Difference between revisions of "Journal of Morphology 11 (1895)"

From Embryology
Line 25: Line 25:
===No. 3, DECEMBER, 1895===
===No. 3, DECEMBER, 1895===
I. Contributions to the Structure and Development of the Vertebj'ate Head. William A. Locy. 497-594
I. {{Ref-Locy1895}} Contributions to the Structure and Development of the Vertebrate Head. William A. Locy. 497-594
II. The Musculature of CJiiton. Lilian V. Sampson 595-628
II. The Musculature of CJiiton. Lilian V. Sampson 595-628
Line 32: Line 32:
IV. The Cleavage of the Egg of Virbius Zostericola {Smith). Frederic P. Gorham .... 741-746
IV. The Cleavage of the Egg of Virbius Zostericola {Smith). Frederic P. Gorham .... 741-746
===No. 1. — May, 1895===
===No. 1. — May, 1895===

Latest revision as of 22:06, 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 11 (1895)

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



With the Co-operation of

EDWARD PHELPS ALLIS, junr., Milwaukee.

Vol. XI



No. 3, DECEMBER, 1895

I. Locy WA.Contribution to the structure and development of the vertebrate head. (1895) J. Morphol. 11(3): 497-595. Contributions to the Structure and Development of the Vertebrate Head. William A. Locy. 497-594

II. The Musculature of CJiiton. Lilian V. Sampson 595-628

III. A Study of the Operative Treatment for Loss of Nerve Substajice in Peripheral Nerves. G. Carl Huber, M.D 629-740

IV. The Cleavage of the Egg of Virbius Zostericola {Smith). Frederic P. Gorham .... 741-746

No. 1. — May, 1895

I. Bashford Dean. The Early Development of Gar-pike and Sturgeon 1-62

II. J. Playfair McMurrich. Embryology of tJie Isopod Crustacea . . 63-154

IIL Edward G. Gardiner, Ph.D. Early Development of Polychoems Caii datus, Mark 155-176

IV. Albro D. Morrill. The Pectoral Appendages of Prionotus and their Innervation 17 7- 191

V. Fanny E. Langdon. The Sense-organs of Lnmbricns Agricola, Hoffm 193-234

No. 2. — October, 1895

I. George Wilton Field. O71 the MorpJwlogy and Physiology of the Echinodei'ni Spermatozoon .... 235-270

II. Gary N. Calkins. The Spermatoge7iesis of Lumbricus . . 271-302

III. W. B. Scott. The Osteology ajid Relatio7is of Protoceras 303-374

IV. Alvin Davison. A Coittribntion to the Anatomy and Phy logeny of Aviphiiima Means {Gardner) 375-410

V. Grant Sherman Hopkins. On the Entcron of Avierican Ganoids . 411-442

VI. Edmuxd B. Wilson. Arc/ioplasvi, Centrosome, and CJiromatin in the Sea- Urchin Egg 443-4/0

VII. Thos. H. Montgomery, Jr. The Derivation of the Freshwater and Land Nemerteans, and Allied Questions 479-484

VIII. Edward Phelps Allis. The Cranial Mnscles atid Cranial and First Spinal Nerves in Amia Calva . 485-491

IX. Louis Murbach. Preliminary Note on the Life-Jiistory of Gonionemiis 493-496

No. 3. — December, 1895

I. William A. Locy. Contributions to the Structure and Development of the Vertebrate Head .... 497-594

II. Lilian V. Sampson. The Mnscidatw^e of CJdton 595-628 '

III. G. Carl Ruber, M.D. A Study of the Operative Treatment for Loss of Nerve Snbstatice in Peripheral Nerves 629-740

IV. Frederic P. Gorham. The Cleavage of the Egg of Virbius Zoste ricola {Smith) 741-746


BASHFORD DEAN, Columbia College, New York City.

A MORE PERFECT KNOWLEDGE of the development of the Ganoids seems to be especially needed to enable the relations of the Teleostomes to be better understood. The study of fossils has at the present time given the materials for the better understanding of many important questions ^ relating to the descent of fishes, but its results must yet conform with those of the embryologist. It is thus especially unfortunate that the development of Polypterus is at present totally unknown, and that the study of even the more accessible forms, Amia and Polyodon, has as yet been neglected. Indeed, of those Ganoids which have hitherto been studied, the developmental history seems to have presented so many difficulties to observers that many years must go by before it may be satis 1 As for example, the relations of early Teleostomes to a stem essentially Elasmobranchian ; the descent of Mesozoic types of Ganoids from Crossopterygians; the lines of ganoidean specialization ; the evolution of Sturgeon from palaeoniscoid ; the affinities of physostome and caturid.

2 DEAN. [Vol. XI.

factorily understood. Every investigator who has been fortunate enough to secure the spawning fish, seems to have either failed in completing his material of developmental stages, or by faulty methods of preserving it has been unable to carry out a detailed study. The Sturgeon has not unnaturally been made an object of national investigation by Russians, and our present knowledge of its early development is to be obtained only in the more detailed works, in Russian, of Salensky ('77-'8l) and Peltsam ('87). In Germany the studies of v. Kupffer ('91) on Acipenser were based upon material which unfortunately was lacking in younger stages. Of Lepidosteus embryonic stages, as is well known, have been obtained by different investigators five or six times, and within comparatively recent years: it has been owing mainly to technical difficulties that little has yet been published of its earlier developmental history.

In the following paper it is the writer's object to describe the early development of Gar-pike and Sturgeon, and by examination of these forms side by side to permit more definite comparisons as to the mode of development of Ganoids.

The material for the study of Acipenser was obtained by the writer during the spring of 1893 at Delaware City, Del., and has afforded a satisfactory and well-preserved series of general developmental stages. The mode of acquiring it has been already recorded ^ ; spawning fish were taken several times during the writer's visit, and a number of experiments were made to determine a successful mode of artificially fertilizing the eggs, and a practical method of hatching them. As a result of the experiments it became possible to obtain an abundant supply of embryos, and considerable care was taken both to secure as perfect a set of the developmental stages as possible, and to preserve them in several reliable ways. As a fixing agent, the picro-sulphuric mixture should be mentioned as having proven in the main most satisfactory for sectioning. The developmental stages of Lepidosteus, as will be noted, were recently obtained during a visit to Black Lake, St. Lawrence Co., N. Y., made in company with Prof. E. B. Wilson, of

1 Dean : Sturgeon Hatching on the Delaware. U. S. F. C. Bulletin, 1893, No. xviii.


Columbia College, who had undertaken to study the Gar-pike in the lines of experimental embryology. The themes of the present paper are arranged as follows:

I. The breeding and feeding habits of Lepidosteus.

II. The early development of Lepidosteus, — more accurately, that beginning with segmentation and ending with establishment of organs.

III. The early development of Acipenser, parallel in discussion with that of Lepidosteus.

IV. A general comparison of the early stages of these kindred forms, and a discussion of the developmental relations of Ganoid, Elasmobranch, and Teleost,

I. Mode of Occurrence of the Gar-pike; its Feeding

Habits and Spawning.

The spawning habits of the Gar-pike, Lepidosteus osseus, have already been recorded by Garman ('78), Beard ('89), and Mark ('90). Their observations had in all cases been made at Black Lake, St. Lawrence Co., N. Y., the locality which has now acquired a well-merited reputation as a spawning region of this remarkable fish. To their accounts the writer would add the following notes collected at the same locality during the present season as preliminary to his discussion of the fish's early development.

The Gar-pikes of Black Lake are exceedingly abundant, and on account of their peculiar breeding habits there is little difficulty in securing their developmental stages. For several weeks during the spawning season they appear in shallow water, often in great numbers, and may be observed as they deposit their eggs. Matured fish, to yield eggs and milt for artificial fertilization, are then to be readily taken.

The general disappearance of the fish shortly after spawning is doubtless the reason that so little is known of its usual life habits. Garman states that after the spawning season the fish are " seldom seen, remain in the deeper parts of the lake away from shore, and are more or less nocturnal in habits.

4 DEAN. [Vol. XL

Occasionally one may be taken with a minnow bait." This note, as far as the writer is aware, is the only definite reference that occurs in literature as to the usual habits of the fish. Belief, however, is very current that Gars are exceedingly voracious, killing and eating the larger fish, and not hesitating to attack even a swimmer who has ventured in their neighborhood. In the South, where examples have been taken as large as seven feet in length, their shark-like fierceness is credited more plausibly, and certainly to a degree which prevents the negroes from bathing in waters {e.g., several branches of the Edisto River) where Gars are known to be large and abundant.^ It is also currently believed that Gars are migratory, appearing at different places at different times, an idea probably to be traced to the general appearance of the fish only at spawning time.

Black Lake, however, furnishes conclusive evidence that the Gar is in no way migratory. The lake is but a small body of water whose communication with the St, Lawrence has long been cut off, and, though land-locked, has afforded especially favorable conditions for every life-stage of the fish. Their disappearance after spawning is, moreover, by no means complete. As stated by Mr. H. J. Perry, who for many years has carefully observed the fish at Black Lake, they may be seen near the surface at any time during the summer and fall. As a rule, their feeding habits appear then nocturnal and numbers are usually taken by night lines.

The actual mode of feeding has been observed by the present writer. The Gar approaches its prey (young dace) cautiously, advancing without perceptible movements ; when within three or four feet it pauses, as if accurately to direct its aim, then, without seeming effort, it shoots quickly forward, secures the fish, stops suddenly, and is again motionless. An occasional bending of the head adds not a little to the apparent dexterity of movement. The food in all cases examined consisted exclusively of small soft finned fishes, mainly dace, none of which were

1 The writer, while in South Carolina, was unable to obtain any direct evidence of the Gars' attacking even larger fish, nor has he seen any efforts on the part of specimens he has taken to warrant any belief in their fierceness.


longer than three and a half inches ; young perch and sunfish, abundant in the locality, do not appear to be eaten. The number of small fishes each Gar had taken was especially large and fully justifies the idea of the fish's rapaciousness ; from the stomach of a male (24 in.), caught while spawning, eleven cyprinoids were taken ; of another (27 in.) the stomach contained the remains of thirteen fishes, while in addition three (of 2 in., 2^ in.) were taken from the pharynx. From all observations it would appear that the Gar is reasonably to be looked upon as only indirectly injurious to food fishes — i.e., bass, perch, pickerel, pike, catfish — in reducing the general food supply. It is also worthy of note that there appeared throughout no evidence of the fish's having taken food that had been torn or cut ; the function of the straight, close set teeth seemed rather to prevent the escape of the prey than to kill or cut it.

Gars are extremely tenacious of life. Spawning fish taken by snare remained out of water two hours in the bottom of a boat : they were still alive and active, and supplied eggs and milt for artificial fertilization. Others after a similar journey were tethered by wire through mouth and gills and were kept alive for a week or more during the remainder of the writer's stay. Fish that had been speared and later placed in water remained alive for four or five hours. The strength of the fish is remarkable. This is perhaps in no way more evident than in the movements of swimming — its power to advance rapidly without apparent effort, to change at will from a position of rest to one of most rapid motion, and then to be able to regain instantly its position of rest. When taken from the water a large fish (4 ft.) in its efforts to escape can with difficulty be held in the hands. Its movements are varied and exceedingly strong, bending both horizontally and (slightly) vertically, almost serpent-like in its flexures. The movements of the head are especially noteworthy.

At Black Lake, as elsewhere, it is only at the time of spawning that Gars become noticeable. They are seen by the fishermen during the early part of May rising about their boats, thrusting their jaws out of water, often making a marked

6 DEAN. [Vol. XI.

"smacking" sound as they emit bubbles of air. Their appearance is at first general, notably in the regions of the deeper parts of the lake. Soon afterward they are seen basking near the surface, moving slowly away as a boat comes within a rod's distance. They are next noticed in schools of often twenty or more, sunning themselves in the middle region of the bays in whose shallows they will later spawn.

Spawning Habits.

The season of spawning appears a little more extended one than is usually stated, beginning about the middle of May and ending about the fifteenth of June. During this time spawning takes place intermittently, so that in all there may not be more than six or seven days in the entire actual ' run.' According to Mr. H. J. Perry there is usually an earlier and a later *run '; the former occurs at favorable localities, induced by unusually warm weather, and lasts but two or three days ; while the general spawning occurs about three weeks later. The variation in the time of spawning as far as the writer knows is indicated in the following table :


First Noted.

Spawn (and disappear).




May 14

May 18-19

May 31-June 2




June 10-12 (?)




June 3-5 (?)




May 24

June 8, 9


H. Virchow


June 8-10 (?)

1894 ]

[ E. B. Wilson j ! and Dean j

May 3

May 14-18 and


June 10-12

Water temperature has doubtless with Gars, as with other fishes, an important relation to the time of spawning. The temperature of the shallow waters where spawning was taking place during the present season varied from 66^^ to 70° F. It is to be noted that the most active spawning occurred when the temperature was as low as ^(i^ (Lower Deep Bay). Coolness in air temperature together with strong winds was found to have no immediate effect on the spawning fish, and even during a heavy cold rain-storm spawning was found not to


be discontinued. The writer notes that spawning occurs not merely "during the heat of the day between 12 and 3 o'clock," ^ but was observed at intervals from 8.30 morning to 7.30 night.

Especial localities have long been noted by the fishermen of Black Lake as favorable spawning grounds of the Gar, and certain particular shore spots or "points," often of but a few feet in diameter, as those in Upper and Lower Deep Bays, have been found year after year, to receive the first eggs deposited during the season. Later, when the height of the breeding season arrives, the fish may be seen spawning on almost every shore of the lake. "Points" at which early spawning occurs are hardly such as the term usually implies. They are little more than rocky shore strips, or rather particular portions, two or three yards in diameter, of a rocky shore. They are by no means prominent, nor are their rough rock fragments cleaner apparently than those of a part of the shore but a few yards away. There are thus but three spots in Lower Deep Bay where the eggs of the Gar are deposited, although the natural characters of the rocky shore are apparently uniform from the mouth to the head of the bay.

The behavior of the fish when spawning is worthy of especial note. When approaching the shore they are seen to be already divided into parties, each female readily recognized by her larger size (about 3 ft. 6 in. ; 4 ft. 6 in.), attended by several males (from two to eight). All are pressed closely together, and in the slow advance and circlings of the party scarcely a movement can be detected. The snouts of the males, lighter in color, probably in sexual coloration, may be seen pressed under rather than over the sides of the female. All fins are spread, dorsals and anals widely erect, the former, together with the upper half of the tail, often protruding from the water and recognizable several rods from shore. The fish in the meanwhile enter very shallow water (five or six inches), so shallow in fact that the backs of the spawning fish are sometimes exposed. The arrival of the fish is followed by a period of quiescence ; then after slowly moving to and fro, circling nearer and further

^ Beard, ref. 6.

S DEAN. [Vol. XL

from shore, a few minutes later a sudden and active splashing takes place. A number of eggs have at that moment been scattered and fertilized, and the water is for the time clouded with milt. There will then follow a period of quiescence, often many minutes in length, followed by circlings and a second oviposition. The eggs, it may be noted, are not deposited at a particular spot, but appear to be sown evenly over the general spawning ground. It is also to be noted that during subsequent circlings of the female several males may often await her return, and from their movements it is not impossible that they are still emitting milt over the freshly deposited eggs. In no case was observed evidence of rivalry among the males, and an examination of all taken during the writer's visit could not detect any injuries caused by fighting. Their breeding colors are, however, prominent, and the richly pigment-spotted sides are accented by the bold markings of anal, dorsal, and caudal, made very evident by the habit of erecting the fins. The paired fins, too, appear of selective importance, and are widely spread during mating. Each is centered with a large ashcolored spot, especially prominent when seen under water. And the writer has noted that a male when unable to secure a place near the female has swum backward before the party, expanding his fins to the utmost, and making side and upward motions with head and paired fins.

It is probable, from observation of the spawning fish, that all eggs are not deposited by the female during a single day. Spawners were noted which deposited four or five batches of eggs and which then did not reappear during the day. Examination of the ovaries shows that but small portions of the eggs have at one time become detached from their follicles, but that ' the ripening process appears in every, and not a particular, region of the ovary. With Gar, as with Sturgeon, it seems evident that a large proportion of the ovarian eggs may not be duly ripened. Many examples were found whose abdominal cavity contained in quantity flaccid and slate-colored eggs which proved incapable of fertilization.

The fishes when actually mating are not easily alarmed. They may be approached closely, and even touched with the

No. I.]


hand. At the height of the spawning season large numbers (thirty) have been snared and speared at a single locality with but temporary alarm to the rest. Early in the season, when the fish were scarcely ripe, the writer found, however, that rarely more than two or three could be taken before the disappearance of all, — at least for a period of several hours. And the snaring of a female was found to give a greater alarm than the removal of several males. The note of Garman as to the cautiousness of the fish in scouting to determine the whereabouts of an enemy was fully confirmed. Timidity of the fish seems, moreover, indicated in the writer's attempt to lure male fish by means of a recently captured spawning female tethered over the spawning bed. Males were shortly attracted, but after one had ventured for a moment close to the female a general alarm was taken, and the fishes did not return.

The eggs almost immediately after fertilization become, as Garman and others have noted, excessively sticky, and acquire firm attachment wherever they happen to lodge, on stones (mainly), sticks, or water weeds. Over these collectors they are found well scattered, as the accompanying drawing (Fig. i)

Fig. 1.

indicates. The stone fragments, loosely piled together, serve to attach eggs on every face, and not infrequently their under sides, being the cleanest, collect the greatest number. The

lO DEAN. [Vol. XI.

eersrs arc in fact sifted anions: the crevices of the rocks, and are to be found many tiers below the uppermost layer. The slime-covered character of the rocks seems little favorable to the prolonged fixation of the eggs, and within two or three days the majority become detached and are sifted deep among the rocks.

The color of the eggs at the time of extrusion is like that of the sturgeon's, slaty gray, and is strong in color contrast to the green-black rocks. About three hours later {early segmentation) the eggs become creamy yellow ; two days later (end of gastrulation) the color has changed to a dull shade of greenish brown, although part of its duskiness may be due to the sediment collected on the outer membrane. The lightness in gravity of the animal-pole is early apparent, and when the axis of the eggs is disturbed the germ-disk regains its uppermost position in a surprisingly brief time, very much as in Teleost or amphibian.

As to the means adopted for securing embryological material ; former experiments (Garman, Mark) had shown that egg-bearing rock fragments taken on the spawning ground might be retained in pans of water (renewed twice daily), and the eggs successfully hatched. A more convenient method was that of detaching the eggs from the stones while at the spawning ground and hatching them subsequently in earthenware dishes (Virchow^) or in a salmon hatcher (Beard). These plans were found by the writer safe and convenient, none the less because the natural hardiness of the eggs rendered them little subject to attacks of fungus {Achlya): and by their use there could certainly be no simpler way of securing developmental stages, were it not that a single objection proves most important : the eggs when collected are found to be of different ages, and in this confusion cannot well be reduced to a complete series of stages. By use of artificial fertilization, on the other hand, the writer found that every stage could be secured conveniently and in suitable quantity.

Artificial fertilization was first attempted at Lower Deep Bay, the writer employing the same mode of procedure and

^ According to Mr. Perry.


the same hatching cases (v. U. S. F. C.BulL, 1893, No. xviii, PP- 336-337) as he had used at Delaware City in the hatching of Sturgeon. Fish were freshly captured : the eggs were extruded by pressure and received in an earthenware dish. Here they were retained dry and shortly fertilized by a few drops of milt from a ripe male. They were then stirred ; several minutes later water was added and as the eggs became adhesive they were quickly spread under water over netting-covered ^ hatching-trays. The thick, glairy mass which surrounds the eggs of Acipenser was not noted. Attachment was speedy; and the egg-covered trays, taken out of water, were taken back to Edwardsville — a two hours' row — with no further precaution than that of moistening the trays to guard against the drying of the eggs. This mode of transport the writer notes to emphasize the hardiness of the eggs, since it was later found that the proportion of eggs lost on these trays was scarcely greater than of those which were at once placed and retained in the floating hatching-boxes.

Artificial fertilization was repeatedly tried and always with favorable results ; for this purpose fishes were transported living from the spawning grounds to Edwardsville. In obtaining the eggs it was found most convenient to excise the entire ovaries ; from these the eggs were readily taken, and though not appearing to separate freely from their follicles they proved nevertheless capable of fertilization. The hatching boxes with their enclosed trays were floated near the shore in about five feet of water at the end of the wharf of Mr. Perry. Here the water was muddy and far from pure, but the eggs proved hardy and the loss from fish fungus was usually inconsiderable. On a tray where a loss of about 50/0 of the eggs was recorded, the mortality seemed mainly due to the crowded condition of the eggs. For convenience in securing developmental stages a mass of eggs were scraped from the trays and kept in the laboratory in earthen pans until required. As a fixing reagent (alcoholic and aqueous) picro-sulphuric acid was

1 In his present experiments tiie writer employed "canopy netting" for the hatching frames. This material is of much finer texture than " mosquito netting," and its circular perforations seem more favorable for the attachment of eggs.

12 DEAN. [Vol. XI.

employed : this had been found to give the best results of all reagents the writer had used with eggs of Acipenser.

During development the water temperature varied from 62°-y2° F. Hatching occurred in 200-220 hours.

In the following table the writer records the time occupied in the growth stages of Lepidosteus compared with that of Acipenser : the more rapid development of the latter was doubtless due to a slightly higher mean temperature of the water as well, perhaps, as to a smaller amount of food-yolk. At Delaware City the mean temperature was about 68°, while that of Black Lake was about 64° F. The writer has learned recently from Mr. Perry that eggs of Lepidosteus deposited on June 10 hatched by favorable water conditions in as short a time as 125-130 hours.


Lepidosteus. Acipenser. First cleavage i i

Second cleavage ......... 2 1.25

Third cleavage ......... 3 2.30

Blastoderm cap grows down to the level of -J- diameter of egg 32 16.30

i " 38 24

" f « 42 26

I " 44 27

Gastrulation begins ......... 37 ig

Blastopore closes 46 58

Embryo appears ......... 60 31

Central nervous system outlined ...... 70 43

Pronephic ducts appear 70 42

Brain vesicles distinguishable 80 46

Optic vesicles distinguishable ....... 80 46

Embryo surrounds circumference of egg 90° (approximate) . 70 ^^

110° " . 85 43

140° " .110 49

" " " 180° « , 125 53

" " " 200° " . 140+ 56

" " 240° " . 146+ 61

" " " 280° « .152+ 68

320° " .160+ 76

Movements of embryo first noted 149 82

Embryo hatches 200 + 94

Yolk material absorbed 1000 + 250 +

No. I.]



The accompanying figure (Fig. 2) shows in side view the natural position assumed by the embryo of Lepidosteus during its development : a, b, and c represent early stages of segmentation ; d, early growth of blastoderm cap ; e, f, and g, gastrulation ; //, early embryo (the heart region is now at the animal pole of the e.gg and the blastopore in a position a little below the equatorial line) ; i, embryo of 92 hours (sagittal axis of embryo has now become horizontal); j\ the horizontal projection of the same embryo ; /, embryo of 130 hours (region of embryo

Fig. 2.

near pronephic tubules is now uppermost); k, same in horizontal projection. The positions assumed by the developing Acipenser are essentially the same as in the above figures ; the only stage of which the author is doubtful is that of the early embryo, h.

The eggs of Lepidosteus and Acipenser compared with those of Teleosts are exceedingly large in size, and in their pigmented character and mode of early development they are strongly suggestive of amphibian. The following comparison may be given of their general characters : —



[Vol. XI.

V) )-<


> o



"3) .S

^ £ rt 3

E o





H 2



c ..

1) 1

S ^

to .

a, >


^ -^ .2 bo

>_ k-i

bo 3




<U o

g s

£ T3

be C

•s, "<

ni E

CD ho






bO "B. >>






cr £

2 1




G 3

+-» 3 43

G cu



< > <


"rt c3

^ ?

D -S

3 O

a; ^

o 2


.s ^

"o "rt


>% 3


bjo i2

oi ^

3 a


> O


.■S 3


o ><

h-1 I'

f^ ^


1-1 <u

"o S


boXl rt O



— 1 <u

cS 3


•a o




-a c




a £

C eo



fc. o



C/3 lU

(J u 3


3 G

C/2 <u



bo _G

'u •4-*









.S ^







u >

E c o ^


3 3


be t.




T3 (J











< 13




- -a £ 5

G 3






C5j 4)



(0 rt -#-* (1)


N rl





3 w







6 o W


t. G 3 O

w ba


II. The Early Development of Lepidosteus.

The Q^'g after being deposited becomes pale cream yellow in color, and for a short time does not exhibit the protoplasmic germ area. The condition of its membranes has been made the subject of the exhaustive research of Mark, who figures and describes for the first time the single micropyle and its contained plug of granulosa cells. The maturation of the ^%g has not been sufficiently followed by the writer to warrant discussion ; he notes that the Q.g% at the moment of extrusion shows a well-marked maturation spindle in the immediate neighborhood of the micropyle, of which Mark has given an excellent figure ; a polar body is here given off almost immediately after fertilization ; its appearance is similar to that figured by Bohm ^ in trout ; it has not, however, like the latter, been seen to undergo a second division.

The change in the egg-substance which permits the heavier yolk to sink to the lower pole of the ^gg is not apparent to the eye during the first ten minutes after fertilization ; the lower hemisphere is then noted to be of a darker color. Soon afterward the ^gg presents somewhat the appearance of PI. I, Fig. I, — its germ area, however, is somewhat larger than figured and its marginal limit less clearly marked. The ^gg one hour after fertilization, seen in the figure, is on the point of undergoing the first cleavage : its germ area is roundly oblong, of flattened surface, slightly raised above the yolk. Sections, however, indicate that at this stage germ disk and yolk are more intimately connected, and in PI. II, Fig. 21, it will be seen that the protoplasm of the animal pole is hardly to be separated from the coarse granular yolk until nearly in the equatorial region of the ^gg. In the figure the nucleus is seen dividing at about one third the diameter of the ^gg from the surface.


First Cleavage (PI. I, Fig. 2). — The first segmentation furrow appears in the living ^gg as a thinning away of the germ

1 A. A. Bbhm: Die Befruchtung des Forelleneies. Sitz.-Ber.d. Gesell. f. Morph, u. Physiol. Miinchen. 5 Mai, 1S91.

1 6 DEAN. [Vol. XI.

disk in a vertical plane ; it results in a trench-like groove showing the yolk mass below separating the blastomeres ; at either end it bows outward, rounding the corners of the germ disc, and extends no further down the side of the eggs in any example the writer has examined ; the margins of the furrow are boldly marked, highest at the animal pole of the ^^%. Sections at this stage indicate that the germ disk is more clearly to be distinguished from the underlying yolk, the nuclei occupying a relatively higher plane than in the preceding figure ; the furrow is seen (PI. II, Fig. 22), to leave below it undivided a well-marked layer of the germinal protoplasm. The first furrow has been observed ^ to divide the germ disk into segments of unequal size, an abnormality of cleavage wellknown in Teleosts,^ which was found to influence in no way subsequent development.

Second Cleavage (PI. I, Fig. 3), in all cases examined, occurs in a vertical plane approximately at right angles to the first. It is expressed in the germ disk only, and like the former furrow could not be traced in the yolk region of the egg. The polar corners of the blastomeres are the most prominent, sharply cut, and but slightly rounded. The nuclei remain in the horizontal plane of those of Fig. 2, and no change is seen to occur in the layer of protoplasm underlying the furrows. During cell division the nuclear changes are not prominent : the spindles are seen with difficulty, the chromosomes are small and obscure, and in the resting stage the outline of the nucleus can hardly be determined. In general the position of the dividing nuclei with respect to the blastomeres is similar to that of amphibian : shortly after division the nuclei are seen to be separated by a thick transparent disk of protoplasm, a segmentation plane which is later expressed in the surface furrow (PI. II, Fig. 23).

Third Cleavage (PI. I, Fig. 4) is also in a vertical plane : in general its direction is parallel to the first furrow ; ^ often,

1 By Prof. E. B. Wilson.

2 E. g. Ryder in the Cod and H. V. Wilson in Serranus.

2 This was accurately determined by E. B. Wilson in his experiments upon eggs attached to glass plates whose various cleavages were recorded.

No. I.]



however, as seen in the figure, it diverges somewhat, tending to appear meridional. In depth and lateral extension its furrows are entirely similar to those of earlier cleavage. The position of the nuclei in this stage is shown in PL II, Fig. 24. It will be noted that the blastomeres are now more widely separated by the first and second furrows j these interstices are the first indications of the segmentation cavity. A view of this stage as seen from the side is given in PI. I, Fig. 5.

FoitrtJi Cleavage (PI. I, Fig. 6) is again a vertical one ; it is in general parallel to the second furrow, resembling in other regards the third cleavage. The extent of the cleavage fissures may be made out in PI. II, Fig. 25, a section in which several nuclei are to be seen dividing for the fifth cleavage. During this stage many variations occur in the size and shape of the blastomeres ; they may readily be reduced, however, to the normal plan of segmentation. A vertical section of this stage (PI. II, Fig. 26) shows the depth of the furrows, and indicates as well the relation of blastodisc to yolk.

Fifth Cleavage (PI. I, Fig. 7), results in a normal stage of thirty-two cells : it is carried out often with wide variations, and the lineage of the blastomeres is to be followed only with difficulty. By serial sections of the more regular examples of cleavage in the late i6-cell stage, the nuclear figures (as for example those of PI. II, Fig. 25) permit, however, the ideal plan of the fifth cleavage to be understood. This the writer expresses in the accompanying figure (Fig. 3), in which is indicated the lineage of eight, sixteen, and thirtytwo blastomeres. Those derived first from the four original blastomeres are denoted by a, those from their derivatives by aa, those from their derivatives by aaa ; a second derivative from the original blastomere is denoted by b, a third by c, and

Fig. 3.

1 8 DEAN. [Vol. XI.

their subsequent derivatives, as bb, bbb, or cc, ccc. The writer believes, accordingly, that the plan of the fifth cleavage is carried out in the marginal cells' budding off their polar ends, giving rise to a circle of twelve cells ; the four cells of the animal pole dividing obliquely in a meridional plane. This mode of fifth cleavage corresponds clearly with that of the ideal type in Teleost segmentation (H. V. Wilson). In many examples of this stage the present writer notes that the four cells of the animal pole undergo horizontal cleavage, so that in surface view but twenty-eight blastomeres may be counted.

From the fifth cleavage onward the division of cells could not be satisfactorily followed, and the outward appearance of similar stages presents so many variations that they seem valueless to record. In the sixth cleavage (PI. I, Fig. 8), the only cell divisions that were noted as generally constant were those of the marginal cells : these undergo meridional cleavage, similar to the former one, its furrows extending no further than the margin of the cell-cap. ^ It is in fact in this stage that the cell-cap is largest in outward size. Horizontal cleavage has taken place irregularly, and has caused the lower layer of the germ protoplasm to be broken up into irregular cells whose upper and marginal limits are readily traced, but whose cytoplasm is seen clearly in many cases confluent with the yolk. This layer of yolk cells is represented in the vertical section of the ^%% at this stage (PI. II, Fig. 27). It will here be seen that the cell-cap consists of a loosely piled cell mass, of two (or three) tiers in depth, inclosing an irregular segmentation cavity. In the floor of this blastoderm cells are distinguished which below open into the yolk mass, and which, most important, may be seen splitting off nuclei into the yolk below. This process in the formation of merocytes the present writer regards as essentially the same as that occurring in Elasmobranchs, and believes that in it the ancestral

1 The writer has found no segmentation stages in which the furrows extend much lower than the equatorial region of the egg. He does not, accordingly, confirm the note and figure of Balfour, and is inclined to believe that the total segmentation of Lepidosteus occurs only as a variation.


condition of the periblast (of Teleosts) is most nearly represented. In this regard he would call especial attention to the segmentation stages immediately succeeding this, for it seems to him evident that the marginal cells undergo a metamorphosis similar to that concerned in the formation of periblast which H. V. Wilson 1 has figured.

After the sixth segmentation continued divisions result in the production of a flattened cell cap (PI. I, Fig. 9), irregular in surface and outline. Its marginal cells are irregular and inconspicuous, and their lateral angles obscure. The cell-cap increasing by continued horizontal division comes next to present an irregular but prominent summit ; its margin is irregular and slightly depressed. At a later stage (PI. I, Fig. 11), the marginal depression has become converted into a moat-like groove. Cells are still seen prominently along its sides although at the summit of the cell-cap continued divisions have rendered them so small that their outlines can be discerned only by means of sections. The noteworthy point, however, at this stage is that at certain points of the outermost cell-ring the cell-walls are found to have entirely disappeared and the nuclei are seen to have entered into the peripheral yolk. This change will at once be seen to correspond closely to that in Serramts which Wilson has figured in his memoir, Figs. 21 and 22. At the same time, furthermore, a median vertical section of this stage (PI. II, Fig. 28), indicates that it is not at the periphery of the cell-cap alone that this modification occurs ; the entire floor underlying the cellcap is now smoothly and distinctly differentiated ; it is no longer filled with the rough and half detached cell outlines of an early stage ; the nuclei scattered through it are seen in continued division, occupying a deeper and deeper plane in the subjacent yolk, and budding off at the surface additions to the cell-cap only at irregular points. The cell-cap is now seen to have a thickness of four cells, and their loose clustering has given rise to an irregular segmentation cavity.

1 H. V. Wilson, On the Embryology of the Sea Bass, Serramts atrarius. U. S. F. C. Bull., No. IX, 1S91.

20 DEAN. [Vol. XL


A very late stage of segmentation is shown in surface view in PI. I, Fig. 12, and in vertical section in PI. II, Fig. 29, Contrasting it with the stage last figured, the following differences are apparent : the summit of the cell-cap is larger and rounder ; division has caused the remaining cells of its margin to become indistinguishable ; its outline has become sharply defined, and its margin, if anything, slightly projects above the marginal groove : of the latter the outer boundary cells have almost disappeared, and their nuclei are now found dividing and migrating into the underlying yolk. The cell-cap has now a thickness of seven or eight cells ; its segmentation cavity has become larger and more distinct ; its line of separation from the yolk is even clearer than in the former stage. At certain points, as before, it receives increments of cells from the underlying yolk. The yolk nuclei are now exceedingly numerous in both central and peripheral regions, and of them as many as four or five tiers may be counted. The entire periphery of the cell-cap is now clearly distinct and separate from the yolk.


From these conditions the transition to early gastrulation is readily followed. In PI. I, Fig. 13, the cell-cap of the former figure has flattened out as a disk whose margins bend around and enclose the animal pole of the egg. The boundary between blastoderm and yolk is clearly drawn, although at one side, the future tail region of the embryo, its demarcation is' more sharply drawn. Pigment is now noticeable in the yolk region, and becomes still more evident as the blastoderm continues to enclose the yolk ; a light olive color is acquired which gives a sharper outline to the adjacent blastopore. A sagittal section of this stage permits the following growth changes to be noted : the cell-cap of Fig. 1 2 has resolved itself into a well defined roof for the segmentation cavity, forming a homogeneous layer of small cells, whose thickened margins are continuous


with the yolk ; at the animal pole it consists of about five tiers of cells, and of twice as many at its periphery (PI. I, Fig. 30). At the posterior region of the blastoderm occurs the sharp surface indentation which marks the dorsal lip of the blastopore ; there is as yet, however, no fissure beneath its margin ; the floor of the segmentation cavity has undergone important changes ; it is now formed of a (single) layer of round, many-sized cells, which the writer has seen in cases budded off from the underlying yolk, but which he believes are mainly derived from the peripheral region, where cell division is now most active ; the line demarking the yolk from this layer is at nearly every point most sharply drawn ; below it, however, no prominent yolk nuclei are to be seen as in the former stage, but the entire upper region of the yolk seems to have acquired a quite different character ; it now consists of elements which suggest those of the white yolk of the chick.

A slightly later gastrula is figured in PI. I, Fig. 14, a stage that is prominent on account of the indentation which the blastoderm shows in its hindermost margin ; it is here that the rim of the blastoderm is thickest and most sharply separate from the yolk ; its thickening is not widely different from the shield-shaped embryonic mass of Teleost, although wider and less evident. Sections show but slight changes from the earlier conditions ; in the margin of the blastoderm the cells have become more numerous ; the anterior lip is still closely in contact with the yolk, the posterior is free for a short distance under its immediate margin ; the embryonic thickening near its margin is due almost entirely to an increase in the number of cells of the blastoderm in what has now become the axial region of the embyro.

In PI. I, Figs. 15 and 16, are figured two later gastrulas : the yolk is in these only to be seen through the rapidly constricting blastopore, which in its latest stages is usually circular.i In the first the blastopore is already circular, all traces of the marginal indentation of Fig. 14 having disappeared ; this was observed to take place not by the concrescence of the

1 The writer was first led to suppose that the circular type of blastopore was abnormal ; this he now believes is the usual form at this stage.

2 2 DEAN. [Vol. XI.

sides of the indentation, but by a mode of unequal growth which caused the indentation to become less and less prominent until it disappeared. In Fig. i6 the marginal indentation has been retained and it is found to be lost only by the time the diameter of the blastopore is reduced about one-half.^ In Fig. 1 6 the margin of the blastopore is seen to be prominent, and the outline of the embryo may be traced in the lightcolored region anteriorly. A sagittal section of this stage, PL II, Fig. 31, shows the extent to which the lips of the blastopore are now separate from the underlying yolk ; it shows clearly the point of union on either side of the yolk and the cells of the inner germ layer (* of the figure). Its coelenteron {c) is similar to that of amphibian, and the writer believes that its mode of growth will upon careful study prove corresponding. The section just noted shows in addition the boundary of the segmentation cavity {s), and around the blastopore the partial separation of the outermost epiblastic cell stratum, « Deckschicht.' Of this the differentiation is later found extended over the entire surface of the egg. The yolk material slightly protrudes through the blastopore as a rounded yolk plug. No beginnings of the middle germ layer have as yet appeared.

From the conditions above described it seems to the writer that the gastrulation of Lepidosteus might be looked upon as an intermediate type, — primitive, moreover, inasmuch as from its beginning till the time of the closure of the blastopore but the two primary germ layers are concerned in its formation. It is intermediate inasmuch as it presents a striking similarity on the one hand to the gastrula of an Elasmobranch ; on the other that it parallels the holoblastic type of amphibian, and that it further indicates in its structures the most essential characters of Teleost. It resembles the gastrula of Elasmobranch in (1) its meroblastic origin, (2) the presence of generally diffused yolk-nuclei in a granular superficial layer of yolk, (3) the sharply marked character of the yolk surface below the segmentation cavity, or more accurately beneath its

^ Many variations occur; examples were noted in which the blastopore was elliptical, ovate, or even slightly constricted in its ventral margin.


irregular floor of loosely associated cells, (4) the even thickness, homogeneity, and sharp demarcation of its germ disk (outer germ layer), (5) the mode of its invagination, especially in the mode of union of the invaginated layer with yolk, (6) the formation of the indentation at the hinder margin of the blastoderm. In short, the addition of abundant yolk would seem in the opinion of the writer to render the type of gastrulation of Lepidosteus in every essential regard similar to that of Elasmobranch. On the other hand, the resemblances to amphibian gastrula might be looked upon as only superficial ; its most decided similarity seems but the result, of its approach to a holoblastic condition, as seen for example in the formation of a circular blastopore, and in the outward growth changes of the early embryo (90-130 hours). To the Teleost gastrula its resemblances may be summarized in (i) the origin of yolk nuclei from the marginal blastodisc (= periblast), (2) the plan of early segmentation, (3) the shield-shaped embryonic mass, and (4) as will later be seen, the absence of a true neurenteric canal.

The establishment of the outward shape of the embryo has its beginning in the stage figured from above and from the side in PI. I, Figs. 17 and 18. The embryo is here a light-colored cell-mass, whose rounded sides slope gradually to the yolk below : a minute pit is the disappearing blastopore, marking the tail region of the embryo ; the head end is sharply raised and slightly projecting ; immediately in front of it a circular and somewhat depressed area denotes the anlage of the vascular system. A sagittal section of an embryo of a slightly earlier stage, PI. II, Fig. 32, is worthy of careful study : it shows the condition of the closing blastopore, and the origin of the middle germ layer. According to the writer's observations the closure of the blastopore takes place in the following manner : at the stage figured in PI. I, Fig. 15, the rim of the blastopore is becoming thickened and rounded, and the dorsal lip is already considerably the thicker ; this mode of increase in the thickness of the blastopore rim is continued until its closure ; at about this time the relative thickness of both poles may be well seen in the section, PI. I, Fig. 32 : the outer

24 DEAN. [Vol. XI.

epidermic layer {cp), ' Deckschicht,' now appears to aid in the process of closure ; at the margin of the blastopore it thickens and appears especially distinct, and protrudes over the rim of the blastopore, as it does in fact in Teleost : its innermost cells, however, are still connected, but loosely, with those of the inner layer : the continued reduction of the blastopore now results in a funnel, closed at its pointed end by the inwedging epidermic cells ; closure and fusion of the lips of the blastopore next follow by what appears a concrescence of the sides of the funnel, an obliteration completed as the caudal eminence of the embryo is in process of forming. It will accordingly be seen that the observation of Beard as to the absence of a neurenteric canal is thus confirmed.

The view of that author, however, as to the origin of the middle layer "from the epiblast on each side of the middle line, and from the epiblastic region at the lip of the blastopore " has not been verified. It will be seen in the above section that the mesoblast (/;/) is clearly to be traced to its union with the hypoblast (or more accurately, perhaps, the undifferentiated tissue) in the underlying region of the blastopore. Its growth thence extends on all sides, but most rapidly forward in the direction of the embryo's axis : in the transverse section, PI. II, Fig- 33> gastral mesoderm is present ; its connection with the inner layer was observed only in the hinder portion of the embryo.

In a sagittal section of an embryo of about 60 hours, PI. I, Fig. 18, the region of the origin of the vascular system is of considerable interest. The outer layer, after its thickening at the head end of the embryo is seen suddenly to taper away as it slants downward continuously over the depressed area ; it now consists of the epidermic stratum, which has already been noted, and an irregular single-celled layer of formative epiblast, and in this condition it is continued around the yolk, PI. II, Fig. 34 (in this figure the drawing has not been accurately reproduced). The segmentation cavity was last seen in PI. II, Fig. 31 ; its flattened condition in that stage has now given rise to the vascular disc-shaped enlargement immediately in front of the embryo's cephalic eminence ; in other regions it has become


SO exceedingly flattened that it can be discerned only with difficulty. The floor of the vascular enlargement is the region of especial interest ; it is already covered with spherical mesenchyme elements which are seen in process of being budded off from the yolk ; they have certainly this origin, although the writer believes that they may also be derived from the marginal cells of both the upper and lower layers : the conditions however that are here presented appear strikingly similar to those figured by Riickert in Selachian. The fate of these vascular elements will later be referred to.

Early Embryo.

The stage in which the outward shape of the embryo begins to appear is figured in PI. I, Fig. 19. Its axis is almost a straight one, and the slender embryo is sharply constricted off with head and tail eminences especially prominent above the surface of the rounded yolk. The head eminence is the embryo's widest part, its outline is bluntly lanceolate, its dorsal surface is flat, but shows the presence of a faintly marked cord of cells in its median line. The tail eminence is less prominent ; it is well rounded, highest in its posterior part. The middle region of the embryo remains as yet at the ftg^ surface. A slight rounding in of the surface at the sides of the embryo is the first indication of a parietal zone. The following structures take their definite origin during this stage : notochord, primitive segments, central nervous system and blood vessels. Of the pronephric duct early traces are found but its more definite appearance is at a later stage. The notochord takes its origin from the hypoblast in a manner very similar to that of the Teleost ; by the time the entoderm has spread under the embryo its thickening is noticeable in the axial line ; the ridge that is thus formed is more prominent near the hinder body region and it is here that the chord is first seen separate from the entoderm. Its appearance at this stage is shown in PI. II, Fig. 33, c. In the embryo described four primitive segments are present ; the mode of their formation is to a degree shark-like inasmuch as the visceral and parietal

26 DEAN. [Vol. XL

layers of the middle layer which form them are early separate, and as traces of a lumen in early segments are to be found. The solid character of the later muscle plates appears strikingly Teleost-like. These the writer believes furnished the basis of Beard's observations ; they certainly appear, however, long before traces of the auditory capsules appear, and rather in the mid-region of the embryo's axis ; their increase is then as Beard states at the expense of the hinder tissue.

The definite origin of the central nervous system does not appear in a stage much earlier than that of PI. I, Fig. 19, and it is here only in the anterior region of the embryo that its relations may be determined. In the hinder trunk region the formative epiblast is found to be generally thickened but otherwise undifferentiated in the median line ; in the mid-region its layer becomes thinner and flatter ; in the head eminence, on the contrary, it becomes thicker, its boundaries are well marked, and it is deeply implanted in the median line ; a median cell cord is now to be recognized as its outward expression ; as yet a lumen has not appeared and the general appearance of the spinal cord is teleostean : no sense organs have as yet made their appearance. The prominent margins of the head eminence are formed of the anterior extension of the mesoblast ; and the writer believes that it was this mesoblastic rim of the head eminence that Balfour and Parker have figured as the early brain vesicles. The writer notes that the anterior end of the central nervous system in this stage ends not in a point of contact with the cells of the ectoderm (lobus olfactorius impar^) but in a thin continuous cell sheet which underlies the ectoderm and forms the roof of the vascular area (segmentation cavity). The relation of the mesenchyme elements, which were earlier seenoccupying the vascular area, to the formation of blood and vessels is by no means easy to determine ; and the wide divergence among observers in their studies on the origin of the vascular system in well studied types seems to render inadvisable what can be but an imperfect note on the conditions of Lepidosteus. The writer would record, however, that as far

1 As Kupffer has demonstrated in the Sturgeon, and as others have recently described in other vertebrates.


as his observations have extended, the characters of this form show a close agreement with those which Ruckert has described in Selachians ; the present writer finds, for example, that the vascular mesenchyme cells become early associated in loose clusters in front of the head region of the embryo, and these he has traced assuming definite shape as a blood tube ; he notes that the early blood cells resemble closely the cells of the early mesenchyme and that these seem to have their origin in the anteriorly extending yolk-margin of the vascular cavity.

The early embryo has clearly attained its outward form in the stage figured in PI. I, Fig. 20, in which seven primitive segments are present. The outline of its trunk has become constricted off from the yolk ; the eminences both of head and tail are rounded and prominent, the mid-region of the trunk is well raised above the yolk, and its parietal lamellae are now sharply marked as a groove on either side of the embryo. In the dark lateral margin of each parietal layer the pronephric duct has become established. The advancing structures of this stage which are to be briefly noted are (i) the brain vesicles, (2) optic capsules, (3) the establishment of a lumen in the anterior nerve axis and (4) pronephric duct.

The changes of the anterior part of the central nervous system of a stage slightly later than this have been described by Balfour and Parker. The first appearance of a lumen in the embryonic brain is to be found in the present stage. The faintly marked axis of the head region denotes the wedge-like insinking of the brain region, and its irregular margin is at present the only outward indication of the vesicular enlargements. Sections show, however, that the anlage of the optic vesicles exists in a grouping of cells on either side of the axis at the broadest part of the cephalic plate ; and further that the lumen which is found within the central axis first occurs in the mid-region of the brain, arising, as Balfour and Parker believed, from the disassociation of cells : that this cavity, however, occurs before that of the optic vesicles is demonstrable, and even in a stage 24 hours older than this. The English observers have figured a similar condition {Ref. 5, PI. XXII, Fig. 26), although they state the lumen as there probably occurring,

28 DEAN. [Vol. XI.

but not to be seen, " the section merely passing through them to one side." The origin of the auditory invaginations has been well figured by Balfour and Parker: it is to be noted that their appearance is about lo hours later than of the optic vesicles. The pronephric duct in its earliest stages seems to the writer more closely comparable with that of Elasmobranch than earlier writers have inferred : the cell end, budded out of the parietal middle layer was noted (in embryos of 1 1 somites) as at first solid ^ later acquiring its lumen through the disassociation of cells; its irregular ridge-line projecting towards the outer layer suggests closely a condition of Pristinrus, and its early connection with the epidermis in the region of the fifth somite is an additional character of comparative interest. The pronephric duct occurs early in development, before even the appearance of the lumen in the optic vesicle, or in the hinder neuron, and before the constriction off of any part of the gut, in strong contrast to its retarded development in Teleosts. The history of the pronephros, as given by Balfour and Parker, has not been followed in detail : this together with the discussion of the inner germ layer might, perhaps, be better followed in a study of the later development of the embryo.

