by a distinct groove and since it is not segmented like the plate. It now remains to discuss in this connection the development of two still more important structures, the lateral eyes and the optic lobes.

The first appearance of the lateral eyes has already been described as a pair of pigment spots lying just lateral to the anterior part of the neural plate in the peripheral groove. As the neural crests are raised up to form the neural tube the retinal spots come to lie in the lateral walls of the primary fore-brain vesicle (rs, fig. 8, A). The walls of the brain immediately begin to bulge out to form the optic vesicle (e, fig. 8, B), the direction of growth being backward as well as outward, as may be seen by comparing the figures of this stage with those of the next stage. The optic vesicle is comparatively large, as in most vertebrates, and seems to form at this stage the whole lateral wall of the fore-brain vesicle.

The histology and histogenesis of the retinal spot in Amblystoma and other amphibians, Rana palustris for example, has been described in detail by Eycleshymer ('93). He found that in Amblystoma the pigment in the early stages is not so well developed as in Rana palustris, neither do the retinal cells show the early differentiation into the characteristic columnar form. My own observations confirm Eycleshymer's work. It seems

472 Leland Griggs.

to be true that scattered groups of pigment-bearing cells are found everywhere in the lining layer of the neural canal but the cells of the retinal spot bear more pigment than is found in any other region. Section A, fig. 11, shows the retinal spot (e) drawn to a large scale and section B shows the pigment-bearing cells still more highly magnified. These cells are just beginning to assume a colamnar form such as Eycleshymer has figured for relatively younger embryos of Ran a palustris and other amphibians. This pigment is collected at the ends of the cells along the walls. At this stage then the anlage of the retina may be clearly identified in sections as well as in surface views.

Of perhaps still greater significance is the development of the optic lobes from a region lying lateral to the neural plate. A deep groove, the lateral infolding, has already been described as forming on each side of the procephalic lobes (li, fig. 4, A). As the neural crests fold over, these grooves disappear but at the same time in the same region between the neural crests and the neural plate appear a pair of low ridges or lobes (ol, fig. 8). Each lobe is bounded anteriorly by the optic vesicle and posteriorly by the cerebellar crest. At first rather indistinct in outline it soon becomes very prominent and clear cut (ol, fig. 9). As its growth continues it bulges backward as may be inferred from the pear-shaped form which it assumes (fig. 9, B, D). It may push back slightly under the cerebellar crest (fig. 9, D). It becomes bilobed as can be seen better in the next stage although the beginning is shown in fig. 9, D.

A study of sections confirms this account of the early appearance of the optic lobes. Fig. 10, B, is a section cut longitudinally through the fore-brain vesicle and optic lobes in a plane shown by the diagram, A. The optic lobes appear as thickened walls of the neural tube lying directly behind the retinal spots. The walls of this region are slightly thicker than those of the fore-brain vesicle and the diameter of the canal is considerably less, a condition already shown in fig. 9, C, a drawing of a halved embryo. Section C, fig. 10, is cut through the same plane as B but the embryo is a little older, as appears from the fact that the neural canal is now closed. In comparing the two sections

The Nervous System of Amblystoma. 473

it is seen that the optic lobes have become thinner and also a little longer. The cavity, too, has become larger. This enlargement of the optic lobes is seen in surface views of fig. 9. Sections E, F, G are longitudinal sections cut in a plane at right angles to sections B and C. Their position is shown by the line in the diagram at D. Section E shows the optic lobe extending posteriorly from the optic vesicle very much as in B. In section F the lobe has become slightly divided into two parts and in section G it is very clearly divided (mbl, mb2). This last section shows also how closely the cerebellar crest is related to the general neural crest as has already been shown in fig. 9.

The histology of this region presents no very distinctive features as the histology of the optic vescile showed none. Section C, fig. 11, shows the cells of the peripheral groove (pg), the region in which the optic lobes arise. The cells present the same appearance as those of the neural plate but they differ strikingly from the cubical cells of the adjacent neural crest. There is slightly more pigment in the peripheral groove than in the plate. Section A, fig. 11, shows the cell structure of the optic lobes at about the time of the closure of the neural canal. The cells are of the same general forn as those of the optic vesicles but they contain less pigment. This early differentiation of the optic lobes, therefore, is a differentiation of outer form not of cell structure.

Thus not only may the anlage of the retina be located on the surface of the embryo while the neural plate is still open but also what appears to be its ganglion, the anlage of the later visual center in the mid-brain, may be located just behind the retinal anlage. It seems proable that the ancestors of the vertebrates possessed an eye and optic ganglion lateral to the brain on either side and that these organs were later infolded into the brain when the central nervous system was transformed into a closed canal. This theory offers a reasonable explanation of the fact that the mid-brain unlike the rest of the tube has at the very outset a thickened layer of nerve elements on the roof and sides, the tectum opticum. Indeed such an explanation has before been hinted at by those authors who have endeavored to


474 Leland Griggs.

find some morphological reason for the peculiar character of the mid-brain. Johnston ('05) suggests that at the anterior end of the nervous system there is included in the brain an area which at the level of the acustico-lateral anlage is supposed to be left out." It has certainly, however, never before been demonstrated that the peculiar structure, the optic lobe, lies originally external and lateral to the neural axis.