III. The Early Development of Acipenser.

The egg of the Sturgeon when extruded is slaty-gray in color : its general surface is pigmented and a rich mass of pigment darkens the animal pole, often displayed as a cross, or star-like marking (PI. Ill, Fig. 35). The germ-disc is already clearly outlined, and its lighter color is made especially prominent by an encircling darkly pigmented zone of the yolk.

The ^%% membranes at this stage may be compared with those of Lepidosteus as figured by Mark.^ The zona radiata is similar in relative thickness, and the villous layer is clearly distinguishable, although staining but lightly. The outermost layer is distinctly separate from the others: as Salensky notes, it stains readily (but irregularly) with haematox3din ; it is many-layered and thick, often four times the thickness of the

1 The present writer confirms the observations of Beard. 2 ^gj, 17.


radiate and villous layers ; its outer surface is irregular and raised in projecting bosses of many sizes. The outer layer, which may prove the equivalent of the gramdosa, is clearly the adhesive envelope of the ^g^ : before immersed in water its thickness is less than that of the combined inner layers ; its structure is lamellar, and its specialized hygroscopic character is doubtless the cause of its cellular elements being difficult to distinguish. The glairy mass which the fertilized eggs give off after some minutes' immersion in water seems clearly the product of the outermost envelope. The writer, it will be seen, differs in the interpretation of the ^gg membranes from Mark, who regarded it " probable that the outer layer would be found to correspond to the villous layer of the Gar-pike," and "that the middle layer was simply the differentiated outer half of the zona." The present writer is, however, by no means convinced that the outer layer is to be regarded as the granulosa, although he regards it probable. During the later developmental stages, when partly detached and greatly reduced in thickness, it becomes most similar to the outer membrane of the Gar-pike : its irregular thickened and distended character during the period of the eggs' fixation may accordingly prove a specialization of its outermost layers to acquire an hygroscopic function, or may be even, but less probably, due to a direct secretion of its glandular cells.

The outward changes which the Q.gg has undergone by the time of the first cleavage are not noteworthy. Sectioned at this state the ^gg presents but slight differences from the conditions of Lepidosteus, cf. PI. I, Fig. 21, and PI. Ill, Fig. 55. The line of demarcation of the deutoplasm is less clearly drawn and the deutoplasm itself is more coarsely granular, its elements usually containing a store of more finely differentiated yolk. The nuclei are seen dividing at a similar niveau, and are larger, though less distinctly marked. In preparations of all early stages the chromosomes are not readily determined. Certainly the most characteristic feature of the e^gg is the presence of abundant pigment : it is seen massed at the animal pole as if in a vortex, its granules sinking in thread-like clusters downward and outward.

30 DEAN. [Vol. XI.


The first cleavage is vertical and occurs at about the same interval after fertilization as in A. rutJicnns. As Salensky has determined, it separates the blastomeres only in the region of the germ-disc. It is deepest at the animal pole but traverses the blastodisc sharply, PL III, Fig. 36, causing a narrow surface fissure different from the trench-like furrow that has been noted in Lepidosteus. Until the appearance of the second cleavage its marginal limit is in the pigmented zone of the yolk bordering the blastodisc. The present writer notes that in transverse vertical section it differs little from that figured in PI. II, Fig. 22 ; the furrow is more sharply cut and deeper but does not penetrate into the yolk. The pigment at the animal pole is now restricted to a circular area comparatively regular in outline. In the earlier stage the outline and extent of the surface pigmentation has apparently no influence in foretelling the direction of the first cleavage plane.

The second cleavage, again vertical, corresponds closely with that of the Gar-pike, It is approximately at right angles to the first plane, exceptionally oblique, as in the figure, PI. Ill, Fig. 37, and may intersect the first furrow with noteworthy Polflucht, Its first outward appearance is in the region of the pole, thence it may be followed as it extends marginally. When it has traversed the blastodisc its depth is approximately that of the first furrow, although outwardly the latter is readily distinguished by its greater extent ; it now surrounds (almost) the entire circumference of the Q^^g, but in the yolk region, as Salensky notes, its furrow is of the most superficial character. The position of the nuclei in this stage is entirely similar to that of Lepidosteus, and has been figured in vertical section by Salensky, Rcf. 29, PI. XV, Fig. 7: his drawing illustrates the depth of the cleavage and the line of demarcation between germ and yolk. A horizontal section (slightly oblique) through the region of the nuclei in this stage is given in PI. IV, Pig. 56; in this figure the distribution of the pigment is to be traced ; at the surface it has accumulated in the rounded and

No. I.]


irregular corners of the blastomeres, and below the animal pole it has extended deep into the germ-disc.

The tJiird furrow, PI. Ill, Figs. 38 and 39, is of interest on account of its irregular appearance : although usually vertical,^ parallel to the first plane of cleavage, it may be meridional and even strictly horizontal, with a range of intermediate variations. As a large proportion (about fifty per cent) of the eggs examined conformed to the plan of cleavage of Lepidosteus the writer believes that this is the normal mode of cleavage of sturio. It is first expressed, like the second cleavage, in the immediate region of the animal pole, thence extends both centrad and peripherad, but does not pass the limits of the blastodisc.

Fig. 4.

Marginally, however, it may attain the bordering pigmented zone. It is in general readily distinguishable from earlier furrows : the first cleavage has by this time completed its circle at the yolk-pole of the ^^^ while the second has passed downward through the pigmented zone and is half surrounding the yolk mass. The niveau of the nuclei and the general pigmentation do not differ from the earlier conditions. A horizontal section, PI. IV, Fig. 57, may be compared with that of the eight-celled stage of Lepidosteus, PI. II, Fig. 24. It will be noted that in Acipenser the pigment has penetrated the first and second

^ The observations were made upon eggs taken from a tray of which about 90% of the remaining eggs was successfully hatched.

32 DEAN. [Vol. XI.

furrows to the niveau of the nuclei, that the blastomeres are of a more uniform texture, and that their nuclei are larger, more regular in outline, and more clearly differentiated in structure.

Variations of the eight-cell stage are shown in the accompanying figure, Fig. 4, the meridional form, A and PI. Ill, Fig. 39, is not uncommon ; more usual variations are C and D, less common are B, E and F, and forms, as G and H, showing an irregular horizontal cleavage are the rarest, occurring in from 4-6fo of the specimens of this stage which the writer has examined. A wide range of variation in the cleavage of this particular form, in view of its supposed affinities is naturally suggestive, although logically perhaps, of little morphological importance : for cleavage changes have been recorded in Petromyzon,^ Teleosts^ and amphibians,^ and are, as yet, of doubtful significance, in cases referable to mechanical causes, alterations of temperature or of water density.

In the case of the Sturgeon, however, it should be stated that the variations in cleavage occurred in a general rate of proportion among the normal eggs of the hatching trays, and on this account appear to be worthy of especial interest. It would certainly appear evident that a variation had occurred in the amount of the yolk material, in cases sufficiently great to permit the third cleavage plane to become horizontal. This peculiarity of the Sturgeon egg will later be referred to, in the discussion of the relationships of Ganoids, as suggesting a tendency toward the evolution of a more perfect holoblastic condition. In the Gar-pike similar variation of cleavage does not occur; of all the eggs examined — to the number of several hundred — no widely-marked differences from the conditions figured in PI. I, Fig. 4, have been recorded.

The variations (Figs. A-F) are disposed symmetrically with reference to the first plane of cleavage (which passes from

1 McClure, ("93) Zool. Am., XVI, p. 429.

2 Ryder, H. V. Wilson, Agassiz and Whitman, and others.

8 Rauber, Morph. Jahrbuch, 1S83; Jordan, J. of Morph., 1S93; v. Ebner, Festschrift f. A. RoUett, Jena, 1S93; ^^^ results of experimental studies of Roux, Morgan, and Hertwig.



right to left in all figures), and are in this respect sufficiently noteworthy to suggest that the first cleavage plane in Sturgeon as in amphibian (as Roux^ recently appears to have reconfirmed) determines the sagittal plane of the embryo.

FourtJi cleavage is again vertical, and the i6-cell stage is not unlike that of Lepidosteus. According to Salensky there occurs in this stage in riitJiemis horizontal cleavage ; but on account of the many variations which probably occur here^ as well as in sturio, the writer is still inclined to believe that the normal fourth cleavage^ in the sterlet may prove to* be vertical. In sturio at this stage the four central blastomeres, PI. Ill, Fig. 40, are irregular in outline, often projecting above the surface, and are usually separated by well-marked fissures : they are still, however, connected with the underlying layer of the germ disc as in the earlier stage. Wide variation in the i6-cell stage is to be noted in the size and shape of the four central blastomeres, in the area and distinctness of the polar pigmentation, and in the degree of obliquity of the fourth cleavage furrow, the latter often tending to become meridional, and in cases equatorial. On the lower pole of the 0:%% the first and second cleavage furrows have by this time intersected.

In the stage of fifth segmentation the writer's material is deficient.

The sixth cleavage is represented in PI. Ill, Fig. 41, and may be compared with the corresponding stage of Lepidosteus in PI. I, Fig. 8. Its lower pole is shown in PI. Ill, Fig. 42. Cell division has by this time turned the blastodisc into a cap of cells of irregular size and outline ; horizontal cleavages have occurred, notably in the region of the animal pole, but these cells have not been caused to be lifted above the surface of the ^%^ ; meridional cleavage has occurred in the marginal cells, but is expressed in an irregular manner, often by becoming oblique, separating only a corner of the marginal blastomeres ;

^ Anat. A)!z., 1S94.

2 Thus Salensky states, " Quand la partie superieure de I'oeuf (germe) est divisee en ^//x parties — the italics are the present writer's — il apparait dans le germe des sillons transversaux."

3 The writer has been unable to compare the results of Peltsam (J?ef. 20).

34 DEAN. [Vol. XI.

sometimes, as in the left of the figure, cleavages remaining meridional, seem unable to penetrate the pigmented zone of the yolk and thus for a time remain undifferentiated. The connection between yolk and cytoplasm which thus maintains seems accordingly equivalent to the condition of the marginal cells in a corresponding stage of Lepidosteus, whose significance has already been discussed (p. 19). In the surface view of the animal pole the pigmentation, in early stages so characteristic of the Sturgeon, is hardly visible, and the fissures between the blastomeres are greatly reduced. The lower half of the Qg^ is traversed by the third cleavage furrow, which intersects the first and second furrows at moderate angles. There are thus, therefore, in the 64-cell stage but 6 cells visible on the lower hemisphere, a condition which differs somewhat from that of r7itJieims, where, according to Salensky, "lorsque Ton pent distinguer douze segments dans la partie superieure de I'oeuf, sa partie inferieure n'en montre encore que 6." Nor is there to be seen in the yolk region the boldly rounded cell outlines which Salensky has figured, Ref. 29, PI. XV, Fig. 10.

A stage about one hour later, PI. Ill, Figs. 43 and 44, already exhibits a great advance in development. The entire light-colored pole is now composed of finely divided cells, even to the margin of the pigmented zone ; at the animal pole the outlines of the segmentation cavity may be faintly determined through the fissures between the cells, and the surface of the yolk-half of the egg is now seen subdivided into many-sized polygonal cells. A cross section of this stage, PI. IV, Fig. 59, and of the same stage at the side, Fig. 58, shows important differences from the corresponding stage of Lepidosteus {cf. PI. II, Fig. 28) ; the blastoderm is growing within the limits of the egg's spherical curve, instead of on its surface by increment from below. It will be seen that the segmentation cavity is here separated from the surface by but a single layer of cells, instead of by the cell-mass of the Gar-pike ; that it is of more definite size and shape, that its floor is now cellular, and not of the clearly marked, merocyte-bearing, Elasmobranch type of Lepidosteus.



In this blastida of Acipenser, however, the relations between yolk and blastomeres may best be understood by referring them to the conditions of the Gar-pike. The periblast-Hke floor of the segmentation cavity in the latter form has now become reduced to a few irregular yolk-bearing cells, which in number and position correspond closely with the merocytes of PI. II, Fig. 28. A reduction in the amount of yolk material might clearly account for this change. As in Lepidosteus the lowermost cells of the blastoderm are derived from cells whose cytoplasm is connected with the yolk, and are budded off, as in PI. II, Fig, 29, from central as well as from marginal regions. The cells connected with the yolk are seen undergoing nuclear divisions, and appear to represent in function the merocytes of Lepidosteus. The irregular polygonal cells of the yolk hemispheres are found to possess dividing nuclei close to their outer surface, but are altogether indistinguishable in sections. Naturally, however, the elaboration of the yolk material, although occurring on every side, is most actively carried on in the uppermost part in connection with the cell growth of the blastoderm.

In a slightly later blastula the increased size of the segmentation cavity and its more definite outline may be seen in surface view, PI. Ill, Fig. 45, and in section, PI. IV, Fig. 60 (with this cf. Ref. 29, PI. XV, Fig. 11). By study of the nuclear figures of the cells roofing the segmentation cavity in the preceding stage, it is found that the present two-celled segmentationroof has been the result of horizontal cell cleavage. The thickened sides of the cavity appear due to the normal increase of marginal cells together with additions from the underlying yolk cells. These in the floor of the segmentation cavity have now produced a double layer of cells, whose upper elements are in size, texture, and nuclear character not to be distinguished from those at the sides. The region, in other words, of the yolk cells is retreating centrad. The writer notes that his description of this stage of the blastula does not agree with that of a corresponding stage of the sterlet given by Salensky.

36 DEAN. [Vol. XI.

The latter author has figured an early blastula whose segmentation cavity has a roof three cells in thickness, the sides one cell in thickness, and a floor lacking in differentiated cells. Salensky, in addition, figures the yolk mass totally traversed by one cleavage plane and partially by a second.

In a slightly later stage the walls of the segmentation cavity have greatly thickened. The roof and sides of the cavity are now of 8-9 cells in thickness and its floor is of 4-5 tiers of cells which preserve their yolk-cell characters. Differentiated cells in the lower hemisphere of the Q.gg are not to be distinguished. Pigment is plentifully scattered in all cell layers of the roof of the segmentation cavity.

A later blastula has been figured in surface view in PI. Ill, Fig. 46, and in vertical section in PI. IV, Fig. 61. It is comparable to the stage which Salensky has figured in PI. XVI, Fig. 12, but differs from the latter notably. Thus its yolk mass is not as yet divided into cells, the segmentation cavity is larger, more definite in outline, its roof thinner at the animal pole, its floor, slightly concave, composed of definite and uniform cellular elements. The stage figured immediately precedes gastrulation, and it is doubtless on this account that a marked asymmetry maintains in the roof of the segmentation cavity : below the thicker side will occur the dorsal lip of the blastopore : outwardly this stage is of finely finished appearance, in contrast with that of Fig. 43 ; continued division has rendered the cells of the animal pole so minute that this region appears waxy ; the marginal pigmented zone has in addition lost its well marked appearance, and the cell outlines of the lower hemisphere are distinguishable but faintly. The pigment which now appears concentrated at the animal pole is in the region where the roof of the segmentation cavity is thinnest {cf. PI. IV, Fig. 61).


Gastrulation begins at a stage represented in PI. Ill, Fig. 47, and in vertical section in PI. IV, Fig. 62. Outwardly this stage is closely similar to that last figured ; it possesses the dark pigmentation at the animal pole, the white zone of minute


unpigmented cells, and the lower hemisphere of dusky color and faint cell outlines. But in the later stage each of these characters has become different in details: the pigmented tract at the animal pole is of lighter color and of obscurer boundary ; the lower margin of the white zone is now more distinctly drawn and at one side is already to be recognized as the dorsal lip of the blastopore ; the lower hemisphere is darker in pigmentation and its marking of cell outlines is more obscure. The section shows but minor changes from the conditions of PI. IV, Fig. 61. The roof and floor of the segmentation cavity are thicker in cell layers : the cavity itself has received an angular extension in the region above the dorsal lip. The latter is observed as a well marked notch between the small and regular cells of the upper hemisphere and the large and irregular cells of the region of the pigmented zone. These now, for the first time, are to be clearly distinguished, and are seen to be largest at the region of the lip and smallest at the opposite side of the embryo. It is to be noted that the outermost stratum of the cells of the upper hemisphere has now differentiated as the ' Deckschicht.' This has been figured by Salensky, although not mentioned in his text. Further differences from the conditions described in riitheniis are to be found in (I) the roof of the segmentation cavity ; this in nithemis is constituted mainly of a layer of cells like those of the yolk region {cf. Ref. 29, PI. XVI, Fig. 13), (II) the shape of the blastopore, which in the sterlet is described as crescentic, (III) the character of the blastoporic "invagination," since in mthenus the lip of the blastopore appears to be widely separated from the yolk cells, and (IV) the qualitative difference, as maintained by Salensky, between the cells of the upper and of the lower hemispheres. The present writer would here note that in his studies of stiirio, he finds nothing to warrant the qualitative distinction that the Russian author has drawn ; he was able to observe in the cells of the upper hemisphere neither zone corticale of the cytoplasm nor any indications of the "petites bosselures qui ressemblent a des pseudopodes lobes," and believes it probable that these were due to an imperfect method of preservation.

38 DEAN. [Vol. XI.

In PI. Ill, Figs. 48 and 49, are figured two later gastrulas, and vertical sections of those stages are given in PI. IV, Figs. 63 and 64. Outwardly these stages arc conspicuous, on account of the sharp color contrast they present in their unpigmented and pigmented zones ; the vegetative pole of the o,^^ is seen to darken in color as it becomes reduced in outward size by the constricting blastopore. The uppermost part of the embryo is in these stages somewhat depressed, and is slightly darkened by pigment. In the earlier gastrula the dorsal lip of the blastopore is more sharply drawn, — shown in the left of the figure, — while in the later form the dorsal lip is often indicated by a slight nick-like indentation of the rim of the blastopore. The thickening of the dorsal lip in the median plane, which shortly appears, is the first indication of the embryo's axis. Examination of sections of these stages enables us to better understand the advances which have taken place in the development. In the earlier gastrula the blastopore's dorsal lip has separated from the yolk cells and has enclosed about 45° of the egg's circumference ; it is thicker comparatively than that of Lepidosteus, and already includes the three germ layers which are seen to become confluent near the margin of the lip. Its exact mode of growth can of course only be determined experimentally, but from the blunted end of the coelenteron it appears that to some degree a growth of the entire lip has taken place. The segmentation cavity has greatly flattened, to a degree in fact which renders it difficult to be determined in the anterior region of the embryo ; its floor has also extended and flattened, and its four or five layers of loosely associated cells are often seen clearly distinct in many regions from the yolk mass. As yet only a trace of the anterior lip of the bias-, topore has appeared ; here the cells of the lower hemisphere are seen to increase in size and irregularity in the direction of the dorsal lip. In this region pigment occurs plentifully, and is present in notable quantity in the cells lining the coelenteron.

The older gastrula may be directly compared with that of Lepidosteus, PI. II, Fig. 31. It presents a marked advance in the growth of the blastopore ; its dorsal lip now surrounds a quadrant, and its ventral lip about 20° of the egg's circumfer

No. I.]



ence. The segmentation cavity has acquired a pronounced deepening in the region of tlie end of coelenteron, but its anterior limits, as before, remain difficult to be determined ; its roof is now greatly reduced in thickness, in a large part of its extent a single-celled layer ; it is densely pigmented and its cell boundaries may be distinguished only with the greatest difficulty. The immediate rim of the blastopore is thickened, especially in the dorsal lip : here the thickened rim, by being pushed against the yolk cells, has acquired a notch immediately in front of it, which will later be discussed as the homologue of Kupffer's vesicle. Pigment again occurs in all invaginated


Contrasted with Lepidosteus, this stage of the Sturgeon will at once be seen to present differences which appear interpretable as of specialized character ; thus we may contrast :

Gastrula. Segmentation cavity.

Roof of cavity.

Lepidosteus. Two-layered.


Thick, little pigmented.

Acipenser. Three-layered.

Broadly dilated near end of coelenteron, its anterior limit almost indiscernible.

Thin, single-celled for a large extent of its surface, heavily pigmented.

Walls of coelenteron. Unpigmented Rim of blastopore.

Thin, somewhat tapering distally.


Thick, blunted, thickest at dorsal lip, where an anterior notch (Kupffer's vesicle) is present.

A late gastrula, PI. Ill, Fig. 50, shows the continued reduction of the blastopore, and the appearance in surface view of the embryo's axis. The region surrounding the blastopore appears slightly conical in its surface curvature, the opposite end of the ^^g flattened and finely pigmented. The light-colored outer layer and the blackened yolk plug of Sturgeon present a well known contrast to the pigment conditions of the amphibian gastrula. The embryo is in this stage but faintly discernible ; it appears as

40 DEAN. [Vol. XI.

a somewhat opaque tract of cells, whose anterior end, broadening widely, represents the cephalic plate, and whose obscure dusky line, perpendicular to the rim of the blastopore, is the anlagc of the hinder neural axis. The blastopore is now circular ; it is only at a stage represented in Fig. 49 that any indication of an indented rim has occasionally been found. In a slightly later stage are to be noted changes in the shape and roof of the segmentation cavity, in the growth of the dorsal lip and the rim of the blastopore, and in the coelenteron in its dorsal region. The dilated region of the segmentation cavity has become deeper, its anterior and posterior margins closer together and more sharply marked. The roof of the dilated area has thickened and is irregularly pigmented. While the ventral lip of the blastopore has remained as in the preceding stage, the dorsal lip, PI. IV, Fig. 65, has increased greatly in length, now enclosing about 130° of the egg's circumference; in thickness, however, it has remained as in the last figure, except at the blastopore's rim. In this entire region an increase in cell material has become marked, least noticeable at the ventral and greatest at the margin of the dorsal lip. Pressure against the yolk plug seems here to have been reasonably the cause of the inflected rim. The coelenteron below the dorsal lip of the blastopore is notably deeper (ecto-entad) than in the former stage, its fundus is more broadly rounded, and its recessus immediately below the dorsal lip, i.e., Kupffer's vesicle, is deeper and sharply trench-like. In the figure the limits of the germ layers are shown ; the inner layer is here thin, but in the axial line becomes the thickened anlage of the notochord. The thickening of the outer layer which caused the appearance of the embryo is found to be exceedingly slight and can only be seen satisfactorily in transverse sections. The Deckschicht, which was earlier noted, cannot be satisfactorily distinguished in this stage.

Further growth of the embryo is illustrated in PI. Ill, Fig. 51. Here the cephalic region is seen to have spread out, ovate in outline, as if flattened upon the rounded surface of the egg. The spinal axis is prominently marked, and at its hinder end communicates with coelenteron ; a small light-colored band


bridging its hinder margins forms an imperfect roof for the neurenteric canal. The elliptical margin of the blastopore appears thick and opaque (PI. IV, Fig. 66). From it, passing forward and slightly diverging, terminating near the hinder region of the head, are the pronephric ducts. The closure of the blastopore and the detailed establishment of the neurenteric canal may at this point be most conveniently understood. A condition later than that of PL IV, Fig. 64, is given in the same plate, Fig. 65 : it illustrates the continued reduction in diameter of the yolk plug and its great increase in (ecto-entad) length : the inflected rim of the blastopore is now exceedingly deep, the recessus (Kupffer's vesicle) under the dorsal lip, prominent, deep but somewhat rounded : a general growth in extent of the entire outer wall of the coelenteron appears to have taken place since its distance from the yolk is seen to have greatly increased. At a subsequent stage. Fig. 66, when the blastopore has become reduced to but half the diameter of that last figured, the yolk plug loses its distinct cylindrical character : it comes to conform to every irregularity of the thickened blastopore. Further reduction of the diameter of the yolk plug takes place regularly : in Fig. 6^, it has become greatly narrowed, and as the section is slightly oblique its connection with the neural canal may be seen : it still touches the surface at a narrowed point. An outward view of these conditions is to be seen in PI. Ill, Fig. 53 : the deepest point of the indentation between the tail folds is the entrance of the neurenteric canal ; immediately below it, at the darkly shaded point, is the disappearing remnant of the yolk plug. In this figure may in addition be seen the hinder limits of the parietal zone, the outline of the neural axis, the undifferentiated tissue of the primitive segments, and at the embryo's extreme margin the hinder part of the pronephric ducts : the tail folds are greatly flattened, but at the hindermost margin are slightly raised above the surface. A final stage in the fate of the blastopore is to be seen in Fig. 54 : the tail folds have fused in the median line, leaving a slit-like opening into the neurenteric canal to be seen at the surface. In sections, PL IV, Fig. 68, and Fig. 69, it seems evident that the lower part,

42 DEAN. [Vol. XL

if not a large part, of blastopore is retained as the hindermost portion of the neurenteric canal, its greatly diminished yolk contents disappearing. In the surface view, PI. Ill, Fig. 54, may be seen prominently the neural canal, primitive segments, and, now depressed and passing under the flattened tail mass, the pronephric ducts.

The origin and fate of the vesicle of Kupffer appears closely connected with the closing blastopore. Within the entire circle of the blastopore's rim occurs the trench which in section has been noted in the different stages of gastrulation. When seen in vertical section it appears under the dorsal lip as the deep recessus which the present writer regards as Kupffer's vesicle. In Sturgeon the fate of this structure (which may be followed in the sections, PI. IV, Figs. 64, 65, 66y 6Z, 69) shows conclusively that it can be regarded as but a growth adaptation of the gastrula, a condition due to an extreme thickening of Randwulst. It is most marked in the mediati line in the tail region, since it is here that the embryo has acquired a keel-like thickening which exerts a mechanical influence in deepening the Randwulst and causing the underlying cavity to become enlarged. In the Teleost where the concentration of the embryo in the median plane is extremely marked, Kupffer's vesicle may well represent an expression of its mode of growth : the germ ring is clearly the rim of the blastopore — or more accurately perhaps the circumcrescence margin — of Acipenser. This interpretation of the vesicle, the writer believes, would adequately explain its peculiar conditions in Salino or Esox : in Salnio fario he has examined the vesicle and verifies the observation of Henneguy as to its possessing a flooring of entoderm cells : this layer, loosely separate from the undifferentiated cells in the anterior region of the tail mass, might, the present writer believes, readily remain in contact with the periblast below, as the mechanical growth change caused the vesicle to appear.

The outline of the early development of the Sturgeon might now best be completed by a discussion of the differentiation of the germ layers.

The origin of the outer layer has already been traced (pp.


37, 38). In PI. Ill, Fig. 51, the head outline is entirely an epiblastic thickening : anteriorly it is a pad of almost uniform thickness ; caudad the thickness of epiblast tapers away, but at the rim of the blastopore suddenly increases, dipping down into the undifferentiated tissue. These characters of the epiblast may be well seen in the (nearly) sagittal section of PI. IV, Fig. 66.

The central nervous system is earliest developed at its extreme ends, brain and neurenteric canal. Between the condition figured in PI. Ill, Fig. 51, and in section PI. IV, Fig. 66, and that of the stage of PI. Ill, Fig. 52, and of the sagittal section PI. IV, Fig. 71, — slightly earlier — notable changes have occurred : there has been a concentration of formative epiblast in the brain region ; its anterior end has dipped deeply down, and has acquired a lumen, whether originally by cell disassociation or by process of invagination the writer has not been able to determine. He is certain, however, that in the initiatory process the cell thickening was deep and sharply marked in the median line. In the section given it is of especial interest to note the point of union, x, of the front end of the brain with the formative epiblast : it is situated at a remarkable distance tailward. A cross section at this point is figured in PI. IV, Fig. 72. It is evident from the longitudinal section that the extension of the brain forward has been caused by the increase in size of its sides and floor. The width and flatness of the ventricle should be noted. By study of serial sections tailward of this stage it is found that the neural canal becomes thicker (shallower) and narrower: there occurs no evidence that the central canal has been formed by cell disassociation ; the formative epiblast rounds abruptly into it, and it appears broadly trench-like : it is roofed, however, dorsally by a strip of cells which appear in contact with both Deckschicht and marginal cells of formative epiblast (PI. IV,

Fig. 73) In the present paper it is not intended to discuss the studies of Kupffer on the morphology of the head of vertebrates. His admirable description of the later development (beginning with the 45th hour) of the brain of Acipenser has been followed by

44 DEAN. [Vol. XI.

the present writer and his observations generally confirmed. His studies of stages earlier than Kupffer had secured do not, however, permit him to believe with the German author, "dass sich beim Stor das urspriingliche Vorderende des Neuralrohres mit Sicherheit hat feststellen lassen," although not doubting that in the Lobus olfactorins impar exists the last connection between brain and epiblast. The earliest Vorderende of the neural canal which the writer has observed occurred in a stage about twelve hours earlier in development than that figured by Kupffer. It has already been noted in PI. IV, Fig. 71. Between this stage and that of Kupffer it has been found that connection exists with the epiblast in front of this point, which corresponds in position with that of Ep"^ of Kupffer's Fig. 13. Sometimes in fact the entire roof of the brain in this region can hardly be regarded as separate from the epiblast : and even a short time before the Lobus comes to be distinctly marked, the anterior region of the brain is for a broad extent {x-x) firmly fused with the epiblast (PI. IV, Fig. 74): that this fusion together with that of the later occurring Lobus is of secondary origin, the present writer is satisfactorily convinced. A discussion of the mode of origin of the hypophysis is reserved for future publication. It may in summary be said that the mode of development of Acipenser appears peculiarly specialized, and that many of its essential features in earlier stages seem entirely due to its flattened conditions of growth.

The latest stage in the development of the central nervous system to be noted in the present paper is that figured in PI. IV, Fig. 52, several hours earlier in development than that of Kupffer's earliest stage (Fig. i). The rather prominent outline of brain and optic vesicles may in sections be shown to be partly due to pigment contained in the cells of the floors of these cavities. The fore-brain is seen continued far forward, the mid-brain is proportionately large, and the hind-brain is as yet unwidened. The tapering tip of the fore-brain indicates indistinctly the Lobus olfactorins ivtpai% which is here the last point of connection between brain and epiblast. The lightcolored margins of the central canal are now clearly marked ; a


surrounding and indistinct zone of dusky color is mainly mesoblastic, while outermost the dark-colored parietal region indicates the general boundary of the fore-gut : in its foremost margin is the anlage of the heart. Pronephric ducts appear at the sides of the neural axis, terminating in a somewhat obscure way in the neighborhood of the hind-brain. As yet traces of neither gill slits nor auditory vesicles have appeared.

The inner germ layer m.3.y next be considered. In the stage of PI. IV, Fig. 64, the entoderm may be traced from within the rim of the blastopore, lining its margins, continuous with the cells of the surface of the yolk mass. Upon the closure of the blastopore, Figs. 66, 6"], 70, the entoderm of the outer wall of the coelenteron is a layer of cells distinctly separate from outer layers : at its lateral margins it gradually becomes continuous with the cell layer of the yolk which constitutes accordingly the inner wall of the gut. The yolk is therefore to be interpreted as identical in its relations with that of Elasmobranch, Teleost, or urodele, and is not to be compared to that of Ichthyophis, as suggested by Ryder, Ref. 22. At this stage the extent of the gut is to be seen in the sagittal and cross sections above referred to. The outer limit of the parietal zone of Figs. 50, 5 r marks in general the boundary of the gut. The notochord arises in the normal manner, a rod-like thickening of the hypoblast in the line of the embryo's axis. Beginning at the rim of the blastopore, thick and wide, as it extends forward it diminishes in size. Fig. 70, and is finally indistinguishable in the region of the hind-brain. The writer can find nothing in the mode of origin of the notochord to suggest the mesoblastic derivation maintained by Salensky.

The mesoblast in Acipenser was regarded by Salensky as derived from the entoderm in as early a stage as that figured in PI. IV, Fig. 62. Here the entire non-entodermic portion of the rim of the blastopore seems to be identified by the Russian author as 'mesoderm,' the entoderm extending around the yolk mass but reaching no farther than the ring-like end of the coelenteron. This ' mesoderm ' was accordingly described as growing

46 DEAN. [Vol. XI.

"aux d^pens des cellules de I'entoderme, qui est tout au moins le principal facteur dans sa formation, et non pas aux depens du bourrelet marginal." Of the entoderm cells at the margin of the ingrowing coelcnteron the more superficial are in their turn continually becoming mesodermic ; " apres s'etre multipliees, elles prennent des caracteres tr^s semblables a ceux des cellules du bourrelet marginal, et de cette fagon, tant que la cavite digestive primitive s'etend, le mesoderme s'accroit de bas en haut aux depens de I'entoderme." From sections of well preserved material a quite different interpretation of the origin and growth of the mesoblast is to be obtained. In PI. IV, Fig. 6'^, it will be seen that in a very early stage of gastrulation the lip of the blastopore is already divided into its three germ layers ; the outer layer thickens and is deeply inflected at the lip of the blastopore, to a point where it becomes connected with the middle and inner layers : here the inner layer is thinnest, thence it widens, but near the end of the coelenteron, merging with the yolk-laden cells, its boundary is no longer to be traced : it is now indistinguishable from the middle layer which from the lip of the blastopore up to this point has remained distinct. At a later stage, Fig. 64, the inner layer of both dorsal and ventral lips is seen to be continuous with the large entoderm cells of the yolk mass : at the Randwulst the three layers merge ; in the region ectad of the end of the coelenteron the mesoderm passes into the yolk mass. In Figs. 65, 66, the germ layers are seen more widely separated, and their confluence in the Randwulst is more prominent. In the section of the greatly reduced blastopore, Fig. 6^, the middle layer in this region is to be clearly seen separate from the inner layer and confluent with the yolk cell mass. In transverse sections of this stage, F'ig. 70, the presence of gastral mesoderm may be established for a short distance in front of the blastopore (equivalent to about one-sixth of the length of the embryo of PI. Ill, Fig. 51); further forward than this the layer of mesoblast decreases notably in thickness, but always maintains its connection laterally with the yolk mass.

The mode of early development of the middle germ layer of

No. I.]



Acipenser presents a suggestive contrast with that of Lepidosteus (p. 24).

Middle layer appears

Its early character

Its mode of growth


Late : about the time of the closure of the blastopore.

A discoidal cell mass, almost symmetrical, its elements alone in contact with inner and outer germ layers at its center, the blastopore.

At first peristomal ; becomes greatly thinned : at its periphery single-celled, mesenchymatous. Gastral mesoderm early apparent in hinder region of embryo's axis thickened notably near the median plane (in stages of PI. I, Figs. 18, 19), connected with entoderm in the hinder region of the embryo's axis.


Early : shortly after the appearance of the ventral lip of the blastopore.

A ring-like cell mass, notably asymmetrical, confluent with inner and outer layers at the rim of the blastopore, and confluent at its outer (peripheral) margin with the yolk mass.

Notably peristomal, a layer of almost uniform thickness from rim of blastopore to undifferentiated yolk tissue. At stage of PI. IV, Fig. 66, its thickness has become reduced in its peripheral region. Gastral mesoblast differs little in thickness from neighboring peristomal mesoblast: the region of its connection with entoderm is restricted to that of the hinder onesixth (about) of the embryo's axis.

The earlier appearance of the mesoblast in Acipenser seems to the writer to be probably due to the smaller amount of food yolk in this form. The difference, however, in the plane of early mesoblastic growth is hardly to be ascribed to this cause alone : the extremely compressed or rather horizontal mode of growth of the Sturgeon would, however, seem adequate to explain this character. The mesoblast has a normal separate growth till in the region of the end of the coelenteron : a possible mechanical cause may here prevent its separate extension ; and a similar reason would account for the uniform thickness of the layer of gastral mesoblast.

48 DEAN. [Vol. XI,

IV. General Comparison of the Early Stages of Garpike AND Sturgeon, and Conclusions.

In the foregoing paper the similarity in the mode of development of these kindred Ganoids has been noted in some detail from stage to stage. There have thus been compared their rate of development (p. 12), the positions assumed during the growth of the embryo (p. 13), the size, adhesiveness, mode of pigmentation and deposition of the ^-^^ (p. 14), the cleavage (pp. 30-34), gastrulation (p. 39), and the origin of the mesoderm (p. 47). There yet remains to be given a summary of the results of the present writer which shall contrast the general processes of growth, z>., segmentation, the development of the primary germ layers and the establishment of the embryo's external form.

The segmentation of Lepidosteus appears clearly of a more meroblastic character than hitherto described : in none of the writer's material have cleavage planes been observed traversing even superficially the yolk pole of the Q-ZZ- The germ substance is sharply separate from the yolk ; its specific gravity appears notably lighter, and at the animal pole it often projects above the surface curvature of the yolk mass. Cleavage is early expressed in regular planes, giving the blastomeres almost conventional outlines ; it is only after the fourth cleavage that variations are observed.

In Acipenser, on the other hand, segmentation is holoblastic, although the furrows traversing the yolk region are of an exceedingly superficial character.^ The consistency of the substance of the germ is more nearly that of the yolk, and the surface curvature of the early blastomeres corresponds with that of the Q.gg. In general cleavage planes are deepest in the region of the animal pole, becoming more and more superficial as they pass around the Q.g^ to its lowermost point : they agree in general direction with those of Lepidosteus but are irregular in their mode of occurrence from the second plane onward.

1 As noted by Salensky and others.


In pigmentation the eggs of Acipenser present the strongest contrast to those of the Gar-pike.

The stages of late segmentation may thus be contrasted. In Lepidosteus the division of the blastomeres results in a cap of cells of irregular sizes : this appears as a distinct prominence of the egg surface : below it a smooth surfaced periblast-like layer contributes at irregular points to the growth of the blastoderm, — a condition maintaining till the time of gastrulation. The cell cap of Acipenser is far more closely continuous with the yolk ; cells large and irregular in size are readily distinguished in an intermediate zone between blastoderm and yolk : of this zone the number of tiers of cells agrees in a general way with that of the merocytes in corresponding stages of Lepidosteus.

The general differences in segmentation appear due to either a greater or decreased amount of the egg's reserve of yolk material. But, everything considered, it seems to the present writer more probable that the mode of segmentation of Acipenser, although holoblastic, is more readily to be derived from that of Lepidosteus than vice versa. The reasons for this view of the relationship of the Sturgeon are based : (i) on the knowledge of its descent as afforded by palaeontology,^ (2) on the curiously superficial character of its total segmentation, (3) on its marked irregularity of cleavage, — a character by no means conclusive in itself, but important in connection with (i) and (2). If the evidence be ultimately accepted as to the derivation of the Sturgeon and of its mode of segmentation, it is obvious that the results of Beard in his paper on the Interrelationships of the Ichthyopsida would be seriously disarranged.

In the development of the primary germ layers the differences of these forms appear in like manner due to the alteration in quantity of yolk-material. In Lepidosteus the clearly marked surface between cell cap and yolk maintains until the time of gastrulation ; in Acipenser a transitional zone of irregular yolk cells exists from an early stage. In one form the cap of cells is of irregular character, in the other it is reduced to a

1 Cf. esp. A. Smith Woodward, On the Palaeontology of Sturgeons, Pro. Geol. Ass., Vol. IX, Nos. I and 2.

50 DEAN. [Vol. XI.

definite layer. In Gar-pike the segmentation cavity, when definitely formed, undergoes the smaller changes in extent and outline, and its roof remains of more uniform thickness. The alterations which these conditions have been seen (p. 38) to undergo in the Sturgeon seem of more specialized character. Early gastrulation is closely comparable in both forms : that of Lepidosteus, although hampered in development by the presence of a greater amount of yolk material, seems more generalized : in this form for example, as previously noted, p. 22, a two-layered condition maintains until the time of the blastopore's closure, the inner and outer layer of the dorsal lip are essentially identical in structure, and as in the elasmobranchian gastrula there occurs no growth modification of the rim of the blastopore to give rise to a Kupffer's vesicle. It also approaches the type of the Elasmobranch in the character of the superficial cell stratum, which it early develops, but diverges more widely than Acipenser in the matter of neurenteric canal. Of the development of the middle germ layer, p. 46, the conditions that have been noted in Acipenser seem more specialized than those of Lepidosteus ; thus the early peristomal mesoderm in the latter form is independent in its distal margin from either outer and inner layers, and presents a strong contrast to the restricted and abbreviated layer of Sturgeon, The gastral mesoderm, moreover, although similar in both forms in its apparent mode of origin, suffers in Sturgeon a flattened growth which can only be regarded as of specialized character.

The growth processes establishing the outward form of the embryo furnish perhaps the strongest ground of contrast in the early development of Gar-pike and Sturgeon : in the former the form outline of the embryo is constricted off from the yolk mass ; in the latter the embryo, in a mode of growth greatly flattened, continues up to a late stage to surround the yolk, and to preserve outwardly the spherical curvature of the egg.

The greatly flattened growth conditions of the Sturgeon embryo are certainly unique among vertebrates ; and among fishes — using the word in its broadest sense, — they may reasonably be regarded as peculiarly aberrant. The mode of establishment of the early external form of the Gar-pike is not


unlike that of Teleost, and is not strikingly different from that of Elasmobranch, or even of urodele ; and in view of the undoubted kinship of Lepidosteus and Acipenser it seems impossible to regard the peculiar growth of the latter as essentially the more primitive and generalized. What the conditions have been that have caused the flattened form-growth of the Sturgeon is not easily to be established, but the greatly enlarged outline of the embryo's head indicates the stages when they are most prominently marked. The brain region in these stages does not appear, accordingly, particularly suited for the study of conditions which are to be regarded as undoubtedly primitive: secondary characters, e.g., fusions of epiblast with the roof of the fore-brain, or with the entoblast of the fore-gut, appear to occur most confusingly.

In concluding the above summary of the essential characters of the early development of Lepidosteus and Acipenser an especially noteworthy result should be emphasized — that Acipenser in its main development features seems the more specialized type, but that it suggests in essential regards its descent from a form not unlike Lepidosteus. The evidence thus furnished seems to add the needed confirmation to the results of Smith Woodward and Traquair, who have derived the Sturgeons from a Palaeoniscoid stem, and recognized in this a close kinship {e.g., through Dictyopyge and Catopterus) with the scaly Ganoids. The Sturgeons (Smith Woodward) ' descending from this stem have evolved an increased size and have degenerated or specialized in conditions of exoskeleton.'

As to the evidence of the kinships of Ganoids afforded by study of their early development : —

Both A. Agassiz and Balfour and Parker have already emphasized the nearness developmentally of Gar-pike and Teleost, Cf. Ref. 5, p. 430, (1H5).

In addition to the teleostean characters previously noted the present writer would add : —

(i) Similarity in the mode of the first four cleavages.

(2) Kindred relation of merocyte zone of germ disc to periblast, suggesting that in the latter a thinning away of the

52 DEAN. [Vol. XL

merocyte zone has occurred in the middle of the floor of the segmentation cavity.

(3) Comparison of the gastrulas {cf. p. 39) ; the rim of the blastopore becomes the germ ring ; the "ventral mesoderm" of H. V. Wilson, the primitive hypoblast of the blastopore's ventral lip : of both lips the marginal fusion with periblastyolk is in Teleost obliterated, — probably on account of the enlarged size of the yolk, and the more perfect relations of periblast to embryo. The interpretation of the Teleost gastrula of Ziegler becomes accordingly slightly modified : coelenteron extends under the rim of the blastopore from the free end of the "ventral mesoderm" to that of the "primitive hypoblast" of H. V. Wilson.

(4) The presence of Kupffer's vesicle in Acipenser, v. p. 42. But it is especially significant that the nearness of Ganoids

to Elasmobranchs in their adult conditions is also to be emphasized in their early development. In Lepidosteus the shark-like characters in partial segmentation, the formation of merocytes, the mode of origin of the germ layers, have already been noted. Of Acipenser the conditions, not v^^idely different, have retained in addition a neurenteric canal. In this comparison it is now evident that difference in size of egg or in the character of its membranes cannot be adduced as an insurmountable objection. Laemargus has shown that an egg capsule is not infallibly an elasmobranchian feature ; and by an analogy the difficulty in accounting for the increase or decrease in the number of eggs (as urged for example by Beard ^) seems practically solved in the case of Bdellostoma and Petromyzon, forms universally regarded as of close genetic kinship. In Bdellostoma ^ the few and large encapsuled eggs, moreover, are probably extremely meroblastic.

The nearing of the phyla of Ganoids and Elasmobranchs on the evidence of early developmental characters seems worthy of especial consideration. Morphology would long since have established the stem of the sharks as most primitive and ancestral of existing gnathostomcs, had not marked differences been

1 Anat. Anz., 1890, pp. 146-159 and 179-188.

2 Ayers, 1894. Lectures of Woods IIoll Marine Biological Laboratory.


adduced in ontogeny. Where characters of primitive holoblastism were sought, a condition extremely meroblastic was found ; and until the question of the possibility of a gain or loss in food yolk should be decided, the segmenting stages of a Ganoid made its ancestry appear more primitive than that of existing sharks. What were the developmental conditions of Cladoselachid or Pleuracanthid can never be understood, but so closely did many palaeozoic sharks approach existing genera in the smallest details of exoskeleton that their development could with but little probability have been widely different. But it would now seem evident that the Ganoid which retains in the main the skeletal and exoskeletal features of the palaeozoic types, possesses as well characters in development which are suggestively elasmobranchian. And on the other hand a Ganoid (Sturgeon) which has widely diverged from its palaeozoic kindred is now found to represent a type of development which is in the main to be referred to the simpler characters of the more ancient Gar-pike, — but is at the same time more nearly holoblastic. That total cleavage is in this case of secondary nature seems a not illogical conclusion, and should certainly strengthen the position of Rabl in the question of the loss of food yolk.

The study of the early development of Lepidosteus and Acipenser leads to the conclusion that the ancestors of the Ganoids were meroblastic, rather than holoblastic, and that their kinship, as far as our present knowledge can decide, was most nearly with the Elasmobranchs.

Laboratory of the Department of Biology, Columbia College, August 25, 1894.

54 DEAN. [Vol. XI.


The following references relate to the breeding habits and developmental history of Ganoids :

1. 1S78. Agassiz, a. The Development of Lepidosteus. Proc. Am.

Acad. A. and S., xiii, pp. 65-76. 5 PI.

2. 1889. Allis, E. p. Anatomy and Development of the Lateral

Line System in Amia calva. Jour. Morph. and in Proc. Amer. Assoc. (l 878-1 879), pp. 296-298.

3. 1885. Balfour, F. M. (Summary of development of Ganoids.) A

Treatise on Comparative Embryology, ii. Macmillan, London.

4. 1881. Balfour, F. M. and Parker, W. N. On the Structure and

Development of Lepidosteus. Proc. Roy. Soc, xxiii, No. 217, pp. 1 1 2-1 1 9.

5. 1882. Balfour, F. M. and Parker, W. N. On the Structure and

Development of Lepidosteus. Phil. Trans. Roy. Soc. 9 PI.

6. 1889. Beard, J. On the Early Development of Lepidosteus osseus.

Preliminary notice. Proc. Roy. Soc, xlvi, pp. 1 08-1 18.

7. 1893. Dean, Bashford. Note on the Spawning Conditions of

Sturgeon. Zool. Am., No. 4360.

8. 1893. Dean, Bashford. Sturgeon Hatching. Bull. U. S. F. C,

335-339 9. 1882. Dunbar, G. P. (Notes on habits, breeding of Lepidosteus.^

Am. Nat., May, 1882.

10. 1894. Ehrenbaum, E. Beitrage z. Naturges. einiger Elbfische.

Mittheil.a.d. Deutschen Seefischereiverein. Berlin. No. 10, pp. 1-47.

11. 1894. FiJLLEBORN, F. Bericht ii. eine z. Untersuchung d. Entwick.

V. Amia, Lepidosteus u. Necturus unternom. Reise n. NordAmerica. Sitzungsber. Akad. d. Wiss. z. Berlin, xl, 10571070.

12. 1867. Gegenbaur, C. Ueber die Entwickelung der Wirbelsaule

des Lepidosteus. Jen. Zeitschr., iii, pp. 359-414. 3 PI,

13. 1893. Jungersen, H. F. E. Embryonalniere des Stors. Zool. Anz.,

p. 464. And ('94) Embryonalniere v. Amia calva, I. c. No. 751.

14. 1870. Kowalewsky, Owsjannikow u. Wagner. Die Entwickel ungsgeschichte der Store. Vorlduf. Mitth. Bull. Acad. Imp. Sci. St. P^iersbourg, xiv. col. 287-325. Also in Melanges Biol, tiris du Bull. Acad. Imp. Sci. St. Petersbourg, vii, pp. 1 71-183.

15. 1 891. Kupffer, C. v. Mittheilungen zur Entwickelungsgeschichte

des Kopfes bei Acipenser sturio. Sitzungsber. d. Gesell.f. Morph. u. Phys. zu Miinchen (17. Nov. u. i. Dec. 1891), pp. 107-123.


16. 1S93. KuPFFER, C. V. Studien zur vergleichenden Entwickelungs geschichte des Kopfes der Kranioten. i. Heft. Die Entw. d. Kopfes V. Acipenser. Lehmann. Miinchen, 1893.

17. 1890. Mark, E. L. Studies on Lepidosteus. Bull. Mtis. Comp.

Zool., xix, pp. 1-128. (Habits of young fish, Function of Swim-bladder, Egg membranes.)

18. 1882. Parker, W. K. Development of the Skull in Lepidosteus

osseus. Proc. Roy. Soc. 9 PI.

19. 1S82. Parker, W. K. On the Structure and Development of the

Skull in Sturgeons. Proc. Roy. Soc, p. 142.

20. 1887. Peltsam, E. D. Beobachtungen iiber die Segmentation des

Sterleteies. Mittheil. d. kais. Gesell. f. Fr. d. Naturkunde an der Moskaiier Utiiversitdt. Bd. 1, Heft 1. Protocolle der Sitsungsber. d. Zool. Section. Bd. 1, Heft 1. Moscow, 1 886, p. 206. (Russian.)

21. 1889. Ryder, J. A. On the Development of the Common Stur geon (Acipenser sturio). Am. Nat., xxii, pp. 659-660.

22. 1890. Ryder, J. A. The Sturgeons and Sturgeon Industries of the

Eastern Coast of the United States, with Account of Experiments bearing upon Sturgeon Culture. Bull. U. S. Fish. Comm., viii, 1890, pp. 231-281. 22 PI.

23. 1878. Salensky, W. Entwickelungsgeschichte des Sterlets. Vor Iduf. Mittheil. Beitragen zu d. Protocollett d. 84-8g Sitz. Gesell. d. Nat. an d. kais. Univ. in Kasan (1877), p. 34. (Russian.)

24. 1878. Salensky, W. Same title. 5. Die post-embryonale Ent wickelung. Beitr. zu den Protocollen der gj. Sitz. Gesell. d. Nat. an d. kais. Univ. in Kasan (1878), p. 21. (Russian.)

25. 1878. Salensky, W. Zur Embryologie der Ganoiden. I. Befrucht ung u. Furchung des Sterlets-Eies. II. Entwickelungs geschichte des Skelets beim Sterlet. Zool. Afiz., No. 11, pp. 243-245, No. 12, pp. 266-269, and No. 13, pp. 288-291.

26. 1878. Salensky, W. (History of the development of the Sterlet.)

Man. Soc. Nat. Imp. Univ. Kasan, vii. No. 3, pp. 1-226. (Russian.)

27. 1879. Salensky, W. Abstract of Salensky (1877-1879) by May zel in Hofinan and Schwalbe' s Jahresbuch f. i8y8, Bd. vii, pp. 213, 217-225.

28. 1880. Salensky, W. (History of the development of the Sterlet.)

Pt. II. Post-embryonic Stages and Organogeny. Man. Soc. Nat. Imp. Univ. Kasan, x, Pt. ii, pp. 227-545.

29. 1 88 1, Salensky, W. Recherches sur le developpement du Sterlet

(Acipenser ruthenus). Arch, de Biol., ii, pp. 233-278. 4 PI.

56 DEAN.


The Early Development of Lepidosteus.

All figures have been drawn from material fixed in alcoholic picro-sulphuric acid : the egg membranes of the early stages (Figs. 1-7), not separated readily by needles, have been removed after a brief immersion in 20% Javelle water. X about 20.

Fig. I . Egg immediately before the appearance of the first cleavage furrow. ^^ hour after fertilization.

Fig. 2. First cleavage: it is noted as a trench-like furrow, extending no further marginally than the limit of the germ disc, i hour.

Fig. 3. Second cleavage : limited as before to the germ disc, whose margins now merge slightly into the yolk mass. 2 hours.

Fig. 4. Third cleavage : similar in marginal extension to first and second. 3 hours.

Fig. 5. Third cleavage : seen in side view. The margin of the germ disc is indicated.

Fig. 6. Fourth cleavage : similar in its limits to the third cleavage. t,% hours.

Fig. 7. Fifth cleavage. 4jt^ hours.

Fig. 8. Sixth cleavage. Horizontal cell division occurs for the first time: it is to be noted irregularly within the ring of marginal cells. 5^ hours.

Fig. 9. Seventh cleavage. Meridional furrows, which in the preceding stage have attained almost the equatorial region of the egg, appear in this stage greatly reduced. 6% hours.

Fig. 10. Eighth (?) cleavage. In this stage an irregular marginal depression and an elevated mass of cells of the animal pole are to be observed. 8 hours.

Fig. II. Late segmentation, showing marginal groove and flattened cell cap. 20 hours.

Fig. 12. Late segmentation. The thickened cell cap is shown in a stage in which it is coming to overspread the marginal groove. 25 hours.

Fig. 13. Early gastrulation. 32 hours.

Fig. 14. Gastrulation. In front of the blastoderm's marginal indentation the embryonic thickening is to be faintly seen. 38 hours.

Fig. 15. Late gastrulation, showing circular blastopore. 44 hours.

Fig. 16. Late gastrulation, a stage slightly older than that of Fig. 15. An indented blastopore rim is here exceptionally present.

Fig. 17. Early embryo, showing the flattened vascular area in front of the head eminence, and the closed blastopore. 60 hours.

Fig. 18. Side view of the embryo of 60 hours.

Fig. 19. Embryo, showing an early stage of the establishment of its external form. 80 hours.

Fig. 20. Embryo of 90 hours.

Journal of Morphology. I hi A'/

! '








lah. .*<■ i'if-.-r,,;

58 DEAN.


The Early Development of Lepidosteus.

Fig. 21. Vertical section of egg of about i hour after fertilization, showing the first division of the nucleus. X 20.

Fig. 22. Vertical section of stage of 2 blastomeres. X 20.

Fig. 23. Horizontal section through the nuclei of 4-cell stage. X 20.

Fig. 24. Similar section of 8-cell stage. X 25.

Fig. 25. Similar section of i6-cell stage. X 25.

Fig. 26. Vertical section of i6-cell stage. X 20.

Fig. 27. Vertical section of stage of sixth cleavage, showing yolk nuclei, m. X 20.

Fig. 28. Vertical section of stage of 20 hours, showing zone of yolk nuclei. X about 55.

Fig. 29. Vertical section of stage of 25 hours. X about 65.

Fig. 30. Vertical (sagittal) section of stage of 32 hours. X about 45.

Fig. 31. Vertical (sagittal) section of stage of 40 hours. Coelenteron, c ; dorsal lip of blastopore, d ; segmentation cavity, s ; innermost limit of coelenteron, X, X. X 30.

Fig. 32. Vertical (sagittal) section of region of blastopore in stage of 42 hours. Coelenteron, c ; dorsal lip of blastopore, dl; epidermic stratum of outer germ layer, e\ inner germ layer, i; middle, m ; outer, o\ yolk, j. X about 100.

EiG. 33. Transverse section of hinder trunk region of embryo of 80 hours (.= 7 primitive segments). Chorda, c ; epidermic stratum of outer germ layer, e\ middle germ layer, m, inner, i ; neuron, n.

Fig. 34. Sagittal section of embryo of 44 hours. Location of closed blastopore, bp ; epidermic stratum of outer germ layer, e ; inner layer, / ; middle layer, m ; outer layer, o ; here of head region ; yolk, y. X 75.

The material of the above sections was fixed in alcoholic picro-sulphuric acid ; Figs. 21, 22, 23, 35, Delafield's haematoxylin, 24-29, concentrated alcoholic picric and acid fuchsin, the remainder, haemacalcium. Egg membranes when present in preparations have not been represented. With few exceptions the figures are from camera drawings

Jon null of Morphology Vol. JO.













. -.r




.Tl .


60 DEAN.


The Early Development of Acipenser sturio.

Egg membranes of the stage of Fig. 43 and later, were removed with needles before fixation ; those of the earlier stages were separated in dilute Javelle water. X about 23.

Fig. 35. Egg immediately prior to fertilization.

Fig. 36. First cleavage, i hour after fertilization.

Fig. 37. ij^ hours.

Fig. 38. Third cleavage, of normal type. 2^ hours.

Fig. 39. Third cleavage, of meridional type, seen in side view. The second furrow is to be observed traversing the yolk half of the egg.

Fig. 40. Fourth cleavage, of extremely regular type. 3^ hours.

Fig. 41. Sixth cleavage. 4 hours.

Fig. 42. Sixth cleavage. Lower pole of egg.

Fig. 43. Late segmentation. 6 hours.

Fig. 44. View of animal pole of stage of 6 hours.

Fig. 45. Blastula. View of animal pole, showing through its transparent roof the outline of the segmentation cavity. 8^ hours.

Fig. 46. Blastula in latest stage. 16 hours.

Fig. 47. Early gastrula. The dorsal lip of the blastopore is seen at the left of the figure. 19^ hours.

Fig. 48. Gastrula, in side view, dorsal lip at the left. 26 hours.

Fig. 49. Gastrula, showing faintly marked indentation of the dorsal lip. 28_^ hours.

Fig. 50. Late gastrula, indicating obscurely the outline of the embryo. 32 hours.

Fig. 51. Embryo of 43 hours. Eight primitive segments are present but are not distinguishable in surface view. In this stage are to be noted : the marked lateral expansion of the head region, the early neurenteric canal, the mode of closure of the blastopore — caused apparently by the more rapid growth of the ventral lip.

Fig. 52. Embryo of 46 hours: view of head and anterior trunk region. As yet the primitive segments — of which about 14 are now present — are not noticeable in surface view. The brain and optic vesicles are shown in an undifferentiated condition, their outlines prominently marked on account of pigmentation. The dark parietal zone of the embryo indicates the limit of the fore-'gut. The region of the pronephric tubules is obscure.

Fig. 53. Embryo of 46 hours : view of tail region, showing the location of the neurenteric canal and of the almost closed blastopore.

Fig. 54. Embryo of 48 hours : view of tail region, showing the late appearance at the surface of the neurenteric canal. About 20 primitive segments are here present.

Joiiriuil of Morpholofft). Vol XI.

I'l III.














ishf&rd Dear del



The Early Development of Acipenser stiirio.

Fig. 55. Vertical section of egg immediately before first cleavage. X 22.

Fig. 56. Horizontal section (slightly oblique) of 4-cell stage, through the region of the nuclei. X 22.

Fig. 57. Horizontal section of S-cell stage, through the region of the nuclei. An irregular cleavage is to be seen in the left side of the figure : the first cleavage plane of this and of the preceding figure passes right and left. X 25.

Fig. 58. Vertical section of 5-hour stage, showing cell outlines in the floor of the segmentation cavity. X 25.

Fig. 59. Vertical section of a stage of nearly the same age as that of Fig. 58, showing in greater detail the relation of the yolk cells to the mode of growth of the blastoderm. X about 65.

Fig. 60. Vertical section of 6-hour stage. X 25.

Fig. 61. Vertical (sagittal) section of stage, immediately prior to gastrulation. The posterior region of the embryo is at the right of the figure. X 25.

Fig. 62. Sagittal section of early gastrula (17 hours). The region of the dorsal lip is notably pigmented, and the outlines of the yolk cells in its vicinity may now be outlined. X 25.

Fig. 63. Sagittal section of gastrula of 26 hours, in which the ventral lip of the blastopore appears as but an indentation at the right side of the figure. The middle, inner, and outer germ layers are seen to be present in the dorsal lip. X 25.

Fig. 64. Vertical section near sagittal plane of gastrula of 29 hours. The inner, middle, and outer germ layers of the dorsal lip are lettered, /, m, o ; the thinness of the inner layer is here accounted for in its proximity to the chorda ; segmentation cavity is lettered s\ coelenteron, c ; and the vesicle of Kupffer, /■.

X 25.

Fig. 65. Section near the sagittal plane of region of the blastopore in stage of 32 hours. The structures of the dorsal lip are lettered as in preceding figure, and are seen readily comparable with those of the ventral lip. X about 65.

Fig. 66. Vertical (nearly sagittal) section of embryo of 43 hours. The thickening of the outer layer in the head region is lettered k, and although the yolk plug is now greatly constricted, the structures of dorsal and ventral lips and the enlarged coelenteron may readily be made out by comparison with Fig. 65. As yet the Deckschicht is not clearly defined and has been outlined in the figure.

Fig. 67. Transverse (slightly oblique) section through the region of the neurenteric canal of an embryo somewhat older than that of Fig. 66. Germ layers are indicated as before : neurenteric canal, n ; yolk cells, y ; yolk plug, yp, now disappearing; on either side of its position is the undifferentiated cell mass.

X4S Fig. 68. Section almost sagittal of neurenteric canal of embryo of 46 hours. It is now seen definitely established, n ; the site of the yolk plug is indicated as a slight diverticulum of its hinder wall, jj//. X about 100.

Fig. 69. Sagittal section of neurenteric canal of embryo of 58 hours. It now opens widely into the gut, ;/, and its cellular wall is well established ; the caudal

62 DEAN.

mass is seen at the left; at the right the notochord has separated from the inner layer ; the epidermic stratum of the outer layer is now clearly marked, e. X about 100.

Fig. 70. Transverse section of the hinder trunk region of an embryo of 43 hours, showing the relation of the inner layer to the chorda and mesoblast.

Fig. 71. Sagittal section of anterior end of neuron of embryo of 11 primitive segments, slightly older than that of Fig. 66, showing the point of connection of the anterior end of the embryonic brain with the formative epiblast, *. The germ layers are lettered as before ; the Deckschicht of the outer layer, d^ is now clearly differentiated.

Fig. 72. Transverse section through the point * of the preceding figure. Lettering as before.

Fig. "ji. Transverse section of region slightly posterior to that of Fig. 72.

Fig. 74. Sagittal section of anterior end of neuron of embryo of 46 hours. The lettering corresponds with that of Fig. 71. Between the limits x-x occur points of union between brain wall and formative epiblast, o ; the anterior third of this region is broadly fused with the outer layer. Fusion also occurs in front of the embryonic brain between the formative epiblast and the cells of the inner layer, /.

Figures, with exception of Figs, i, 2, 3, 46-48, are from camera drawings. Material was fixed either in picro-sulphuric or in corrosive glacial acetic. Haemacalcium and Delafield's haematoxylin were the staining agents.

.JoiwnaJ, of Morphol ogy Vo l.Xl.






«tt tnstvVimerit/mUi: Frankfarf'-il





Part I. The Phenomena of Impregnation and the Formation of

THE Egg-membranes in Jaera 66

1. The process of impregnation in Jaera 66

2. The formation of the egg-menih-anes in Jaera 68

Part II. The Segmentation and Formation of the Germ-layers 73

1. The segmentation and formation of the gei'm-layers in Jaera.. 73

2. The segmeiitation and formation of the germ-layers in Asellns

comtmcnis 88

3. The segmentation and formation of the gerjti-layers in Porcellio

and Armadillidium 95

4. General consideration of the segmentation 107

Part III. The Later Development of the Germ-Layers 115

1. The later history of the Mesoderm 115

2. The formation of the digestive tract and the later history of the

vitcllophags 124

3. General considerations on the formation of the germ-layers in

the Crustacea 128

Part IV. Notes on the Development of Certain Organs 137

The observations recorded in the following pages were commenced in the summer of i8go at the Marine Biological Laboratory, Woods Holl, Mass. The object in view was to follow out for the Crustacea the cytogenetic development after the manner which has yielded such important results in the case of Clcpsinc (Whitman) and Nereis (Wilson), and after examining some early stages of several different species of Decapods, Isopods, and Amphipods, I finally chose for my purpose the ova of Jaera marina (Fabr.) Mobius, which seemed to present several advantages, notably in the early differentiation of the germ-layers. In addition, at that time the embryology of the Lsopods had received, on the whole, less attention than it was entitled to, and considerable differences seemed to exist in the modes of segmentation of the forms which had come under ob


servation. A preliminary notice of my observations upon this form was published in the Zoologischer Anseiger in the summer of 1 892, and seemed of sufficient interest to lead me to extend my studies to Ascllus communis, Say, and later to Porcellio scabcr and Armadiliidium vnlgare, the greater portion of the material being collected at Woods Holl. To my friend, Dr. W. M. Wheeler, I am indebted for a considerable quantity of material, which came originally, I believe, from Naples, and illustrates the development of CymotJioa, and to Dr. F. H. Herrick for material from Ligia, collected at Beaufort, N. C. In neither of these cases, however, were the earlier stages of segmentation represented, and I could make use of them only for the later stages of development.

In the present paper I do not propose to enter in detail into the organogeny of the group, my observations on this department of the subject not yet being sufficiently advanced. I shall have occasion, however, to refer to certain facts which have been made out concerning the development of some of the organs, and consequently shall divide the paper into four portions, the first of which will treat of certain phenomena connected with the impregnation and formation of the egg-membranes of Jaera ; the second, of the segmentation and the development of the germ-layers ; the third, of the later development of the mesoderm and endoderm ; while in the fourth, some scattered notes concerning the development of certain of the organs will be discussed,

A word is necessary with regard to the identification of Jaera marina. In my preliminary notice ('92) I referred to it z& Jaera albifrons, on the authority of Harger ('78). In a more recent paper by Sye ('87) it is stated that the form described by Fabricius in his Fauna Groenlandica as Oniscus mariniis is identical with that later described by Montagu as Oniscus albifrons. In 181 5 Leach united these two forms in the genus Jaera, and adopted for the species the name of /. albifrons, which is the authority for the name employed by Harger. If, however, the two species are identical, then Fabricius's specific name has the priority, and the entire name should be properly that employed by Mobius in 1873, Jaera marina. As regards


the identification of the Porcellio and Armadillidiiini studied, I can only say that they are the common forms of these genera occurring in New England, and lacking the literature necessary for a definite identification, I cannot state positively that they are the species named above.

The methods employed were comparatively simple. After experimenting with a number of fixing reagents, such as corrosive sublimate, hot water, picro-sulphuric acid, and others, I finally adopted an alcoholic picro-sulphuric acid, which gave admirable results, producing no distortion of the ova, and preserving the cells in an almost perfect manner. Picric acid is dissolved in 70% alcohol until saturation, and two volumes of sulphuric acid are added to every hundred volumes of this solution. The reagent is, indeed, simply Kleinenberg's strong solution made with 70% alcohol instead of with water. For staining the early stages I employed Kleinenberg's haematoxylin, generally deeply overstaining, and then carefully washing out in 70% alcohol, acidulated with hydrochloric acid. The stain is washed out more rapidly from the yolk than from the protoplasm and nucleus, which thus became very distinct. In the early stages the eggs thus treated could be cleared in oil of cloves and thus studied, but I found that in later stages, after the cells had reached the surface of the yolk, it was preferable to study them as opaque objects, using direct illumination. When the blastoderm was well formed it was possible, in the larger ova, such as those of Poirellio, Armadilliduun, Ligia, and CymotJioa, to remove it from the yolk and study it as a transparent object after clearing, but in the minute eggs of Jaera, and to a certain extent in Asellus, this method of procedure was not feasible. For sectioning the ova of Jaera, and the later stages of the other forms, the usual paraffin method was employed. In the early stages, before the formation of the blastoderm, the brittleness which the yolk assumed under this method prevented its application to the larger eggs, and for these, after the removal of the egg-membranes, the celloidin method was employed.


Part I. — The Phenomena of Impregnation and the Formation of the Egg-membranes in Jaera.

I. The process of Impregnation injaei^a.

On examining a series of transverse sections through an adult female Jaera my attention was arrested by a peculiar chitinous body, circular in section and composed of a series of concentric layers (PI. V, Fig. i, Sp), which occurred on each side of the mid-line, just below the tergum of the fifth thoracic segment, — a short canal, at the bottom of which it lay, running forwards and upwards to open between the fourth and fifth segments. On sectioning other individuals similar structures were found, varying, however, considerably in position, sometimes having a more lateral and sometimes a more median position, sometimes near the dorsum of the body, or at other times seated more deeply and more posteriorly, but always surrounded by a cellular investment. The impression which one obtains from the examination of the relations of these structures in a number of specimens is that they are bodies which are attached to the dorsum of the female in the region indicated above, and which gradually migrate downwards and backwards into the body. In other words, they appear to be intrusive bodies, and do not, strictly speaking, belong to the category of female organs.

Sye ('87) mentions the occurrence of these structures, but has not followed their relations sufficiently to arrive at any conclusion as to their significance. He states, however, that they do not occur in the male, and were found in the female only during the breeding season, not occurring either before or after that period.

The investigation of this structure does not strictly fall within the scope of the present study, but incidentally some facts regarding it have been discovered which throw some light upon its significance, and may be recorded here in the hope that the attention of other observers will be directed towards a fuller investigation of the subject.


I believe that the chitinous capsule is a portion of a spermatophore which is deposited upon or inserted into the dorsal surface of the female, passing downwards and backwards to reach the oviduct just where it joins the ovary. I have not investigated the structure of the spermatophore in detail, nor have I observed the manner of its formation by the male, but believe that the evidence I have gathered warrants the interpretation here given. In one set of sections the capsule lay just beneath the hypodermis in the fifth thoracic segment, and extending backwards from it was a tube formed of a single layer of cells, and containing a granular mass which did not take a carmine stain. This I take to be the body of the spermatophore undergoing degeneration, the spermatozoa which it originally contained being found in the oviduct and especially in a dilated portion of it near its proximal end, which may be termed the receptaculum seminis. Fertilization takes place apparently in the oviduct ; at all events I have not found spermatozoa in the ovary, though it is possible v S-^y.'--- ! ?' ^^ that they may pass into it during the moult which ov rs od

precedes OVipOsition, a S Relationships of the Various Parts.

they are stated to do in

Porcellio and other land-forms by Friedrich ('83). The relationships of the various parts concerned are represented in the diagram above.

The varying position of the chitinous capsules is to be explained as stages in their expulsion from the body after the spermatophores have discharged their contents. They appear to migrate down the canals originally occupied by the bodies of the spermatophores, these latter seeming to degenerate, and later they reach the oviducts, passing to the exterior finally by these. At least I have found them in sections close to the proximal ends of the oviducts, and also lying in the oviducts just internal to their external opening. I could not find any trace of the bodies of the spermatophores when the capsules had reached the oviducts, and, furthermore, the canals extend


ing from the dorsum to the oviducts along which they passed had also disappeared in this stage, a fact which suggests the origin of the canals as an in-pushing of the hypodermis in front of the spermatophore.

If this interpretation of the observed phenomena be correct, we have an example among the Crustacea of a phenomenon which resembles not a little that termed by Whitman ('91) hypodermic impregnation, although the conveyance of the sperm to the oviduct is somewhat more direct than in the typical cases he records. Examples of the phenomenon have been found in various groups of Invertebrata, as will be seen from Whitman's paper, but, so far as I know, nothing similar to what occurs vsijaera has been recorded for other Crustacea. Spermatophores occur in many forms, such as the Copepods, Thysanopodous Schizopods, and some Decapods, and furthermore, copulation, during which the male rests on the dorsum of the female, is very frequent, but I know of no previously recorded case where the spermatophores are inserted upon the dorsal surface of the body.

2. The Formation of the Egg-Memhranes in Jaera.

The newly-extruded tgg of Jaera is of a bright grass-green color, similar to that which Rathke ('37) has described and figured for Bopyrus and Janira, and is somewhat irregular in shape, later, however, becoming oval and measuring on the average about 0.2 mm. in length by 0.19 mm. in breadth. In the center of the o^g^ is the nucleus enclosed within a stellate mass of protoplasm, a thin peripheral layer of this same material enclosing the yolk spherules, which are mainly albuminous with a few oil globules. Sections of about half-grown ovaria-n eggs (PI. V, Fig. 2) show that a delicate network of protoplasmic fibrils {pn) extends from the central {cp) to the peripheral protoplasm, the yolk granules lying in the meshes of the network in the manner described by P. Mayer ('77) for Eupagnriis; in later stages and in the mature ^gg the network cannot readily be made out, being obscured by the great development of the yolk, but the peripheral protoplasm (Fig. 3, //) can readily be seen during the early stages of segmentation.


The number of ova contained in the brood-pouch varies considerably in different individuals ; the smallest number found was three and the greatest twenty-two, the average being about eleven. A moult apparently precedes the oviposition, as has been observed for a number of other species, but I have not endeavored to follow in detail the phenomena accompanying the extrusion of the ova. An attempt to discover the conditions which governed this process was without result; apparently it is not simply a question of time of day, i.e., of light and temperature, nor can the stimulus be supplied by impregnation, since in a number of females isolated at the same time, from which later developing ova were obtained, and which must therefore have been impregnated prior to the isolation, the time of extrusion of the ova varied too much to allow of its reference to this cause.

Attention may be called to the fact that the period of egglaying of Xh^ Jaera marina of our coast differs materially from that of the European form. According to Sye ('87) the extrusion of the eggs occurs in the latter in March and April, while at Woods HoU I found that it took place throughout the entire summer. At least it was going on at the middle of June when my first observations were made, and continued without interruption until the end of the first week of September, the time of my departure from the laboratory.

Eggs taken from the brood-pouch before the polar globules have formed possess but a single membrane, the chorion, which shows a decidedly wrinkled appearance, and is separated from the ovum by a considerable space. The chorion is somewhat sticky at this stage, the eggs adhering to each other and to the bottom of the watch-glasses in which they are being examined, but later this property is entirely lost. As to the origin of the membrane, my results are not very definite, since I have not been able to satisfy myself as to whether a delicate membrane was not present in ovarian eggs. In one set of sections through ovaries containing almost ripe ova I thought a very delicate membrane could in some places be discerned, and though the uncertainty was too great to warrant a definite statement, I am inclined to believe that the


chorion is a product of the follicle-cells, and is formed within the ovary.

The ova of Aselhis, Porcellio, and Armadillidium agree perfectly with those oijaera as regards the presence of membranes, the chorion being the only envelope before the extrusion of the polar globules.

Soon after the ova oijaera reach the brood-pouch, the polar globules are given off, fertilization apparently taking place either in the ovary immediately before the ova leave it or else in the oviduct. How the spermatozoa pass through the chorion I could not discover, there being apparently no micropyle. Two polar globules are extruded (Fig. 4, pg), and one or both may subsequently divide, since not unfrequently three globules were visible, two being smaller than the third, and in some cases I observed four.

During the period occupied by the maturation of the ovum as indicated by the formation of the polar globules a second membrane, the vitelline membrane (Fig. 4, ym) is formed, and its relations to the polar globules possess considerable interest. In the majority of cases in which the relations were observed both polar globules were between the chorion and the vitelline membrane (Fig. 4), but not infrequently both were within the latter membrane ; and in some cases one was between the two membranes and the other between the vitelline membrane and the ovum.

It seems to be a very general rule that the ova of Crustacea when extruded are enclosed within a single membrane, the chorion, which is almost certainly formed as a secretion by the follicular cells, though certain authors have referred its formation to the walls of the oviduct. A few cases occur in the literature to which I have access which form an exception to this rule. Grobben ('79) states that the egg of Moina possesses no membrane when extruded, but that some time after it reaches the brood cavity one is formed, which, since it must, under the conditions, be a product of the &%g protoplasm, is to be regarded as the vitelline membrane. The same author ('si) has described a similar absence of a chorion in the ova of Cctochilus, and Delia Valle ('89) finds the same conditions in


those of Garmnariis ptilcx. In both these cases, as in Moina^ a membrane forms later which must be identified as the vitelline membrane.

In regard to this latter membrane the data as to its occurrence are not quite so thorough as could be desired, partlyowing to a membrane formed at a much later period, the blastoderm membrane, having been confounded with it in some cases. It seems necessary to distinguish between the two structures, the vitelline membrane being, in the strict use of the term, a membrane which is formed by the protoplasm of the Q.%^ about the time of the formation of the polar globules, and connected, as will be shown later on, with the process of fertilization. It is not my intention to enumerate the various instances in which the occurrence of this membrane has been described, but a few cases may be referred to. In Asellus van Beneden ('69) observed the membrane only when the process of segmentation had advanced to the eight-celled stage, but I have seen it in the egg of A. coinni2inis immediately after the formation of the polar globules. In Onisais Bobretzky ('74) found two membranes in the youngest ova which he obtained, these membranes being apparently the chorion and the vitelline, but on the other hand Bullar ('78) finds but a single membrane in the newly extruded ova of Cymothoa, a second one appearing only when the embryo is fairly formed, there being apparently no true vitelline membrane, though it is possible that it may have been overlooked, since Bullar did not have an opportunity of examining ova in the early stages of segmentation.

In the Decapod Crustacea there are more definite statements as to the absence of this membrane. Ishikawa (85) found two membranes in AtyepJiha, and Lebedinski ('90) has described the same number as occurring in EripJiya, and it may be presumed that in both these cases we have to do with a chorion and a vitelline membrane. In the Lobster, however, Bumpus ('9i) found no trace of a vitelline membrane, and the same result followed the researches of Kingsley ('8?) on Crangon, Cano ('93) on Maja, and apparently those of Mayer ('77) on Eiipagunis and Herrick ('92) on A/pheiis, to mention only some of the more recent observations.


What all the conditions may be which determine the formation of the vitelline membrane is at present obscure, but the primary condition is understood more especially through the observations of Fol and the Hertwigs ('87) to be normally a stimulus imparted to the egg protoplasm by the spermatozoon. Herbst {'93) has shown that it is possible to imitate this natural stimulus and call forth the membrane in unfertilized eggs by subjecting them to the action of various chemical substances such as Benzol, Toluol, Xylol, and others ; and the Hertwigs found that exposure to the action of certain poisons such as cocain and chinium sulphate deprived the protoplasm to a greater or less extent of the power of responding to the normal stimulus. The membrane appears to be the outer layer of the protoplasm of the ovum, hardened or thrown off as a result of a stimulation of the protoplasm. Such a cause for its formation explains its relations to the polar globules of Jaera. It is entirely unrelated to their formation, as their varying position within or without the vitelline membrane demonstrates, though it is possible that their formation, that is to say, nuclear division, may also be more or less influenced by the penetration of the spermatozoon. Why the membrane does not form in Honiarus and other forms it is difficult to say ; it can hardly be a case of inhibition of the stimulus due to the pressure of a considerable amount of yolk, since the ova of Crangon have a less abundant yolk than those of Porcellio, yet the former do not develop a vitelline membrane while the latter do. Still it is possible that the yolk may have some effect, since it seemed that in Porcellio the formation of the membrane was in some cases very much belated, not appearing until after segmentation had begun ; opportunities for making conclusive observations on Porcellio were, however, not afforded me.


Part II. — The Segmentation and Formation of the Germ-Layers in the Isopods.

I. The Segmoitation and For7nation of the Genn-Ldyers in


Shortly after the extrusion of the second polar globule the first segmentation occurs, its plane passing in the usual manner through the point at which the polar globules were extruded, and lying, therefore, at right angles to the longer axis of the egg. The division affects, however, only the nucleus and the protoplasm which immediately surrounds it, there being no indication of the division upon the surface of the ovum, a state of affairs which persists through several divisions. For the sake of convenience in description, however, the two nuclei with their surrounding protoplasm (PI. V, Fig. 5, A, C) will be spoken of as cells, it being understood that the ovum at this stage and throughout several subsequent stages is in reality a syncytium.

The second division results in the formation of four cells and is likewise confined to the nuclei and the protoplasm in their immediate vicinity. The axes of both spindles are evidently at right angles to the direction which the spindle of the first division held, but after the division is completed the lines which may be imagined as joining the four cells into pairs are not in any case observed by me parallel to each other. Thus, in some cases when two of the cells, derived from the same parent cell, were in clear focus, the other two were also visible, one being in almost equally clear focus while the other was somewhat obscure, and the lines joining the members of each pair were evidently inclined to one another at an angle which varied considerably in different cases. In other eggs, however, when two of the cells were in clear focus only one other could be perceived, the fourth being directly below it and only brought into view by rolling the egg through an angle of about 90°; in other words in these cases the lines joining the cells in pairs are exactly at right angles with one another. These two conditions are shown in Figs. 6 and 7 of PL V, the two


cells A and B being the products of the original cell A, and C and D the result of the division of the original C of Fig. 5. It would be interesting to know definitely whether these two arrangements represent two modes of segmentation ; or whether the cases in which the inclination of the axes of the two pairs to each other is comparatively slight, are simply stages in rotation through an angle of 45° of each of the two pairs of cells formed by a typical meridional division. I did not succeed in obtaining ova in the spindle stage preceding this division, and therefore cannot answer this question positively, but from the circumstantial evidence at my disposal I am inclined to believe that there is a rotation, but also that in a number of ova it remains incomplete.

In the cases in which the rotation has reached its fullest extent some interesting features are brought to light when a comparison is made with the corresponding stages of other ova. The existence of a cross-furrow has recently attracted considerable attention in connection with its significance as indicating a " spiral " form of cleavage. The rotation of the ^g% oijaera may be regarded as an extreme form of "spiral" cleavage, and it becomes of interest to note the relationships of the furrows which separate the various cells. In this particular egg no crossfurrows exist in reality on account of the nature of the segmentation, but it is not difficult to imagine what their relations w^ould be were the cleavage holoblastic. Practically the same condition would obtain which has been described by Ludwig ('82) for the ovum of Asterina in the four-celled stage, and this condition may be regarded as due to the rotation of each pair of cells through an angle of 45°, the arrangement thus being essentially similar to that described by Wilson ("92, p. 452) for Nereis. In the figures of Jaera the conditions appear to be slightly different, but this is due to the slightly different position in which the ^gg is drawn, a position which may be imitated by rotating the Nereis ^^'g through 45°, when A and B will lie in one plane and C, D, in another plane at right angles to the former.

In Jaera, then, we have to deal with a rotation of each of the two pairs of cells through an angle of 45°, and the arrangement


is strictly comparable to that of Nereis. That this is so is evidenced by the relation of the antero-posterior axis of the future embryo to the first segmentation plane, which is the same as in Nereis, a fact which can, however, be more readily perceived when the next segmentation has been completed.

In Fig. 8 is shown a preparation in which the division into eight cells is not quite completed, the cells being still united in pairs by a short band of protoplasm. It will be seen that A and B have divided in such a way that their spindles must have been parallel, one of the daughter-cells of each lying immediately below the other, and being thus concealed from view. This division is practically an equatorial one. The spindle of cell Cy however, assumed a position at right angles to those of A and B, and thus underwent what may be considered a third meridional division, while the spindle of D was inclined at an angle of 45° to the other three. As the result we get the arrangement which is represented in Fig. 9, taken from an ^^^ which was rotated slightly so as to bring all four cells resulting from the division of A and B into view. It will be seen from this that the cells have arranged themselves around the long: axis of the &gg, an axis which represents the longitudinal axis of the future embryo, and which stands at right angles to the axis indicated by the polar globules. Around one of the new poles of the ^%Zi which the later development shows to correspond to the anterior extremity of the embryo, four cells. A, a', B, b\ are arranged at almost equal distances, and near the posterior pole is a second circle of three cells, C, c\ and d\ also almost equally spaced, while the eighth cell, D, occupies an almost polar position, lying, however, a little to one side of the actual pole. This cell is destined to give rise to the vitellophags, by which term its descendants may hereafter be denoted.

We have seen that the first division-plane was parallel to the shorter axis of the ^g%, at the extremity of which were the polar globules, while the second plane was parallel to the larger axis, and we now see that the longitudinal axis of the future embryo corresponds with that of the second division-plane, an arrangement which agrees with what occurs in Nereis, Crepiditla,


and Umbrella. It has been shown, however, by the observations of Miss Clapp ('91) that too much importance has been attached to the question of the axial relations of the ovum and the embryo, and further consideration of this point may be omitted. But not only is the future longitudinal axis marked out in the eight-celled stage, but the dorso-ventral axis is also clearly indicated by the position of the vitellophag cell, that side of the &gg upon which it lies being the future ventral surface of the embryo. Unfortunately I have not been able to ascertain what relation this surface bears to the point of extrusion of the polar globules, since these structures cannot be discovered in the preserved ova upon which I worked ; it would be interesting to know if in this particular, also, Jaera resembles the forms mentioned above.

The relations of the cells to the yolk and the peripheral protoplasm in the eight-celled stage are shown in the section represented in Fig. lo. The peripheral protoplasm (//) is still to be seen as a distinct layer covering the surface of the yolk mass, and at a short distance below it, imbedded in the yolk, are masses of protoplasm surrounding the nuclei. During the segmentation these structures have passed peripherad, and in the eight-celled stage the protoplasmic masses surrounding them have almost reached the surface, their connection with the peripheral protoplasm by means of slender processes being very distinct. Other processes radiate into the yolk, and though actual anastomoses of the processes from different masses were not observed, the presumption is very strong that they exist. The central portion of the yolk presents a very different appearance from that which encloses the cells, having broken up into small masses, from which apparently all protoplasmic matter has been withdrawn, there being \w Jaera, as in many other Crustacea {e.g., Moina), a central mass of yolk entirely destitute of protoplasm.

In the next stage a division of all the cells again takes place, and a sixteen-celled stage is the result. The greatest interest attaches to the cells situated at the posterior pole of the egg, and to these alone our attention need be directed. In a number of eggs in which the segmentation had just been completed.


or in which traces of the spindle were still visible, I found at the posterior pole the arrangement which is represented in Fig. II. The two cells vi are the products of the division of D of the preceding stage, and represent the vitellophags ; they are surrounded by a circle of seven cells, six of which result from the division of C, c', and d' of the eight-celled stage, while the seventh is derived from one of the cells of the anterior pole, probably A. In a number of eggs, however, I found the number of cells forming the ring to be six only (Fig. 12), a number which agrees with what is found in the next stage. It seems that shortly after the completion of the division there is a migration or rearrangement of certain cells, one of those which at first formed the ring leaving it and migrating towards the anterior pole of the egg. My reason for supposing this to be the case is that I have found the seven-celled ring in ova in which the division was not quite completed, while the six-celled ring was only observed in fully divided eggs, and furthermore it will be seen that the next stage of development can be derived directly only from a six-celled ring. The cell which leaves the ring is not that {A') which came into it during the division from the anterior-pole, but so far as I can discover it is one of the cells produced by the division of c {i.e., c^ of Fig. 1 1). From Fig. 12 it might be supposed that it was C of Fig. 1 1 that had migrated out of the ring, but this appearance is, I believe, due to an alteration of the position of C after c^ has moved away. From the observation of a number of eggs, and noting the position of the cell which persists in the ring w^ith relation to the two endoderm cells, I have come to the conclusion that c"^ is the one which disappears.

What the significance of this migration may be I cannot even suggest. Why a cell belonging to the anterior hemisphere should enter the posterior one, and vice versa, I cannot understand. The result of the process is, however, a differentiation of the germ-layers at this early stage. At or in the near vicinity of the posterior pole are the two vitellophag cells ; the six cells surrounding them will give rise to the mesoderm and the liver endoderm ; while the eight remaining cells occupying the anterior hemisphere of the ovum will form the ectoderm.


In the 32-celled stage a remarkable difference of appearance supervenes, depending upon the cells having at length reached the surface of the yolk and fused with the peripheral layer (see Fig. 20), thus allowing a superficial indication of the segmentation to appear. The eight ectoderm cells of the i6-celled stage have given rise to sixteen cells (Fig. 13, Ec), which have a characteristic appearance. Each has a more or less perfect hexagonal outline in surface view, and in the center of the area enclosed by cell-boundaries is the mass of protoplasm surrounding the nucleus, which I have spoken of hitherto as the cell. This mass of protoplasm is not, however, sufficiently large to cover the entire surface of the cell, but sends off numerous radiating processes which come into contact with those of other cells, so that, even at this stage, when the true cellboundaries are distinguishable, the ovum is a syncytium, a fact of which sections give ample evidence. The six dark mes-endoderm cells of the i6-celled stage divide in such a manner as to form a circle of twelve cells (Fig. 13, MEn), placed somewhat obliquely to the antero-posterior axis, while the posterior pole is occupied by four vitellophag cells (^z), which are not, however, arranged exactly around the posterior pole, but incline slightly towards the ventral surface. A marked difference in capacity for staining also becomes evident in the cells composing the egg at this stage. The twelve mes-endoderm cells, as the cells marked MEn in the figures may be termed, are distinguished by the very deep tint which their protoplasm shows, as well as by the apparent absence of processes radiating from it, while the vitellophag cells show, on surface views, hardly any protoplasm at all, the darkly-stained nuclei standing out prominently upon an almost unstained ground. The ectoderm cells are intermediate between these two extremes, their protoplasm being distinctly visible, but assuming a much fainter stain than the mes-endoderm. This peculiarity is in part due to the different concentration of the protoplasm around the nuclei in the three kinds of cells. As sections (Fig. 20) show, the nuclei of the mes-endoderm cells [MEn) are surrounded by a large quantity of protoplasm, from which slender processes pass outwards to unite with a thin layer of protoplasm which


forms the cell-boundary and from which other delicate processes extend into the yolk, probably uniting with processes of other cells. The ectoderm cells {Ec) have much less protoplasm around their nuclei, and the protoplasmic processes are, as a rule, longer and stouter than those of the mes-endoderm while a layer of protoplasm marking the cell-boundaries is not so distinct, though indications of it may be seen. The vitellophag cells, on the other hand (vi), seem to consist simply of nuclei imbedded in the peripheral protoplasm, and in sections I could distinguish no processes in connection with them. The segmentation, though appearing from the surface to be total, is in reality not at all so, the greater portion of the yolk taking no part in it, and indeed being apparently destitute of protoplasm.

In the next stage (Fig. 14) the vitellophag cells iyi) have increased to eight, the mes-endoderm {MEn) is composed of two circles, each consisting of twelve cells, and the number of ectoderm cells has also been doubled, so that a 64-celled stage is reached. This stage does not call for any more detailed description, except that it may be pointed out that the various cells still retain the staining peculiarities which distinguished them in the preceding stage.

The 64-celled stage is the last one in which the division involves every cell, that is to say, in which the increase of cells follows a geometrical progression. In the next stage the vitellophag cells do not take part in the division, and, in fact, do not divide again for some time. The next stage, consequently, is composed of 120 cells, a view of the posterior pole of an Qgg passing into it being given in Fig. 15. It will be there seen that the cells of the two mes-endoderm circles are dividing in such a manner that each circle will eventually be formed of twenty-four cells. It will be observed, however, that one of the cells of the posterior circle is dividing in a somewhat different plane from all the others, so that one of the daughter cells {l.en) will encroach upon the area occupied by the vitellophags. The significance of this cell is a little doubtful, since I have not been able to follow the fate of its products. I believe, however, that it is destined to give rise to the endoderm which will form the so-called liver lobes, and


consequently have indicated it in the figure as the liver endoderm il.en). My reason for this belief is derived from the analogy which may be traced between this cell and the cells which give rise to the liver in Astaciis, in which form the origin of the liver has been clearly traced by Reichenbach ('86). After the invaginated cells have begun to assume their vitellophag function, there is to be seen in the ventral wall of the invaginated sack a layer of columnar cells which do not ingest the yolk-granules, and which form what Reichenbach terms the entoderm plate. From these cells the liver lobes arise. If, as I think must be done, the vitellophag cells of Jaera be compared to the invaginated cells of Astacus, which form the secondary yolk-pyramids, then the relations of the liver endoderm of Jaera correspond very closely with those of the endoderm plate of Astacus, a point which may be more clearly seen in the succeeding stages {e.g., Fig. 17). \xv Astac7is \.\iQ vitellophag cells and the cells of the entoderm plate form a continuous layer, but in Jaera the vitellophags scatter irregularly through the yolk, and the continuity of the two parts is broken. Makins: due allowance for this difference, I think there is sufficient similarity between the two structures, the liver endoderm of Jaei'a and the entoderm plate of Astacus, to suggest their identity. It is anticipating somewhat to enter upon a discussion of the origin of the liver lobes here, but it may be stated that the cells from which they arise in situ \rv Jaera and other Isopods cannot readily be distinguished from the mesoderm cells which lie in their immediate vicinity. As will be seen later, there is a migration forward of the majority of the mesoderm cells to form the mesodermal tissues of the anterior or naupliar portion of the body, and in this migration the cells .of the liver endoderm, it may be imagined, share, taking up their position on each side of the body and proceeding to ingest yolk. In their origin they are closely related to the mesoderm cells and resemble them in appearance, whence the difficulty of distinguishing them, when they have reached their definite position, from the adjacent mesoderm cells.

In my preliminary notice ('92) of some of the observations recorded in this paper, I described the row of dark cells in the


32-celled stage as mesoderm, while the vitellophags I regarded as representing the entire endoderm. I now believe that this view was incorrect, one of the so-called mesoderm cells containing endodermal material as well as mesodermal. This cell is evidently situated near the ventral mid-line, the remaining cells of the circle being purely mesodermal. In the stage at present under discussion the endoderm {l.eji) and the vitellophags {vi) become finally separated from the mesoderm (ine) by the oblique separation off of the liver endoderm i-cell, and the differentiation of the germ layers is thoroughly established. Fig. 16 is a side view of an ovum of this stage in which the division has been completed, and it is seen that the staining properties of the cells remain very nearly the same as in earlier stages, though some of the ectodermal cells in the mid-ventral line just anterior to the mesodermal circles are no longer readily to be distinguished by their staining powers alone from the mesoderm cells.

Beyond this stage I have not been able to follow the divisions stage by stage, and the next stage figured is somewhat more advanced than the last, and probably represents the results of two cleavages. Interesting changes which will result in the differentiation of the embryo are now beginning. In Figs. 17 and 18 are represented two views, a ventral and a dorsal, of the same egg. In the ventral view one sees a noticeable increase of the area occupied by the mesoderm {vie), or at least by darkly-staining cells which appear to be mesodermal, though it is possible that a few of them may be ectodermal cells which have stained more deeply than usual. For the most part, however, these cells are mesodermal, and in the mid-line there projects backwards from them the liver endoderm {l.en), which is now composed of several cells. The number of vitellophag cells still remains at eight, and in front of the mesoderm are seen the ectoderm cells, which show the anastomosing processes so well-marked in the 32-celled stage (Fig. 13). In comparison with the stage represented in Fig. 16 the presence of these processes is very marked, but it seems probable that that figure represents an ovum preparing to divide, and that in the resting stage the syncytial condition


of the ectoderm becomes as well-marked as in Fig. 17. The dorsal view (Fig. 18) presents a very different appearance from the ventral. In the first place, the ectoderm cells are much more widely separated from one another, and secondly, the mesoderm no longer forms a complete ring, but a break in its continuity has appeared in the mid-dorsal line, and, furthermore, it will be seen that the band thus formed tapers rather suddenly toward either extremity. These appearances can only be explained satisfactorily by supposing that a migration of the ectodermal and mesodermal cells is taking place towards the ventral surface, or to express it slightly differently, that all the mesoderm cells and part of the ectoderm is concentrating upon the ventral surface to form the embryo.