From this history of the anlage of the lateral eyes and their centers, the optic lobes, it might be expected that the anlage of the parietal eyes could also be found near the open neural plate. Locy ('95) has claimed to have found them in the embryos Squalus as a series of accessory optic vesicles" lying behind the true optic vesicles of the lateral eyes. Hill ('00) has confirmed the presence of these accessory optic vesicles in his study of the chick. Nothing of this kind however seems to appear in Amblystoma. The pineal eye forms very late. No new facts could be discovered either in support or in refutation of Locy's theory.

The post-cephalic region at the time the neural crests close over presents a very characteristic appearance due to the presence of narrow shallow grooves running across the plate transversely and obliquely in no regular pattern (fig. 8, A). This is one more factor which aids in marking off the procephalic lobes from the remainder of the plate.

Stage 9 (fig. 12). A few figures of older embryos are introduced here in order to carry the work to a point where the more important landmarks of the adult brain may be readily recognized. Fig. 12, B and C, are side views of embryos just old enough to show the otic pit (op) and the swellings of the brain tube called by Kupffer secondary neuromeres. The changes that take place between the closing of the neural tube and the condition represented by these older embryos are readily understood. Fig. 12, A, is a view of an embryo in which the neural crests have already fused. The optic vesicle is divided by a faint groove or constriction into eye stalk (es) and eye vesicle proper (e) . The mid-brain is bilobed as in the preceding stage (mbl, mb2). The cerebellar swelling (ml) now begins o separate a little from the neural crest. A series of new swell

The Nervous System of Amblv stoma.


ml ,rnt>J mi\ mhz'. ^

mf .miPJ

Fig. 12. Side views of the neural tubes of older embryos, stage nine, e, eye; es, eye stalk; ml, m2, m3, neuromeres of the hind-brain; mbl, mb2, divisions of the mid-brain; op, otic pit.

ings now appears in the hind-brain back of the cerebellum forming in order from in front backward, the completion of which is seen in the older embryos B and C. (ml, m2, m3). Between the second and third swelling appears at this time a pigmented crescent shaped depression the otic pit (op). The presence of these various swellings along the neural canal gives a very characteristic segmental or beaded appearance.

This description of the swellings or '^neuromeres" of the closed neural tube agrees essentially with the results of other authors. The agreement will be made apparent by a more detailed discussion of the important features.

In the embryonic fore-brain of vertebrates Kupffer ('03) distinguishes three regions or, as he calls them, ' ' secondary neuromeres," the telencephalon, the parencephalon and the synencephalon. Locy ('95) and Hill ('00) found three primary neuromeres in the fore-brain before the development of the secondary divisions but they could not trace the primary divisions into the secondary ones. Neal ('98) considered the fore -brain to be one large neuromere which later showed secondary divisions. The relation of the first procephalic neuromere to the fore-brain has already been discussed. In Amblystoma the telencephalon appears as Kupffer has described it but it develops so late that no satisfactory explanation of its relation to the primitive segmented neural axis can be given. Behind the telencephalon lies a division called by Kuffer the parencephalon. The lower

476 Leland Griggs.

half of it comprises the infundibulum or infundibular depression and the base of the eye stalk. Between the parencephalon and the mid-brain is a small wedge-shaped region, the synencephalon, a region which in fish embryos is prominent but whiph is hardlj^ noticeable in Amblystoma. The temporary appearance in the mid-brain of two swellings is, according to Kupffer, almost universal among vertebrates. It has been found in various types by Locy, Hill and Johnston. It is undoubtedly present for a short time in the optic lobes of Amblystoma but it soon disappears and the mid-brain becomes an unsegmented region with a narrow ventricle and thick walls as shown in the older embryo seen in fig. 12, D. In the hind brain the most important swelling has been called the cerebellar crest. As first described it lies directly behind the mid-brain but in older stages it migrates slightly backward leaving a triangular region in which the adult cerebellum arises. This triangular region is called by most authors a distinct cerebellar neuromere. Amblystoma, and according to McClure ('90) Kupffer ('05) and Johnston ('05) all the Amphibia as well, lack the so called "blank neuromere" which is found in the embryos of many vertebrates just behind the swelling described above. The second swelling in the hindbrain of Amblystoma resembles the first. Behind the otic pit lies a third swelling resembling the first two. These swellings of the hind-brain may well be called true neuromeres since as Johnston and Locy have shown they are like the undoubted segments of the spinal cord, and since there are no accessory structures behind the cerebellum to produce secondary divisions.

There are then in the closed neural tube of Amblystoma a series of swellings extending from the anterior end of the brain to the otic pit, but since these divisions are of varying morphological significance they cannot rightly be called neuromeres. If no such general term is applied to them as a group considerable confusion may be avoided, for then the morphology of each may be investigated on its own merits as it has been the aim of this paper to do.

The Nervous System of Amblystoma. 477

General Conclusions

The foregoing description, whatever may be the value of the results attained, emphasizes the value of breaking away somewhat from the traditional methods of attacking the problem of cephalogenesis. Two new lines of work have been followed.

First. The number of neuromeres is not the all important part of the problem. Since the time of Goethe students of cephalogenesis have bent their energy mainly to finding out- how many segments the head contained. Even if the search had led to a definite result, which has been far from the case, the main problem is far from being solved. The grouping of neuromeres into tagmata may prove to be more importa