This concentration has become more marked in the stage represented in Fig. 19. The preparation is figured from the ventral surface, and it is seen that the mesoderm band has become broader along the ventral mid-line, or, at any rate, that it is composed of a greater number of cells in that region than it was in the preceding stage, and at the same time the bands have thinned out considerably at the sides where before they were broadest. The liver endoderm (/. en) is still visible, though less prominent than before, and the vitellophags have divided, as may be seen from their marked approximation in pairs. The ectodermal cells on the ventral surface have now come into close apposition so as to form a plate lying in front of the mesoderm band, the front edge of the plate being somewhat deeply notched in the mid-line. The cells composing it are well defined, and in the preparations are distinctly separated from each other, possessing no protoplasmic processes as in preceding stages, and sections show that they now give off no processes into the yolk, which seems to be entirely destitute of protoplasm and forms a central purely nutritive mass, upon the surface of which the cells lie. So far as the ectoderm of the plate and the mesoderm are concerned the syncytial condition no longer exists, though apparently the dorsal ectodermal and the vitellophag cells still are united through the intervention of the peripheral protoplasm. As regards the ectodermal cells, they are rather widely scattered, and their protoplasm is


hardly visible in surface view, though more distinct than that of the vitellophags.

Sections through this stage, or through one very slightly later (PI. VI, Fig. 21), show that during the concentration of the mesoderm towards the ventral surface it becomes several layers thick, forming a thickening of cells projecting into the yolk, all the cells, however, being well separated from this latter material. In front of this mesodermal "plug" lie the cells of the ectodermal plate {cc) arranged in a single layer, and it is further to be noted that the yolk at this stage contains no cells whatever, nor are any to be found in it by a most careful study of serial sections through ova in what may be termed the blastula stages (PI. V, Figs. 13-16). All the cells produced by segmentation reach the surface in Jaera, and in this respect my observations are in harmony with those of the more recent students of Decapod embryology, such as Weldon ('92) and Herrick ('92), and in opposition to those of Kingsley ('87). In one batch of eggs taken from a single individual, and which were in about the same stage of development as those represented in Figs. 17 and 18, I did find in several cases a single cell almost in the center of the yolk below the mesoderm band. The arrangement of the superficial cells, however, differed distinctly from those of other eggs in the same stage of development obtained from several individuals, and I believe these ova to have been abnormal. Whether or not they would have proceeded on to complete development I was, of course, on account of the method employed, unable to determine.

In a preliminary notice of this paper published some time ago (McMurrich, '92) I ascribed the formation of the mesoderm plug to the division of the mesoderm cells parallel to the surface of the ovum, i.e., to delamination. Further observation has, however, led me to doubt the perfect accuracy of this statement. Division of the mesodermal cells forming the plug occurs, it is true, but it seems probable that during the concentration of the mesoderm towards the ventral surface some of the cells are, so to speak, elbowed out of their original superficial position and forced below the surface, and to this process rather than to a tangential division I ascribe the formation of the plug.


When the concentration is complete the mesoderm forms a somewhat oval mass of cells whose anterior edge is surrounded by a row of ectodermal cells which are the most posterior cells of the ectodermal plate. Injacj-a on surface view these cells are not specially differentiated from the other ectodermal cells of the plate, but the further development shows that they possess a somewhat special function, and that they correspond to the ectodermal teloblasts which become evident later on, and which have been described in CymotJioa by Patten ('90). They now begin to divide in a direction at right angles to the long axis of the embryo, and by repeating this division give rise to rows of cells extending forward from them. By this process the teloblasts are themselves forced backward over the mesendodermal plug, which thus becomes covered by ectoderm. About the time, however, that the mesendodermal plug is about half covered, the vitellophag cells, which, 'up to this time, have retained their original position, begin to migrate into the interior of the yolk, as is shown in the section represented in PI. VI, Fig. 22. In this section one sees anteriorly a row of ectoderm cells {ec) which represent the ectodermal plate, and behind them two cells {tt) which form one of the rows of cells produced by the division of the teloblast {T). This cell appears considerably larger than the other ectodermal cells, but so great a difference as occurs in this particular case is not as a rule noticeable. Below the teloblast and behind it are found a number of mesoderm cells {MEn) representing the mesendodermal plug, but which are much more loosely aggregated than they were in the section represented in Fig. 21. The chief interest of the section, however, lies in the arrangement it shows for the vitellophag cells {vi). Two of these ar-e seen ; one still occupies a position at the surface, while the other has sunk into the yolk a short distance. The example of the latter is followed later by the other vitellophags, and it may be again remarked that when they begin their immigration there is no trace of cells in the interior of the yolk, nor have I succeeded in finding evidence that any of the vitellophags come from other regions of the blastoderm. In later stages the vitellophags are certainly more numerous than they were


at the time of immigration, but I believe that this is entirely due to a subsequent multiplication, and that all the vitellophags trace back their origin to the cell D of the eight-celled stage.

By the time the ectoderm has grown back over the mes-endoderm plug the embryo has the appearance represented in Fig. 22, in which the area occupied by the mesoderm is indicated by the dark shading. The ectoderm cells covering this region are manifestly arranged in rows as they are also towards the sides, while anteriorly they become more scattered, though this is not well shown in the figure. The penultimate row of cells seems to be the teloblast row, though I could not make out a perfect correspondence between them and the rows of cells extending forward from them. It will be noticed, however, that the multiplication of those nearer the middle line has taken place somewhat more rapidly or to a greater extent than that of those situated towards the sides, and the posterior edge of the ectoderm plate has now become convex backwards instead of concave as it was originally. In front of the mesoderm and to the sides a series of rows of cells passing forwards and outwards can be seen, and at the extremity of these is a patch of larger cells which will later give rise to the eyes and which are clearly seen in the side view (Fig. 23, E). These rows of cells leading towards the eyes were not always, however, as clearly marked as in the embryo from which this drawing was made. One other fact in connection with the preparation requires mention, and that is that the cells of the mesendodermal plug are much less closely aggregated than in earlier stages and seem to have a tendency to scatter themselves, so that the shaded area of the figure has a greater extent than that originally occupied by the plug.

The division of the teloblasts progresses in later stages, and a layer of ectoderm, whose cells are arranged in regular longitudinal and transverse rows, extends backwards over the yolk. In Figs. 24 and 25 is represented an embryo in which the blastoderm has extended about two-thirds of the way around the yolk, Fig. 24 representing the anterior half of the embryo, and Fig. 25 the posterior half. Looking at the anterior half

86 MCaMURRICH. [Vol. XI.

one sees that the cells of the anterior portion of the body are arranged very irregularly, but form three bands which enclose a triangular area whose apex is directed backwards and whose cells are scattered about irregularly. The two bands which form the sides of the triangle are the ventral lateral bands which have been so often figured in the early stages of the Decapods and in which the naupliar appendages and nervous system will form, while the transverse band forming the base of the triangle is apparently not so well marked in the Decapods, though evidently distinct in AlpJiciis according to Herrick's ('92) figures and descriptions. At the basal angles of the triangle are the " Anlagen " of the eyes {E). This anterior region represents then the naupliar region of the embryo, and behind it comes what has been termed by Bergh ('92) the metanaupliar region, whose ectoderm is markedly arranged in rows and has resulted by the growth of the teloblasts. Examining the anterior portion of this region, one finds that there are about eleven rows of cells which extend farther forwards than the others. In the embryo figured there seemed to be only ten of these longer rows, but this seems to have been an abnormality if it can be so called, eleven being probably the typical number. Further back other rows are to be found on each side, those nearer the middle line being longer than those further laterad. At the end of each row is a teloblast (Fig. 25, T) slightly larger than the other cells of the row and frequently showing karyokinetic phenomena. The number of teloblasts and accordingly of the rows is liable to a certain amount of variation apparently. A central teloblast with its row is generally well marked, as may be seen in Fig. 26 {cT) drawn from a somewhat older embryo than Fig. 25, and it may be stated here that the cell row arising from this gives rise to a " Mittelstrang " recalling what has been described in other Invertebrates. On each side of this central teloblast there are about eleven or twelve lateral teloblasts, there being typically the same number on both sides, though not infrequently one finds, for example, eleven on one side and twelve on the other, as in Fig. 26. Behind the teloblast row is a varying number of ectoderm cells arranged somewhat irregularly. They are


somewhat more numerous in later than in earlier stages, as may be seen by comparing Figs. 26 and 25 (i^V), but their exact origin I have not been able to determine ; it seems possible that they may be ectodermal cells which the teloblasts have pushed before them in their growth over the surface, a view which their increasing number in older stages would justify. I cannot, however, advance any definite proof of this idea.

As regards the mesoderm the results obtained from the study of Jaera have been by no means satisfactory, the smallness of the q^^ rendering proper manipulation exceedingly difficult. It is, however, easy to determine that the mesoderm plug no longer exists in the stage represented in Figs. 24 and 25, and that mesoderm cells are to be found in this stage beneath the ectoderm of the naupliar region of the embryo. That these two facts stand in intimate relation with one another I have good reason to believe. Sections show that in stages slightly younger than Fig. 24 the cells of the mesoderm plug are becoming separated and are extending forwards into the region covered by the ectoderm plate. This process continues in later stages, and one finds eventually only a few scattered cells in the region originally occupied by the mesendodermal plug, while anteriorly the cells have become quite numerous. The greater part of the mesoderm contained in the mesoderm plug goes accordingly to form the naupliar mesoderm, and so far as could be discovered all the naupliar mesoderm m. Jaera is derived from this source. The liver endoderm in stages later than Fig. 19, is not distinguishable and apparently accompanies the mesoderm in its distribution, taking up its position at the junction of the naupliar and meta-naupliar regions where later the liver Anlagen appear. Of the origin of the meta-naupliar mesoderm little could be determined. One finds the mesoderm arranged in masses on either side of the middle line in Fig. 26 {Me), and from what is known to occur in other forms it may be supposed that these masses are the result of a teloblastic division such as Patten ('9o) has described for Cymothoa. I was unable, however, to make out the mesodermal teloblasts in surface views qI Jaera, and though


it is certain, I think, from the sections I possess that they exist, yet I have not been able to get a clear idea from Jacra either as to their arrangement or as to their origin.

The development of Jacra has now been traced up to a period in which the germ layers have assumed their general distribution. Further changes are mainly of a histogenetic character and may be postponed for the present until an account has been given of the development of Aselhis, Porcellio, and Armadillidium.

2. Tlie Segmentation and Formatiott of the Germ-layers of

Aselliis Commtmis.

So far as I am aware, there are no published descriptions of the early development of the common Aselhis of this country, first described by Say as Aselhis communis, though several more or less complete accounts have been given of the European A. aquaticus. Rathke ('34), Dohrn ('67), and Sars ('68) did not make any definite observations upon the segmentation, but Van Beneden ('69) describes accurately its general features, showing that it is of the typical centrolecithal character and regular, though he did not attempt to follow it out in detail. The latest author who has written on the subject, Roule ('89), has certainly not advanced our knowledge of it. His statements in the brief contribution he has published are so remarkable and so little in harmony with those of Van Beneden and with what might be expected from analogy with other Crustacea, that one involuntarily distrusts them. They are to be explained probably as due to failure to study in surface views stained ova, the only method by which the phenomena of the early stages can satisfactorily be made out.

My observations on Aselhis commimis are, I regret, not quite so complete as these on Jaera, inasmuch as I failed to obtain a few stages; they are, however, sufificiently connected to allow of a close comparison with Jacra, which was the object I had in view in undertaking them. The early segmentation is step for step identical with that occurring in Jaera, so far as that form of segmentation followed is concerned. I have reason to believe,


however, that in this species also variations occur in the segmentation, certain cells in different eggs dividing in different directions. I have not, owing to technical difhciilties, attempted to follow out these variations, but have confined my attention to that variety which resembled what has been described ior Jaera and which seems to be the most typical.

The center of the yolk is occupied in the unsegmented &gg by a mass of protoplasm containing the segmentation nucleus and prolonged at the surface into a number of branching processes, w^hich undoubtedly are continued through the yolk to form a network in the meshes of which are situated the yolk granules. I was not able to detect this network in mature ova, but a peripheral zone of protoplasm, somewhat thinner relatively than that of Jaera, was readily observable and is probably, as in that form, in organic continuity with the central mass. The first cleavage divides this latter into two portions (Fig. 28, A, C), its plane being in this case apparently also at right angles to the long axis of the future embryo. The second division gives rise to four cells (Fig. 29, A, B, C, D), two of which lie in a plane at right angles to that of the other two, while the third division produces eight cells, four of which {A, a', B, b') are arranged in a circle near the anterior extremity of the embryonic axis, three others {C, c', d') in a second circle near the posterior extremity of this axis, while the eighth [D) lies slightly to one side of the extremity of the axis, the whole arrangement being identical with that of Jaera in the same stage {cf. Fig. 9). In the i6-celled stage the arrangement is slightly different from what obtains in Jaera. All the cells divide as in that form, and there are produced two circles, each of four cells, in the anterior half of the Qgg, while at the equator are two other cells, which arise from two of the cells forming the circle of three in the eight-celled stage. Behind these is a circle of four) cells which correspond in position to the mesodermal circle ^f Jaera, but have not the same fate, while near the posterior pole are two cells {D and X) corresponding in position and history, but not exactly in function with the endoderm cells of Jaera. I have not observed any shifting of cells such as I have described for Jaera^


and the principal difference in the two forms seems to be the smaller number of cells in Ascllus in the circle surrounding the products of D.

The next, or 32-celled stage, I have not, unfortunately, succeeded in obtaining, but from preparations of the 64-celled stage I have been able to determine what the divisions mayhave been. In Fig. 32 is a view of the 64-celled stage in which the cells which arise by the division of each of the cells of the 32-celled stage are bracketed. Disregarding all other cells than D and X of Fig. 31, each of these divides into two, so there are four cells which correspond cytogenetically to the four vitellophag cells of the 32-celled stage of Jaera. At the next division these divide into the eight cells represented in Fig. 32, as D^ E^ D^ D"^ X^'^, which present a well-marked differentiation, the cells indicated by D being much smaller than those which result from X. It will be noticed the X and D groups of cells no longer are situated at the extremity of the oval Q.gg, but occupy almost the middle of one of its faces, that namely which is to be the ventral surface of the embryo. Whether this change of position is due to a shifting forward of all the cells over the yolk, the axes of the ^gg remaining as before, or whether it is only an apparent shifting due to an alteration of the shape of the yolk, I am unable to say, but incline towards the latter idea. It is also noticeable that the segmentation hais now manifested itself at the surface, the surface of the yolk being divided into areas corresponding to the cells, a central mass of yolk, apparently destitute of protoplasm, remaining unsegmented. In Jacra the superficial segmentation made itself manifest in the 32-celled stage; whether it appears in the same stage in A. communis I am unable to state, but it seems not improbable that it does. In A. aqnaticus, according to Van Beneden ('69), it appears in the i6-celled stage.

In the next stage 128 cells are formed, the D group having now increased to eight (Fig. 33, D^-D^) and still remaining distinguishable by their smaller size. The X group I was not able to identify in this stage, at which, indeed, a very unfortunate gap occurs in my observations. This is all the more


unfortunate in that it is just after this stage that the differentiation of the mesoderm takes place, and I am obliged to depend upon circumstantial evidence in attempting to ascertain its origin. Shortly after the 128-celled stage, the concentration of the cells towards the ventral surface, to form the blastoderm, begins. By the time this has become well marked the differentiation of the mesoderm has made itself apparent. In Fig. 34 is represented the youngest blastoderm I observed. It forms a somewhat oval disc, whose longer diameter is at right angles to that of the future embryo. In the center is a patch of cells {vi) which stain more deeply than the rest and which I take to correspond to the vitellophag cells of Jaera ; in front of them lies a row of cells {MEn) somewhat smaller than those still farther forwards, and this row I believe to represent the mesoderm, or more probably, as in Jaa-a, they are mesendodermal. In the blastoderm figured the row seems to be composed of six cells, a number which suggests a close comparison with Jaej-a, in which the same number was found when the mesoderm was first differentiated. As to the orig-in of the mesendodermal cells I can only suppose that they belong to the D group, and that the four D cells of Fig. 32 contain all the mesoderm and endoderm of the future embryo. The small size of the mesendoderm cells, and the fact that they and the vitellophag cells combined are just about sufficiently numerous to account for two divisions of the D group, suggests that this number of divisions have intervened between Figs. 33 and 34, and judging from the appearance of Fig. 34 there can hardly have been a smaller number, though I did not attempt to enumerate the ectoderm cells. This method of determination from circumstantial evidence of this kind is, I am aware, exceedingly unreliable ; but I believe in this case it allows of accurate conclusions, and that from the X cell of Fig. 31 ectoderm cells and from the D cell mes-endoderm and vitellophags result.

In a slightly later stage (Fig. 35) the vitellophags are still clearly distinguishable, and the mesodermal row [Me) has become much longer, now forming a curved line of about fourteen (J) cells surrounding the anterior portion of the endoderm.


In front of these another row of cells (7") is noticeable, consisting of exactly eleven cells. This is the ectodermal teloblast row, and it is always symmetrical, a central cell (cT) being readily distinguishable with five cells on either side of it. At this stage no further differentiation is noticeable, but a little later the arrangement represented in Fig. ^6 is found. Here a central, darkly-staining mass of cells may be found, which consists of both mes-endoderm and vitellophags. It is several cells thick, and represents the mesendodermal plug plus the vitellophags oijaera. About the center of the mass a distinct depression was visible in the specimen figured, but it does not correspond to any deep invagination, but seems to be formed by the withdrawal of several cells from the surface at this point. The mesoderm is no longer distinguishable from the endoderm either by its staining properties or by its arrangement, but it seems highly probable that the anterior part of the mass is mesendodermal, while the posterior is composed of vitellophags. What has happened is that the mesodermal row of Fig. 35 has undergone a concentration towards the middle line similar to what occurred in Jaera, a multiplication of the cells having taken place at the same time. A careful examination of the anterior edge of the mesendodermal mass shows a somewhat irregular row of eigJit cells {meT), a number which suggests the possibility of these being the eight mesodermal teloblasts which, as will be seen later, give rise to the meta-naupliar mesoderm. The ectodermal teloblasts (7^) are even more distinct than in the last stage, and have arranged themselves somewhat differently. The five cells of each side have come closer together, and have separated from the central teloblast {cT), and, furthermore, each of the lateral teloblasts has budded off a small cell (^r'), the first of a teloblastic row. The lateral teloblasts and their progeny form two rows each consisting of five cells lying in front, and somewhat laterally to the mesoderm plug. Extending from these rows laterally can be noticed on each side a band of ectoderm cells, showing a slight tendency to be arranged in rows, but nevertheless with a good deal of irregularity. These two bands are the lateral ventral bands of the naupliar region of the embryo. The cells which are to give


rise to the eyes cannot be distinguished at this period, nor is a distinctly marked transverse band, such as occurs in Jacra, to be made out.

Up to this stage sections show no vitellophag cells imbedded in the yolk, but slightly later the immigration of a considerable number of cells of the posterior mesendodermal region takes place, and at about this same time a stout, backwardly projecting process of the mesendodermal region became visible, and this I take to be the liver endoderm, since it is closely comparable to the process described in Jaera as the Anlage of the liver lobes. Its origin I have not been able to m.ake out, nor is its fate at all certain, since it quickly disappears, most probably becoming indistinguishable from the mesoderm cells.

In Fig. 37 is represented a somewhat later stage of development, in which it is seen that the teloblasts have increased considerably in number, there being now twenty-two of these cells, ten on one side of the center one {cT) and eleven on the other, a disparity of number on the two sides, which is repeated in the specimen from which Fig. 38 was drawn. The source of the teloblasts indistinguishable in the previously described stage (Fig. 36) I have not been able to discover. They do not, however, appear to be formed by the transverse division of the eleven primary teloblasts, and it seems probable that they are due to the addition to the teloblast row of cells lying originally lateral to it, or of the progeny of such cells. Whatever may be the origin of these cells the teloblast row now forms almost a semicircle, enclosing a mass of cells which represent the mesendoderm. In front and at the sides of the teloblasts are to be seen the teloblastic rows, those on either side of the middle line being a little longer than those further towards the sides, a point which is not well brought out in the figure, it being difficult in it to distinguish between the anterior teloblasts and the most posterior cells of the naupliar region.

These latter, and the cells in front of them, are characterized by their irregular and scattered grouping, so that they present a marked contrast to the regular rows of the meta-naupliar region. The more anterior cells, however, are more closely aggregated, and form the so-called cephalic lobes of the embryo.


These, and the cells intervening between them and the anterior teloblastic cells, form the lateral ventral bands, which have approximated much more closely to each other than in earlier stages. No transverse band, such as was found in Jacra, can be made out in this stage (its absence in earlier stages has already been noticed), unless a patch of cells {DO?) lying in the middle line between the two cephalic lobes is its representative. This is possible, though another possibility, that this patch represents a rudimentary dorsal organ, should not be lost sight of. In the embryo figured another patch of cells, which stained quite deeply, was found in the median line immediately in front of the teloblastic cells ; I cannot discover any special significance for these cells, which are indistinguishable in later stages, and regard them simply as naupliar cells of the midline joining the two lateral bands.

In the stage represented in Fig. 38 the teloblastic rows have increased considerably in length, the rows originating from the eleven primary teloblasts being longer than any of the others, which decrease in length the further they lie from the median line {compare in this respect Jaera, Figs. 25 and 26). The teloblast row no longer forms a semicircle as in the previous stage, but is now straightened out, indeed, is slightly concave anteriorly. It has at the same time extended backwards over the region previously occupied by the mes-endoderm, which layer is unrepresented in the drawing, and it may be noted, too, that all the teloblastic rows are parallel to each other, and have an antero-posterior direction. The more lateral ones have not increased in length very much when compared with the previous stage, but the more median ones have, and it would seem that the straightening out of the teloblastic row was due to this. excessive growth of the more median rows, which are consequently pushed backwards, the lateral teloblasts remaining fixed or being pushed outwards only. This explanation of the change of form of the teloblast row was suggested in the description of Jaera, but the evidence in its favor is much more pronounced in the present species. Special attention may be directed to the median teloblastic row, the cells of which are markedly different from those of the other rows, and constitute,


as in Jaei-a, the Mittelstrang. Behind the teloblasts is a mass of ectoderm cells more or less irregularly arranged, which represent the hind end of the body, and among which the anus will form.

With regard to the naupliar portion of the embryo at this stage, there are but few points to be noted. The separation between the naupliar and post-naupliar regions is clearly indicated by the scattered arrangement of the cells, and the lateral ventral bands have approximated still further than in Fig. 37. On each side, just at the junction of the two regions, is to be seen a small patch of cells which constitutes the Anlage of the lobed (lappen-formige) organ of Aselliis.

3. TJie Segmentation and Fonnation of the Get-m-layers m Porcellio and Arinadillidiwn.

The similarity between the ova of Porcellio and Arniadillidiiim in their development is so great that the two forms may be considered together.

Porcellio has been the subject of several papers, which present most remarkable discrepancies in the descriptions given of the segmentation, and, in considering them, papers on Onisciis may also be included, since my own observations on forms belonging to this genus and on the nearly related genus Philoscia, though most fragmentary, nevertheless clearly show that the type of segmentation is identical with that of Porcellio. The earliest paper is by Bobretzsky ('74) on Oniscns nmrariiis, and the results recorded in it have been accepted and copied into many general works. According to Bobretzky the segmentation of Onisciis is of the epibolic variety, and resembles that described by Van Beneden ('69) for Mysis. According to his account a quantity of protoplasm collects at one pole of the tz%y forming a colorless, transparent, rounded mass, which, in the next stage observed, is represented by a plate or disc of cells lying sometimes at the pole of the &^^, sometimes near the pole, and sometimes upon the side. In later stages this disc extends over the yolk, the mes-endoderm forming from it before it reaches the equator in cases where it originally occu


pied a polar position. It will be seen that there is a large gap in Bobretzsky's observations between his supposed earlier stage and the next he describes, and I believe from my observations on an undetermined species of Onisacs that he is entirely in error, that his disc represents a comparatively late stage of segmentation, and that his first stage was an abnormal Qgg. In other words, the Oniscus which I observed had a typical centrolecithal segmentation closely similar to that of Asellus, certain of the cells aggregating at an early stage to form a blastoderm or disc, the rest of the yolk being covered by scattered stellate cells whose existence Bobretzsky probably overlooked. Nusbaum ('86), who also studied Oniscus nmrarms, did not observe the segmentation stages, but states that, in ova in which the disc was not yet present, the central portion of the yolk stained deeply and was coarsely granular, and seems inclined to regard this as indicating the presence at that point of the "plasme formatif," though he was unable to distinguish the first segmentation nucleus.

Close on the heels of Nusbaum's paper appeared one by Reinhard ("87) on Porcellio scaber, for which he described a typical centrolecithal segmentation, the cells formed by the division of the original centrally situated segmentation cell gradually migrating towards the surface of the yolk, upon which they are scattered, forming what Reinhard terms " Inselchen." Up to this point this author's observations tally perfectly w^ith my own, but his account of the formation of the mes-endoderm is, I think, incorrect. He says : " Unter diesen Inselchen konnen sich an verschiedenen Stellen mehrere Lagen von Zellen befinden . . . Der Raum zwischen den Inselchen fullt sich allmahlich an, und bald zeigt sich ein Theil der-Eioberflache mit einer dichten Lage des Ectoblasts bedeckt. Die unter dem Ectoblast liegenden noch indifferenten Zellen stellen das primare Entoderm dar. Nur allmahlich differenziren sich dieselben in Mesoderm- und Entoderm-Zellen." In how far my results differ from these will be seen later.

Roule has contributed several short notices on the development of the Isopods, and especially of Porcellio. In the first of these ('89) he describes the egg of Porcellio as having at


one end a patch of hyaline plasma which segments into large cells extending over the surface of the yolk by the addition of new material added from the yolk. So far as I can understand the brief notice, he believes the mesendoderm to form very much in the manner described by Reinhard. He speaks, however, of nuclei which " naissent spontanement " (!) at the periphery of the vitellus, the yolk particles in their neighborhood assuming an altered appearance. In a general way he agrees with Bobretzsky regarding the segmentation, and with Reinhard in the formation of the germ layers, unfortunately being in error with regard to both processes. In a second paper (Roule, '90) the same author evidently mistakes an ^^'g in a somewhat advanced stage of segmentation for one which has not yet divided. He describes the occurrence of islands of formative plasma at the surface of the yolk, but asserts that only one of them, situated at one pole of the &g%, contains a nucleus ! This nucleated island divides and forms a disc which gradually extends and fuses with the non-nucleated islands, which do not divide until the migration (!) into them of nuclei or particles of nuclear matter from the disc, and the extension of the latter is thus continued. The disc also increases in thickness by the deutoplasm in its vicinity assuming the character of formative plasm, into which nuclei migrate from the disc cells. A truly remarkable series of phenomena ! In this paper, however, he corrects one gross error of his earlier one, denying the spontaneous formation of nuclei, and on the whole comes a little nearer to the truth, though still almost hopelessly in error. His third paper (Roule, '91) deals with the formation of the germ layers, and he advances the idea that the endoderm, i.e., the liver endoderm, appears in place as the result of a proliferation of the blastoderm cells on each side of the body a short distance behind the head, while the mesoderm arises from the proliferation of the blastoderm cells at various regions, practically, indeed, in every region, a view which is emphasized in later contributions (Roule, '92, '92^). Roule's statements are evidently assumptions based upon imperfect observations, and if proper use had been made of surface views, m.any of them would never have been published.


It is evident that considerable difference of opinion exists as to the mode of segmentation and the development of the germ layers in the Oniscidas. I have not been able to follow out the development of Porcellio and Armadillidiitm quite so. thoroughly as that of Asellus or Jacra, but my observations are sufficient to allow of a close comparison with these forms. The fertilized but unsegmented egg of Porcellio or Armadillidiiim resembles very closely that of Asellus, though considerably larger. It is enclosed in a chorion and a vitelline membrane, and the surface of the yolk is covered by a thin layer of protoplasm, while in the center is an amoeboid mass of protoplasm surrounding the first segmentation nucleus (PI. VII, Fig. 39). In the next stage (Fig. 40) two such amoeboid cells are to be found in the egg, which, it may be here remarked, is much more irregular in shape than either Asellus or Jacra, orientation being accordingly much more difficult than in these forms. This 2-celled stage is succeeded by one in which there are four cells (Fig. 41), the arrangement being practically the same as that described m.Jaera ; for instance, the line joining two of the cells {A and B) being at right angles or nearly so with that joining the other two {C and D). I was not able to obtain ova which showed the original two cells in the spindle stage, so cannot state how the arrangement is produced, or even that the cells are correctly identified with those of Jaera; they are identical in position, but whether they are also genetically identical I cannot state. One further point in connection with this stage deserves mention. The peripheral protoplasm, which in earlier stages was uniformly distributed over the yolk, at this stage becomes concentrated to a certain extent at one region of the egg {pp), a region which will later become the posterior 'end of the naupliar portion of the embryo, and may be regarded, therefore, as the posterior pole of the ^'gg. This aggregation of the peripheral protoplasm is an exceedingly interesting process, marking out as it does thus early a polar differentiation of the egg, and giving also a suggestion as to one of the possible explanations for the belief in an epibolic segmentation in Porcellio. In order that there may be no question as to the centrolecithal character of the segmentation, I represent (Fig.


42) a section through an ovum in the four-celled stage. The section has passed through two of the cells, and shows clearly their position in the yolk exactly as in Jacra or Asellus, or any other form which segments in the typical centrolecithal manner.

In the eight-celled stage there seems to be considerable irregularity in the arrangement of the cells in different ova, but in some an arrangement comparable to that oijaera was found. In Fig. 43 is represented an ovum of Arniadillidhim which is just passing into the i6-celled stage, all the nuclei but one being in the spindle stage. This one, which has not yet started to divide, lies near the posterior pole of the ^gg and corresponds to the D cell of Jacra and Asellus so far as its position is concerned. In other eggs both of Ponellio and Arviadillidiwii this same arrangement was found, and it seems to be normal for some eggs, though in others it could not be made out. The i6-celled stage derived from an ^gg of this arrangement shows as vcvjaera two cells at the posterior pole (Fig. 44, D and X), surrounded by a circle of apparently four cells in Porcellio (Fig. 44) and five in Annadillidiinn (Fig. 45). Whether this difference be due to the slightly different stages represented, the Porcellio ovum represented having just completed division, while that of Armadillidiinn is just passing into the 32-celled stage, I cannot say, but just as the number of cells composing the circle differ in Jacra and Asellus, so too they may do here without any appreciable alteration of the similarity of the general features of development. In fact I may state that in one ^^^ of Porcellio I saw six cells in a circle, and I believe that even in different eggs from the same individual there may be considerable differences of arrangement of the spherules at this stage, although the two polar cells are probably always distinguishable. Sections of ova of Porcellio in this stage (Fig. 47) show that the cells have reached the surface of the yolk and are imbedded in the peripheral protoplasm, so that the ^^^ is a syncytial blastula whose cavity is completely filled with yolk, there being certainly no nuclei, and apparently no protoplasm, scattered among the yolk granules.


The 32-celled stage of Porcellio I have not seen, but in Armadillidiimi (Fig. 46) the two polar cells have divided so as to form four cells arranged as shown in the figure, and around them is a circle of eight cells whose origin I have not made out. The appearance of the ova at this stage is markedly different from that of Jacra, and presumably Asellus also, on account of the total absence of superficial segmentation lines. Indeed, in neither Porcellio nor Ai'inadillidiinn does a segmentation of the yolk ever occur, a condition no doubt explained by the greater amount of yolk presented in these forms. It was probably some such stage as this which Roule mistook for an unsegmented ovum.

The next stage which I have observed corresponds to the 64-celled stage of Asellus, though whether there are actually sixty-four cells present I cannot state. The most striking feature of the stage (Fig. 48) is the commencement of a concentration of a number of the cells around a point which in some cases is on the side of the egg, and in others at its extremity. The concentration is indeed but feebly marked at this stage, but is nevertheless indicated, and becomes pronounced in the next stage. In the center of the region around which the concentration is taking place _/i?//r cells (D^-D^) may be observed which differ somewhat from the others in their staining properties. These may be the four polar cells of the preceding stage or else four of the eight cells which would be produced from their division. Which of these alternatives is correct I cannot state, but from analogy with what occurs in Asellus I am inclined to believe the second one the more probable. These cells give rise to the mes-endoderm, and in the next stage (Fig. 49) have increased to eight in number, -the concentration of the other cells increasing. These last two figures (Figs. 48 and 49) represent ova of Porcellio; I did not find Armadillidiimi ova in corresponding stages, but the next figure (Fig. 50) represents an egg of that genus in a slightly later stage. The mes-endoderm is formed of a much larger number of cells, and the concentration of the ectodermal cells around it is very evident. In Figs. 51 and 52 slightly later stages of Armadillidiu'm are shown, presenting a still greater


increase in the number of mes-endoderm cells as well as of the surrounding ectoderm cells, the latter increase being due mainly, if not entirely, to the division of the cells which have already undergone concentration in Fig. 50, there being probably no further migration of cells from the dorsal surface of the Q.^^ to form the blastoderm. In these last two figures a row of four cells {Me) not at all well-marked off, however, lies at the anterior margin of the mes-endoderm zone, and it is possible that they correspond to the row of mesoderm cells occurring in Aselhis, though it has been impossible to distinguish them in later stages, and accordingly their significance is uncertain.

An Q^g of Porcellio, in a slightly later stage than Fig. 50 though earlier than Fig. 51, is represented in Fig. 53. From this it will be seen that the process of development proceeds in this species as in Armadillidhim. There is a similar continuation of the concentration of ectoderm cells around the mesendoderm, to form the blastoderm, and in Fig. 54 a later stage is shown, in which the cells of the ectodermal portion of the blastoderm are considerably increased, and in which the concentration is somewhat greater. In the preparation from which this drawing was made the blastoderm had been dissected from the surface of the yolk, and in the operation was slightly torn ; enough of the yolk was left adherent, however, to support a few of the scattered ectoderm cells which do not take part in the formation of the blastoderm. Up to about this stage in Porcellio^ and in Arinadillidm7n also, all the cells are at the surface of the yolk, but now careful focusing of the mes-endoderm region shows that there are a number of cells below the surface, a fact which is also shown by sections of the blastoderm (Fig. 55). I found no indication of spindles perpendicular to the surface in my preparations of this stage, and am inclined to believe that the cells of the lower layer arise by immigration rather than delamination, although I am not in a position to maintain that the latter process does not occur. It is certain, however, that at this stage lower layer cells occur only in the mes-endoderm region.

In Fig. 56 a somewhat older blastoderm of Porcellio is shown, s'r.i'l.ir pr2prj"?.t'ons of ArniadiUidittm having also been ob


tained. Here the region occupied by the mes-endoderm has greatly increased, being represented in the figure by the shaded area, though it is probable that a portion of the area really lies outside the true mesendodermal region, the anterior margin of the area appearing dark in the preparation on account of the lower layer mesendodermal cells having begun to migrate forwards and to scatter beneath the general surface of the blastoderm. This forward migration and scattering has proceeded further in Fig. 57, which is from Arniadillidiuni, though preparations similiar to this and the succeeding figure were obtained from Porcellio. The concentration of the cells to form the blastoderm has now reached its greatest extent, the nuclei are crowded together, and the entire area covered by the blastoderm, notwithstanding the increased number of cells, is little if any greater than that which it covers in the stage represented in Fig. 53. A little behind the middle of the blastoderm is seen a very dark area {MEn) which represents the mesendodermal region in which the lower layer cells are heaped together to form a mesendodermal plug projecting into the yolk, and anteriorly, as well as posteriorly, less dark areas are to be seen, which are due to lower layer or mes-endoderm cells which have migrated forwards and backwards from the mes-endoderm region. I say migrated, because I have not been able to find any evidence for the assumption that they originate in situ by delamination from the ectodermal portion of the blastoderm. Numerous spindles are to be found in blastoderms of this stage, but I have never seen any directed perpendicularly to the surface, and believe that here as in Jaera and Aselhis all the mes-endoderm comes from the cells of the mesendodermal region.

In this figure there is to be seen just at the front edge of the mesendodermal region an imperfectly defined row of cells which are the first indication of the ectodermal teloblasts (7^). They are not sufficiently well marked to determine their exact number, though they seem to be more numerous than eleven, which was the number first differentiated in Aselhis. They serve to mark accurately the anterior edge of the mesendodermal region.


The scattering of the mesendodermal cells and the more perfect differentiation of the ectodermal teloblasts continues in later stages, producing the appearance shown in Fig. 58. Here the entire area of the blastoderm presents a dark appearance due to the presence under the ectoderm cells of lower layer cells, the mesendodermal region being but little darker than the remaining portions. The scattering of the mesendodermal cells is complete ; and careful focusing, as well as sections, show the presence of vitellophag cells in the yolk. No definite differentiation of these cells from the remaining mesendodermal elements is, however, to be distinguished, and, taking the three forms of cleavage here described into comparison, it will be seen that we have in them three gradations of differentiation of the mesendodermal elements. In Jacj-a there is first of all a distinct differentiation of the vitellophags from the rest of the mes-endoderm and later a differentiation of the liver-endoderm for the mesoderm, so that three distinct portions of mes-endoderm are visible. In Asellus the differentiation of vitellophags from the remaining mes-endoderm is at certain stages distinct, though later becoming inconspicuous, and a recognizable differentiation of liver-endoderm is questionable. Finally, in Armadillidiuin and Porcellio no differentiation into the three portions can be made out, and it is necessary to speak simply of the mes-endoderm. There can be no doubt, however, that this mes-endoderm is equivalent to the mesoderm, liverendoderm, and vitellophags of Jaera, giving rise to cells which play the parts taken by these various elements in the development of the embryo. The significance of this variation in the differentiation of the mesendodermal elements will be considered more in detail in a later portion of this paper.

In the stages represented in Fig. 58 the further differentiation of the ectodermal teloblasts is also shown, there being thirteen recognizable in the preparation figured. One cell of the row is hardly on a level with the others, and one is tempted to consider it the central teloblast, in which case there is a marked disparity in the number of cells present on either side of it, there being but five on one side and seven on the other. The number five is significant in view of what has been de


scribed as occurring in Asellus, but there seems to be less definiteness in the number of the cells composing the teloblastic row when it is first recognizable than in that form.

The development of Porccllio and Arviadillidium has now been carried to a stage in which the germ-layers may be said to be differentiated, though as already indicated it is as yet impossible to distinguish the liver-endoderm from the mesoderm. This differentiation, as is essentially the case with Asellus, only supervenes at a much later stage, and a, halt may be conveniently called here. In order, however, to carry the account up to the stage at which it was left in Asellus, an additional figure (Fig. 59) is given, in which the embryo is distinctly recognizable. The figure was drawn from an ovum viewed by direct illumination as an opaque object, and shows the optic lobes distinctly indicated. Posteriorly are a number of scattered cells, in front of which is the row of ectodermal teloblasts (7^), which are smaller in comparison with the cells of the teloblastic rows (7>) to which they have given rise than is the case in Asellus. There are about twenty-three teloblasts in the preparation figured, and it is noticeable that there is no indication of the existence of eleven primary teloblasts such as occurred in Asellus and probably also in Jaera, the number of cells in the most anterior of the four teloblastic rows which are present being but slightly less than that of the posterior row. This is in harmony with the indefiniteness of the number of cells composing the row in earlier stages, and it may be supposed that in Armadillidium and Porcellio, the teloblasts do not enter upon their characteristic method of division until nearly the entire row has differentiated. No difference can be distinguished between the central and the remaining teloblasts, nor does the median teloblastic row of cells differ from the remaining rows; and furthermore it seems that in the species under consideration the scattering of the mes-endoderm takes place relatively earlier in comparison with the formation of the teloblastic rows than in either Asellus or Jaera, since in the stage figured there is no indication of any special mass of mes-endoderm remaining in what was originally the mes-endoderm region, but the scattering has been complete. In the naupliar region of the


embryo certain definite dark patches occur, indicating the presence of special masses of lower layer cells, which mark out imperfectly the naupliar somites. Finally it is to be noticed that in front of the embryonic region proper there is a dark patch {DO ?) situated in the mid-line, which recalls the similarly situated patch occurring in Asellus (Fig. 37, DO?) and may represent the dorsal organ.

As was stated at the beginning of the paper, I have had for study a certain amount of material representing stages in the development of Cyviothoa and Ligia, but I have not considered it necessary to devote space to a special description of the results obtained from it, both on account of the incompleteness of the material and the general similarity which exists, so far as the material went, between the development of these forms and that of the species just described. Cyviothoa has been the subject of a paper by Bullar ('78), and more recently Nusbaum ('33) has published an account of the development of Ligia occanica, unfortunately, however, concealing his results under cover of the Polish language, the majority of embryologists being obliged to rely, for information as to the contents of the paper, partly on the brief abstract of it which appeared in the Biologisches Centralblatt, and partly on the preliminary notice published by its author in the same journal ('9i).

In the earliest stage of Cymothoa which I possess the development has reached a stage which corresponds in all essentials to that of Porcellio represented in Fig. 56, and the later stages up to the formation of the embryo likewise resemble what has been described for Armadillidium in all essential particulars. Thus there is the same absence of differentiation of the mesendoderm, the same early scattering of the mesendodermal elements, and apparently the same absence of a distinct differentiation of eleven primary ectodermal teloblasts. In fact the preparations are so similar to those of Porcellio and Avinadillidiicm that I have not considered it necessary to figure them. The earliest stage I possess is identical in age with that figured by Bullar, and like that author I cannot make any conclusive statements as to the nature of the segmentation. This much, however, is certain, that there occur scattered over the surface


of the yolk at this stage numerous cells just as in the forms whose segmentation I have described, and it seems probable that in Cymothoa the segmentation follows the centrolecithal type, notwithstanding the large amount of yolk which is present, the Q^g measuring about a millimetre in length with a diameter but slightly less.

In Ligia oceanica Nusbaum apparently believes that the segmentation is of the discoblastic type, and Van Beneden ^ maintains the same view. So far as Nusbaum's observations are concerned it must be pointed out that the material at his disposal was hardly sufficiently representative of the earliest stages to allow of perfect certainty on this point, and in view of what certainly occurs in other Oniscidas a certain amount of doubt as to the accuracy of his interpretation of the observed phenomena is legitimate. My material, unfortunately, is likewise too imperfect to settle the question, though the earliest stage I have serves to emphasize the doubt. A section through an egg of this stage is represented in Fig. 60, and shows distinctly four nuclei and a fragment of a fifth, the protoplasmic masses in which they are imbedded lying flush with the surface of the yolk. In a surface view of the same (t^^ before it was imbedded the nuclei were seen to be evenly scattered over the surface, and in running through the series of sections, which was complete, over eighty nuclei were counted. There were none whatever in the interior of the yolk, and none of the cells had separated from the yolk any more than those figured. If the segmentation had been discoblastic one would expect to find some of the cells more or less aggregated at one region of the egg but this does not occur, and the appearance presented is identical with that found at a somewhat earlier stage of development in Porcellio. Further observation is necessary, I believe, before it can be definitely decided whether the segmentation of Ligia is discoblastic or superficial, i.e., centrolecithal.

1 Van Beneden's paper, " Recherches sur la composition et signification de I'oeuf," published in the Alhn. de PAcad. roy. dc Bclgiqiie, T. xxxiv, 1870, I have not seen, the above statement being made on the authority of Korschelt and Heider ('91).


The next stage which I have seen is one apparently slightly younger than that of which Nusbaum has represented sections in Figs. 18 and 20 of his Polish paper. In surface view there is seen to be a distinct blastoderm which resembles closely the stage of PorcelUo shown in Fig. 54, except that a greater number of cells enter into the composition of the blastoderm, a fact probably explained by the greater amount of yolk present in the &gg of Ligia. Certain of the cells of the dark area, i.e., according to my. interpretation the mesendodermal area, have already sunk beneath the surface, and in the center of the area is a distinct depression. Other material of the blastodermic stage which I possess shows but little difference from this, the cells in some of the ova examined being somewhat more numerous, and between this condition and one in which the embryo is distinctly outlined there is a gap. Nusbaum ('93) has represented two later blastodermic stages in his Figs, i and 2, showing in addition to a central thickening two lateral ones, and assumes that the former gives rise to the endoderm and the latter to the mesoderm. Sections which he gives, repeated in the abstract, show that in the region of the lateral thickenings lower layer cells occur, and he believes that they have been produced in situ. For this belief, however, it seems to me he has given no conclusive proof, and it is quite possible that we have to do here, as in other Isopods, with a migration forwards of mesoderm cells, both the endoderm and mesoderm differentiating from the central dark area of the blastoderm and from, this alone. I do not assert that this is so, but the fact of the close resemblance of the two stages I have figured with what I have found in Porcellio and Armadillidiinn suggests the supposition that all the processes of development of Licria resemble those of Porcellio.

4. General Consideration of the Segmentation.

In the ova of the four species whose development has been thoroughly studied we have to do with typical cases of centrolecithal or superficial segmentation, and certain facts have been described which I believe have important bearing upon


some of the problems which have of late been exciting no little attention from embryologists. Before passing on to consider these facts, however, I wish again to emphasize what has already been said as to the syncytial nature of the developing Q.'gg up to the stage at which the cells completely separate from the yolk, since it is to the existence of the syncytium that the interest which attaches to the centrolecithal segmentation as seen in the Isopods is due. In dealing with holoblastic ova in which there is apparently a formation of distinct spherules, embryologists have been too apt to concentrate their attention on the individual spherules, and to regard them as more or less independent units and not as parts of a continuous whole, and it is largely to the assumption of this standpoint that the mosaic theory of development owes its existence. There is, however, an increasing tendency towards the rejection of this theory and towards a return to the view long ago (1867) enunciated by the botanist Hofmeister, according to which the growth of the individual cells is determined by the growth of the entire organism, an idea admirably expressed by DeBary in the aphorism " Die Pflanze bildet Zellen, nicht die Zelle bildet Pflanzen." Rauber ('S3) in an exceedingly suggestive paper extended this idea to animals ; Heitzmann has carried it to its full limits, and to-day it is upheld by such authorities as O. Hertwig ('92), Whitman ('94), and to a certain extent, Wilson ('93).

In developing holoblastic and meroblastic ova it is difficult to demonstrate actual continuity of protoplasm throughout the various spherules, and so far as they are concerned the idea just referred to rests upon a theoretical basis. With centrolecithal ova the case is different, and it is not difficult in the ova of the species I have described to demonstrate the existence of a syncytium. It is not improbable that such a condition will be found in all centrolecithal crustacean ova, though up to the present, so far as I am aware, it has been described by a single author only, namely Samassa ('93) in his paper on the development of the Cladoceran DapJmella. In this form, as well as in the Isopods I have described, a considerable amount of yolk is present, and there is no indication of total cleavage. There


are, however, numerous Crustacean ova in which such a form of cleavage does occur, as, for instance, those of Ljtcifcr (Brooks, '83 ), and between the holoblastic cleavage of this form and a typical centrolecithal cleavage, numerous gradations are to be found (see Korschelt and Heider, '91). The exact character of the cleavage of the ova of the primitive Crustacea is of course a matter of speculation, but the fact that centrolecithal cleavage seems to be characteristic of the Crustacea, occurring in practically all the groups in some grade or other, this fact indicates that cases such as those of Lucifer are secondary. If it be true, then, as I suppose, that all typically centrolecithal ova are syncytia in the early stages of development, are we to believe that with the loss of yolk and the assumption of a holoblastic cleavage, all direct continuity between the spherules is dissolved } Are we to believe that there is no continuity in Lucifer, notwithstanding that in all probability there was continuity in the ova of its ancestors .-'

In this connection the ovum of Peripatus capejisis is of no little interest. It has been pointed out by Sedgwick ('86) that the ova of P. Novae Zealandiae, P. capensis and P. Edivardsii form a series, so far as the amount of yolk which they contain is concerned, the first named having a considerable amount while the last has practically none at all. Now, according to the statements of Miss Sheldon ('88) the ovum of P. Novae Zcalandiae undergoes a segmentation which is essentially centrolecithal and forms a syncytium, while in P. capensis the cleavage approaches the holoblastic form and yet a syncytium again results. The spongy character of the capensis ovum strongly suggests that the ancestors of that form possessed yolk-laden ova, and that the loss of the yolk has been comparatively recent. This loss has not, however, resulted in the dissolution of the continuity of the spherules, and furnishes some support for the supposition that, even in such cases as Lucifer, there may be also a continuity of protoplasm, the separation into distinct spherules being only apparent.

In the ova of many Decapod Crustacea the formation of what are termed yolk pyramids occurs, and it has generally been supposed that the various pyramids were separated by


distinct "segmentation planes." In the Isopods there is no very distinct formation of yolk pyramids, though the appearance of segmentation planes upon the surface of the yolk in the 32-celled stage for instance of Jacra, is the equivalent of the process. Physiologically, the occurrence of the yolk pyramids is exceedingly interesting, but morphologically it has no further significance than is to be found in typical cleavage. I wish, however, to refer here to the appearance presented in section by ova of Jaera in which the yolk cleavage has occurred. In Fig. 20 such a section is represented, and from it the cause of the appearance of the yolk cleavage can be clearly seen. Below the large mes-endoderm cells {MEtt) can be seen a distinct line of protoplasm, also indicated beneath the ectoderm cells {Ec). This line corresponds, when it reaches the surface, with the lines of yolk cleavage, and there is no reasonable room for doubt but that it is the presence of this line which produces the appearance of the cleavage. The line, however, is really a section of a membrane which behaves to reagents in the same way as the general protoplasm and in addition is connected with this by branching protoplasmic filaments. Furthermore, and this is an important point, processes run off from the membrane centrally into the yolk, undoubtedly uniting with similar processes from other cells and producing a protoplasmic network throughout a portion of the yolk outside the limits of the membrane. This membrane cannot, I believe, be regarded as a thin membrane of dead material, i.e., the ordinary conception of a cell wall cannot be applied to it ; it is in reality living protoplasm and consequently it is evident that tJie cleavage of the yolk does itot iiiterjnipt the protoplasmic continuity. Whether or not this idea applies also to cases such as Astacns, in which there is a formation of typical yolk pyramids, remains to be seen.

For some time after the cells have reached the surface of the yolk, their stellate outline indicates the continued existence of the syncytium, but later they appear to round off and actual continuity cannot be demonstrated. This change of appearance can be seen by comparing Fig. 48 or 49 of Porcellio with Fig. 53, which shows a later stage in the development of the


same species. To what extent this apparent discontinuity of the protoplasm of the various cells is a reality, and how far it is of importance for the complete histological differentiation of the cells I do not propose to discuss, but wish to point out that the existence of a syncytiiun is no bar' to a certain amount of differentiation. Thus to confine our attention to Jaera, in which early differentiation is most marked, in the eight-celled stage the cell D has been separated off and, as has been seen, gives rise solely to vitellophags, and at the sixteen-celled stage the ectoderm is thoroughly differentiated from the mes-endoderm. In this stage, however, there is no visible difference in the cells, and that there is a differentiation can only be determined by tracing their future history. In the next stage, however, which, as has been shown, is still a syncytium, histological differentiation is distinct, the vitellophags presenting a very different appearance from the mes-endoderm cells, and these from the ectoderm.

This syncytial differentiation reminds one very forcibly of the differentiation found in the ciliate Infusoria ; it is a differentiation which proceeds independently of the existence of cell boundaries, the force which compels it being at present beyond our ken, and not to be regarded, it seems to me, as resident in the nucleus. The fact of the occurrence of cytoplasmic differentiation in uninuclear Infusoria stands in opposition to any such view, and a phenomenon which has been described on a preceding page as occurring in the early stages of development of Porcellio and Armadillidiitm seems to demonstrate that cytoplasmic differentiation may occur independently of definite nuclear influence. I refer to the remarkable concentration of the peripheral protoplasm which occurs in ova of these forms at the four-celled stage. (See Figs. 41-44.) The region towards which the concentration takes place is that in which later the blastoderm will be developed and at the time of its appearance the nuclei present in the ovum are still some distance from the surface, embedded in the yolk, and only in the 32-celled stage do they pass into the peripheral protoplasm, the cytoplasm which surrounds them fusing with it. We have to do here with a precocious segregation of a portion of the cytoplasm which

I 1 2 MCMURRICH. [Vol. XI.

is to take part in tJic formation of the blastoderm, and this segregation supervenes not in aceordance zvitk any previous loeation of a nueleus, but independently. I do not of course mean to assert that the nuclei may not possess a coordinating or even a trophic action upon the cytoplasm, but that they are directly responsible for the segregation or concentration seems to me an unwarranted assumption. The phenomenon stands apparently in relation to the growth of the entire organism rather than to that of a part of it, and is an instance from the animal kingdom indicating the distinction so forcibly pointed out by Sachs ('87) as existing between growth and cell division.

It has already been pointed out (see p. 74) that in the fourcelled stage of Jaera there is an arrangement of the spherules, apparently comparable with that found in holoblastic ova which undergo a " spiral " cleavage. In later stages, however, the cleavage departs widely from the typical mode of progress of the spiral method, probably in harmony with the precocious differentiation which has already been shown to exist. The Isopod segmentation agrees strictly with none of the three types of cleavage defined by Wilson ('93), though it bears most resemblance, in early stages at least, to the " spiral " method, and the peculiarities of later stages are to be ascribed to the same causes as Wilson proposes in explanation of the bilateral type. In his earlier discussion of the forms of cleavage ('92) Wilson held that the spiral type of cleavage was determined by the spherules having a tendency to arrange themselves along the lines of least resistance, assuming therefore a purely mechanical cause acting from the exterior as an explanation. In his later paper, however, he modifies this extreme view, stating his opinion that " eleavage forms are not determined by mecJianieal conditions alone." Assuming that by " mechanical conditions " he means conditions extrinsic to the ovum, I believe that in Jaera we have practically a demonstration of the correctness of this view.

Let us briefly recall the rearrangement of the protoplasm which obtains \w Jaera. Completely surrounding the exterior of the ovum is a thin layer of peripheral protoplasm, corresponding to the " Kcimhautblastcm " of the Qgg of Insects, and


within this is the yolk scattered in the meshes of a protoplasmic network extending from the peripheral protoplasm to a more or less centrally situated mass containing the nucleus, this central mass and the nucleus alone undergoing cleavage, at least in the early stages. As cleavage proceeds the central masses formed by it separate from one another, remaining connected, however, by strands of the network. It is difficult, therefore, to understand how any extrinsic forces can determine the position assumed by the central masses, and since they are not in contact with one another, but are suspended, as it were, in a protoplasmic reticulum, like knots in a fish-net, it is also difficult to understand how Berthold's ('86) law of minimal contact areas can determine their position. I see no escape from the conclusion that the cleavage form of Jaera is determined entirely by intrinsic conditions, and though we cannot exclude the action of extrinsic forces in holoblastic ova, yet the presumption is allowable that even in these the intrinsic forces are important. In discussing the cause of the direction of cell division, it must be remembered that the forces which primarily determine karyokinesis reside in the cell, since recent observations have strengthened the view which regards the archoplasm and the centrosomes as of the greatest importance in this respect. The generalization of O. Hertwig that the spindle tends to form in the direction of least pressure is one to which there are exceptions, and in centrolecithal ova, such as those of the Isopods, it cannot come into the question. We are left, then, no choice but to refer the vis essentialis which determines the direction of the karyokinetic spindle, and therefore the cleavage form of Jaera, to the constitutio7ial peculiarity of the ovnm.

I do not, however, in the least intend to imply that external conditions, such as pressure, etc., do not influence the direction of the spindle in holoblastic ova ; indeed, the evidence we have indicates that they do ; my remarks refer simply to the ova of Jaera, and those of similarly developing forms. It may be pointed out, however, that the formation of the second polar globules, if not of the first also, is apparently determined by intrinsic forces, and the relation of the first segmentation plane to the point of emergence of the polar globules is also probably

114 MCMURRICH. [Vol. XI.

due to these same forces. We must assume, accordingly, that intrinsic forces reside in all ova, though they may be overshadowed by external influences in some cases. In fact this idea may be carried further, and it may possibly be that the action of external forces may be sufficient to interfere with the arrangement of the cells which would result if they were excluded, and the histological differentiation of the ovum be retarded thereby.

One other point which the study of Isopod cleavage has suggested seems to be worthy of notice here. It is connected with the question of the value of cytogeny as a basis for homology, a question which has already been discussed by Wilson ('92), who points out that " cells having precisely the same origin in the cleavage, occupying the same position in the embryo, and placed under the same mechanical conditions, may nevertheless differ fundamentally in morphological significance." A comparison of the cleavage oi Jaera with that of Aselhis brings support to this view. Thus in the eight-celled stage of the former species (Fig. 9) we find at one pole of the Qgg a cell D, which gives origin to all the vitellophags of later stages ; in the eightcelled stage of Ascllns (Fig. 30) we find an exactly comparable arrangement of the cells, but the cell D has by no means the same morphological significance, inasmuch as it does not contain vitellophag elements alone. Similarly mjaem, in the sixteen-celled stage (Fig. 12), there are six cells. A', c, c', and d\ which give rise to the liver-endoderm and mesoderm of later stages, while in Aselhis it will be seen that the same cells, the same, that is, so far as their cytogeny is concerned, are entirely ectodermal. Indeed the difference between the two species may be expressed by stating that the differentiation of .the germ-layers takes place practically at one stage later in the cleavage in Asellus than in Jaej'a.

The significance of this is by no means certain, though it is interesting to note that an explanation of it may be found in the different size of the two ova. Thus the egg of Asellus is several times the size of that oijaera, and it maybe that the extra amount of yolk acts in retarding the differentiation, or, in other words, a certain spatial relationship of the blastula cells


may be necessary before differentiation can take place, this relationship, on account of the greater quantity of yolk in Asellus, supervening later in that form than in Jaera. This suggestion is tempting, but it must be noted that a further retardation of the differentiation does not take place in Arniadillidiiim or Porcellio, whose ova are again several times the size of those of Asellus, and yet the differentiation is identical in time with that of Asellus.

Part III. — The Later Development of the Germ-layers.

I . The Later History of the Mesoderm.

As has been already pointed out, two distinct regions are to be recognized in the Isopod embryo, an anterior one corresponding to the naupliar region and a posterior or meta-naupliar region. In these two regions one finds a very different behavior of the mesoderm. In very young embryos only the naupliar region is represented, the blastoporic region lying immediately behind it, the front edge of the blastopore being formed by a row of ectoderm cells, in some cases at least, eleven in number, while the ectoderm cells in front of this are arranged more or less irregularly, though the orthogonal curves described by Reichenbach ('86) in Astaciis are more or less plainly visible.

In the earliest stages this ectoderm rests directly upon the yolk, there being no trace of mesoderm or ectoderm below it. As described in preceding pages, the cells of the blastopore area, the mes-endoderm and the vitellophags, immigrate and multiply, forming a plug projecting into the yolk, and later they scatter, principally forward, so that at this stage numerous cells begin to be found beneath the naupliar ectoderm. In none of the forms studied is there to be observed any difference in form or appearance between the mesoderm and the endoderm cells at this stage ; in Jaera the vitellophags are early distinguishable from the other mes-endoderm cells, but in other forms they do not become distinguishable until the beginning of the scattering of the mes-endoderm plug when


they assume their characteristic function. For convenience, I distinguish here these vitellophag cells from the remaining elements of the mes-endoderm ; their significance will be discussed in another section of this part of the paper.

The mes-endoderm cells do not remain irregularly scattered under the naupliar ectoderm very long, but one soon finds them arranging themselves in two bands which diverge as they are traced forward, and extend from what was originally the blastopore region to beneath the ocular lobes, forming what have been termed the mesoderm bands, and lying beneath the regions of the ectoderm in which the cells of that layer are somewhat concentrated, as shown in Fig. 25 for Jaera and in Fig. 38 for Asellus. In addition in Jaera a transverse band extends across from about the anterior end of one mesoderm band to the other, so that occupying the central part of the naupliar region of the embryo there is a triangular area in which there are practically no mesoderm cells (see Fig. 25) and in the anterior portion of which, it may here be stated, the stomodaeal invagination will take place. I have not been able to follow the development of these mesoderm bands in Porcellio or Arinadillidmm, but in these as well as in the other forms studied their existence is clearly indicated in later stages when the naupliar limbs begin to form.

When these stages are reached, the mesoderm cells are found to have arranged themselves into groups, which, as the limbs grow out, migrate into their interior, and multiplying there form solid mesodermal axes for them. Other mesodermal cells are scattered beneath the Anlagen of the nervous system and beneath the region where the stomodaeal invagination is preparing, but no special mesodermal groups were found immediately external to the points of origin of the naupliar limbs. At this stage a differentiation of endoderm cells appears owing to the formation of a special group of cells on either side at about the level of the first maxillae, these cells forming the Anlage of the liver lobes. They early arrange themselves in the form of a hollow sphere, which is not complete, however, being open towards the yolk, which the cells proceed to digest. I may say here that I have found


no distinction of a secondary mesoderm from a primitive one at this stage, though later, I believe, cells homologous with Reichenbach's secondary mesoderm are to be found.

The later development of the naupliar mesoderm I have not attempted to follow, and shall turn now to a consideration of what takes place in the meta-naupliar region.

At the time of the scattering of the mes-endoderm cells the ectodermal teloblasts begin to divide in the manner described and grow backwards over the blastoporic region, the entire meta-naupliar ectoderm on the ventral surface being produced by this teloblastic growth. In embryos which are removed from the yolk, it can readily be seen that the meta-naupliar mesoderm, as well as the ectoderm, is produced by a teloblastic growth, though the rhythm of division of the mesodermal teloblasts and their arrangement is very different from that of the ectoblasts. I have succeeded in making out these mesoblasts in Porcellio, Ligia, and CymotJioa with great distinctness, and sections of Jaera and Aselliis show that essentially the same arrangement of them occurs in these forms.

In Fig. 62 is represented a preparation of Ligia, which shows the arrangement occurring in that species, and I have chosen Ligia for this purpose on account of the discrepancies which my preparations show when compared with the figures given by Nusbaum ('93, Fig. 6). Preparations exactly similar in all essentials have been obtained from CymotJioa and Poixcllio, and also by Bergh from Mysis, a fact which strongly suggests that Nusbaum's observations are somewhat incorrect. Lying beneath the ectoblasts, and slightly behind them, are eight mesoblasts arranged in a very definite manner. One is situated on each side of the median line of the embryo; at a slight distance from these on each side are three others, which are separated from one another by intervals shorter than that which separates the innermost one from the cell which lies near the middle line. Immediately in front of each one of these mesoblasts is to be seen a single cell, the products of their last division, and further forward are to be seen three other transverse rows, each consisting also of eight cells, arranged as in the two posterior rows. In the sixth row,


counting from behind forward, the two cells on either side of the median line remain as before, but in the three lateral rows of either side a division of the original cells has occurred, there being on one side six cells arranged in two parallel rows, while on the other side the middle cell has not yet divided, so that only five cells are seen.

In the next row, or since the various transverse rows have evidently a segmental significance, in the next segment {aV), the multiplication of the mesoderm cells has progressed still further, there being now four cells in a single row representing the original two cells on either side of the median line, while the arrangement indicated for the lateral cells in the segment next behind is complete, and two parallel rows consisting of three cells each are to be found. In the next segment anteriorly (///") there are still four median cells, but the lateral rows are again in division, the division apparently affecting the anterior of the two rows already present and leading to the formation of three parallel rows each consisting of three cells. In this segment there is to be seen the first indications of the budding out of the limbs, these structures becoming more pronounced as one passes forward, and at the same time the multiplication of the mesoderm cells progresses rapidly, so that in the jDreparation at present being described it is impossible to trace any regularity in the division of these cells. In the most anterior of the segments figured, however {tJ?), as well as in the two which succeed it five masses of mesoderm cells are to be distinguished: (i) a mass occupying the median line of the segment and unquestionably derived from the two mesoblasts situated on either side of the median line, (2) a mass on either side which corresponds to the limb bud, below which it l.ies, and (3) a mass, also paired, which lies just lateral to each limb bud.

The regularity of the division of the mesoderm cells is shown more satisfactorily in Fig. 63, which represents the posterior extremity of a very slightly younger embryo of CymotJioa. Posteriorly the cells were not perfectly distinct, some of them apparently having remained adherent to the yolk when the embryo was removed from it. In the fourth segment, counting


from behind forwards, one of the median cells is in process of division, and in the fifth segment its division is complete and its fellow is in process of division. In the sixth segment there are four median cells lying all on the same level, an arrangement which indicates a certain amount of migration of the daughter cells, since the division is oblique as is shown in the two segments lying behind. In this sixth segment also two rows of lateral mesoderm cells are found on each side, each row consisting of three cells. The only difference seen from what has been described in Li^ia is that the median cells beofin to divide before the lateral ones, while the reverse seems to be the case in Ligia. In the seventh segment the median cells are again in division, each of the four dividing off a cell anteriorly, the eight cells so formed arranging themselves later somewhat irregularly, as may be seen from the eighth segment, which also shows that there are nine cells in each of the lateral masses, arranged in three rows of three cells each and probably formed in the manner indicated in the eighth segment of the figure of Ligia.

The regularity of the entire process of growth of the metanaupliar region of the Isopods is most remarkable, and the more one studies it the greater is the wonder it excites. The regular rows of ectoderm and mesoderm cells are wonderful in themselves, and when there is added a more or less definite number of rows for all the species, we see that we are dealing with laws of growth which are at present far beyond our powers of explanation. It is true that the number of ectodermal teloblasts is not always quite constant, though approximately so, but it is exceedingly interesting to find that where, as in Aselliis, they can be traced back to their earliest differentiation, there is a definite number of them, namely eleven. And this definiteness of number is not confined to the Isopods, but is found also in Mysis (Bergh, '93). As regards the mesoblast, however, the number is more constant, eight and eight only occurring in Cymotkoa, Ligia, and Porcellio ; and again we find exactly the same number in Mysis. Nusbaum ('93) in the figure he gives of their arrangement in Ligia, as well as in that he gives of Oniscus, shows neither the constancy of number nor the regularity of the

1 20 MCMURRICH. [Vol. XI.

earlier divisions which I have found, and the harmony of my observations with those of Patten ('90) on CymotJioa and those of Bergh on My sis indicate that his observations on this point have not been conducted as carefully as might be desirable. As is well known, Patten ('90) was the first to describe the mesoblasts and their arrangement in the Isopods, and my results agree essentially with his, except that I have not been able to discover the syncytial connections which he figures as existing in the younger rows. I am not prepared to say that the connecting strands of protoplasm do not exist, indeed, I see no reason why they should not, but in none of the numerous preparations I have made of CyviotJioa have I been able to discover them.

The relation of the ectodermal and mesodermal transverse rows to the segments of the meta-naupliar region is of some interest, and I have been able to determine that eacJi roiv of inesode7'7n cells is eqiiivalent to a segment. Thus, in Fig. 62, of Li^ia, it is clear from the formation of the limbs and the relation of the mesoderm masses to these structures that the mesoderm masses are segmental, and it can also be seen that the masses have been produced by the multiplication of a single transverse row of eight mesoderm cells. In fact the most anterior of the segments represented in the figure is the third thoracic, and the last in which the limb rudiment is visible is the seventh thoracic. Following upon this are seven transverse rows of mesodermal teloblasts which correspond to the six abdominal segments and the telson, the last row of mesoderm cells which, it may be noted, may be regarded as the original mesoblasts, giving rise to the mesoderm of the telson, just as in the annelids the mesoblasts in all probability give rise to the mesoderm of the anal segment.

Does each transverse ectodermal row also represent a segment .-* This is a question more difficult to answer than when the mesodermal rows were concerned, but I believe that the preparation represented in Fig. 64 indicates what the relation is. This preparation is from a younger embryo of a Cy77iotJioa than that represented in Fig. 63, and though it is difficult to be sure whether or not all the mesodermal segments are complete, yet I have reason to believe that they are. This point is


not, however, essential for our present purpose. What is of immediate concern is the relation of the mesodermal and ectodermal transverse rows. The anterior four rows of mesoderm cells are each separated by an interval corresponding to an ectodermal row, and each corresponds to an ectodermal row, while the row immediately in front of the mesoblast corresponds practically to the row of ectoderm cells immediately in front of the ectoblasts. Certain of the ectoblasts are, however, in process of division, and it is clear that this division will result in the interpolation of a row of ectoderm cells between the ectoblasts and the row above which the penultimate row of mesoderm cells lies, so that eventually the same relations of the mesodermal and ectodermal rows as exist more anteriorly will be brought about. The conclusion is, then, that two rows of ectodermal cells primarily correspond to each segment and that therefore the rhythm of division of the ectoblasts and inesoblasts is not the same, the ectoblasts dividing tzvice as rapidly as the mesoblasts. It may be pointed out that while the figure which supports this conclusion is slightly diagrammatic, yet nevertheless the relations of the ectodermal and mesodermal cells were carefully indicated by means of the camera, and the figure therefore represents these relations accurately. It is hardly necessary to state that the occurrence of a greater number of ectodermal rows between each of the mesoderm rows in older segments is due to the division of the cells composing the rows ; the numerical relation just described is to be seen only in segments which have been but recently formed.

Not only is there this definiteness in the number of cells concerned in the formation of each segment, but the number of the segments themselves is definite throughout the entire group of the Isopods. As is well known, there are seven thoracic and six abdominal appendages (not counting the telson) in the Isopods, and if to these be added the segment which bears the maxillipeds and the two maxillary segments, we have in the meta-naupliar region of the Isopods sixteen segments. Since what may be regarded as the primary ectoblasts and mesoblasts are finally located in the telson, it is clear that diwing development the mesoblasts mnst divide ex

122 MCMURRICH. [Vol. X I .

actly sixteen times mid the cctoblasts tJiirty-two or probably iliirty-tJiree timeSy before tJiey relijiqiiisJi their teloblastic mode of division.

What determines the cessation of the teloblastic mode of division is a puzzle, but it is accompanied by very decided changes in the appearance of the ectodermal teloblasts. This difference is shown in Figs. 63 and 64. In the latter the cctoblasts are very large and readily distinguishable, while in the former this distinction is entirely lost and they have the same size as the remaining cells of the ectodermal rows. Bergh ('93) has shown that in My sis this is accompanied by a change of position of the spindle in the cell ; so long as the mode of division is teloblastic the equatorial plate lies in front of the middle of the cell, but when it ceases the plate lies at the middle and the cell divides into two equal parts. This stage I have not succeeded in finding, and can only state that a reduction in size of the teloblasts takes place, and that they, together with the mesoblasts and the ectoderm cells lying behind them (the anus making its appearance in the center of these cells), form the telson.

In a transverse section through one of the anterior thoracic segments of a CymotJioa embryo (Fig. 65) in which the limbs {th) are represented only by ectodermal thickenings, one finds immediately below each of the nerve ganglia {N) a mass of mesoderm cells {cM) which are evidently the result of the division of two mesoderm cells which were budded off from the mesoblasts which lie on each side of the middle line. Peripherally to these are found some scattered mesoderm cells {mM) which represent that portion of the mesoderm which will give rise to the mesodermal tissues of the limb ; in the next sections these are somewhat more abundant, the similarity to Fig. 62 being thus more pronounced than in the section figured. Still more peripherally is to be seen a compact mass of mesoderm {IM) which corresponds to the lateral masses seen in the more anterior segments of Fig. 62. In a later stage the arrangement of the mesoderm is very similar. Fig. 66 shows a section through the first thoracic appendage of such a stage, and from it it will be seen that the limbs have be


come separated by a considerable distance from the nerve ganglia, a separation which increases markedly in the succeeding stages. Beneath the point of origin of each limb is to be seen a portion of the limb mesoderm, and in the transverse section of the maxilliped a core of mesoderm occurs which has evidently wandered into the limb as it grew out from the body. Laterally to the limb mesoderm on each side, is a portion of the lateral mass of the segment, still distinct, and the only marked difference from the section shown in Fig. 65 is the absence of the more median mesoderm masses, which are represented only by a few scattered cells situated on either side of the nervous cord. This is perhaps to be explained by the separation of the limbs from the median line which has already been mentioned, the mesoderm having scattered so as to cover the greater surface now developed.

As to the ultimate fate of the various portions of the mesoderm, it may be stated definitely that the limb mesoderm and the mesoderm of the lateral masses become converted almost entirely into muscle tissue, though the possibility of a certain amount of connective tissue being formed from them cannot be excluded. The fate of the median masses could not be followed, however, though I am inclined to think that they form connective tissue rather than muscle fibers. Consequently I have refrained from applying Bergh's ('93) term of myoblasts to the mesodermal teloblasts of the meta-naupliar region, since, though applicable to the three lateral mesoblasts of each side, it is questionable whether it is as properly applied in the case of the two median cells.

In his preliminary paper, Nusbaum ('91) describes with a figure an embryo of Ligia, in which the abdominal limbs are not yet formed, and calls attention to the existence of a transitory exopodite in the thoracic appendages. This observation I can confirm, and may add that the exopodites may also be seen in embryos of other Isopods, though less distinctly than in Ligia. Nusbaum also calls attention to a patch of cells in each segment situated just laterad of each limb rudiment, describing them as ectodermal thickenings and homologizing them with epipodites. These patches are readily discernible in the

I 2 4 MCMURRICH. [Vol. X I .

embryos of all the Isopods I have studied, but I must take exception to Nusbaum's interpretation of them. In preparations of entire embryos, such as he figures, they might readily be taken for ectodermal thickenings, but sections show at once that they are really the lateral mesodermal masses shown in my Figs. 65 and 66 {IM). An homology of these structures with epipodites seems out of the question ; they do not stand in relation to the limb, but are situated in the pleura when these are developed, and, as stated, give rise to the lateral muscles of the body.

So far nothing has been said regarding the origin of the heart, practically all the teloblastic mesoderm going to form muscle-tissue. Inasmuch as the heart formation concerns the vitellophag cells, a description of its origin will be postponed to the next part of the paper, in which the fate and significance of the vitellophag cells will be discussed.

2. The Formation of the Digestive Tract and the Later History

of the Vite Hop hags.

It has been stated in an earlier part of the paper that what are probably to be considered endoderm cells are distinguishable in an early stage oi faera and perhaps also of Asellns, but they are recognizable only by their position and not by any histological peculiarities. In the Oniscidas studied they could not be determined, the entire aggregate of mesodermal cells being identical in appearance. Even in the Asellids they become indistinguishable in the later stages when the scattering of the mes-endoderm takes place, and it has not been possible to trace their later history and their conversion into the liver-lobes, which is, I believe, their fate.

However that may be, the liver lobes first make their appearance as a mass of cells on either side at about the level of the first maxillary segment, and the cells of each lobe early arrange themselves in an epithelium, forming a more or less spherical body which is open towards the yolk, a certain amount of which is enclosed by them and apparently undergoes within them a digestion. At an early stage in Cymothoa the spherical


sac becomes drawn out on its posterior surface into three fingerlike processes, while in the Oniscidas and Asellidae only two processes are developed, and appear at a somewhat later stage of development. These processes give rise to the liver-lobes of the adult, continuing to increase in length with the growth of the embryo. For a considerable time the liver-lobes lie free in the yolk, eventually uniting anteriorly with the posterior end of the stomodaeal invagination, by which the anterior portion of the digestive tract, as far back as the posterior extremity of the stomach, is formed, while the intestine is formed almost completely by the proctodaeal invagination.

In other words I find, as Bobretzky ('74) found in the case of Oniscus and as Nusbaum ('93) also finds, that the digestive tract is formed almost altogether from the two ectodermal invaginations, the mesenteron being represented principally by the liver-lobes, only a very small portion of the intestine, just where the liver-lobes unite with it, being possibly endodermal.

The stomodaeal invagination makes its appearance relatively early, and when first clearly distinguishable lies almost on a level with the rudiments of the antennules, if anything a little behind them, slightly later (Figs. 69 and 70 st) being distinctly behind them, though in front of the antennae. The invagination presses deeper and deeper into the yolk, and at its posterior extremity enlarges to form the stomach, the liver-lobes, as already stated, uniting with it at its posterior extremity. The proctodaeum appears slightly later than the stomodaeum and when first observable appears as a patch of cells lying a short distance behind the teloblasts (Figs. 62 and 63, a) and therefore having the same relation to these structures as had the blastopore in a much earlier stage, though from the mode of growth of the teloblasts backwards over the blastoporic cells, there can be no identity of these latter with the proctodaeal cells. The invagination at first forces its way through the yolk, but tends towards the dorsal surface of it where it lies in the more anterior abdominal segments. It extends far enough forward to unite or almost unite with the stomodaeum, a very small amount of endoderm, as already stated, possibly being interposed between the two invaginations.

126 MCMURRICH. [Vol. XI.

Such being the mode of formation of the digestive tract, what is the ultimate fate of the vitellophags, and which of the three germinal layers do they represent }

At the time of the migration of the blastopore cells the vitellophags sink into the yolk, which divides up into masses as has been so frequently described in other Crustacea, there being, however, no formation of secondary yolk pyramids such as occur in Astacus. The yolk persists for a long time, even up to hatching in the thoracic region of the body ; but it disappears in slightly earlier stages in the abdominal region and here phenomena which accompany its disappearance may be studied. In Fig. G'j is represented a transverse section through the anterior portion of the abdomen of an embryo of Cymot/ioa in which the yolk is beginning to undergo disintegration. Towards the ventral surface on either side is to be seen one of the lateral mesoderm masses whose cells are beginning to be converted into muscle tissue {mu), and in their vicinity are to be seen a number of scattered cells, whose origin is I believe indicated on the left side of the figure. Here at the sides above the lateral mesoderm mass are a number of vitellophags still scattered through the yolk, which throughout the whole section has broken up into small and somewhat separated particles, a preliminary to its breaking down into minute granular particles such as are seen in the upper left-hand portion of the figure {dy). As this ultimate disintegration occurs the vitellophags are set free and form the scattered cells already alluded to. This setting free of the vitellophags is not confined to the lateral regions of the body, but is to be seen also towards the dorsal surface. Here one finds in the section figured in the middle line the proctodaeum {pr) and on either side of it. is a row of cells which correspond to Nusbaum's cardioblasts. Their situation in the yolk seems to indicate that they may represent vitellophags, though I have not been able to trace either their origin or their ultimate fate. Above the proctodaeum and on either side are to be found numerous other cells, some of which and probably all are freed vitellophags, and various stages in their aggregation and separation from the yolk may be found by tracing the series of sections forwards.


Some of these cells are becoming muscle cells {imi) and will eventually produce the longitudinal dorsal muscles of the abdominal region, while others do not become thus transformed, but retain their original character in much later stages. Whether or not they become blood corpuscles I cannot certainly state, though from what I have seen in sections through later stages I am inclined to believe that that is, in part at all events, their fate.

In the oldest embryos of CymotJioa which I possess the organs are all fully formed, and there is no trace of yolk in its original condition to be found. In slightly earlier stages the digestive tract is complete, and there is still a certain amount of yolk present in the thoracic region of the body, and within it vitellophags are to be seen, and there does not seem to be any possibility that these cells enter into the formation of the digestive tract ; in fact I have seen no indications that any of the vitellophags do so. In the latest embryos (Fig. 68) the spaces between the various organs are occupied by a granular mass {dy) which resembles in appearance the granular disintegrated yolk seen in the dorsal region of Fig. ^J {dy), and I believe this matter is of the same nature and represents the final condition of the yolk. It occupies the place of the blood plasma, and has imbedded in it a certain number of amoeboid cells (vi), the blood corpuscles, which are, there is every reason to believe, persistent vitellophags. This result agrees perfectly with the observations of Nusbaum on Mysis i^on), and I can confirm for CymotJioa the statements contained in the last two paragraphs of page 192 of his paper.

Not all the vitellophag cells take part, however, in the formation of persistent tissues. Some appear to break down, being enclosed within the liver lobes, and others seem to disintegrate independently of the action of the liver, giving rise to particles of chromatin such as may be seen in Fig. 6^ {Cr) scattered among the cells at the surface of the yolk, and resembling the chromatin nebulae, which have been described by various students of Decapod embryology. So far as I have seen, however, such disintegrating cells are few in the Isopods, and are to be found only in the latest stages of embryonic development.

128 MCMURRICH. [Vol. XI.

My belief, then, with regard to the vitellophags, is that they take no part in the formation of the digestive tract, that some of them disintegrate and disappear, but that the majority take part in the formation of persistent structures, such as connective tissue, muscle tissue, blood corpuscles, and perhaps even of the heart itself. These are what are generally recognized as mesodermal structures, and I do not believe that the vitellophags should be regarded as being anything but mesoderm,

3. General Considerations on the Formatioji of the Germ-layers

in the Crustacea.

I wish, in the first place, to emphasize once more the distinction existing between the mode of formation of the naupliar and meta-naupliar regions of the Isopods, a distinction already pointed out by Bergh ('93) as obtaining in Mysis. The latter portion, so far as its ectoderm and the greater bulk of its mesoderm are concerned, is produced by a teloblastic growth, while in the naupliar region no such general method of growth is pronounced. If the egg-embryo proper be regarded as consisting of the blastopore region and the portion of the embryo immediately in front of this, then the meta-naupliar region may be regarded as a portion of the body normally developed after hatching, but in the Isopods developed, in accordance, probably, with the occurrence of a brood-pouch, before the embryo begins to lead a free life. In other words, the development of the Isopods points back to a period when a free-swimming N'anplins occurred in the development of the ancestors of the group, the egg-embryo being a Naupliiis, if one may so express it, just as it is in the Penens, for example, in which the post-mandi'bular segments develop only after hatching. In the Annelida a teloblastic mode of growth has been described, more especially in connection with the mesoderm, and if the development of such a form as Polygordins be taken as a type of the larval method of development in the Annelids, an interesting comparison may be made. Thus in Polygordius the Trochophore may be regarded as the egg-embryo proper, the addition of new segments being associated with a teloblastic growth of the mesoderm,


and consequently the same relations obtain between the trochophoral and meta-trochophoral regions of Polygordiiis as have been noted as obtaining between the naupliar and meta-naupliar regions of the Crustacea as represented by Mysis and the Isopoda.

Whether these similar relations indicate an homology of the two regions in the two groups or not I am not prepared to state. If the teloblastic growth is homologous in the two groups, that is to say, if it has been derived from a common ancestor, then it follows that the trochophore and the Naitpliiis are homologous, but, on the other hand, there is no evidence that this is the case ; indeed, the marked dissimilarity in the details of the arrangement of the teloblasts points strongly against any such assumption, and suggests that the teloblastic mode of growth has been developed independently in the two groups, and is to be regarded merely as a provision for rapid growth. The fact that even certain of the organs, such as the nerve ganglia, also show teloblastic growth in the Crustacea and Insects (Wheeler, '91), lends support to this view of the question.

I have indicated that my results as to the formation of the germ-layers agree essentially with those of Bergh on Mysis. In one important particular, however, there is a difference in our interpretations of the phenomena. As I understand Bergh's statements, the entire mass of blastoporic cells becomes converted into vitellophags, endoderm, and eight mesoblasts, and the conclusion is that the mesoderm of the naupliar region is produced by the mesodermal teloblasts. This is certainly not the case in the Isopods, in which, as I have stated, a considerable number of the blastoporic cells give rise to mesoderm cells, independent of the mesodermal vitellophags, and teloblastic growth is found in connection with neither the mesoderm nor the ectoderm in the naupliar region of the body. We may consider the naupliar region to be the embryo proper, and the blastopore being an embryonic structure gives rise to the embryonic mesodermal and endodermal tissues, making provision, however, by the formation of the mesodermal teloblasts for the rapid growth of the later larval mesoderm.

130 MCMURRICH. ■ [Vol. XI.

So many accounts of the formation of the germ-layers in the Crustacea have been published, and the various accounts have been so frequently brought together in resiim^, that I may be pardoned for refraining from entering into a lengthy, critical review of the Avorks of my predecessors in this line. It may be pointed out, however, that the various accounts may be divided into two groups: (i) Those which derive all the mesodermal and endodermal cells primarily from the blastoporic cells, and (2) those which assign a portion at least of the mesendodermal formation to extra-blastoporic regions. My results upon the Isopods belong to the first group, and are in harmony, in this respect, with those of the authors, such as Grobben ('79 and '81), Samassa ('93), and Brauer ('92), who have studied the formation of the germ-layers in the lower Crustacea, as well as with those of Bobretzky ('74) on OniscnSy of Bergh ('93) on My sis, of Brooks ('83) on Lucifer, and of Reichenbach (86) on Astaciis, not to mention other students of the Decapods. In the last forms, however, several authors, as for example, Kingsley ('87), Ishikawa ('85), have described the occurrence of cells imbedded in the yolk before the differentiation of the blastoporic cells, a condition probably due to the migration or delamination of some of the blastula cells at an early stage, as has been described by Herrick ('92). This is a phenomenon apparently peculiar to the Decapods, and I do not propose to discuss its significance here; with the exception of this peculiarity the mes-endoderm formation of the Decapods is localized in the so-called blastoporic region.

To the second group belong the observations of several authors whose results require more detailed notice. Nusbaum has described in Mysis iizi) and Ligia ('91) the formation of mesoderm from the ectodermal cells along the entire length of each of the lateral bands of the naupliar region of the embryo, and Lebedinski ('90) holds essentially the same view as to the origin of the mesoderm in EripJiya. As regards Mysis Bergh has given a very positive statement as to the incorrectness of Nusbaum's observations on this point, and as already stated there seems to me to be no evidence in the Isopod that such a mode of formation of the mesoderm obtains. Nusbaum evi


dently has not carried his observations far enough back to observe the scattering forward of the mesoderm from the blastopore, and finding mesoderm cells below the lateral ventral bands has imagined that they have split off from them. It remains to be seen whether further observations on the Brachyura will confirm Lebedinski's statements ; it may be noted, however, that the recent work of Cano ('93) on Maja does not afford any support to them.

In this connection mention may be made of Weldon's ('92) observations on Crangon, in which he finds a patch of mesoderm cells on either side of what he takes to be the blastopore and beneath apparently the posterior ends of the lateral embryonic bands. I have observed the same arrangement in Palaemonetes and VirbiiLS but cannot agree with the interpretation Weldon puts upon it. He regards what he has found as a partial confirmation of Nusbaum's views, but in reality the conditions are quite different, since Nusbaum derives the mesoderm from the ectoderm of the ventral surface of the naupliar region, while, even granting that Weldon's views are correct, it is formed from the meta-naupliar region in Crangon, in my opinion a very important difference. But, in addition, Weldon identifies what he terms the ventral neuro-muscular plates of his Fig. 6 with the thoracico-abdominal plates of Astaciis, and this is where I believe he has fallen into error. As I interpret the similar appearance in Virbuis these neuro-muscular plates are in reality part of the blastopore which is elongated laterally, and they are entirely composed of mesoderm cells, not yet being covered in by ectoderm, a view which, it seems to me, is corroborated by the appearance which is seen in section and which Weldon represents in his Fig. 16. His Fig. 6 is not at all comparable, as he supposes, to Reichenbach's Fig. 3, but is of a much earlier stage. Why the mesoderm should arise as two lateral masses rather than as a single mass situated immediately in front of the vitellophag-endoderm mass, may perhaps be explained by the separation of the two lateral ventral bands in early stages.

The remaining observations which belong to the second group of results require but little discussion. The views of

12,2 MCMURRICH. [Vol. XI.

Reinhard ('8?) and Roule ('9i, '92, '92 a) have already been stated (p. 96), and it is very clear that they receive not the slightest confirmation from the forms I have studied, and are at variance with the results of the other observers who have studied the embryology of the Isopods. As regards the Amphipods much has yet to be done before a proper idea of their early development is obtained; the results of Dr. Sophie Pereyaslawzew ('88) and her co-workers Mesdames Rossiiskaya - Koschewnikowa and Wagner, are evidently quite inadequate, the mesoderm being stated to arise in connection with the limbs, evidently not having been traced even approximately to its origin. The results of Delia Valle ('93) though much more carefully worked out, still leave much for later investigation, it being impossible to harmonize them with what occurs in other Crustacea.

There are reasons then for doubting the correctness of the views of those authors who ascribe an extra-blastoporic origin for the mesendodermal elements in the Crustacea except in the case of the Decapods, and since these are the most highly specialized of all the Crustacea, the precocious formation of vitellophags which they show may with justice be regarded as a secondary phenomenon. It remains to consider what is to be regarded as the blastopore in the Crustacea, and what the significance of the various phenomena which have been described in connection with it.

One naturally turns to the simpler forms to get an idea as to the primitive character of the blastopore, and in the Phyllopods one is at once met by two striking facts, (i) the mesoderm and endoderm have a common origin and cannot at the time of their formation be distinguished from one another, and (2) they arise by the immigration of certain cells into what would be the blastocoel were the yolk not present, there being no indication of an invagination. These two facts are true of Daphnia and DapJmella (Samassa, '93) and of Branchipus (Brauer, -92), and according to Samassa's account of Moina also, though Grobbcn (-79) describes for this form an invagination, and also an early differentiation not only of the mesoderm from the endoderm, but also of two distinct portions of the mesoderm.


These results, however, since Samassa's observations, stand in need of confirmation before they can be accepted. In CctocJiilus Grobben ('8l) describes an invagination of the endoderm and a distinct differentiation of the mesoderm from this, but in description after description of Crustacean development one reads of a solid endodermal plug, a condition which seems to be of the most frequent occurrence.

However, not unfrequently it is possible to distinguish a differentiation of the blastoporic cells, if the term blastopore can be applied here, and in this respect the phenomena described in preceding pages are exceedingly interesting. In CyniotJioa, Ligia, and the other Oniscidas, and practically in Asellus, no differentiation can be made out, but mjaera a wellmarked differentiation of a portion of the mesoderm, namely of the vitellophags, occurs. The differentiation of the endoderm can be disregarded, since it is one of position only and does not persist, so that we have in that form an anterior mesendodermal mass and a posterior mesodermal group of vitellophags. Such a differentiation is peculiar ; it is not a differentiation of endoderm and mesoderm, but a specialization of a certain part of the mesoderm from the remainder, and it is interesting as indicating a process of precocious segregation which may be carried to extreme lengths, and to which the mesoderm, on account of the multiplicity of organs which arise from it, is especially susceptible. Thus the remarkable differentiations of teloblast cells which are found in such cases as Ltunbricns (Wilson, '89) and Clepsine (Whitman, '78) are to be regarded as cases of this kind. From the general ectoderm are differentiated two teloblasts which give rise to the ventral nerve cord ; and from the general mesoderm are differentiated the nephroblasts. We find, too, the interesting peculiarity that the nephroblasts pass into the blastocoel at a later period than do the mesoblasts, and the conclusion has been drawn that the nephridia therefore are of ectodermal origin in these cases. Is there any necessity for such a conclusion } It seems to me that the phenomenon is simply the culmination of the process of differentiation of portions of the germ-layer, the segregation of the nephroblasts from the mesoblasts having proceeded so

134 MCMURRICH. [Vol. XI.

far that they act independently of them. In fact we see this same phenomenon in various degrees of perfection in the differentiation of the mesoderm from the endoderm, for phylogenetically the mesoderm is derived from the endoderm, as may be seen for instance in the group of the Turbellaria and as is indicated in the ontogeny of many forms. In some cases, e.g., the Echinoderms, the two layers separate perfectly only after they have passed into the blastocoel ; in other cases they are differentiated from one another in the blastula and pass inward together, while in other cases again the migration or invagination of the endoderm may precede that of the mesoderm, or the mesoderm, either in whole or in part, may pass inward before the endoderm. To speak of the origin of the mesoderm in one case from the ectoderm and in another from the endoderm shows a want of appreciation of the phylogenetic significance of the mesoderm and of the possibilities which result from differentiation.

Primarily, then, in the Crustacea there was the formation of a blastula whose cavity was more or less completely filled with food yolk, and there was a differentiation of mesendodermal cells by immigration, the distinction of mesoderm from endoderm only supervening later. I do not mean to assert that the more remote ancestors of the Crustacea may not have shown an invagination of the epibolic or even of the embolic type ; in reality the distinction between invagination and immigration is so slight that one may readily become converted into the other, and the only question of interest in connection with the two phenomena is as to which is the most primitive phylogenetically. I see no reason to withdraw from the position I have taken on this question in earlier papers, in which I announced myself as a supporter of Metschnikoff, and do not intend to discuss the question here. I believe, firmly, however, that the cases of invagination to be found within the Crustacea are secondary.

If these cases be examined, some interesting relations are to be found, relations which involve a correct appreciation of the significance of the vitellophag cells. I have already stated my opinion that these structures are to be regarded as mesoderm


cells, and may point out that this view is in harmony with those of Samassa ('93) and Nusbaum ('87). Most authors who have studied the development of the Decapods have considered the vitellophags to be endoderm, though Herrick ('92) comes to the conclusion that some of them at least give rise to mesodermal structures. When we come to examine critically the reasons for the belief that they are endodermal, one finds that they do not rest on direct observation, but rather upon analogy with what is supposed to take place in Astac7ts. Here we have an invagination lying behind the region from which the ordinary mesoderm arises, and this is represented in the majority of other Decapods by a plug of cells which give rise to the vitellophags, and may in certain cases show indications of a cavity, as in Crangon for instance (Weldon, '92). If, then, the invagination of Astaciis gives rise to the mesenteron, as it is supposed to do, what more natural than to suppose that the vitellophags assist in the formation of the endoderm in other forms ? I know of no case, however, in which they have been actually traced to such a final destination, and the whole question comes down to the significance of the invagination of Astacits, granting the homology of the vitellophags with the cells which form the secondary yolk-pyramids of that form. Let us examine, then, the relations in Astaciis, as shown in Reichenbach's beautiful monograph ('86).

It seems to me that we have to distinguish, according to Reichenbach's descriptions, two sets of elements in the invagination, namely, the cells which absorb the yolk and form the secondary yolk-pyramids, and which, for exact comparison, we may here call the vitellophag cells, and certain cells which do not absorb the yolk, and which form what Reichenbach terms the entoderm plate. From the entoderm plate the liver-lobes arise, and it takes part, also, in the formation of a part of the mesenteron. The important question is how much of the mesenteron is derived from it, and what evidence is there that the yolk-pyramids contribute to the formation of the mesenteron. In the oldest stage which Reichenbach figures a considerable amount of yolk is still present. The stomodaeal and proctodaeal invaginations are well developed, the latter being in

1 36 MCMURRICH. [Vol. XI.

contact with a posterior prolongation of the mesenteron, which Reichenbach describes as being formed from the entodermplate cells. The ventral wall of the mesenteron is formed of similar cells, as is also a certain amount of the dorsal wall, and it would require but a comparatively slight extension of the yolkless cells to complete the mesenteron and shut off from it completely the yolk-pyramids.

Unfortunately, observations on the later stages of Astacus are wanting to render such an interpretation of the formation of the mesenteron a certainty, but I hold that it is one which is more in harmony with the cases in which the fate of the vitellophags has been fully traced up to the disappearance of the yolk, than the view which is generally held. And, furthermore, it harmonizes with certain observations of Reichenbach himself on the yolk-pyramids. It is easily recognizable that as development proceeds in Astacus there is a marked diminution of the number of secondary yolk-pyramids, and there is good reason to suppose that this is associated with the formation of the so-called secondary mesoderm cells ; indeed, Reichenbach has traced the formation of these cells to the yolk-pyramids. In part, then, the yolk-pyramids give rise to mesoderm elements, and their entire conversion into such elements does not seem to me at all improbable.

If these views be correct, then we have in Astacus, and probably in Eupagurus as well (Mayer, '77), an invagination which includes both endodermal and mesodermal elements, and there is a further formation of mesoderm by immigration immediately in front of the invagination. This condition is derived from one in which no early differentiation of endoderm, mesoderm, and vitellophags can be recognized, and from this same condition is to be derived the arrangement seen in Jacra, where the ordinary mesoderm and endoderm form a common undifferentiated mass while the vitellophags are sharply marked off, as well as the arrangement described for Lucifer by Brooks ('83), in which there is an invagination of endoderm and the differentiation of two cells which migrate into the blastocoel before invagination, and apparently represent both the mesoderm and vitellophags.


Whether or not these views are correct later observations will determine, but they permit of a reduction of the developmental phenomena of the Crustacea to a common type, and indicate how the various modifications described have been brought about.

Part IV. — Notes on the Development of Certain


The observations recorded in this portion of the paper are fragmentary, and consist simply of a few facts which have been incidentally noticed concerning the development of certain organs, and which do not fall strictly within the limits of this paper as originally laid out.

The development of the nervous system I hope some time in the future to work out thoroughly, and shall merely notice here two points concerning it. The first of these is the occurrence of a teloblastic mode of growth of the nerve ganglia in the Isopods, of the same character as that described by Bergh ('93) as occurring in Mjsis, and similar to what Wheeler ('91) has observed in insect embryos.

The second point is the occurrence in the embryos of all the Isopods I have studied of a pair of ganglia, unaccompanied by corresponding limbs, and lying in front of the antennular ganglia. These are shown in Figs. 69 and 70, of which Fig. 69 represents an embryo of CymotJioa, and Fig. 70 one oi Jaera in a slightly later stage of development. From these figures the following arrangement of the ganglia can be seen. Anteriorly on either side are the large optic ganglia {op), and nearer the median line two elongate oval masses, which are to be regarded as the cerebral ganglia proper (cc), the optic ganglia being phylogenetically probably a secondary differentiation from these. Behind and externally to the cerebral ganglia is to be seen on either side a thickened mass of tissue {G), which is the ganglion to which special attention is directed here, and behind it lies the antennular ganglion {ciG), situated in front of the mouth invagination {st), the other two naupliar ganglia {at G and mG) lying behind this structure.

1^8 MCMURRICH. [Vol. XI.


In later stages the ganglion G as well as the antennular ganglion, and later still the antennary,fuse with one another and with the cerebral ganglion to form the syncerebrum of the adult. The occurrence of this supernumerary ganglion, whose presence indicates the existence of a segment between the eye-bearing and the antennular segments, has already been described by Bumpus ('91) as occurring in the Lobster, and in that form what may possibly be transient indications of a pair of appendages corresponding to the ganglia are developed. I have not been able to detect in the Isopods any indication of appendages, but there can be no doubt that the ganglia I have described are homologous with those seen by Bumpus. From the abstract of Nusbaum's paper ('93) it seems that he has observed what he takes to be a pair of praeantennular ganglia in Ligia, which are not, however, identical with those I have found, having an entirely different position and being formed by a splitting off of a portion of the antennular ganglia. The praeantennular ganglia of Homarus and those I have described here arise quite independently of both the cerebral and the antennular ganglia, and have, I believe, a segmental value. I have seen thickenings in the region in which they are indicated by Nusbaum, but what their significance may be I am not prepared to state.

I do not intend to follow out the questions of homology which the presence of a praeantennular segment suggests, especially as they have already been discussed by Kingsley ('94). It seems to me that further observations, especially upon the developments of the Entomostracan forms, are necessary before any certain conclusions can be formulated.

In front of the mouth in Fig. 70 two elevations {cp) are seen, which later stages show uniting to form the upper lip, and behind the mouth on a level with the mandibles two other elevations {Mt) are developed, which give rise eventually to the metastoma. There has been a certain amount of discussion as to whether the metastomal elevations are to be regarded as of metameric value and as having the same significance as limbs. The evidence furnished by the Isopods points to a negative answer to this question, since no nerve-ganglion or neuromere which can be assigned to them is distinguishable.


and furthermore it is to be noted that they do not appear contemporaneously with the limbs, making their appearance long after the limbs are readily distinguishable.

Finally I have introduced two figures {Figs. 71, 72) representing late stages in the development oi/aera, which illustrate a point of some significance in connection with the phylogeny of the Isopods. In the adults of these forms there is no indication of the presence of a carapace, but in Fig. 71 a fold {car) is clearly seen which extends backward to behind the first pereiopod {t/P) and represents a rudimentary carapace. In the stage represented in Fig. 72 this fold covers relatively a smaller extent of the body, its posterior margin being on a level with the interval between the maxilliped and the first pereiopod. I think there can be no question but that this fold represents a rudimentary carapace, and points to the derivation of the Isopods from ancestors possessing a more or less perfectly developed carapace fold. Whether these ancestors are most accurately represented to-day by the Schizopods, as some have maintained, or by the Cumacea remains to be seen, contributions on the development of the latter group being much required, the observations of Blanc ('85) leaving many points in connection with their development unsettled.

Marine Biological Laboratory, Woods Holl, Mass., August, 1894.



'93 Bergh, R. S. Zur Bildungsgeschichte des Keimstreifens von Mysis.

Zool. Jahrb., Abth. fiir Morph., vi. 1893. '86 Berthold, G. Studien iiber Protoplasmamechanik. Leipzig. 1886. '85 Blan'C, H. Ddveloppement de I'oeuf et formation des feuillets primi tifs chez la Cuma Rathkii, Kroyer. Recueil de Zool. Suisse^ ii.

1885. '74 Bobretzky, N. Zur Embryologie des Oniscus murarius. Zeitschr.

fiir iviss. Zool., xxiv. 1874. '92 Brauer, A. Ueber das Ei von Branchipus Grubii, v. Dyb. von der

Bildung bis zur Ablage. Abhand. k'dnigl. Acad. IVz'ss., Berlhi.

1892. '83 Brooks, W. K. Lucifer, a Study in Morphology. Phil. Trans. Roy.

Soc. Lottdoii, clxxiii. 1883. '92 Brooks, W. K., and Herrick, F. H. The Embryology and Metamorphosis of the Macrura. Proc. Nat. Acad. Sci., Washington.

1892. '78 BuLLAR, J. F. On the Development of the Parasitic Isopoda. Phil.

Trans. Roy. Soc. London., clxix. 1878. '91 BuiMPUS, H. C. The Embryology of the American Lobster. Journ.

of M or ph., V. 1 89 1. '93 Cano, G. Sviluppo e Morfologia degli Oxyrhynchi. Mitth. a. d.

Zool. Stat, zu Neap el, x. 1893. '91 Clapp, Cornelia M. Some Points in the Development of the Toadfish (^Batrachns tati). fourn. of Moiph., v. 1891. '89 Della Valle, A. Deposizione, fecondazione e segmentazione delle

uova del Gammarus pulex. Atii Soc. Nat. Modena, 31^6 sdr., viii.

1889. '93 Della Valle, A. Gammarini del Golfo di Napoli. Fauna und Flora

des Golfes von Neapel. Monogr. xx. 1893. '67 DoHRN, A. Die embryonale Entwickelung des Asellus aquaticus.

Zeitschr. fiir wiss. Zool., xvii. 1867. '83 Friedrich, H. Die Geschlechtsverhaltnisse der Onisciden. Zeitschr.

fiir A^attir'wiss.,\v\. 1883. '79 Grobben, C. Die Entwickelungsgeschichte der Moina rectirostris.

Arbeiten a. d. Zool. Inst. Wien, ii. 1879. '81 Grobben, C. Die Entwickelungsgeschichte von Cetochilus septentrio nalis, Goodsir. Arbeiten a. d. Zool. Inst. Wien, iii. 18S1. '78 Harger, O. Report on the Marine Isopoda of New England and

Adjacent Waters. Report of ike U. S. Commissioner of Fish and

Fisheries for 1878.


'93 Herbst, C. Ueber die kiinstliche Hervorrufung von Dottermembranen an unbefruchteten Seeigeleiern nebst einigen Bemerkungen iiber die Dotterliautbildung iiberhaupt. Bi'olog. Ceniralbl., xn. 1893. '92 Herrick, F. H, See Brooks, W. K., and Herrick, F. H. '87 Hertwig, O. and R. Ueber den Befruchtungs- und Theilungsvorgang des thierischen Eies unter dem Einfluss ausserer Agentien. Jetiaische Zeitschr. 1887. '92 Hertwig, O. Urmund und Spina bifida. Arch, fiir mikr. Anat.,

xciii. 1892. '85 ISHiKAWA, C. On the Development of a fresh-water macrurous Crustacean, Atyephira compressa de Haan. Quar. Journ. Micr. Sci., xxv. 1885. '87 Kingslev, J. S. The Development of Crangon vulgaris. Second

paper. Bull. Essex Inst., xvm. 1887. '94 KixGSLEY, J. S. The Classification of the Arthropoda. American

Naturalist., xxviii. 1894. '91 KoRSCHELT, E., and Heider, K. Lehrbuch der vergleichenden Ent wickelungsgeschichte der wirbellosen Thiere. Jena. 1891. '90 Lebedinski, J. Einige Untersuchungen iiber die Entwickelungs geschichte der Seekrabben. Biol. Centralbl., x. 1890. '91 Lebedinski, J. Die Entwickelung der Daphnia aus dem Sommereie.

Zoo I. Ans., xiv. 1891. '82 LuDWiG, H. Entwickelungsgeschichte der Asterina gibbosa, Forbes.

Zeitschr. fiir wiss. Zool., xxxvii. 1882. '92 McMurrich, J. P. The Formation of the Germ-Layers in the Isopod

Crustacea. Zool. Anz., xv. 1892. '77 Mayer, P. Zur Entwickelungsgeschichte der Dekapoden. Jenaische

Zeitschr., xi. 1877. '86 NusBAUM, J. L'embryologie d'Oniscus murarius. Zool. An2.,\x. 1886. '87 NusBAUM, J. L'embryologie de Mysis chameleo (Thompson). Arch.

de Zool. exp. et gen., s^e sdr., v. 1887. '91 NusBAUM, J. Beitrage zur Embryologie der Isopoden. Biol. Centralbl.

xi. 1 891. '93 NusBAUM, J. Materyaly do embryogenii i histogenii rowononogow (Isopoda). Abhandl. der Krakauer Acad. Wiss., xxv. 1893. (Abstract in the Biol. Centralbl., xiii. 1893.) '90 Patten, W. On the Origin of Vertebrates from Arachnids. Quart.

Journ. Micr. Sci., xxxi. 1890. '88 Pereyaslawzew, Sophie, et al. Etudes sur le d^veloppement des Amphipodes. Bull. Soc. Imp. Nat. Moscou. 1888, 1889, 1890, and 1 891. '34 Rathke, H. Recherches sur la formation et le developpement de I'Aselle d'eau douce (Oniscus aquaticus, Linn.). Annal. des Sci. Nat., 2me ser., ii. 1834.

1^2 MCMURRICH. [Vol. XI.

'37 Rathke, H. Zur Morphologic. Reisebemerkungen aus Taurien.

Riga and Leipzig. 1837. '83 Rauber, a. Neue Grundlegungen zur Kenntnis der Zelle. Morph.

Jahrb., viii. 1S83. '86 Reichenbach, H. Studien zur Entvvickelungsgeschichte des Fluss krebses. Abhandl. Senckenbg.A^aturf. Gese/lsch., xiv. 1886. '87 Reinhard, W. Zur Ontogenie des Porcellio scaber. Zool. Anz., x.

1887. '89 RouLE, L. Sur I'dvolution initiale des feuillcts blastodermiques chez les Crustaces Isopodes (Asellus aquaticus, L., et Porcellio scaber, Latr.). Coinptes Reiidiis, cix. 1889. '90 RouLE, L. Sur le developpement du blastoderme chez les Crustaces

isopodes (PorcelUo scaber). Compics Rcndiis, ex. 1890. '91 RouLE, L. Sur le developpement des feuillets blastodermiques chez les Crustaces isopodes (Porcellio scaber). Coinptes Retidtes, cxii. 1891. '92 RouLE, L. Sur le developpement du mesoderme des Crustaces et sur

celui de les organes derivees. Coinptes Rcndits, cxiii. 1892. '92a RouLE, L. Sur les premieres phases de developpement des Crustaces

edriophthalmes. Compies Rendtts, x\\\. 1892. '93 S AMASS A, P. Die Keimblatterbildung bei den Cladoceren. I und II.

Archiv. f. ?nikr. Attat., xli. 1893. '68 Sars, G. O. Histoire naturelle des Crustaces d'eau douce de Norv^ge.

Livr. I. Les Malacostraces. Christiania. 1868. '86 Sedgwick, A. The Development of the Cape Species of Peripatus.

Part II. Quart. Journ. Micr. Sci., xxxx. 1886. '88 Sheldon, Lilian. The Development of Peripatus Novae-Zealandias.

Quart. Journ. Micr. Sci., xxviii. 1888. '87 Sye, C. G. Beitrage zur Anatomic und Histologic von Jaera marina.

Inaug. Dissert. Kiel. 1887. '69 VAN Beneden, E. Rechcrches sur rembryogdnie des Crustacds. I. Observations sur le ddveloppement de 1' Asellus aquaticus. Bidl. Acad. Roy. de Belgique, 2me sdr., xxviii. 1869. '69a van Beneden, E. Recherches sur rembryogdnie des Crustaces. II. Developpement du Mysis. Btdl. Acad. Roy. de Belgique, 2me s^r., xxviii. 1869. '87 VON Sachs, J. Lectures on the Physiology of Plants. (Translated by

H. M. Ward.) Oxford. 1887. '92 Weldon, W. F. R. The Formation of the Germ-layers in Crangon

vulgaris. Quart. Journ. Micr. Sc/., xxxiii. 1892. •91 Wheeler, W. M. Neuroblasts in the Arthropod Embryo. Journ. of

Morph., iv. 1891. '78 Whitman, C. O. Embryology of Clepsine. Quart. Journ. Micr. Sci., xviii. 1878.


'91 Whitman, C. O. Sperm atophores as a Means of Hypodermic Impregnation. Journ. of Morph. 1891.

'94 Whitman, C. O. The Inadequacy of the Cell Theory of Development. Jourti. of Morph., ix. 1894.

'89 Wilson, E. B. The Embryology of the Earthworm, foiirn. of Morph., iii. 1889.

'92 Wilson, E. B. The Cell Lineage of Nereis, fourn. of Morph., vii, 1892.

'93 Wilson, E. B. Amphioxus and the Mosaic Theory of Development. Jourti. of Morph.., viii. 1893.




Lettering used uniformly throughout the figttres.

a = anus.

fue = mesoderm.

ad = abdominal limb.

MEn = mes-endoderm.

aG^ antennular ganglion.

w^ 7"= mesodermal teloblast.

At'^ ^antennule.

mG =^ mandibular ganglion.

At"^ = antenna.

mlif^Mmb mesoderm.

AtG = antennary ganglion.

Mn = mandible.

B = dorsal blood vessel.

ms = Mittelstrang.

Car = carapace.

vtt = metastoma.

ce = cerebral ganglion.

mit = muscle.

cA/= central mesoderm mass.

Mx 1 = first ma.\illa.

cp = central protoplasm.

Mx 2 = second maxilla.

cT= central ectodermal teloblast.

J\fxp = maxilliped.

D.O. = dorsal organ.

JV=z nerve ganglion.

dy = disintegrated yolk.

op = optic ganglion.

E := eye.

pti = protoplasmic network.

Ec = ectoderm.

pp = peripheral protoplasm.

Ep = upper lip.

pr = proctodaeum.

G = praeantennular ganglion.

j/ = stomodaeum.

k = heart.

T^^ ectodermal teloblast.

Z = liver.

td = transverse band of embryo,

/ ^ liver lobe.

te = telson.

l.en = liver endoderm.

th = thoracic limb.

IM= lateral mesoderm mass.

tr = teloblastic row.

L.O. = lateral organ of Asellns.

y = yolk.

l.v.b. = lateral ventral band of embryo.



All the figures on this plate are oi Jaei-a marina.

Fig. I. Transverse section of adult 9- Ow = ovary ; C^- = germinal region of Ovary; ^ ^ chitinous portion of Spermatophore ; /) = intestine ; DC = digestive ccecum ; N'= ventral nerve cord ; B = dorsal blood vessel.

Fig. 2. Section of immature ovum showing protoplasmic network. For the sake of clearness the yolk granules have been omitted, /c ^= follicle cells.

Fig. 3. Section of mature ovum showing the peripheral and central protoplasm, the latter containing the nucleus. The network not visible.

Fig. 4. Ovum just after the expulsion of the polar globules {/^) drawn from living specimen. The dark structure in center is the segmentation nucleus with its surrounding protoplasm seen indistinctly through the yolk, c/i = chorion ; ym = yolk membrane.

Fig. 5. Optical section of ovum in the two-celled stage.

Figs. 6 and 7. Optical sections of ova in the four-celled stage showing different stages in the rotation of the cells.

Fig. 8. Ovum during the formation of the eight-celled stage. The figure is from a cleared preparation so that all the nuclei are visible.

Fig. 9. The eight-celled stage completed, from a cleared preparation.

Fig. 10. Section of ovum in the eight-celled stage.

Fig. II. Surface view of posterior pole of ovum in the i6-celled stage. The two terminal endodermal cells are surrounded by a circle of seven cells.

Fig. 12. Surface view of posterior pole of the i6-celled stage in which the number of cells in the circle has been reduced to six.

Fig. 13. Surface view of the 32-celled stage seen from ventral surface.

Fig. 14. Surface view of the 64-celled stage seen from the side.

Fig. 15. Surface view of posterior extremity of ovum passing into the succeeding stage.

Fig. 16. Surface view of ovum in the succeeding stage seen from the side.

Fig. 17. Ventral view of ovum in which the concentration of the cells towards the ventral surface is taking place.

Fig. 18. Dorsal view of the same ovum.

Fig. 19. View of ovum in which the concentration is complete.

Fig. 20. Longitudinal section of ovum in the 32-celled stage.

!,mrtml of Morphology Vol.XI.




Fig. 21. Sagittal section through an egg of Jaera showing the mesoderm plug (ine) and the teloblastic cell of the ectoderm plate (Z").

Fig. 22. Sagittal section through egg of Jaera showing the beginning immigration of the endoderm cells (en) to form the vitellophags. The cells of the mesoderm plug have begun to scatter and the teloblastic growth of the ectoderm has produced short rows of cells (tr).

Fig. 23. Ventral view of embryo of Jaera in which the teloblastic growth has carried the ectoderm tract over the mesoderm plug.

Fig. 24. Side view of the same embryo showing the patch of optic cells {E) and the rows of cells extending to it from the ectoderm plate.

Fig. 25. View of anterior half of an embryo of Jaera in which the yolk is about two-thirds overgrown, showing the scattered arrangement of the cells in the naupliar region and the teloblastic rows in the meta-naupliar.

Fig. 26. View of posterior half of the same embryo showing the row of teloblasts (Z") the central one of which (cT) is very well marked.

Fig. 27. Portion of posterior half of an older embryo somewhat broken by pressure showing the row of teloblasts (T) and the " Mittelstrang " (7ns). The dark transverse bands (nte) in the anterior part of the figure represent mesoderm masses showing through the superjacent ectoderm.

Fig. 28. Egg of Asellus Communis in the two-celled stage viewed as a transparent object.

Fig. 29. Egg of Asellus in four-celled stage.

Fig. 30. Egg of Asellus in eight-celled stage.

Fig. 31. Egg of Asellus passing into the i6-celled stage.

Fig. 32. Egg of Asellus in the 64-celled stage viewed as an opaque object.

Fig. 33. Egg of Asellus in the 1 28-celled stage.

Fig. 34. Blastoderm of Asellus represented as if removed from the ovum.

Fig. 35. Egg of Asellus at the time of the differentiation of the eleven primary ectodermal teloblasts.

Fig. 36. Egg of Asellus just after the first division of the ectodermal teloblasts.

Journal of Morphology. Vol XI.











-0-^' ^'



e> _ ^





B ■<b'



II' If





. .T















Fig. 37. Embryo of Asellus represented as spread out flat.

Fig. 38. Embryo of Aselliis in stage later than Fig. 37 showing the teloblastic growth of the meta-naupliar portion.

Fig. 39. Ovum of Porcellio in the one-celled stage. The thickness of the peripheral protoplasm (//>) in this and the next figure is slightly exaggerated.

Fig. 40. Ovum of Armadillidiiim in the two-celled stage.

Fig. 41. Four-celled ovum of Arrnadillidiiun.

Fig. 42. Section through a four-celled ovum of Porcellio.

Fig. 43. Eight-celled stage of Armadillidium.

Fig. 44. i6-celled stage of Porcellio. The lines joining the various cells indicate their origin from the eight-celled stage.

Fig. 45. i6-celled stage of Armadillidium, viewed from the posterior pole.

Fig. 46. 32-celled stage of Armadillidium, viewed from the posterior pole.

Fig. 47. Section through an ovum of Porcellio in the 32-celled stage.

Fig. 48. Ovum of Porcellio in the 64-celled stage.

Fig. 49. Ovum of Porcellio in the succeeding stage when the concentration of the blastoderm around the mes-endoderm (nt-eii) is beginning.

Fig. 50. Ovum oi Armadillidium in a stage slightly later than that of Fig. 49.

Figs. 51 and 52. Still later stages of Armadillidium.

Fig. 53. Blastoderm of Porcellio in situ.

Mini id of Morphology Vol. XI.








-_• IP





Fig. 54. Blastoderm of Porcellio at stage when the mes-endoderm cells are beginning to sink beneath the surface.

Fig. 55. Section through a blastoderm of Porcellio of the same age as that of the preceding figure, showing the lower layer cells in the blastoporic region.

Fig. 56. Blastoderm of Porcellio at stage just before the scattering of the mes-endoderm cells.

Fig. 57. Blastoderm of Armadillidium at the time of differentiation of the ectodermal teloblasts.

Fig. 58. Blastoderm of Armadillidium with the ectodermal teloblasts more perfectly differentiated.

Fig. 59. Embryo of Armadillidium, commencement of teloblastic growth.

Fig. 60. Section through ovum of Ligia at the time when all the cells have reached the surface.

Fig. 61. Blastoderm of Ligia at stage when the mes-endoderm cells are sinking beneath the surface.

Fig. 62. Posterior portion of embryo oi Ligia showing the mesoblasts. IllF7/ = thoracic segments ; 1-6 = abdominal segments ; t = telson.

Fig. 63. Posterior portion of embryo of Cymothoa showing the mode of division of the mesoblasts.

Fig. 64. Posterior portion of embryo of Cymothoa showing the ectodermal and mesodermal teloblasts.

Joiirnnl of Morphology Vol. XI



.-.6. .,:. '©.^f $©■ ^



-J MEp




'■' til 'is ; '■ -^






H -^9«^\*«i!^..°J


^il'©!®^®!®®^^ ?^ 3,5^>? .*^,


®*9 Ih'



a 9





. o:a.. :-0

n©'^' Q G



■ r.tnh'ixin:- F' ••



Fig. 65. Section through the thoracic region of an embryo of Cymothoa in a stage somewhat later than that represented in Fig. 62.

Fig. 66. Section through the second thoracic appendage of an embryo of Cymothoa older than that from which Fig. 65 was taken.

Fig. 67. Section through the abdominal region of an embryo of Cymothoa in which the yolk is beginning to disintegrate, en == chromatin nebula.

Fig. 68. Section through the thoracic region of an embryo of Cymothoa in which the disintegration of the yolk is complete.

Fig. 69. Anterior portion of an embryo of Cymothoa.

Fig. 70. Anterior portion of an embryo oi/aera.

Fig. 7 1 . Lateral view of an advanced embryo of Jaera.

Fig. 72. Lateral view of an embryo oi Jaera almost ready to hatch.

Journal of Morphology. VolXL

Pl. IX.






mt Mn







ini: ■ L


.•;/' /■


pr B mil

m. hsL y. WtrnitlLWntir. Franifiul W



This worm, first described by Mark (i), may be found in abundance in the waters of Woods Holl harbor and vicinity. During the summer months it lives on the eel-grass and on dead shells in shallow water. In the northwest gutter of Hadley Harbor, where there are extensive sand and mud flats covered by but a few inches of water at low tide, large numbers of dead shells of clams, pectens, and mussels are scattered over the surface. On the under and thus darkened side, these beautiful little red worms collect in quantity, often as many as a dozen on one shell. That they prefer the dark is evident even when placed in a clear glass dish on a table near the window, as they invariably congregate on the surface away from the light ; or if a small portion of the dish is shaded they seek the darkest corner, crowding close together and often one on top of another to secure a position sheltered from the light. Also if shells are placed in the dish they cluster on the lower and thus shaded side of the shell. They seem to prefer a clean, fresh shell, and are to be found more frequently on the shaded surface of a white clam shell than on that of those darkened by age and marine growths.

It is on these sheltered portions of their favorite abiding places that the eggs are to be sought, for when kept in a glass dish they lay their eggs quite freely. It is noticeable, however, that the eggs are more frequently deposited on the under surface of a shell, if there be one in the dish, than in the light ; also that they lay more frequently in the night than in the day-time if no place sheltered from the light rays is offered to them. That it is darkness rather than the hour of night which induces them to deposit their ova, is shown by the following experiment at the season when they were laying

156 GARDINER. [Vol. XI.

most abundantly. Two dishes of salt water, both containing these Turbellaria, were placed side by side near the window in a well lighted room, and one of these dishes was covered by a dark felt hat. At the end of the afternoon nearly twice as many clusters of ova were found in the darkened dish as were in the dish exposed to the light.

The process of ovulation has frequently been watched with a lens when the ova were being deposited on the side of a fins:er-bowl or small dish. When the individual is about to deposit its ova, it draws in all projecting portions of its body, assuming the form of an oval disk so that at the first glance it is difficult to distinguish the anterior from the posterior end. Then the central part of the body is elevated so that a space is left between the body and the glass. As the ova pass out of the female genital pore they appear at first unsurrounded by any capsule, but in the course of a few minutes they appear to be contained in a clear transparent mucus mass. From two to ten ova are thus deposited in each capsule. They are never deposited in rows, but seem scattered irregularly through the central part of the capsule, which is of much firmer consistence on the surface than in the interior.

Not infrequently the whole mass rotates several times before it adheres to the glass. In capsules which are deposited on the surface of disintegrating shells the upper surface is often rendered partly obscured by the presence of fine particles of the shell which adhere to the surface of the capsule during this rotation. On one occasion a capsule was deposited on the side of a glass dish where some finely powdered carmine had been smeared, and the result was that the whole surface of the capsule was covered with the fine colored particles.

During the process of ovulation the animal is reluctant to move, and on being disturbed with a needle or other instrument contracts and shrinks from the contact, but leaves the place it occupies much less willingly than under ordinary circumstances. When the disturbance is persisted in, the animal moves off, trailing after it a line of mucus and eggs. Eggs deposited thus unprotected by the full capsule seldom develop normally.


The capsule is soluble in weak hydrochloric and nitric acids.

Sections through the animal containing mature and nearly mature ova show that a small, inconspicuous polar globule spindle is formed some time before the ovum is laid. In some cases two polar globules may be seen after the ovum has been laid, adhering to its surface on the macromeres A or A^ close to the point where the second cleavage plane intercepts the first. While the ovum is still within the parent, and shortly after the polar globules have been formed, a remarkably large and distinct spindle is formed, which, since it appears directly after the formation of the polar bodies, must be the first segmentation spindle. It may be readily seen by placing the animal under a cover slip and by drawing off the excess of water with filter paper, so that the cover slip exerts sufficient pressure to cause the animal to fully expand, but at the same time does not rupture it. In ova examined in this manner the nucleus may appear as a round almost transparent spot of good size in the center of the ovum ; in other ova it is somewhat elongated, while in others it appears as a large dumb-bell-shaped structure (Fig. 31) which occupies the greater part of the ovum. Indeed, the elongation of the nucleus, the formation of the spindle, and drawing apart of the polar suns, may be observed in one specimen examined at intervals of a few hours.

That spindles presenting this peculiar appearance are common in the ova of many Turbellarians is shown by von Graff (2), who in his monograph figures individuals belonging to three different genera, in all of which ova containing such spindles are shown. He does not, however, discuss this matter in the text.

A very remarkable phenomenon occurs in connection with this spindle ; for if the animal be kept too long under the cover slip or placed under certain abnormal conditions, the polar suns grow dimmer and draw closer together and the dumb-bell-shaped structure disappears entirely until the nucleus appears to return completely to its resting stage, and remains unchanged till after the ovum is laid, when again the spindle is formed, heralding in this case the first stage in segmentation. The formation of the polar globules, the fertilization of

158 GARDINER. [Vol. XI.

the ovum, the formation of the first spindle and its disappearance under certain conditions, present many problems of interest on which I am now at work, and which will form the subject of a future paper.

Selenka (3) first called attention to a spindle in the uterine ova of Thysanosooii Diesiiigii, which from the description given resembles very strongly, in size and general appearance, that just described in the ova of P. caudatus. He describes in detail the formation of the spindle seen by him, and states that after the equatorial plate of chromosomes is formed, the polar suns fade in distinctness and draw closer to one another, while the chromatosomes melt together, and the whole nucleus returns to its resting stage. Selenka suggests that this partial karyokinesis occurs in order to effect a rearrangement of the yolk particles. Lang (4) states that among the Polyclads, he has observed even in the uterine eggs a similar phenomenon, but cannot accept Selenka's explanation as to the object for which this spindle is formed, for no such rearrangement of the yolk globules occurs at this stage. Wheeler (5) has also observed a similar spindle to occur in the uterine ova of Planocera Inquilina, and also describes it as disappearing before oviposition. All of these authors state that in these Polyclads, the polar bodies are formed and the fertilization of the ovum effected after ovipositio7t ; therefore the spindle in these forms cannot be the same in origin as that just described in Polychoerus, in which form the polar bodies are formed before oviposition.

The early stages of segmentation may be studied without removing the ova from the capsule, by placing the whole capsule in a concave slide. This method is, however, unsatisfactory, for it is impossible to rotate the ovum under inspection. When taken from the capsule, which is easily done with needles under a dissecting microscope, the ova generally segment abnormally, and soon die. All of the ordinary killing reagents such as Perenyi, corrosive sublimate, and chromic acid in its various combinations as recommended by different authors, are perfectly useless for these ova, for any of these reagents destroy all trace of segmentation. The most effective


reagent is a mixture of equal parts of absolute alcohol and glacial acetic acid, and even this is very uncertain in its effects, some ova being fixed and losing their color almost immediately, while others are affected very slowly. Examination shows that in these latter the cell structure is much injured, or, more generally, totally destroyed. The reason for this uncertainty is probably due to the difficulty of removing all of the mucus capsule by which they are protected from the salt water. The inner part of this capsule is so transparent that its presence can hardly be detected.

Ova which have been cut out of the parent are of a delicate yellowish white color, and no trace of pigment is to be seen in them. When laid they are about .06 X .04 mm., and a few flecks of a reddish yellow pigment may be seen scattered over the surface. Examination with an oil immersion shows that these pigment granules are between 2 and 3 /la. in size, and are, roughly speaking, double spheres in shape. Fig. 32 represents these granules, a, seen from the side, and b from the end. The form, as well as the curious movements and change of position which they undergo, suggest that they are some form of alga, but I have not been able to demonstrate that such is the case.

In the newly laid ovum they are few in number, but may be seen, even with a fairly low power, lying on or near the surface of the ovum. After the ovum is laid they begin to multiply and apparently to migrate from within toward the surface of the cell. As will be related in the description of the early segmentation, they form a girdle about the ovum which heralds the first cleavage, and in every successive cleavage up to the ten-cell stage the line through which the cleavage plane will pass is thus marked out by pigment granules before the cell divides. Nusbaum (6) has shown that the pigment granules in the cells in the tail of the tadpole are moved about in a somewhat similar manner, always moving or being moved, within the cell in which a spindle is forming, in such a manner that they form an equatorial plate in the same plane as the chromatosomes. Further, when the cell divides, the surfaces of the two daughter cells lying in contact are covered with

l6o GARDINER. [Vol. XI.

pigment granules while the rest of the cell is free from them. The action of these granules in the ova of PolycJiaenis is apparently very similar, but unfortunately they are entirely dissolved in alcohol, so that the history of their movements within the cell is beyond reach by sections. Not infrequently, while examining the surface of an ovum with an oil immersion, I have seen one of these granules come up from within the ovum to the surface, and move across the field of vision, but I have never yet seen a granule pass from one cell into another ; yet they disappear in some cells and later on appear in quantity in others. When the ovum is so viewed it is clearly suggested that there are wonderfully active forces at work within, for the surface fairly scintillates with movements of the protoplasm and these pigment granules. The manner in which the granules are moved about from place to place indicates the powerful nature of the force within. In regard to the disappearance in one part of the ovum and reappearance in another. Figs. 2, 3, 4, and 5 show these granules massed in cells B, B', C and C, while later on these same cells (Figs. 1 1 and 12) are almost devoid of them, while on the directly opposite pole of the ovum they are so crowded that the cleavage lines are almost obscured. Fig. 13 shows the pole at which the granules were first massed, and Figs. 14, 16, and 18 show the opposite pole, which at an early stage was almost destitute of color. Fig. 9 gives the appearance of this pole when the granules first collect there, and also shows the color-effect produced when they are closely packed together. That the granules actually migrate from one cell to another seems impossible, for, as is shown in Figs. 23, 24, and 25, the cells are bounded by clear, distinct membranes through which the granules would have to pass. It seems much more probable that they fade out and become disintegrated in one part, while others are organized in other parts.

If an ovum is crushed they begin to fade almost immediately and in a few minutes are no longer distinguishable. When the segmenting ovum has reached the stage shown in Fig. 22 the pigmentation is almost confined to the few cells A, A', B^, B'^, and £"3, E'^, and may be said to have reached its maxi


mum. With a very low power lens this stage is easily distinguishable by its red spot, which becomes later a little more diffused, and then gradually fades out before the embryo is formed.

The pigment granules which characterize the adult, as described by Mark (i), are of a totally different form and color, and bear no resemblance whatever to those which occur in the ovum. I have examined many animals by crushing them under a cover slip, but have found nothing resembling these pigment granules. If they occur at all in the adult, which I doubt, they are very few in number and therefore difficult to find. So, also, in the immature ovum these pigment granules are rare if they occur at all. The young worm just hatched is marked by granules similar to those formed in the adult, while none or almost none of the above described granules are to be found in it.

In the ova of an undescribed species of Aphaiiostoma {)), which is found in this vicinity, and to which my attention was first called by Professor H. C. Bumpus, a precisely similar arrangement and movement of pigment granules, if they be pigment granules, occurs. They are distinctly larger and redder than those found in the ova of PolycJioerus, and the dividing line between the two halves (Fig. 32, a) is more distinctly marked. This Aphanostoma (?) is a dark green color, and no red pigment is to be found in it, and the ova when newly laid, have a color very similar to that of the parent. Soon, however, the red pigment appears on the upper pole, and as segmentation progresses is finally collected on the opposite pole, forming the red spot similar to the one just described, but more brilliant in color. Experiments similar to those detailed by Haberland (7) to prove whether these granules might not be independent organisms, were attempted, but the results were perfectly negative. The ova died very soon and the pigment granules disappeared entirely.

A very remarkable phenomenon during the segmentation of the ova of Polychoenis is the manner in which the ovum changes its form and becomes distorted, as if some internal forces were pressing in different directions ; also the marked difference

1 62 GARDINER. [Vol. xi.

in shape at times when the karyokinetic activity is at its highest as compared to that when the ovum is in its resting stage. The ovum shown in Fig. lo soon rounds up so as to be indistinguishable from that shown in Fig. 12. Not infrequently, however, one part will round up some time before the other, suggesting most strongly the absence of the anteroposterior and bilateral symmetry which characterizes this form of segmentation.

To turn now to the segmentation of the ovum. Shortly after the ova are laid, the pigment granules form a complete girdle round the smaller axis of each ovum (Fig. i) and mass themselves also in the region where the cells {C, C) will be formed, and then the ovum divides into two cells of equal size {A and A). The second cleavage plane is in the region of this pigmented area. It is at right angles to the first and divides the cells A, A', into A, B, and A', B' (Fig. 2). Immediately after these cells (B and B') have been budded off, they begin to rotate from left to right, so that B, instead of lying directly above A, and B' directly above A', both B and B' cover the line of the first cleavage plane and rest on both A and A' (Fig. 3). Fig. 4 shows the position attained by B as seen from the side, while B' passes to the opposite side of the ovum. After this four-celled stage has been reached a period of rest follows, the ovum becomes again oval, and all cell outline disappears. If it were not for the bright girdle of pigment and the massing of the granules in the neighborhood of B and B', the ovum might easily be mistaken for one just laid and unsegmented.

The appearance of the third cleavage plane is preceded by the reappearance of the cell outline and also by a massing of the pigment granules on the surface of A, A' in the neighborhood of B and B', which cells move apart to right or left so that the line of the first cleavage plane may be seen between them. At the same time the cells C and C are budded off from A and A' (Fig. 5), exactly as were B and B'. No rotation, however, takes place, but the cells C and C round up and draw closer together in such a manner that B and B' are forced still farther apart and come finally to lie on opposite sides of the ovum, as is shown in Figs. 6, 7, 8 and 1 1. j


Again follows a period of rest and disappearance of all cell outlines. Renewed activity in cleavage is heralded by changes in the pigmentation, and the line of the next cleavage plane is indicated by a distinct line of red granules dividing off the upper portions of A and A^ in the same direction as that of the second and third cleavage planes and presently D and D^ are budded off from A and A^ respectively (Fig. 7).

After the usual period of rest, the fifth cleavage plane is, as lisual, first indicated by the pigment granules, and then the cells E and E^ are budded from A and A' respectively (Fig. 8), and in a plane parallel to the first but at right angles to all other cleavage planes which have been described. Hence we have a ten-celled stage, eight of which cells have been budded off from the original macromeres A and A\ which exist now as mere remnants. At this stage we have the cells which will form the germinal layers already differentiated, for a little later, by a process which will be described A and A^ pass into the centre of the ovum and form the mesendoderm, while the lineal descendants from B, B\ C, C, D, D\ E and E' form the ectoderm.

Thus far the segmentation has proceeded with great regularity, but from now on the order in which the cells divide varies so much that it is often difficult to determine exactly in which generation certain cells are to be classed. Although the bilateral symmetry is on the whole maintained, it frequently happens that on one end or side, the cells divide more rapidly than elsewhere, so that an ovum presents the appearance of irregularity ; but when this occurs, the cells on the corresponding end or side divide in an exactly similar manner shortly afterward, so that the symmetry is soon regained, and the general scheme of segmentation is always the same. In order to make the description of the process more clear a short explanation of the manner in which the cells are lettered may be necessary. The cells A and A^ are the result of the first division of the ovum, and from these B, C, D, E and B\ C, D', E' are given off respectively. In all cases the sign ' indicates that the cell is descended from A' and its absence that it is descended from A. Thus we have Ei.r.i descended from A

1 64 GARDINER. [Vol. XI.

through E, and E'l.r.i descended from A' through E'. All cells marked by letters bearing the sign ' lie on the right of the shorter axis as viewed in the figures, except B' and its descendants. The cells B and B' originated from A and A' respectively, and when first formed B was to the left and B' to the right (Fig. 2), and though in a later stage they have come to lie directly on the shorter axis, the letter B is still applied to the cell which was budded off from A, and B' to the cell budded off from A'.

The cell B divides subsequently into Bi and Bz. B\ will divide into B\.\ and B1.2, and B2 will divide into Bz.x and B2.2. The derivatives from B^ on the opposite side of the ovum will be numbered similarly, but all marked with the sign '. The cells C, D, E, C, D', and E' are, however, so placed that in the subsequent cleavage some cells derived from each of them will lie on the right and some on the left of a line drawn through the long axis of the ovum. The cells lying on the right will be distinguished by an r, and those to the left by an /; thus Cr and CI, Dr and Dl (Fig. 13). As these cells further divide, those derived from them will be indicated by numbers which will show the generation to which they belong ; thus, Z>r divides into Dr\ and Drz. In the next generation Dri gives rise to Dr\.\ and Dri.z ; while Drz divides into Drz.x and Drz.z.

The system by which the cells arising from E and E' will be distinguished, necessarily differs slightly from that applied to B, C, D, and B', C, D' , for E and E do not give rise to cells which lie to the right or left of the long axis till a somewhat later stage. In the first generation these cells give rise to cells E\, E2, Et^, E4, and E'l, E 2, E t^, E\, some of which lie directly on the line of the longer axis which will divide the ovum into right and left halves (Figs. 14, 15 and 16). In the next generation, however, when certain of these cells divide into right and left, they will be distinguished by the letters r and /; thus ^srand E2I (Fig. 18).

To return now to the process of segmentation when the tencelled stage has been reached. As a rule the next cells to divide are B and B', although it often happens that their division is delayed till later. In either case, whether they divide now or


after D and D' , E and E have divided, they invariably give rise to B\, Bz, B'l and ^'2 respectively, as is shown in Fig. 10. This figure shows the pigment granules dividing the cell B into approximately equal parts, and the plane of cleavage follows this line. Occasionally the line of granules and cleavage plane is somewhat oblique to the line as represented in this figure, and in such cases a rotation takes place, so that eventually the cells Bi and Bz lie as is shown in Figs. 15 and 20. The cells B'l and B'z are derived from B' in an exactly similar manner, and lie on the side of the ovum opposite to that drawn. At about the same time B^ and B't, are budded off from B2 and B'2 respectively.

The cells D and D' are generally the next ones to divide. Figs. 8 and 10 show the cell D from the side, and Fig. 11 shows both D and D' from above. The division takes place, as is shown in Fig. 13, along the long axis of the ovum, thus giving rise to Dr, Dl, Dr, D'l. Also C and C divide in a similar manner (Fig. 13), giving rise to Cr, CI, Or, C'l. At about the same time as this takes place, E and E divide in such a manner that we have E\, E2, and E'l, E'2, as is shown from the lower pole in Fig. 14 and from the side in Fig. 15. In Fig. 14 it will be seen that the pigment in Ez and E'2 is massing close to the cells A and A', and this heralds the formation of Et^ and E' t, which are shown in Fig. 1 5 from the side and in Fig. 16 from below. In this latter figure it will be noticed that ^3, E' y By and B' ^ form as it were a cross on the lower pole, with the cells A and A' in the center of it. In Fig. 17 (the same stage as shown in Fig. 16) the ovum is viewed from the lower pole and drawn as if transparent. The cells Ey E' y E2, E'z, By B' y A and A' lie on the surface and partly conceal Ei, E'l, Bi, B'l, Bz and B'z, which lie below them, while Dr, Dl, D'r, D'l, Cr, CI, Or, C'l are completely hidden and indicated by dotted lines.

Presently the cells Ei, Ez, E'\, E'z, divide in a plane parallel to the long axis of the ovum so that Eir, Eil, Ezr, E2I, E'lr, E'll, E' 2r, E'zIqxq formed as is shown in Fig. 18. The relative position of these cells is best shown in Fig. 19, in which the ovum is again treated as if transparent. Almost

1 66 GARDINER. [Vol. XI.

coincident with these changes the cells E^^i', E^l, E\r, E\l are formed. Fig. 20, which is a view of the right side of the ovum, shows Errand E'^r, and in Fig. 21, which is a diagram of this stage, and treated as if transparent, E4/ and E'4/ are shown by dotted lines. Apparently these cells arose from a division of Eir, E\l, E' ir, ^"'1/ respectively.

After this the ovum having attained a thirty-eight-celled stage goes into its characteristic resting stage and the cell outlines become obscure. During the next period of activity the ovum reaches a sixty-six-celled stage, apparently in the following manner. It is, however, almost impossible to identify with absolute certainty the order in which the cells divide. It appears, however, that Di'i divides so as to give rise to Dri.i and Dn.2 (Fig. 22 and 26); Drz into Drz.i and Drz.z ; Dr^ into i^?'3.i and Dr^^.z ; Dh into Dli.i and DI1.2 ; BI2 into DI2.1 and DI2.2 ; DIt^ into i^/3.1 and Dl^.z ; D'ri into D'ri.i and B'ri.2 ; D'r2 into D'r2.i and D'r2.2\ D'r-^ into D'ryi, D'ry2 ; D'li into D'h.i and D'li.2 ; D'h into D'l2.i and D'12.2 ; D'l^ into DHyi and D'ly2 ; making in all twenty-four cells derived from D and D' . The cell Eir divides into Eiri and Eir2 ; Ezr into E2i'i and ^2r2 ; £"4^ into E^ri and £"4r2 ; ^1/ into ^i/i and £"1/2 ; E2I into ^2/1 and ^2/2 ; E^l into ^4/1, ^4/2 ; E'lr into ii'in and E\r2\ E 2r into ^'2^1 and ^'2^2; ^'4;' into if'4ri and E' ^1-2 ; ^%/ into E \l\ and £1/2 ; £"'2/ into E zh and £"'2/2 ; E\l into ir'4/1 and E^h. The cells -£"3 and ^'3 have not divided, but including them the total of the derivatives from .£■ and E is twenty-six cells. B\ also divides into B\.\ and Bi.2 ; B2 into ^2.1 and ^2.2 ; E i into ^'i.i and E 1.2 ; ^'2 into Z>"2.i and ^'2.2, which together with ^3 and ^'3 make ten cells derived from B and B^ .

Fig. 26, which is a diagrammatic representation of the sixtysix-celled stage (Fig. 22), shows the manner in which this stage has been derived. The dotted lines represent the cells on the opposite side of the ovum. It will be seen that / represents the cells A and A^ from which B, B\ C, C, D, D\ and E, E have been budded off successively. The ten derivatives from B and B^ are shown in // ; in /// are the four derivatives from C and C^ \ in /F the twelve derivatives from D\ in F the


thirteen derivatives from E ; in VI the twelve derivatives from Z>'; in VII the thirteen derivatives from E \ maldng in all sixty-six cells, all, except A, A', which are much the largest, of about the same size.

Thus we have, as it were, two very different methods of segmentation shown in different periods during the formation of the sixty-six-celled stage. At first four pairs of cells, B, B\ C, C, D, D\ and E, E', were budded off from the two macromeres A and A' in successive generations. From now on segmentation progresses by the division of these eight cells, while the remnants of the two macromeres undergo no further division till much later. Further, the first four generations follow always in the order in which they are named {B and B', C and C, D and B', E and E') but from now on the generations are less easy to distinguish. For instance, in some cases D and B' give rise to Br, Bl, B'r and B'l before the cells E and E' divide at all, while in other cases E and E' give rise respectively to El, E2, -£"3, E'l, E'2, E' 2,, at the same time or even before B and B' divide. Hence it is impossible to distinguish with accuracy the relative ages of the derivatives of the eight cells. They appear, however, to be formed most generally in the order above related, which is also shown in the diagrammatic cell lineage shown on p. 168. From the appearance of the first cleavage plane on, both bilateral and antero-posterior symmetry are maintained. By subjecting the ten-celled stage ova to pressure by crushing them slowly, and also by sections, a better idea of the relative size of the cells can be obtained than by merely surface views, and it is apparent that C and C are the smallest of those budded from A, A' ; B, B' are next larger in size, B, B' next, and E, E' the largest.

To form the sixty-six cells

C, C have given rise to 4 cells B, B' " " " 10 "

D, D' " " " 24 "

E, E' " " " 26 "

5 The remnants of the ") ^4 cells of about the samesize (Fig. 22).

i two first macromeres \ ^' ^' " " ^ ^^^^^ ^^g^ ^^an the above.

66 cells.

1 68


[Vol. XI.

Beyond this stage I have not been able to follow the cell lineage, for surface views show that all visible parts of the cells are so nearly of the same size and form, that it is impossible to distinguish one from another. It appears, however, as if all


Bi.i. B2.1. Baj.



^ \

E2X E2<r.

Ea.1.4: E.2.1,l.E2.r.2.

E'2.U, , ^■'■*- ^'■•'•^E2.r.2.

^^-Ea.lr E'ai.



t.rx. hA.


y \ D.1>L.

D.l.a.1. D.1-3-2- ^^D.1.22.

Dfr. ND.'r.3.l.


Dfr.,. ^I^-'-"

BiA B'ti.

DiAORAM OF Cell Lineage.

cells (except A and A') underwent division at about the same time, for soon the whole surface of the ovum is made up of exceedingly small cells.


To understand the fate of A, A' it is necessary to return to the description of the ovum in its earlier stages. When the ovum consists of but four cells a distinct segmentation cavity exists, as is shown in section, Fig. 23. As the ovum returns to its resting stage this cavity becomes obliterated by the rounding of the outer surfaces and consequent sinking in of the cells (Fig. 24). This disappearance of the cavity is, however, merely temporary, for with renewed karyokinetic activity of the cells it is again formed. In the ten-celled stage (Fig. 8) the cavity reappears again in section (Fig. 25), but to be obliterated as before by the sinking in of the cells A, A', and also by the change of form of all the other surrounding cells (Fig. 27), particularly E, E'. At this stage the ovum becomes much elongated and quite deeply arched (Fig. 10), the cells A, A' being always on the concave side.

Fig. 28 shows a horizontal section through the sixteen-celled stage in which no trace of the cavity is to be found, while in the next figure (Fig. 29), a cross section shows a large cavity. This alternate formation and obliteration of the cavity continues until the ovum has obtained the magnitude of upward of sixty-six cells, when the cells A, A' sink into and completely fill the cavity, while the cells ^3, B' ^ and ^3, E' ^ (Figs. 16 and 18) draw together and cover the space in the surface left by A, A'. Sections of later stages show no trace whatever of this segmentation cavity. On the contrary. Fig. 33, a cross section through a distinctly later stage, shows an outer layer of small cells, while within are a mass of much larger cells evidently derived from A, A'. These inner cells form the mesentoderm of the adult, while the outer ones, all descendants of B, C, D, E, and B', C, B', E', form the ectoderm. At a still later stage (Fig. 34) the cells of both mesendoderm and ectoderm are all of about the same size, showing that the division of the cells A, A' and their descendants must have been quite rapid. The ectoderm is already quite diflferentiated from the central mass. Soon, however, it becomes distinctly differentiated into a two-celled layer, and the central mesentodermic mass seems to undergo a process of degeneration. Open spaces occur, and in the center the cells are less numerous, as is shown in

I 70 GA R DINER. [Vol. X I .

Fig. 35, a horizontal section through a very advanced embryo. This section suggests in its appearance a larval coelenterate planula of some kind. From the remnants of this mesentodermal mass the parynchym of the body is formed.

No trace of an alimentary tract is to be foimd at any time during the development, for I have sectioned with great care every stage up to and including the free-swimming young. During the summer of 1894 I obtained the ova of the undescribed species of ApJianostoma ( .-^ ) referred to above and compared its segmentation with that of Polychoeriis. In every detail the processes are alike in both : the formation of the eight cells by the continuous division of the original macromeres ; the subsequent division of each of these cells so as to form a thirtyeight-celled stage ; the existence of a segmentation cavity and its obliteration by the sinking in of the remnants of the two original macromeres, and throughout all of this the maintenance of the antero-posterior, as well as bilateral symmetry. All of these points have been carefully observed and compared, though no attempt was made to study the later stages of ApJianostoma.

The only work which I have seen which treats of the development of any of the Acoela is a paper by Mile. S. Pereyaslawzew (8) in which she describes the segmentation of ApJianostoma, Nadina, Proporus, Convoluta, CyromorpJia. She finds that in all of these ge7tera the segmentation process is identical. Unfortunately no figures are given in her paper, and it is impossible to follow the process from description alone with absolute certainty. It is apparent, however, that while my observations may differ from hers in a few minor details, yet we agree as to the general plan of segmentation. ■

W. Repiachoff in a " Nachtrag zu Pereyaslawzew" states that he also has studied the segmentation of the same form as Mile. Pereyaslawzew described, and finds that his observations with very slight exceptions confirm those made by her. He, however, states that he has seen in the " Archigastrula ahnliches Stadium" " eine deutliche Urdarmhohle." By " Archigastrula ahnliches Stadium " he probably refers to the stage shown in Figs. 14, 15, 16, and Fig. 27, which latter is a longitu


dinal section through the ten-celled stage. As will be seen in these figures, the ovum is much arched, and the cells abutting on A, A' appear to be pressing them into the cavity.

" Eine deutliche Urdarmhohle " doubtless refers to the blastopore-like opening left when A, A' finally pass into the segmentation cavity.

Very shortly after this takes place the cells ^3, ^3, ^'3, and B' T^ close over this depression, and it is completely obliterated except for the pigment which remains for some time. This, however, disappears, and it is impossible to determine whether the mouth of the future embryo is to be formed at this point or elsewhere.

Marine Biological Laboratory, Woods Holl, Mass., September 12, 1894.



1. Mark, Edward L. Polychoerus caudatus. Nov. gen. Nov. spec.

Festschrift zum siebenzigsten Geburtstage Rridolph Leiickarts.

2. Graff, Ludwig von. Monographie der Turbellarien. I. Rhabdo coelida.

3. Selenka, E. Ueber eine eigentiimliche Art der Kernmetamorphose.

Biol. Centralbl. Bd. I, 1 881-1882.

4. Lang, Arnold. Die Polycladen des Golfes von Neapel. Flora nnd

Fauna des Golfes von Neapel. xi. Monographie.

5. Wheeler, William Morton. Planocera Inquilina. four. Morph.,

ix, 2.

6. NusBAUM, JdzEF. Ueber die Verteilung der Pigmentkornchen bei

Karyokinese. Anat. Anz. viii. Jahrg., No. 20, 1893.

7. Haberland, Gottlieb. Ueber den Bau und die Bedeutung der

Chlorophyllzellen von Convoluta Roscoffensis. Published in Die Organisation der Ttirbellaria Acoela von Ltidivig von Graff.

8. Pereyaslawzew, Mlle. S. Sur le d^veloppement des Turbellari^s.

Zool. Anz. viii. Jahrg., 1885.

9. Repiachoff, W. Nachtrag zu Pereyaslawzew's Ddvel. des Turb.

Zool. Anz. viii. Jahrg., 1885.















































Two-celled stage. First cleavage.

Four-celled stage viewed from above.

Four-celled stage after the rotation of B and B' .

Four-celled stage viewed from the side.

Six-celled stage viewed from above.

Si.\-celled stage viewed from the side.

Eight-celled stage viewed from the side.

Ten-celled stage viewed from the side.

Ten-celled stage viewed from below. Natural color. Drawn by R.

Ten-celled stage viewed from the side ; the ovum somewhat crescent Ten-celled stage viewed from above.

Twelve-celled stage viewed from the side as it goes into its resting

Sixteen-celled stage viewed from above.

Sixteen-celled stage viewed from below.

Sixteen-celled stage viewed from the side ; becoming crescent-shaped.

Eighteen-celled stage viewed from below.

Eighteen-celled stage viewed from below as if transparent.

Twenty-six-celled stage viewed from below.

Twenty-six-celled stage viewed from below as if transparent.

Thirty-eight-celled stage viewed from the side.

Journal of Morphologij Vol. XI.


\ i


B B'


C B (•■




c (••■'





A A'


c c



c r



15%. .

Kf. ?:ch. m


D.r: D'r



Dj. a A' B.s







Be Hr,



I-::-, K:.




Br, [!■: /.:;_

I!). B, B..' B- ..J . (l. : .. ,F:.i.r.

20. R, <^^- <■'■ a Cr



b:s jjii jij

£.y r-;;>—^C: \^:b:ii <^-2B:itaB:sji'

£.3 Bs

lUh Arst f Werner i WutUr. Frankfurt V'?



Fig. 21. Diagram of the thirty-eight-celled stage. The cells on the farther side indicated by dotted lines.

Fig. 22. Sixty-six-celled stage.

Fig. 23. Longitudinal section through the four-celled stage. Showing the segmentation cavity, S.

Fig. 24. The same as Fig. 23 during the resting stage. Showing the segmentation cavity obliterated.

Fig. 25. Longitudinal section through the ten-celled stage. Showing the segmentation cavity, S.

Fig. 26. Diagram of the sixty-six-celled stage. The cells on the farther side indicated by dotted lines.

I. The remnants of the original blastomeres which form the mesentoderm. II. The lineal descendants of cells B and B'.

III. The lineal descendants of cell C and C.

IV. The lineal descendants of cell D. V. The lineal descendants of cell E.

VI. The lineal descendants of cell D'. VII. The lineal descendants of cell E'.

Fig. 27. Longitudinal section through the ten-celled stage. Showing the segmentation cavity filled up by A and A' .

Fig. 28. Horizontal section through the sixteen-celled stage.

Fig. 29. Cross section through the sixteen-celled stage.

Fig. 30. Horizontal section through the twenty-six-celled stage.

Fig. 31. An ovum cut from the adult. Showing the first cleavage spindle.

Fig. 32. Pigment granules from an ovum of Polychaerus caudatus.

Fig. 33. Section through an ovum much more advanced than the sixty-sixcelled stage.

Fig. 34. Section through an ovum still older than that in Fig. 33.

Fig. 35. Horizontal section through a ciliated embryo.

Journal of Morph ology Vol. XI.

n. XI.

Oar diner del.

liihAr^uWerrerSimer, Frankfmt^M



The Triglidae have attracted the attention of European anatomists for more than three-quarters of a century on account of the remarkable finger-like processes of the pectoral fins. These processes, which have proved to be free fin rays, were found to be very richly supplied with nerves, and enlargements, or lobes, were found on the dorsal surface of the spinal cord, where these nerves united with it.

Special efforts have been made to discover sense buds or other end organs in the epidermis of these free fin rays.

The strong resemblance to such dermal appendages as barbels, led Merkel (i) to characterize them as wholly analogous in structure and function.

No one has hitherto succeeded in finding sense-organs on these rays similar to those found on barbels, and there are great differences of opinion in regard to the peripheral termination of the nerves in these organs.

In the hope of settling some of these questions, I undertook the study of the Gurnards found along the Atlantic coast.

I wish to acknowledge my great indebtedness to Dr. C. O. Whitman, Director of the Marine Biological Laboratory at Woods Holl, at whose suggestion the work was undertaken, and also to Dr. J. P. McMurrich for many valuable suggestions received during the progress of the work.

The representatives of the Gurnards found along our coast are different from those found in European waters.

Two species which are quite abundant at Woods Holl were studied ; namely, Prionotus palviipcs Storer, and Prionotus evolans Gill (14). These fish may attain the length of fifteen to eighteen inches and weigh one and a half to two pounds, but are generally much smaller.

178 MORRILL. [Vol. XI,

The food of these fish consists of crabs, shrimps, and small fish (14), but they will eat pieces of beef, fish, or shark, and are particularly fond of pieces of clam or snail.

They are very plentiful at Woods Holl during their spawning season, the latter part of May and the first of June. P. palmipes occurs in much greater numbers than P. cvolaiis.

The pectoral fins, in both species, are very large, being about one-third as long as the body and nearly as broad as long. They extend horizontally from the sides of the body, when expanded, somewhat in the manner of wings, and it is owing to this that they have received the common names of brown-winged and red-winged sea-robins, according to the color of the fins.

They are also known as flying-fish. The term "grunter," which is often applied to them, has reference to the peculiar sounds produced when disturbed.

Nervous System.

The central and peripheral nervous system of both species of Prioiiotiis were studied by dissections and macerations.

The differences between the two species are so small as far as the general arrangement is concerned, that a description will only be given of the species figured (/'./^/^wz/^j-, PI. XII,

Fig- 3) The most noticeable feature of the central nervous system

and the only part which will be considered here is the series of

paired enlargements, six in number, on the dorsal surface of

the spinal cord. These enlargements, or " accessory lobes " as

designated by Ussow (10), Fig. 3, ac.L, are associated with the

origin of the sensory roots of the first three pairs of spinal

nerves (I, II, and III). The two anterior pairs of lobes are

very indistinct ; in most cases the third pair is well developed,

and the fourth and fifth are closely crowded together as if they

had at one time been united. The posterior pair is much larger

than any of the others. The first spinal nerve (I) arises from

the first and second pair of "accessory lobes"; the second

spinal nerve (II) from the third pair, and the extremely large

third (III) receives fibers from the last five pairs of lobes.


The sensory root (Fig. 4, s.r.) of the third spinal nerve is more than ten times as large as its motor root (Fig. 4, vi.r^j and extends forward between the motor and sensory roots of the second, and first spinal nerves. The motor and sensory roots of the first and second spinal nerves, and the motor root of the third are about equal in size (Fig. 4).

A nerve of the brachial plexus (Fig. 3, c) unites the second and third spinal nerves near their origin.

The somewhat flask-shaped enlargements of the first three spinal nerves {g',g", and^'") just after they arise from the cord, contain the spinal ganglia.

The second spinal nerve (II) passes through a foramen formed by the clavicle and scapula. At the posterior border of the pectoral fin the nerve bends nearly at right angles to penetrate and follow the triangular space between the proximal ends of the paired bones which form the skeleton of each ray and the distal border of the brachial ossicles where the rays articulate with the ossicles. The nerve extends forward and downward nearly to the anterior border of the fin where it unites by cross fibers with the two or three branches from the nerve (Fig. 3, 3) which supplies the posterior free ray. A branch of this nerve (II) is sent to each half of every fin ray (Fig. 3, n.f.r. of II)' except the two or three anterior ones which are innervated by branches (Fig. 3, n.f.r.) from the nerve of the posterior free ray (Fig. 3, 3) as already noted.

The third spinal nerve (Fig. 3, III) passes under the pectoral girdle and divides into three large branches (Fig 3, 1,2, 3) quite near its origin. Each of these divisions, anterior (Fig. 3, i), middle (2), and posterior (3), is much larger than the entire trunk of either the first (I) or second (II) spinal nerves. They lie on the inner surface of the muscles of the fin, just beneath the skin.

The middle branch (2) divides into two parts (Fig. 3, A and B); the former innervates the second free ray, counting from the head, while the latter passes to the posterior surface of the third free ray and with the posterior branch (3) of the third spinal nerve (III) supplies that ray. The anterior branch of the third spinal nerve (i) furnishes the nerve supply for the first free ray.


In some cases the middle branch (2) of the third spinal nerve does not divide before reaching the base of the second free ray ; no branch is given in such a case to the third free ray. The nerves supplying the first and second free rays divide in each into two nearly equal rather large nerves and one or two small ones near the base of the ray. The larger nerves lie in the anterior and posterior portions of the rays and break up into smaller branches in their course, which can be traced to the papillated surface of the skin. There is considerable variation in the branching of these nerves.

Morphology of the Free Rays.

The three free rays of the pectoral fins in both species of Prionotus have the form of hooked finger-like appendages, the distal third of each being bent almost at right angles to the proximal two-thirds. When the fish is swimming the free rays are held close to the body, and are hidden by the fins viewed from above or at the side. The fish when resting quietly on the bottom of a tank or pool brings these rays to a position parallel to each other, with the somewhat knob-shaped tips near the sides of the head, touching the surface on which the fish is resting and along a line which makes an angle, posteriorly, of about eighty degrees with the long axis of the body.

These rays can be moved through an arc of 180°; a considerably greater freedom of motion than is possessed by the other rays of the pectoral fin. The free rays increase in size from the first or anterior to the posterior, the latter being much longer as well as larger than the first. In P. evolans, the free rays resemble normal fin rays to a considerable extent. They have, however, become imperfectly quadrilateral from the angle outward, and arc slightly enlarged. In P. palmipes there is considerable modification. Distally to the angle of the free ray, it is pentagonal in cross section with a reentrant angle on the faces which look backward when the ray is in the resting position (Figs, i and 2). The anterior face is the narrower and the outer the broader. The breadth of the distal portion of the free ray increases from the angle for about one-third of


its length and then gradually decreases to the end, making this portion of the ray resemble a pair of truncated pyramids placed base to base. Separating the anterior from the outer and inner faces are two narrow ridges (Fig. i, r and /), one on either side, which arise just proximally to the angle of the ray and passing outward gradually increase in height and breadth until they end abruptly in the knob-shaped terminal enlargement {k).

The ridges are covered with conical papillae which are closely confined to them until near the tip of the ray, where they spread out over the narrow intervening portion of the anterior face (Fig, i). Papillae also appear on the outer and inner faces. They gradually increase in number from the posterior edges of these surfaces until they completely cover them a short distance from the tip ; consequently the proximal fourth of the knoblike end of the ray is completely covered with papillae, those nearest the tip being the largest, .15 mm. in diameter. There are no papillae on the faces which form the reentrant angle (Fig. 2). The size of the papillae varies from .08 mm. to . 1 5 mm. in diameter. The smaller ones are scattered between the larger.

The largest papillae occur, as already stated, at the tips, while others of nearly equal size are found on the ridges separating the anterior from the outer and inner surfaces of the ray. The color of these appendages from the angle to the tip is lemon yellow, but is dotted with many stellate black pigment masses, scattered irregularly over their surfaces.

Skeleton of Free Rays.

The skeleton of each of the free rays consists of two tapering parallel osseous rods, each of which is composed of a great number of semi-transparent, cylindrical bodies joined end to end by thin opaque discs of a cartilaginous substance. Near their proximal ends the skeletal rods become completely ossified, the jointed structure wholly disappearing. The rods, which are arranged dorso-ventrally, diverge near their bases to form a triangular space. They are also slightly enlarged at their points



[Vol. XI.

of articulation with the metapterygial bone. The ventral rod has a flattened conical enlargement on its posterior border for the insertion of the muscles which move the ray. This projection has a height more than twice as great as the diameter of the rod. The halves of the skeleton of the ray approach each other distally. Their inner surfaces are quite firmly bound together by dense connective tissue. Each rod is nearly round in section near its base, but becomes considerably flattened dorso-ventrally in its distal half. The connective tissue in the free rays is very abundant and dense.

Muscles and Blood Supply.

Three pairs of quite large muscles control the movements of each free ray.

The similarity of the musculature to that described in Trigla makes it unnecessary to give a detailed description at this time.

The general arrangement of the blood vessels in the free rays will be indicated in describing the cross section.


In a young sea-robin, probably P. evolans, lo mm. in length, the anterior rays were as yet not in any way distinguished from the others, since their separation from the fin had not begun, being still united to them by the web of the fin. During the

first day after its capture this fish increased 2 mm. in length, and the membrane connecting the three anterior rays had begun to disappear (Fig. i). The separation began by the


shrinking away of the web between the rays, apparently by absorption. This gradually increased until first the outer parts of each ray were free (Fig. 2). The fish now measured 14 mm. in length. Toward evening of the second day, when the fish was 17 mm. in length, the rays were almost entirely free, only a very small portion of the connecting membrane remained, (Fig. 3) and this eventually disappeared. The process was almost identical in several other specimens which were examined.


In studying the anatomy of the nerve-supply of the pectoral appendages, the anterior half of the body was macerated from one to three days, according to the size, in 40^ nitric acid. This was rendered necessary by the thick osseous covering of the head of the fish. In the dissection of these acid preparations, porcupine quills were used.

The nerves when freed from the muscles and connective tissue were preserved in 70^ alcohol.

For the study of the histology of the free rays, the following reagents gave satisfactory results : Kleinenberg's picro-sulphuric acid, osmic acid, Merkel's fluid, and Miiller's fluid. The specimens were washed and preserved in alcohol, passing from 50/0-70%, where they remained until used.

The results obtained by the use of Merkel's fluid were in some respects the most satisfactory. The parafflne method was used for obtaining sections which were cut 5 mm. in thickness. The tissues were stained, in mass, with Delafield's haematoxylin or borax carmine. In studying the peripheral terminations of the nerves in these rays, Dogiel's (9) method of using methylen blue gave very suggestive results, but the picrate of ammonia, used to fix the blue, macerated the epidermis so that the ultimate nerve-endings could not be studied as carefully as would have been possible if the epidermis could have been sectioned in place. The best demonstration of the nerve-endings which I obtained, aside from Golgi, was by the use of Mitrophanow's gold chloride method as quoted by Macallum (7). Perfectly fresh tissue was treated with a i % solu

184 MORRILL. [Vol. XI.

tion of gold chloride for one hour, then rinsed in distilled water and placed in a 10% solution of formic acid, and kept in the dark for twenty-four hours. Very little difference was found in the results obtained by treating the tissue first with 10% formic acid for fifteen minutes and then for fifteen minutes with I % gold chloride. The gold was reduced in this case in a weak solution (2 or 3%) of formic acid in the direct sunlight.

Tissues stained by these methods were imbedded in parafifine by the usual methods, and sections 2 mm. in thickness were cut and mounted in Canada balsam.

The rapid Golgi method gave much more complete demonstrations of the distribution of the nerves in the epidermis than any other method.


Cross sections of the free rays of both species of PrionoUis were made about half an inch from the tip of the ray. In P. palmipes the section is somewhat quadrilateral in form, but three of the faces have reentrant angles, making seven imperfectly marked sides as shown in the outline (Fig. 5).

Covering the epidermis of the free rays is a semi-transparent layer composed of hyaline prisms, each forming a cap over a single cell and probably secreted by it. A similar layer is called the cuticle by Jourdan (4) in his paper on Peristedion. The cuticle is easily separated from the underlying epidermal cells by the action of reagents. It is thinnest over the papillae, and thickest on the surfaces which look backward {Fig. 5 c) when the ray is in the resting position. This cuticular layer is very slightly stained by reagents, and shows no nuclei.

The epidermis consists of five or six irregular layers of closely crowded cells. The epidermis will be considered more fully in connection with the nerve endings.

Numerous nerve trunks of various sizes are arranged, as seen in the section (Fig. 5 ;/), nearly parallel to the outer and inner faces of the ray, but lie considerably below the skin. They are closely crowded together around the flattened skeletal bones (Fig. 5, I and 2) which occupy the center of the ray. The sections of the parts of the axial skeleton are somewhat crescent


shaped with their long diameters extending in an antero-posterior direction. The parts of the axial skeleton are separated internally by connective tissue, which holds them firmly in place.

The blood vessels (Fig 5, b) are arranged in four groups about the central skeleton, each face having its group.

In P. cvolans the section is much more nearly quadrilateral in outline (Fig. 6). The same general arrangement of parts exists as in P. palmipcs, as will be seen from the figures.

Over the papillae (Fig. 7) of the free rays the cells forming the outer layer (Fig. 7, /) of the epidermis in P. palmipes are somewhat columnar in shape. The outer portion of each cell stains very imperfectly, while the large nucleus which is found near the inner end takes the stain readily.

The inner row of cells (Fig. 7, b) has at this portion of the ray been transformed into a layer of cylindrical epithelial cells placed perpendicular to the basement membrane on which it rests. This layer constitutes more than one-third of the entire thickness of the epidermis.

There seems to be considerable uniformity in the position of the nuclei of the outer layer of cells, since all are placed with their long axes transverse to the long axis of their respective cells. In the inner row of cells (Fig. 7, b) the long axes of the nuclei and cells coincide. The intermediate rows of epidermal cells have not undergone any modification except that large spindle-shaped, deeply staining cells arranged perpendicularly to the surface of the papillae and broadest on their inner ends are found quite regularly among them, but varying in position.

The epidermis is still more highly modified on the surfaces forming the reentrant angle. The outer layer has become thickened, and the cells have assumed a spindle form. The inner layer is similar to that found over the papillae, but much thinner, and the intermediate cells have become more than twice as numerous as they are elsewhere on the ray. The spindleshaped cells are more abundant than over the papillae. There is no sharp line of demarcation between the epidermis on this portion of the ray and that over the papillae, as one passes gradually into the other.

1 86 MORRILL. [Vol. XI.

There are no papillae on the surface of the free rays of P. evolans. The epidermis closely resembles that already described for the faces of the reentrant angle in the free rays of P. palDiipes, except that the spindle-shaped cells are crowded together at certain points (Figs. 8, g, and lo), and the cuticle is thinner and does not show perpendicular striae.

Nerve Termmations .

The nerves in the free rays form a plexus just beneath the surface of the epidermis (Figs. 7, 9, 10, 13, and 16, sp). The fibers in this plexus are very closely crowded under the papillae in P. palmipcs, and less so beneath the longitudinal ridges in P. evolans. Nerve fibers (Figs. 7, 8, 9, and 10, iif) from the plexus, in both species of Priojiotus, penetrate the basement membrane and pass out between the cells of the inner layer of the epidermal cells, where they divide, sending numerous branches in all directions along the distal ends of the inner layer of epidermal cells. These fibers soon curve outward, a large proportion of them ending free just below the cuticle, while a smaller number are directly connected with spindleshaped cells (Figs. 10, II, 13, 15, 16, and 17) the outer ends of which extend to the cuticle. A single detached cell and nerve fiber is seen in Fig. 17 from a methylen blue preparation.

The gold chloride preparations agreed very closely with those obtained by Golgi's rapid method. The latter (Figs. 12-16), however, showed much greater detail, and brought out the epidermal plexus with great distinctness, a point very imperfectly shown by the use of gold chloride.

The manner of the nerve ending in the epidermis, between the papillae (Fig. 12) in P. palmipes, was similar to that over the papillae, except that the nerve fibers were widely scattered and nerve cells rarely found. The peripheral nerve fibers in P. evolans were similar to those in the general epidermis of P. palmipcs, but are fewer in number and more easily studied by the gold chloride method.



Dr. Harrison Allen (11) found four layers of epidermal cells in P. palviipes, but does not describe the modifications which arise in the epidermis over the surface of the papillae, and does not describe the manner of the nerve endings.

As already noted, the Triglidae have received considerable attention from European anatomists. In 1808 Samuel Collins discovered the enlargements on the spinal cord, and in 181 1 Tiedeman (12) observed that the free rays were in some way related to the spinal enlargements. The latter also described and figured the musculature of the free rays.

Deslongchamp, who was the first to observe these fish in captivity, claimed that the free rays were organs of locomotion. Tiedeman (12) records having seen a Trigla move on the deck of a ship by means of the free rays.

The muscles of the free rays were carefully studied by Deslongchamp, and later by Jobert (2). The latter states that there are many anomalies.

Jobert (2) concludes that the free rays are modified fin rays.

The cross sections of the free ray of Trigla, as shown by Jobert (2) and Zincone (3), are elliptical, with the antero-posterior diameter nearly twice as great as the one at right angles to it. The general arrangement of the axial skeleton nerves and blood vessels is essentially the same as in Prionotiis. No detailed drawings of the epidermis are given.

From the descriptions given by Jourdan (4), Jobert (2), and Zincone (3), the epidermis of the free rays of Trigla resembles that in Peristedioji more than that in Prionotiis.

There is a sub-epidermal nerve plexus as in Prionotiis. Merkel (i) was able to trace the nerve fibers into the epidermis, but could not follow them. Jobert (2) by the use of gold chloride, found each fiber connected with a small terminal body (corps epidermique) in the epidermis which measured .004 to .005 mm. in diameter. Zincone {3) claims to have found nerve fibers continuous with spindle-shaped cells.

Jourdan (4) found the non-medullated nerve fibers of the free rays of Peristedion terminating in small papillae on the

1 88 MORRILL. [Vol. XI.

surface of the dermis insinuated between the cells of the basal layer of the epidermis. In some cases this investigator observed a small terminal enlargement in the papillae, the exact location of which was not indicated.

The drawings of the dorsal surface of the brain and spinal cord of Trigla adriaiica given by Tiedeman (12), together with his descriptions, show considerable difference between the first three spinal nerves in Trigla and Prionottis.

The first spinal nerve in Tiigla is the largest, and arises by three roots, each from the side of an enlargement. The second spinal nerve arises from the fourth enlargement only, while the third spinal nerve arises by two roots, one from the fifth and the other from the sixth enlargement.

The six pairs of spinal enlargements, or accessory lobes (10), are small and appear to be equal in size and distinctness, which is far from being true for Prionotns.

Tiedeman (12) found that the fibers of the posterior root of the third spinal nerve were generally very fine, while those of the anterior root were quite coarse. Most of the nerve trunks contained both kinds of fibers, or, as in the case of the branch to the swimming bladder, the fibers were all of the coarser variety. A similar difference in the size of the nerve fibers was observed in Priouotus.


When specimens of Prionotns were placed in a tank of water in the bottom of which there were several inches of sand, the fish buried themselves in the sand very quickly by a rapid rolling movement as they rested upon its surface. The sand was thrown out on either side and settled over the surface ofthe body so that in most cases only the eyes, top of the head, and the tip of the nose were visible. Two openings, one on either side of the posterior dorsal portion of the gill-covers for the escape of the water which was taken into the mouth, became visible as the water was forced out.

Little or no attention was generally paid to food for a few moments after it was placed in the water, but as the fish swam over it, as it lay upon the sand or bottom of the tank, the free


rays came in contact with it. The fish immediately began to move the free rays much more rapidly than usual, passing them over the piece of meat or fish several times, and then by a rapid lateral movement snapped it up.

After finding two or three pieces of food the fish were guided by sight, apparently, as they swam across the tank to catch the food before it reached the bottom when fresh pieces were thrown to them. In the large fish ponds of the United States Fish Commission the fish often swam ten feet and were able to secure the food before it reached the bottom, the water being from four to five feet deep. The fish frequently became so excited that any light-colored object was taken into the mouth, such as bones, small pebbles, and pieces of shell. In one case two large specimens of P. evolans rushed at the same piece of meat and missing it caught each other's jaws instead.

Inedible objects were quickly dropped. Tainted meat was quickly rejected from the mouth although it might be taken in again almost immediately, only to be thrown out. The same piece of tainted meat was sometimes taken by several fish in succession and then lay unnoticed. Pieces of meat dripping with turpentine were swallowed as readily as the pieces which had not been treated with it.

In order to test the use of the free rays independently of sight the crystalline lenses and cornea were removed from some fish and in other cases the cornea was covered with varnish, balsam, or tar. The repeated experiments were negative in their results, as the fish paid no attention to the food, even when it was placed in contact with the free rays.

Bateson (13) claims that the Triglidae do not take food at night. I have not been able to prove that Prionotus takes food in darkness.

To test the effect of removing the free rays, fish were selected which took food readily and the free rays were all amputated close to the body in some cases, and in others they were left of different lengths.

When the free rays were all removed the fish occasionally detected food by sight. In one case I saw normal movements of the stumps of the free rays when food fell on the bottom of

I90 MORRILL. [Vol. XI.

the tank near the mutilated fish. The ends of the stumps of the free rays were more than an inch from the food. No effort was made to take the food.

The movements of the free rays of PrionoUis resemble walking so closely that it is natural that the free rays should be looked upon as locomotor organs. The early observers of the European Triglidae held this view. The movements of the free rays are exactly the same when the fish swim upward in contact with the glass sides of the tank as when swimming along the bottom.

These fish have been observed several times to turn over small stones and shells, as they swam along the bottom, by means of the free rays.

The place covered by the stone or shell was subsequently thoroughly examined by means of the free rays, apparently in the search for food.

The method of nerve termination described by Ranvier (8) and other authors (5 and 7), as characteristic of organs of touch is very much like that found in Prionotus. This together with the physiological evidence, so far obtained, from the study of the Gurnards, strengthens the opinion that the free rays have been modified for tactile purposes, and that they are mainly if not altogether used in searching for food.

Hamilton College, Clinton, N. Y.



r. Merkel. Ueber die Endigungen der sensiblen Nerven in der Haut der Wirbelthiere. 1880.

2. JOBERT. Etudes d'anatomie comparde sur les organes du toucher chez

divers mammiferes, oiseaux, poissons, et insectes. Annales des Sciences Natiirelles^ v. Ser., Tom. xvi. 1872.

3. ZiNCONE. Osservazioni anatomiche di alcune appendici tattili dei

pesci. Rendiconto della S. R. Accad. di Napoli. 1876.

4. JouRDAN. Structure histologique des barbillons et des rayons libres

du Peristedion cataphractum. Afchiv. de Zool. Expcr. et Gen., ii. Ser., Tom. viii. 1890.

5. Fajersztajn. Terminaisons des nerfs dans les disques terminaux

chez la grenouille. Archiv. de Zool. Exper. et Gen., ii. Sdr., Tom. vii. 1889.

6. Macallum. Termination of Nerves in the Liver. Quar. Jour. Mic.

Soc, vol. xxvii, N.s. 1887.

7. Macallum. Nerve Terminations in the Cutaneous Epithelium of the

Tadpole. Quar. Jour. Mic. Soc, vol. xxv\, ti.s. 1885.

8. Ranvier. Termination of Nerves in the Epidermis. Quar. Jour.

Mic. Soc, vol. XX, N.s. 1880.

9. DoGiEL. Methylenblautinktion der motorischen Nervenendigungen

in der Muskeln der Amphibien und Reptilien. Arch. f. Mikros. Anat., Bd. 25, Heft. 3. 1890.

10. Ussow. De la structure des lobes accessoires de la moelle epiniere

de quelques poissons osseux. Archiv. de Biol., Tom. iii. 1882.

11. Dr. Harrison Allen. Proc. Acad. Nat. Sci. oj Phila., p. 377, 1885.

12. Tiedeman. Von dem Hirn und den fingerformigen Fortsatzen der

Triglen. Meckel's Arch., Bd. 2. 1816.

13. Bateson. Sense Organs and Perceptions of Fishes. Jotir. Marine

Biol. Assoc, i. No. 3, N.s. 1890.

14. GooDE. The Fisheries and Fishery Industries of the United States,

Sect. I.



Fig. I. Anterior surface of free ray oi P. palmipcs Storer, when in the resting position, showing ridges r and r' covered with papillae and terminal knob k. Drawn by R. Takano. X 6.

Fig. 2. Posterior surface of same showing reentrant angle a, marginal papillae /, and terminal knob k. X 6. Drawn by R. Takano.

Fig. 3. Maceration showing dorsal view of anterior portion of central nervous system of P. palmipes Storer, with the branches of the three first spinal nerves, \, II, and III ; Ol.Ji. and Ol.h. nerve and bulb ; Op.it. and Op.l. optic nerve and lobe ; Cb. cerebrum ; chl. cerebellum ; ac.l. six pairs of accessory lobes on .spinal cord ; I, 2, 3, main branches of third spinal nerve (III) : A and B, divisions oi 2 ; a and d, principal secondary divisions of i ; c and </, divisions of A ; e, anterior branch of 3; /a. continuation oi B ; k, I, and m, portions of dermal covering at tips of rays to show distribution of the nerves distally ; n.f.r. nerve fibres of rays, one branch being distributed to each half of the ray; g, h, i, and/, branches of 3; <:, a nerve of the branchial plexus connecting second, II, and third, III, spinal nerves; g"\ ganglionic enlargement of the third, III, spinal nerve. X 2.

Fig. 4. Ventral surface of same preparation ; 771. r. motor root ; s.r. extremely large sensory root ; In. infundibulum ; l.i. lobes inferiores ; IV and V, fourth and fifth spinal nerves. X 2.

Fig. 5. Cross section of free ray of P. pal77tipes Storer, made ^ inch from distal end ; a, p, 0, and /, anterior, posterior, outer, and inner surfaces ; c, thickened cuticle shown on surfaces forming reentrant angle r ; i and 2, bones of ray; d, blood vessels ; «, nerves. X 30.

Fig. 6. Cross section of free ray of P. evolans Gill, ^ inch from distal end. The letters are the same as in Fig. 5. X 30.

Fig. 7. Vertical section of a papilla of free ray of P. Pal77iipes Storer ; c, cuticle ; b, proximal layer of epidermal cells ; /, peripheral layer ; sp, subepidermal plexus ; 7tf, nerve fibers. Obj. 6 Verick ; cam. luc. Abbe. Gold chloride preparation.

Fig. 8. Cross section of epidermis of P. evolans Gill. The parts are indicated as in Fig. 7, b.vi. basement membrane ; s.c. sensory cell. Obj. 6 Verick ; cam. luc. Abbe ; gold preparation.

P'lGS. 9 and 10. Similar to Fig. 8. Figs. ii. Sensory cells from P. evolans Gill, showing nerve fibers, 7if, cuticle, c ; nucleus of cell c and terminal enlargement of cell i.e. Gold chloride preparation. Zeiss, oc. 4, y'j oil im.; cam. luc. Abbe.

Fig. 12. Golgi preparation of epidermis of P. palmipes Storer, showing distribution of nerve fibers. Zeiss, oc. 4, obj. D.

Fig. 13. Similar to Fig. 12, showing portion of subepidermal and epidermal plexii with sensory cell. Zeiss, oc. 4, -^^ oil im., cam. luc. Abbe.

Fig. 14. Similar to Figs. 12 and 13. Oc. 4, obj. D.; cam. luc. Abbe.

Fig. 15. Golgi preparation similar to 12, 13, and 14. sp, nerve fibers as they leave the subepidermal plexus. Zeiss, oc. 4, y^^ oil im.; cam. luc. Abbe.

Fig. 16. Similar to preceding. Zeiss, oc. 4, obj. D.; cam. luc. Abbe.

Fig. 17. An isolated sensory-cell and nerve fiber from methylen blue preparation.

.IiiiiiiKii of .Vorp/ioloyy. I'ol.XI.


R D Momn d/l

Litk. Anst y. Werneri^^Ur, rror :■'■■■



In December of 1891, sections of Lumbricus agricola, Hoffm.2 were prepared in the Laboratory of Animal Morphology of the University of Michigan, in which the epidermal cells presented in many places an arrangement similar to that which is found in the vertebrate "taste-buds." At the suggestion of Prof. J. E. Reighard, the writer and Miss M. F. Randolph undertook an examination of the literature and a study of the structure, nerve-supply, and distribution of these apparent sense-organs. Soon after this, there appeared a paper by Lenhossek ('92), in which it is stated that the sensitiveness of Lumbricus is due to isolated nerve-cells scattered through the epidermis, and that these nerve-cells are never grouped into sense-organs. An examination of the literature brought out the fact, not referred to in Lenhossek's paper, that epidermal sense-organs had already been seen and described by Leydig ('65), Mojsisovics ('77), Vejdovsky ('84), Ude ('86), and Cerfontaine ('90), in Lumbricus agricola, and in other species of Lumbricidae. Our work then resolved itself into an attempt to determine accurately the facts at first hand in order to understand the conflict between the account of Lenhossek and the accounts of earlier writers. As Miss Randolph did not return to the University the next year, I have since carried on the work alone.

For the sake of clearness, I shall give first a continuous account of my own observations, and shall reserve until the end of the paper all discussion of the work of others,

^ Work from the Laboratory of Animal Morphology of the University of Michigan, under the direction of Prof. J. E. Reighard.

2 This species, upon which all my work has been done, has the characteristics of L. herculeus, Sav. as given by Ude ('86).

194 LANGDON. [Vol. XI.


Sections stained with Kleinenberg's haematoxylin have been used for a study of the finer structure of the sense-organs and the course of the nerve-bundles passing to the epidermis. The nerve-supply of the sense-organs and the course of the nerve-fibres in the epidermis and in the central nervous system have been studied by means of Golgi's shorter silver nitrate method. The distribution of the sense-organs has been determined from surface views of the removed cuticula.

The fact that these sense-organs of the epidermis have been so often overlooked by competent observers seems to warrant an account of the methods employed by me, although this account contains little that is new.

It is not difficult to obtain well stained sense-organs in a good state of preservation, if sufficient care is taken in all stages of the preparation. To insure successful cutting of sections, it has been found best to feed the worm on wood-pulp in the preparation of which no chemicals have been used. In killing, great care must be taken to avoid contortion or an excessive discharge of mucus from the gland-cells of the epidermis. The method found most successful is the alcohol method given by Cerfontaine ('90), p. 337. But it has been found best to have the alcohol act more slowly; 70% alcohol was used in place of the 96% recommended by Cerfontaine, and it was made to drop on the filter paper at the rate of sixty drops a minute. In about an hour the worms are so stupefied that all the paper may be removed except the small piece on which the alcohol drops, and the alcohol may be made to drop more rapidly. At the end of two hours the worms are placed in 50% alcohol for an hour, in 70% for twenty-four hours — during which this alcohol should be several times renewed — in 96% for twenty-four hours, and then preserved in fresh 96%. It is better to use for sections worms that have been recently prepared. When these preserved worms are cut into pieces, it is best to remove the wood-pulp from the alimentary canal with small forceps in order to facilitate sectioning. Parts of the worm chosen for study are run through absolute alco


hol, cedar oil, soft paraffin, one-half hard and one-half soft paraffin, and finally embedded in the latter. Each change from one reagent to another must be made gradually, and the temperature of the paraffin bath must not rise above 54° C. The minute structure of the sense-organ does not show well in sections cut over \o\i thick.

The serial sections are straightened out and fixed to a slide by the alcohol method. The paraffin is then dissolved in turpentine, and the sections run back into 70% alcohol. They are then stained for about twenty-four hours in a weak solution of Kleinenberg's haematoxylin, prepared according to the formula given in Lee's Vade Mecum. It has been found best to mix the stain about twenty-four hours before using, to use a weak solution, and stain on the slide. When the sections are stained, they are washed in 70% alcohol and then with a saturated solution of sodium bicarbonate in 70% alcohol. If overstained, they may be treated with weak acid alcohol before using the bicarbonate. The sections are then run into absolute alcohol, cleared in clove oil, and mounted in xylol-balsam.

The silver nitrate method used is, in general, that given by Dr. C. J. Huber in Anat. Anzeiger, Jahrg. 7 (1892), p. 587, Dr. Huber kindly suggested two changes which add to the permanence of the preparations : first, to leave the turpentine on fifteen minutes instead of five ; second, to use pure balsam instead of turpentine-balsam. At the suggestion of Professor Reighard the creosote was replaced by cressylic acid, which is less disagreeable to handle. Care must be taken to remove all traces of the creosote or cressylic acid, otherwise the sections are sure to spoil. For sectioning, pieces of the worm are fastened to a wooden block which can be clamped into the object-holder of a sliding microtome. The end of the block is covered with thin collodion ; the object is taken directly from the silver nitrate, in which it can be left until needed, pressed into the collodion, and then covered with a few drops of the latter. If exposed to the air, the collodion will soon harden so as to fix the object firmly in place. It is well to moisten the collodion over the tissue with a little 96% alcohol to keep it from contracting the tissues while it is hardening. When the sections

196 LANGDON. [Vol. XI.

are cut, the surrounding collodion is easily removed. When it is desirable to use the oil-immersion on these sections, they should be cut about lOyu. thick and the cover-glass pressed down finnly.

The following method was employed in the preparation of the cuticula for a study of the distribution of the sense-organs. A large worm which has been previously hardened in alcohol is placed in 70%. Beginning at one end, the worm is cut into pieces of 20 metameres each. As each piece is cut off it is placed in 50% alcohol, a longitudinal cut is made with a sharp razor through the body wall in the mid-ventral line, and the edges of the cuticula along this cut are loosened with fine forceps. Then by holding the cuticula firmly at one edge and gently rolling the piece of the worm out from under it, the cuticula is peeled off. All this is done under the 50%. The cuticula usually is ready to peel off about as soon as the piece of the earthworm is placed in the 50%. A glass slide is then held in the alcohol, the cuticula floated onto it, and pressed down with a camel's-hair brush ; the slide is removed from the alcohol, and allowed to dry. No cover-glass or further preparation is needed. With a little practice perfect preparations of the entire cuticula may be thus made. Cerfontaine ('90) places the worm in 30% alcohol for three or four days and removes the cuticula entire, afterward cutting out pieces of it for study. This procedure is apt to macerate the cuticula, and the removal of the latter entire renders more difficult the cutting of it into sections for mounting. Cerfontaine says : " La structure de la cuticule ne peut s'observer que sur des preparations fraiches, parce que I'examen doit se faire dans I'eau ou, mieux encore, dans I'alcool." Cuticula dried on the slide as described above has been found to be in perfect condition for study over a year after preparation, and seems to show its structure better than when examined in water or alcohol.

A Study of Haematoxylin Pi'eparatio7is of the Epidermis.

General structure of the epidej'mis. — The epidermis of Lumbricus agricola is covered exteriorly by a thin cuticula composed of at least two layers of fibres ; the fibres of one layer


are at right angles to those of the other layer, and the fibres of both layers are at an angle of 45° to the long axis of the body. The inner surface of the epidermis rests on a thin basementmembrane, which is apparently composed of connective tissue and separates the epidermis from the circular muscle-layer beneath. The epidermis itself is composed of two layers of cells, a superficial and a basal layer, each one cell deep. The cells of the superficial layer are of two kinds, supporting cells and gland-cells. A supporting cell is almost columnar in form ; it has a square-cut top, a base prolonged into several processes, and an oval nucleus at about the middle of its height. A glandcell is "goblet-shaped." It has an external opening through a pore in the cuticula above it for the discharge of its secretion, and the nucleus is usually forced into the base of the cell by the accumulation of the secretion. The base is sometimes broad and entire, and sometimes divided into basal processes.

The cells of the basal layer are imbedded between the basal processes of the cells of the superficial layer. A basal cell is usually rounded and contains a rounded nucleus ; sometimes it extends toward the cuticula for a varying distance between the cells of the superficial layer. Most writers state that the supporting cells are probably changed into gland-cells, but Cerfontaine ('9o) believes that both supporting and gland-cells are formed from these basal cells. Intermediate stages between the basal cells and both kinds of superficial cells are so easily found as to leave in my mind no doubt as to the correctness of his statement. All the cells of the basal layer rest directly on the basement-membrane. All the cells of the superficial layer which do not rest directly on this membrane reach it by means of their basal processes. (See Cerfontaine ('SO) for figures of these various cells.) Imbedded between these various kinds of epidermal cells are groups of sensory cells collected into definite sense-orzans.

The sense-orgajis of the epidermis. — The sense-organs in the anterior metameres have been studied as a type (PI. XIII, Fig. i). Each organ has in general the form of an ovoid, the smaller end of which projects into a raised spot in the cuticula above it; the broader end is flattened, and rests on the basal mem

1 98 LANGDON. [Vol. XI.

brane of the epidermis. This ovoid may be broad or narrow, and is usually more or less irregular; the base is sometimes distorted by the presence under it of a bundle of nerve fibres. The greatest width is usually just above the base, occasionally it is at the base. In a cross section of a sense-organ its outline rarely appears as a circle, but it is usually a little flattened or otherwise distorted.

The sense-organs in the middle zone of an anterior metamere vary in height from 80 to ioo/a. At its summit such an organ may be from 18 to 28/x wide; at the widest part, from 40 to 60 [I wide.

The lateral limit of each organ is clearly defined by the layer of supporting cells around it. This is shown in longitudinal sections, in which the sense-organ is clearly outlined on each side by one or two supporting cells, which follow the outline of the organ and which are much flattened, evidently by pressure from within the sense-organ. In a cross section of a sense-organ, the inner (in relation to the organ) walls of these adjacent supporting cells are found to form a continuous membrane around the organ. It is thus seen that the cells of the sense-organs are enclosed in a cavity whose walls are formed by one or two layers of supporting cells which differ from the other supporting cells only in their flattened form. These may be called the covering cells of the sense-organ.

Within the receptacle thus formed by the covering cells, lie the two kinds of cells belonging to the sense-organ, small basal cells and long slender cells which extend from the basal membrane of the epidermis through the cuticula. The small cells are sometimes evenly distributed along the base of the senseorgan, and sometimes collected into a little group in the center of the base or toward one side. Each cell is irregularly rounded and almost filled by a rounded nucleus — in fact, it does not differ from the small cells of the basal layer of the epidermis. I have not been able to determine the function of these basal cells of the sense-organ; it seems to me possible that they are the cells which are to produce new sense-cells, but I have found no intermediate forms. The long cells, which are the true sense-cells, occupy the main part of a sense-organ; there


are from 35 to 45 in one of the larger sense-organs. These sense-cells stain about the same color as the supporting cells, but appear more finely granular and differ from them greatly in form. I have found no intermediate forms between the sense-cells and the supporting cells. The supporting cells are either of the same width throughout their height, or slightly wider at base and apex; they have clearly defined cell-walls, and the walls of one cell are closely applied to those of surrounding cells; on account of mutual pressure the outline of a cross section of a cell is nearly hexagonal. The sense-cells, on the other hand, are much narrower at base and apex; in the upper part of each cell it is impossible to distinguish any cell-wall, and it is difficult to see one in the lower part ; each cell stands alone, clearly separated from its fellows by a space which appears as if filled by a fluid, and a cross section of a cell is circular or elliptical instead of hexagonal. In a longitudinal section through the center of a sense-organ one might at first suppose these cells had been torn apart in the preparation.

The greater part of the sense-cells have their nucleus at or below the middle of their height; a few cells, usually situated between the center and the lateral surface of the sense-organ, have their nucleus near the cuticula. The part of the sensecell in which the nucleus lies is always the widest part, and is almost completely filled by the nucleus. If the nucleus occurs near the cuticula, the cell tapers into a very slender, fibrelike base. If the nucleus is found at the middle height of the cell or near the base, the base of the cell does not become so slender, and often sends off several basal processes which pass to the basement-membrane between the small basal cells. Above its nucleus, each sense-cell tapers into a slender part which looks like a delicate strand of protoplasm and reaches to the cuticula. Each cell terminates in a still more slender, hair-like process which passes through a canal in the cuticula and projects stififly for about 2/x above the surface. The varying position of the nuclei in the sense-cells seems unimportant, and is probably brought about by the cells accommodating themselves to the space in which they lie. The nuclei

200 LANGDON. [Vol. XL

generally appear elliptical in longitudinal section; some have a truncated top, and a few have one side concave. In a cross section of a nucleus it appears elliptical, circular, or almost triangular. In each nucleus several large nucleoli and usually a few chromatin threads can be made out. The nuclei of the sense-cells cannot be distinguished from those of the supporting cells.

Over each sense-organ the cuticula is elevated and much thinner than elsewhere. These two features render it concave on the side next to the sense-organ, and convex on the outer surface. In longitudinal sections through a sense-organ, the fine canals through which the sense-hairs pass may be easily seen piercing this place in the cuticula. When the sense-cells are straight and their upper parts quite slender, the hairs show plainly above the cuticula ; but when the upper parts of the sense-cells are thickened or thrown into sinuous curves, the hairs are found to be partly or wholly withdrawn into the cuticular pores. This sinuous appearance is often seen in alcoholic material, and may be accounted for in two ways : the gradual stupefaction of the worm in alcohol may cause a contraction of separate portions of the protoplasm instead of an even contraction of the whole cell, or there may be first an even contraction, and afterwards a relaxation of part of the protoplasm. When a sense-hair is drawn into its canal, the part which normally lies in this canal is below the cuticula. This part is then seen to be of the same finely granular protoplasm as the part of the cell below it, and it generally shortens and widens. This contraction of the sense-cells and withdrawal of their hairs into the pores in the cuticula may be a normal process in response to external irritation, but is probably an unnatural contraction caused by the alcohol. The more delicate nature of the sense-cells and their comparative freedom from contact with each other permit reagents to act on them to a greater extent than on the other epidermal cells.

In these haematoxylin preparations, bundles of lightly stained fibres are found passing through the circular muscle-layer and pressing against the basement-membrane of the epidermis. Each bundle is enclosed in a nucleated membrane and its


fibres can often be traced through the basement-membrane into the base of the epidermis, but it is impossible to discover their further course in these preparations. That these bundles of fibres are nerves can be proved by tracing them, in serial sections, to the central nervous system. To avoid repetition, a description of the course of these nerves will be left until the silver nitrate preparation is described.

In these preparations, the sense-organs have been found in the epidermis of every region of the body except in the clitellum, but they are most numerous in the anterior and posterior metameres. In longitudinal sections, these organs are found most often in the cephalic border and in the median zone of a metamere. In cross sections, the most noticeable ones are found in sections passing through the setae. The form of a cephalic border of a metamere — curving down as it does to the intersegmental groove — causes a cross section through this region to pass obliquely through the epidermis. Therefore the sense-organs in the cephalic border of a metamere are easily overlooked in such sections. The sense-organs in the metameres back of the clitellum differ from those described for the anterior metameres merely in size. As the epidermis is thinner in these metameres, the organs are smaller — usually about 45/Lt high and i6/a wide. In a few caudal metameres, the sense-organs again increase in size.

The sense-organs not only occur in the epidermis, but Miss Randolph found them among the epithelium cells lining the buccal cavity. So far as I know, they have not been found in this region before. The sense-organs are here found as far caudad as the median zone of the fourth segment, i.e., almost to the caudal limit of the buccal cavity. In this entire tract the sense-organs are larger and more numerous in the dorsal than in the ventral epithelium. In a given section through the cephalic half of the buccal cavity, the number of sense-organs in the dorsal wall varies from i8 to 31, and in the ventral wall from o to 7. In the caudal half of the buccal cavity the same comparative difference exists, but the organs are fewer in number. This difference in the relative number of orscans in the two walls may be due to a difference which exists in the form

202 LANGDON. [Vol. XL

of these walls. The dorsal wall curves evenly ventrad into the buccal cavity, while the ventral wall is much convoluted longitudinally. The sense-organs in the first part of the buccal cavity are shorter and of greater average width as compared with their height than those of the epidermis, about 'jGii high and from 50 to 6o;U' wide ; toward the caudal end of this cavity they diminish in size, and become only 20\x high and \o\x wide. The sense-hairs in the first half of the buccal cavity are from 4 to 6/x long, over twice as long as those of the epidermal sense-organs. The cuticula over a sense-organ in the buccal cavity is no thinner than that covering the surrounding cells, and the summit of a sense-organ does not project above these cells. Sometimes the cuticula passes evenly over the senseorgan from the surrounding cells; more often, however, it is indented around the sense-organ so that the summit of the latter really projects into an elevated area of cuticula, but this area is not elevated above the general cuticular level. The base of the sense-organ is often rounded and then presses the basement-membrane, which is very thin under the epithelium, out among the muscle-fibres. The sense-cells of the organs in the cephalic part of the buccal cavity do not differ from those of the sense-organs in the epidermis. In the caudal part, they are comparatively wider and have a more rounded nucleus.

Even in the cephalic part, the sense-cells and also the epithelium cells seem less crowded, and the former have larger spaces between them than in the epidermis. This may be due to the absence of the gland-cells from the epithelium. In these clear spaces between the sense-cells one often sees delicate fibres. These are sometimes cut walls of cells, but usually prove to be the very slender bases of some sense-cells. It was suggested' in the description of the sense-organs of the epidermis, that the varying height of the nuclei of the sense-cells was merely due to the necessity of these cells adapting themselves to the space they were in. This receives confirmation from the position of the nuclei of the sense-cells in the buccal cavity. In this region, where the sense-cells have more room, all nuclei are generally found below the middle height of the senseorgans (PI. XIII, Fig. 2).


A St7idy of the Epidermis by Means of Golgi's Silver Nitrate


Effect of the stain on epidermis in general. — In sections prepared by this process, the cuticula was usually stained black and a black precipitate deposited beneath it. This rendered difficult a study of that portion of the epidermis just beneath the cuticula. The supporting cells remained clear or were stained a uniform black ; in the latter case their basal processes were readily seen. The gland-cells were never stained either wholly or in part, but they could be easily distinguished because of their shape and contents. The basal cells were never stained. Sometimes the basement-membrane was obscured by a thick deposit of silver, but it generally showed with perfect clearness as an unstained, apparently structureless layer between the epidermis and the circular muscle-layer.^

Intraepidermal nei've-fibres. — The most striking feature in these sections is the presence of more or less deeply stained fibres betiveen the epidermal cells, and in no instance connected with the bases of these cells (PI. XIII, Figs. 3, 5, and 6). That these fibres are nerve-fibres there can be no doubt. That these are really fibres and not the edges of lamellae formed by the deposit of silver in intercellular spaces may be shown by an examination of sections which pass obliquely through the epidermis. Between the cell walls are seen sections of these fibres appearing as black dots ; as the course of the fibres is oblique to the surface of the section, they may be traced down by focusing, and seen to be really fibres. That these are nervefibres is shown by two facts ; first, by the fact that they present precisely the same appearance as undoubted nerve-fibres which are found among the muscles and in the central nervous system — the appearance of a clear, glistening thread sur 1 In all my preparations the blood-vessels among the circular muscles show distinctly either stained or unstained. These blood-vessels press closely against the basement-membrane, forming here great loops and coils. Lenhossek ('92) stated that the blood-vessels in his preparations enter the epidermis itself. Although I have searched my preparations carefully, I have found no instance of this outside of the clitellum. Lenhossek has a later paper (Die intraepidermalen P.lutgefrisse in der Haut des Regenwurmes, in Verhandl. Naturf. Gesell. Basel, Bd. 10, Heft I, p. 84) which I have not seen.

204 LANGDON. [Vol. XI.

rounded by a black deposit ; secondly by the fact that they may be traced into the central nervous system.

Both cross and longitudinal sections show that these fibres pass between the epidermal cells from the basement-membrane almost or quite to the cuticula. Some of the fibres are simple, some branch one or more times. It is hard to decide definitely as to their ultimate endings. Some become very slender and seem to end freely between the cell walls at varying distances below the cuticula ; most of the fibres pass almost to the cuticula, where their ends are lost in the black deposit which is usually found there. The fibres vary in diameter and depth of stain ; they are sometimes straight and of uniform caliber, but usually a little sinuous and varicose. In a few cases the upper end was found to turn at right angles and to run for a short distance just beneath the cuticula, sometimes again turning and running a very short distance toward the basement-membrane. There is sometimes found an appearance which suggests an anastomosis of these fibres in the epidermis (PI. XIII, Fig. 5). A careful study shows that, in such cases, the fibres merely cross one another.

Sometimes in the basement-membrane, sometimes among the cells of the basal layer of the epidermis, are larger and more deeply stained fibres. In sections which pass obliquely through the epidermis, it is easy to see that the fibres between the epidermal cells arise from these fibres which pass along the base of the epidermis. It is not difficult to find places in cross sections in which the same fact is clearly shown (PI. XIII, Figs. 3 and 6). These epidermal fibres often appear, at first sight, to be connected with the basal processes of the supporting cells, but a careful study shows that in every such case the fibres really run under or over these processes. Sections passing obliquely or tangentially through the base of the epidermis show that these subepidermal fibres form a freely branching network (PI. XIII, F'igs. 8, 9). The fibres of this network often appear to anastomose, but careful focusing shows that in the greater number of such cases the fibres merely cross each other. In a smaller number of such cases there seems to be a real anastomosis (PI. XIII, Fig. 9, a).


In cross and longitudinal sections, there appear bundles of stained fibres passing to the epidermis through the circular muscle-layer (PI. XIII, Fig. 3). These fibres are often varicose and sinuous, and appear imbedded in or surrounded by an unstained substance. In sections in which the fibres show at the very base of the epidermis, it is easily seen that they pass throus:h the basement-membrane and become continuous with the fibres of the subepidermal network. Hence the intraepidermal nerve-fibres are branches of fibres which approach the epidermis in the same position as the epidermal nerves described in the haematoxylin preparations. As will be seen presently, these fibres are efferent nerve-fibres. These intraepidermal nerve-fibres appeared in every preparation made, and have been found in all regions of the earthworm except in the clitellum, which has not been studied for this purpose; but they appear to be more numerous in the caudal metameres. The fibres have been found between the supporting cells, but appear to be more numerous in those regions in which the gland-cells are more abundant. As these fibres have not been obtained by any other stain, it is not known whether fibers have remained unstained in parts of the epidermis where they have not been seen. However, the great readiness with which these fibres took the stain would seem to oppose this idea. No intraepidermal fibres have been found in the sense-organs, and they are not numerous in their immediate vicinity. This latter fact is probably connected with the absence of gland-cells in the immediate vicinity of a sense-organ.

Between the cells of the epithelium lining the buccal cavity are nerve-fibres which are similar to the intraepidermal fibres (PI. XIII, Fig. 10). The epithelium cells and the cuticula covering them is never stained, consequently the endings of these fibres can be seen more readily than in the epidermis (PI. XIII, Fig. 7). Some fork once; many end in what appears to be a flat plate, but it is impossible to state that this is anything different from the "artefacta" often seen in the course of a fibre. These fibres in the buccal epithelium may be traced to the cephalic ganglia or aesophageal ring.

The greater abundance of the intraepidermal fibres in regions

2o6 LAXGDOX. [Vol. XI.

in which the gland-cells are more abundant suggests that the function of these nerve-fibres may be to control the secretion of the gland-cells. But the fact that the fibres are also found among the supporting cells and in the buccal epithelium indicates that, even if the above suggestion be true, this is not their only function. The intraepidermal fibres may have for their function the control of the general metabolism of the epidermal cells, or it may prove that their function is sensory. Sense-organs. — Although no difficulty has been experienced in staining the intraepidermal nerve-fibres, there has been an uncertainty in the action of the stain on the sense-organs. No difficulty has been found in recognizing the latter, whether stained or unstained, in all preparations in which they occur. In sense-organs which are unstained, even in cases in which it is impossible to distinguish the separate cells, the organs themselves may be easily identified by the elevation of the cuticula above them and by the layer of deeply stained covering cells around them (PI. XIII, Fig. 3). In rriany of the sense-organs the sense-hairs, sometimes stained, sometimes clear, could be seen projecting from the cuticular elevation, thus affording an unmistakable evidence of the presence of the sense-organs. When the sense-cells of the sense-organs are stained (PI. XIII, Figs. 1 1 -14), the covering cells are sometimes stained, sometimes unstained. In all cases the basal cells of the sense-organs, as well as those in the epidermis, are unstained. The sense-cells themselves have in my preparations taken the stain so differently from any other cell of the epidermis that they could be easily distinguished even under a low power. The covering cells of the sense-organ and the supporting cells of the epidermis stain black or a blackish brown, and no trace of a nucleuscan be seen in them. The sense-cells stain a reddish brown, and usually the nucleus shows as a clear oval spot. This difference between the color taken by the supporting cells and that taken by the sense-cells must be due to some intrinsic difference between them; the nerve-fibres of the sense-cells have the same brown color, instead of staining black like the other nerve-fibres. Since these sections were prepared, the sensecells have darkened a little, and in some cases the nuclei have


become indistinct. Not only are the supporting cells more nearly black, but their walls have an uneven appearance, and a cross section shows that the silver is deposited on them irregularly and rather thickly. The sense-cells present a smooth appearance, and look as if their walls were really stained. But cross sections of these cells show a delicate unstained wall, which has the same clear, glistening appearance noticed in the nerve-fibre, and, on the outside, a very thin, evenly applied deposit of the silver.

Comparatively few of the sense-cells in a given organ are stained. But a study of a thin section under the oil immersion shows plainly the outline and nuclei of the unstained cells (PI. XIII, Figs. 11-13). All the sense-cells, both stained and unstained, present a plump, rounded form, and taper to both ends from an enlarged part in which the nucleus lies. In short, they are of the same form as the sense-cells described in the senseorgans stained by haematoxylin, and are clearly the same thing. The summits of the sense-cells cannot be traced through the cuticula; the slender, converging peripheral ends of the cells blend into a dark mass, which is continuous with the blackened cuticula. It is impossible to trace a single cell through this mass, but a careful study of the summits of the sense-organs almost always reveals a group of short black or colorless hairs projecting from its surface. The summits of the sense-organs in which the cells were stained are not so elevated as in any other preparations; they appear as if there had been an unusual contraction of the sense-cells. Other facts support this interpretation. The sense-hairs do not project above the cuticula as great a distance as in the unstained organs, and the cells themselves appear more thickened.

The silver nitrate stain permits the character of the bases of the sense-cells to be clearly seen. The cells which have a nucleus near the cuticula extend toward the base of the senseorgan into a slender fibre as in the haematoxylin preparations, and this fibre is a nerve-fibre (PI. XIII, Fig. 15, a and b). Those cells which have their nucleus at the center or near the base of the sense-organs sometimes end abruptly at the base (PI. XIII, Fig. 15, g). In this case a nerve-fibre is attached

2o8 LANG DON. [Vol. XI.

to the cell near the circumference of the base. Some of these cells simply taper into a mere fibre (PI. XIII, Fig. 15, /;■). The larger number of such cells have a base which forks into two or more basal processes (PI. XIII, Fig. i 5, c, d, n, vt). These processes, which are usually varicose, descend in an irregular course to the basement-membrane. In most cases they end at the basement-membrane, but sometimes they extend along the outer surface of this membrane for a short distance beneath the neighboring cells. One of these processes, and only one, is always produced into a nerve-fibre (PI. XIII, Fig. 5, 11, 12). In cross sections it is often impossible to establish the connection of some of the cells with nerve-fibres or to trace the course of these fibres. In sections which pass obliquely through the epidermis the connection of each cell with a nerve-fibre and the course of these fibres along the base of the epidermis can be clearly seen.

Since but few of the sense-cells in a given section of an organ are stained — frequently only one or two — it sometimes appears as if these stained cells were not grouped into senseorgans. But a study of such cases under the oil immersion always reveals the fact that the cell in question is one of the sense-cells of a sense-organ. In every case I have been able to make out that the stained cell is but one of a group of cells of similar shape, the rest remaining unstained, and that this group is enclosed in covering cells which reveal the oval outline of the sense-organ. In almost every case a more or less marltid elevation shows the summit of the sense-organ, and this elevation usually bears a cluster of short sense-hairs. All these facts clearly prove that every sense-cell in my preparations which might be taken at first sight for an isolated cell is but one of the sense-cells of a sense-organ. The sense-cells of some of the sense-organs at the entrance to the buccal cavity were found to be stained. These cells were like those of the epidermal sense-organs, except that they were usually more slender. The sense-organs of the buccal cavity itself were unstained and the characteristic shape of their cells could not be made out ; but the presence of the organ itself could usually be detected by the convergence of the summits of the cells (PI. XIII, V'lg. 10).


The nerve-fibres proceeding from the sense-cells are readily distinguishable from those which supply the epidermis. They are smaller in diameter and consequently of more delicate appearance ; in their course along the base of the epidermis they neither branch nor anastomose, but pass directly to the

\!ltj. Y-^



^mts. ie{t,

\at\t. rvr.

Fig. I. A diagram showing a ventral ganglion and the course of its three pairs of nerves. In this diagram the ventral ganglion is represented in place and the body wall cut through on one side to show the nerves, aiit. nr., anterior nerve-trunk ; a, space into which the ventral rami pass (this space is between the ventral longitudinal muscle-tract and the inner intcrsetal tract ; cir. vius., circular muscle-layer; cut., cuticula ; d. r., dorsal ramus of nerve; dor. loti. tr., dorsal longitudinal muscle-tract ; <?/., epidermis ; ep. nr., epidermal nerve ; mt. Ion. tr., inner intersetal tract of longitudinal muscles ; int. long. tr. 2, outer intersetal tract ; intra./., intraepidermal nerve-fibres ; ints. jfr., intersegmental groove ; /«/^. i(?/., intersegmental septam ; lat. Ion. tr., lateral tract of longitudinal muscles; m. ^r., median groove around metamere , m. h»-., median nerve-trui# (anterior nerve of so-called " double root ") ; nr. r., nerve-ring ; sen. org., sense-organ ; sub. n., subepidermal network ; ven. gan., ventral ganglion ; ven. Ion. tr., ventral longitudinal tract ; v. r., ventral ramus of nerve.

nearest one of the epidermal nerves which traverse the circular muscle-layer ; the sensory fibres from sense-organs on one side of a metamere always enter an epidermal nerve of that side ; they never cross the mid-dorsal or mid-ventral line. In these nerves, the fibres from the sense-cells are never so deeply stained as the efferent nerve-fibres, — they usually present a clear brownish appearance. They are further distinguishable from the fact that they keep more nearly parallel to one another and run in uniformly sinuous lines.



[Vol. XL

Course of the nerves from the cetitral neiiioits system to the epidermis. — As is well known, from each side of each ganglion of the ventral nerve-chain three great nerves take origin. The anterior pair of nerves arises just caudad of the anterior septum of the metamere. The middle and posterior pairs, which lie so closely together that they are often called " double nerve-roots," arise just caudad of the middle of the ganglion

t^ nr.

%\T. miLi.



latloK. {»»

•yen. lui.U

Fig. 2. A diagram showing the course of the median nerve-trunk and its formation of a nervering. This diagram holds good for the other nerve-trunks, except that the sense-organs would not be so numerous as iu this. For reference letters, see text, Fig. i.

(see text, Fig. i). All of these nerves pass latero-ventrad to the inner surface of the longitudinal muscle-layer, and there divide into a ventral and a dorsal ramus (see text, Fig. 2). These* rami pass to the inner surface of the circular muscle-layer, the ventral ramus using the space between the ventral and inner intersetal tract of longitudinal muscles, the dorsal ramus using the space between the inner intersetal and lateral tracts. On reaching the circular muscle-fibres, the dorsal ramus sends a very short branch ventrad between the circular and longitudinal muscle-layers, and then turns and passes dorsad and slightly caudad between the two muscle-layers. As the dorsal ramus


approaches the mid-dorsal line, it leaves the space between the two muscle-layers, runs among the circular muscle-fibres, and finally reaches the base of the epidermis near the middorsal line. The dorsal ramus of one side never crosses the mid-dorsal line to the other side of the metamere. The ventral ramus of each nerve-root turns between the two muscle-layers and passes to a point near the mid-ventral line, but never crosses this line (text, Fig. 2). There are thus formed in each segment three nerve-rings, which are incomplete in the middorsal and mid-ventral line and which lie for the greater part of their course between the circular and longitudinal musclelayers. Each of these nerve-rings is formed of four parts, the two dorsal and the two ventral rami of a pair of nerves. Back of the anterior metameres from which the diagram for text. Fig. 2, was made, an "accessory" tract of longitudinal muscles appears between the ventral tract and the inner intersetal tract. Each of the three nerve-roots then passes directly to the circular muscle-fibres through the space between this accessory and the ventral tract, and does not divide into its dorsal and ventral rami until the circular muscles are reached. In this case no ventral branch is given off from the dorsal ramus, and each nerve-ring appears more complete in that each half can be traced directly from the mid-dorsal around to the mid-ventral line, and is not interrupted in the region of the inner intersetal tract, as shown in the diagram. From each of these nerve-rings, the epidermal nerves before noted pass to the epidermis, and as they are given off the dorsal and the ventral rami of the nerve-rings become smaller. The epidermal nerves leave the nerve-rings at various angles but are generally inclined cephalad ; they rarely approach the epidermis directly at the base of a sense-organ, but they are most numerous in those regions in which the sense-organs are most numerous. The epidermal nerves from the anterior nervering supply the epidermis of the anterior part, probably almost half, of a metamere ; those from the middle nerve-ring supply the epidermis in the zone which includes the setae, and those from the posterior ring supply the posterior part of the metamere.

212 LANGDON. [Vol. XI.

From the ventral nerve-chain, nerve-fibres, apparently arising from the small ganglion-cells, pass out by each of the nerveroots into the three nerve-rings. A careful study has not been made of the nerve-fibres going to the muscles, but I have observed that they may arise from any part of a nervering. Other fibres which cannot be told from the nervefibres of the muscles pass from the nerve-ring through the epidermal nerves and form the subepidermal network previously described. In no case have these efferent fibres been seen to be connected with any cell in their course from the ventral ganglion to their ending in the epidermis. The sensory fibres pass without branching, anastomosing, or changing in diameter, to the central nervous system, which they reach by the same course followed by the efferent fibres in passing to the epidermis. Some of the epidermal nerves contain only efferent fibers, some contain both efferent and sensory ; some have been noticed which contain more sensory than efferent fibres, but it is rare to find one which contains only sensory fibres. Each nerve-ring and each nerve-trunk contains both the sensory fibres and the efferent fibres passing to the epidermis. Since the sense-organs are less numerous in the posterior half of a metamere, the posterior ring and its corresponding trunk contains a smaller number of sensory fibres than the others. Since the sense-organs are usually larger and more numerous around the middle of each metamere, the middle nerve-ring and its trunk contain more sensory fibres than the others.

In the central nervous system, each sensory fibre divides into two branches. One branch passes caudad, one cephalad, into the ganglia of the next metameres. Each branch ends freely. Of course it is impossible to trace an individual fibre from its origin in a sense-cell to its ending in the central nervous system, but these sensory fibres can be so readily distinguished from all other fibres that it is easy to identify them in the great nerve-trunks and in the ventral ganglia. The only cells with which these sensory fibres are connected are the sense-cells of the epidermal sense-organs.


Distribution of the Sense-Organs.

In a surface view of the cuticula the two layers of fibres which form it, the comparatively large openings of the glandcells, the nephridial openings, the cuticular sacs of the setae, and the intersegmental grooves may be readily identified {PL XIII, Fig. 4, and PI. XIV). Attention is usually first directed to the cuticular elevations over the sense-organs by the absence of the gland-pores over a little area around each elevation. These cuticular elevations themselves appear as irregularly rounded raised spots, which contain numerous openings smaller than the pores of the gland-cells (PI. XIII, Fig. 4). Each one of these openings is the outer end of one of the pore-canals seen in sections, and, normally, a sense-hair protrudes through each. Each opening lies at the intersection of two short straight lines, each of which is a line of contact of two adjacent cuticular fibres. The existence of these two intersecting lines shows that both layers of the cuticular fibres are present over the sense-organs. The pore-canals through which the sense-hairs pass are thus spaces left between the fibres of the cuticular layers at their intersection. That is, the pore-canals are such openings as might be made by taking a layer of fibres, laying a second layer over and at right angles to the first, and then forcing a blunt instrument through both layers between their threads. Whether these openings always exist between the cuticular fibres, or are made by an outward growth of the sense-hairs, has not been determined. Even if they always exist, the presence of the cross-like marking at each opening shows that the sense-hair pushes the fibres still farther apart. When the cuticula is removed, the sense-hairs are usually pulled out of these canals. Occasionally the hairs are torn from their cells and remain projecting above the cuticula. A study of the distribution of the cuticular spots over the senseorgans forms the most ready means of determining the distribution of the latter.

The cuticular spots are found on every metamere of the body, and appear on the clitellum metameres after the clitellum has disappeared, and their size and distribution shows the following facts :

2 14 LANG DON. [Vol. XI.

1. The sense-organs are largest and most numerous in the prostomium and first metamere. Here they are irregularly distributed, so that there is no evidence of a distinct grouping.

2. Distinct groups or zones of sense-organs may be recognized over the rest of the body. It should be borne in mind that every metamere contains sense-organs irregularly scattered over it outside of these groups and zones. The latter are merely noticeable from the fact that the organs in them are larger or more numerous than elsewhere. In the metameres back of the first the following arrangements of the senseorgans may be noted (PI. XIV) :

{a) There are always more sense-organs in the cephalic than in the caudal half of a metamere. This is also true of the first metamere.

(b) In the fifth or sixth metamere, the organs are seen to be more numerous in a zone which passes around the cephalic margin of the metamere. Passing caudad, this ccpJialic zone increases in prominence for six or seven metameres. It then becomes reduced to a single irregular row of large senseorgans, and remains as such until near the caudal end of the worm. In the caudal metameres, it gradually loses its prominence by a decrease in size of its organs. In this cephalic zone the smallest organs are generally found on the dorsal surface, and the largest near the nephridial opening.

(f) In the second metamere, there appears a zone of somewhat larger organs, which occupies the elevated caudal border of a slight groove that always passes around each metamere just cephalad of its median circumference. This zone is in line with the setae, and may be called the median zone. In the third and fourth metameres this median zone is well, marked. From this point on, it diminishes in width until, back of the clitcllum, it is reduced to a single irregular row of large organs. Continuing caudad, the organs of the median zone are found to diminish in size until about the last seventh of the body is reached. They then increase in size to the caudal end of the body. As a rule the smallest organs in this median zone occupy its dorsal part, and the largest ones are near the setae.


{d) Around each nephridial opening, there is always a thick cluster of small organs which may be called the nephridial group. The sense-organs of this nephridial group are always found around the nephridial opening, no matter zvhat its positio)i. The nephridial group is more prominent at the cephalic end of the worm, and diminishes in size toward the caudal end.^

It will thus be seen that the sense-organs are most numerous and largest at the two extremities of the earth-worm ; that the most prominent organs at these extremities are those of the median zone; that those most prominent in the middle region of the body are those of the cephalic zone; and that the degree of prominence of the cephalic zone in any one region is inversely proportional to that of the median zone. If the nephridial opening really had the definite position often assigned to it — in front of the outer one of the inner pair of setae — a distinct lateral line of larger organs might be traced along both sides of the body. But the variability in the position of the nephridial opening, not only from metamere to metamere, but also on opposite sides of the same metamere, accompanied as it is by a variability in the position of the nephridial group of senseorgans and the position of the largest organs of the cephalic zone, destroys all lateral symmetry in their distribution. The cuticula from the ventral surface of the prostomium and from the buccal cavity shows the cuticular markings of the senseorgans irregularly distributed in both places, and occurring in as great numbers on the ventral surface of the prostomium as on its dorsal surface. The cuticular spots from the buccal cavity are less distinct, owing to the fact that the cuticula in this region is not elevated over the summits of the sense-organs. A groove often surrounds the cuticular spot, the same groove noted in sections of the sense-organs from the buccal cavity.

In a worm about 19 cm. long, which contained 153 metameres, there was found to be an average of 1000 sense-organs

1 It is difficult to explain the presence of this group around the external opening of a ncphridium unless it serves as a guard against parasites. The cuticular spots are exactly like those found over the rest of a metamere, and sections show that the underlying sense-organs have the same structure as those found elsewhere.

2l6 LAXGDOaW [Vol. XI.

to a mctamere, making about 1 50,000 in the whole worm. In the specimen from which the chart was made, there were about 1900 sense-organs on the first metamere and the upper surface of the prostomium, 1200 on the tenth metamere, and 700 on the fifty-sixth. These numbers are, of course, only approximate, but they give some idea of the great abundance of these organs.

The various zones and groups of sense-organs may be readily seen with a hand lens in living specimens, unstained alcoholic material, and in those stained with haematoxylin, if a large worm is used for the examination. The sense-organs appear as small, slightly elevated spots which reflect the light more strongly than the surrounding surface. Those organs which occur in line with the setae are most readily found. In the study of living specimens it is interesting to note that, when disturbed, the worm often draws the entire prostomium into the opening of the buccal cavity, thus protecting the numerous sense-organs on the prostomium as well as closing the opening. During regular contraction of the earthworm, the cephalic and caudal metameres, especially the former, are very much elevated around the median part, thus bringing the median zone into great prominence, while the border of adjacent metameres are so closely pressed together that the cephalic zone is concealed. This may account for the prominence of the median zone at the two ends of the worm. Further support for this interpretation may be found in the fact that in the metameres of the middle region of the body, where the cephalic zone is prominent, there is much less contraction, and consequently the cephalic zone is not apt to be concealed.

If a transparent worm about 15 mm. long be examined in a watch-glass of water under a cover-glass with a 4 mm. objective and 4 ocular, the sense-organs may be demonstrated. As the worm moves about in the little space thus formed, it is possible to observe continuously particular spots on the surface. The clusters of sense-hairs may then be distinctly seen radiating outward from the rounded elevation of the cuticula over the sense-organs.



The foregoing observations seem to me to warrant the following statements of facts :

1. The epidermis, exclusive of the sense-organs, contains three kinds of cells, arranged in two layers : an outer layer of gland-cells and supporting cells, and an inner layer of small basal cells.

2. The supporting cells and gland-cells are not connected by intermediate forms, but both are connected by intermediate forms with the basal cells.

3. The epidermis is covered exteriorly by a cuticula composed of at least two layers of fibres.

4. The epidermis is separated from the circular muscles by a basement-membrane.

5. There are nei've-fibres ending freely between the cells of the epidermis ; these fibres are more numerous among the gland-cells.

6. These intraepidermal fibres arise as efferent fibres in the central nervous system, and those in each half of a metamere come from the corresponding half of a ventral ganglion, or cephalic ganglion and aesophageal ring if in the prostomium and first metamere.

7. The efferent nerve-fibres leave the ventral ganglia by each of the three great paired nerves which arise from a ganglion, and reach the epidermis by the regular course of these nerves.

8. Each of these nerves passes through the longitudinal muscles to the inner surface of the circular muscle-layer and divides into a dorsal and a ventral ramus.

9. These rami pass between the two muscle-layers to the mid-dorsal and mid-ventral line. The rami of each pair of nerves thus form a nerve-ring which is incomplete dorsally and ventrally, and three of these rings thus exist in each metamere.

10. From these three nerve-rings, epidermal nerves pass through the circular muscle-layer to the epidermis.

2i8 LANGDON. [Vol. XL

1 1 . The efferent nerve-fibres of these epidermal nerves pass through the basement-membrane and form a subepidermal network,

12. From this network, the intraepidermal nerve-fibres pass between the cells of the epidermis.

13. In the epidermis are sense-organs, each of which is composed of a group of sense-cells covered by a layer of modified supporting cells. In the base of each organ are usually a few small basal cells.

14. These sense-organs are also present in the epithelium lining the buccal cavity.

15. The sense-cells taper toward both ends ; the outer end of each cell extends into a fine hair which passes through a pore-canal in the cuticula above it and projects beyond the latter ; the basal end of each sense-cell is always produced into a sensory nerve-fibre.

16. The sense-cells are the only cells with which the sensory fibres are continuous ; they are therefore ganglion-cells.

17. The sensory nerve-fibres pass along the basement-membrane to the nearest epidermal nerve. These fibres then take the same path to the central nervous system that the intraepidermal fibres took coming from it, and they enter a ventral ganglion by each of the three nerve-roots on their own side of the metamere.

18. Most epidermal nerves, each nerve-ring, and each nervetrunk thus contain both efferent and afferent (sensory) nervefibres.

19. In the ventral ganglia each sensory fibre divides into two branches; one passes caudad, one cephalad ; each ends freely in the next ganglion.

20. The sense-organs are distributed over the entire surface of the body, but are most numerous and largest at each end.

21. In most metameres a cephalic and a median zone of larger organs may be distinguished.

22. A nephridial group of small organs occurs around each nephridial pore.

In view of the conflicting accounts of the sensory structures in the epidermis of Lumbricus, I have thought it best to


restate the facts in the above form. All the facts in the foregoing description which have been described by previous investigators have been carefully verified in my own work. In the following critical summary of the literature, I have tried to indicate briefly the facts previously described and also those not before noted.

Summary of the Literature on the Sensory Structures in the

Epidermis of Lumbricus.

The early observers, such as Pontallie ('53), Clark ('57), and Lankester ('65), considered that the upper lip of Lumbricus served as a sense-organ because of its rich nerve supply and its mobility. After the cellular structure of the epidermis had been demonstrated, investigators began to seek an explanation of the well known sensitiveness of Lumbricus by referring it to definite cellular elements of the epidermis. This cellular structure was first perceived by Leydig ('65), who discovered the supporting cells, gland-cells, and sense-organs.

Gland and s7ippo7'ting cells. — Leydig ('65) considered the gland-cells to be sense-organs having the appearance of unicellular glands, because, when isolated, he found an apparent nerve-fibre connected with each. Below the supporting cells of the upper lip, he described and figured, "verastigter Wurzelnerv," which he considered a prolongation from the bases of these cells. Perrier ('74) supported Leydig's observations, although he stated that positive proof was needed as to the character of these apparent nerve-fibres. Mojsisovics ("77) found the same network below the supporting cells. Vejdovsky ('84) considered that these apparent nerve-fibres might be either connective tissue fibre or efferent nerve-fibres passing to the epidermis as to any other organ, and that their presence was no indication of the sensory nature of the common elements of the epidermis. Ude ('8S) considered Leydig's observations on the basal network of the supporting cells probably correct. It seems to me very probable that the fibres attached to the glandcells were really nerve-fibres, the intraepidermal fibres described in this paper ; these might easily come away with the isolated

2 20 LANGDON. [Vol. XI.

gland-cells. The network from the bases of the supporting cells was also probably bits of these nerve-fibres apparentlycontinuous with the true basal processes of these cells. Leydig's figures of these cells show that they are supporting cells, and precludes the possibility of referring this to the true sensecells. The basal processes of the supporting cells are understood when one remembers the layer of basal cells beneath them. Ude ('86) and Cerfontaine ('90) are the only ones who refer to these basal cells. There is no connection of nervefibres with the bases of either gland or supporting cells.

Sense-organs. — Leydig ('65), the discoverer of the senseorgans, saw them as " blase rundliche Flecken," which appeared by focusing below the surface. He described these spots as five or six times larger than the surrounding cells, and often limited by pigment. Of their minute structure he could only see that each was formed by a certain grouping of cells. I have found only the sense-organs of the dorsal surface of the head end limited by pigment. This may be of physiological interest. Perrier ('73) recorded his inability to find any sense-organs in Lumbricus, but he made only surface examination. Mojsisovics ('77) was the first to describe the sense-organs as seen in sections, and to figure their cellular structure. His figures do not show the true form of the sense-cells. He represented these cells as differing from the supporting cells merely in the possession of terminal hairs. He gave F. E. Schulze credit for discovery of the " Porencanalchen," through which the sensehairs pass, but he himself seems to have first seen the corresponding cuticular markings in surface view. Darwin ('82) referred to Lumbricus as " remarkably deficient in the several sense-organs." Vejdovsky ('84) and Ude ('86) verified Mojsisovics' description of the sense-organs and their cuticular markings, but Ude failed to find the sense-hairs. It seems to me probable that in his preparations these hairs were drawn into the pore canals. Vogt and Jung ('88) were not only unable to find any sense-organs themselves, but stated that such organs are unknown. Kulagin ('90) found the " Zellenanhaufungen " described by Ude. He probably refers to the senseorgans. Cerfontaine ('90), although he gave a more complete


description and illustration of the sense-organs of Lumbricus than any of his predecessors, added little to our knowledge of their structure. He likened the mode of arrangement of the cells in a sense-organ to the overlapping scales in an onion. A section through the outer surface of a sense-organ, consequently through the covering cells, does often give this appearance, but a good section through the center of an organ shows that there is no regularity in the arrangement of its cells. Cerfontaine gave the first satisfactory figure of the cuticular markings. Lenhossek ('92) and Retzius ('92, a and b) both denied the existence of definite sense-organs. The titles of some of the papers which contain descriptions of these organs appear in the footnotes to Lenhossek's article, but these descriptions are not referred to in his paper. It seems to me that no observer has correctly described the cells of the senseorgans.^

^ An article entitled " Zur vergleichenden Anatomie der Oligochaeten " by Dr. Richard Hesse has recently appeared in the Zeit. f. wissen. Zoologie, — Bd. 58, p. 394, 1894, — in which the writer correctly describes the sense-cells and also arrives at the same conclusion concerning the nerve-cells of Lenhossek as I have presented. Hesse worked on several species of Lumbricidae, one of which, Lumbricus herculeus, Sav., is the same as that upon which I worked. Our results on this species are in general confirmatory ; they differ in the following points : (i) Hesse finds in a sense-organ supporting as well as sense-cells, but does not note the small basal cells. He describes these supporting cells as of the same width throughout. I have looked through my preparations again, tracing single sense-organs through all their sections, and I feel warranted in believing that such supporting cells do not regularly form a part of a sense-organ. I have found, in organ after organ, nothing but cells which tapey- to both ends from the enlarged part in which the nucleus lies, and which end in a hair-like process passing through the cuticula. Moreover, I have found that the number of sense-hairs seen in a given section of a sense-organ corresponds to the number of cells in that section ; and the number of pores over a sense-organ — as shown in the cuticular markings — corresponds to the number of cells usually present in a sense-organ of the same region. Sections which give cross-sections of sense-organs show that these organs often have one side deeply indented. In such cases, a longitudinal section of the same organ on this side would have shown some of the covei-ing c€^?. between the sense-cells, and these covering cells would appear like the supporting cells described and figured by Hesse. I would therefore consider that, if a sense-organ contains any supporting cells, these do not differ in form or peripheral ending from the sense-cells. (2) Hesse believes that the sense-organs in a given metamere "auf drei Giirteln liegen, die um das Segment herumlaufen." Of these his " mittlere Giirtel " must correspond to my median zone, while his " vordere " and " hintere Giirtel " must be composed of some of the small organs irregularly dis

222 LANGDON. [Vol. XL

Isolated nerve-cells. — Kulagin ('88) described in the epidermis of several species of Lumbricus isolated sense-cells with a hair projecting through the cuticula and a base connected with a nerve fibre. As he gave no figures, and apparently did not use a specific nerve-stain, it is impossible to decide on the character of these cells. Lenhossek (-92), believed the sensitiveness of Lumbricus to be due to isolated nerve-cells which were scattered " an alien Stellen der Korperoberflache mit Ausnahme der intersegmentalen Furchen." He stated that these cells "finden sich weder auf gewisse Gegenden beschrankt, noch an bestimmten Stellen zu besonderen Sinnesorganen angehauft, sondern erscheinen iiber alle Gebiete der Epidermis gleichmassig vertheilt." I have been unable to discover any isolated nerve-cells. I have found many cells which

tributed over a metamere. His " vordere Giirtel " is not near enough to the intersegmental groove to correspond to my " cephalic zone." The different results we have arrived at in regard to the distribution and numbers of the sense-organs is due to the fact that my study was made by means of the cuticular spots and his by means of sections. In the latter the cephalic zone, the nephridial group, and many of the smaller organs are apt to be overlooked. (3) Hesse believes that the two ventral rami of a nerve-ring meet in the mid-ventral line. I have never found this to be the case. He found that the dorsal rami of the anterior and posterior nerve-rings always remained between the two muscle-layers. I have found that all three rings pass into the circular muscle-layer during the last part of their course dorsally. (4) Hesse describes groups of ganglion-cells in the course of the nerves of the prostomium. In discussing their function, he considers it probable

' dass die sensiblen Fasern die von den -Sinneszellen aufgenommenen Reize an die Ganglienzellen iibermitteln, deren motorische Fortsatze an die Ruckziehmuskeln der Oberlippe fiihren und deren Zusammenziehung veranlassen," but he has &\\dently not been able to trace these fibres. I have been able to find ganglion-cells which seem to be the same as those which he describes, but it does not seem to me that the fibres from the sense-organs end among these cells. In the course of the nerve-rings, Hesse has "nur einmal eine solche beobachtet, und zwar eine bipolar." My haematoxylin and silver nitrate preparation have not shown such ganglion-cells in the nerve-rings, but some alum carmine mounts which I have lately examined have revealed them in considerable numbers. I counted the number of these ganglion-cells in one half of each nerve-ring in the first thirteen metameres of one worm. The second metamere had but one on its median ring ; the third had one on its anterior ring and four on its posterior ring. From the third through the thirteenth, the ganglion-cells occurred in every nerve-ring in numbers varying from two to eight. I could see that these cells were bipolar, and in several cases that a fibre started both towards and away from the central nervous system. I e.xpect to continue my work on these ganglion-cells and to give more concerning them in the future.


correspond to Lenhossek's descriptions and illustrations of the nerve-cells, but I have found them to be in every case in one of the sense-organs seen by previous observers. It is my opinion that the isolated nerve-cells described by Lenhossek are the sense-cells of the sense-organs. It will be noticed that his diagram of the appearance of these cells in the epidermis (see Lenhossek, '92, Taf. V, Fig. 6) represents a section through the region of the setae, the very region in which the sense-organs are most numerous, and that he more often figures these cells in groups of two or three than isolated. Lenhossek's illustrations and descriptions of these " Nervenzellen " correspond exactly to the sense-cells of the sense-organs as they appear in my silver nitrate sections, except that I have never found the basal processes of these cells running as far along the base of the epidermis as he figures them. It seems to me not unlikely that some of the long ends of these processes may be parts of the subepidermal network. That Lenhossek failed to recognize these organs may be due to the fact that the silver stains but few of the cells of one organ. And if he used thick sections mounted without a cover-glass, it would probably be impossible to perceive the unstained sensecells and the outline of the sense-organ.

Lenhossek figures his nerve-cells as ending at the cuticula. The difficulty of retaining the cuticula in position and the heavy deposit of silver made along it, seems to me to explain his failure to find the sense-hairs. In all my sections in which the sense-cells were stained, the sense-hairs were almost or quite withdrawn, and might therefore be readily overlooked. My own experience has shown me that a knowledge of the structure of the sense-organ as shown in haematoxylin preparations is necessary for a correct interpretation of the appearance obtained by the silver nitrate method. Lenhossek stated that in haematoxylin preparations his nerve-cells could scarcely be told from the supporting cells; I have found that the sensecells of the sense-organs may be readily distinguished from the supporting cells in all my preparations.

Retzius ('92a) figures and describes the isolated nerve-cells of Lenhossek. He figures at the base of the epidermis a net

2 24 LANGDON. [Vol. XI.

work formed from the basal processes of these nerve-cells ; it seems to me probable that much of this network is really a part of the subepidermal network of the efferent fibres. Of the nerve-cells themselves, he says, " Eine Gruppirung derselben zu dicht gedrangten Gruppen oder Organen, wie von vornherein angenommen werden konnte, scheint nicht vorzukommen." In view of my interpretation of the facts, one statement which Retzius makes concerning an appearance in the cuticula is interesting. He describes, in cross-sections, an appearance of fine lines crossing the cuticula perpendicularly and then says, "an den Stellen wo die Sinnesnervenzellen die Cuticula beriihren, sah ich ferner oft eine kleine hiigelartige Erhohung der letzteren und ihre lineare Zeichnung verstarkt, so dass das Ganze den Eindruck feiner Stiftchen (oder Kanalchen) machte, ohne dass ich die Natur dieser Bildung sicher zu eruiren vermochte." From this description, I should judge that Retzius saw, at least in some cases, the cuticular elevation over a sense-organ and the fine pores piercing it. In the two illustrations which he gives of this appearance {I.e. PI. VI, Fig. 2) the cuticula appears exactly like that over a senseorgan but the cell beneath seems to me to be a very large


Retzius found in the " Mundepithel " what he considered might be " Geschmackszellen." These were apparently fibres which passed through the epidermis and ended in an enlarged part under the cuticula. His figures of these do not admit of their being the sense-cells of the sense-organs.

Netve-siipply of sense-organs. — Vejdovsky ('84) and Ude ('86) both described a direct connection of nerve-fibres with the sense-organs, but neither gave any evidence of this connection or any evidence that the fibres were really nerve-fibres. Lenhossek ('92) proved the direct connection of his nerve-cells with nerve-fibres. If his nerve-cells are the sense-cells of the senseorgans, he was the first to prove the connection of these organs with nerve-fibres. His account of this connection and his description of the course of these fibres has been with one exception confirmed by my own work; I have never found a sensory fibre crossing the circular muscles alone, but always


by way of the epidermal nerves. Retzius ('92) differs from Lenhossek in regard to the central endings of the sensory fibres. He found that these sometimes ended in the same ganglion which they entered, and that their ends were not always tapering, but were "in der Regel etwas knotig-varicos, ungefahr wie bei anderen sehr einfachen Nervenendigungen, gewohnlich etwas gebogen und nicht selten etwas verzweigt." I have not found such endings. Retzius found in each half of the ventral nerve-cord three parallel longitudinal tracts of sensory fibres. Of these the outer and largest one received sensory fibres entering by all three nerve-roots ; the median bundle was next in size and received fibres from the anterior and median nerve-roots ; while the inner and most delicate tract received fibres only from the median nerve-root. That means that the sense-organs around the median zone of a metamere furnish fibres to all tliree tracts, those around the cephalic part to two, and those around the caudal part to but one. Apathy ('92) believed that the sensory fibres described by Lenhossek and Retzius were not sensory but motor fibres, and were not connected with epidermal cells.

Distribiilioji of tJic sense-organs. — Leydig ('65) found the sense-organs on both extremities of Lumbricus. Vejdovsky ('84) extended the distribution over a few metameres back of the head ; Ude ('86) extended it over the whole body, and first noted the median zone of sense-organs. Cerfontaine ('90) added to this the fact that this zone is situated on a crest of the epidermis. His statement that this crest is more prominent on the ventral surface has not been confirmed by my observations. No observer has noted the cephalic zone or the nephridial group, nor has any one attempted a systematic study of the distribution of the sense-organs. No notice has been found of the presence of the sense-organs in the buccal cavity.

Functions of sense-organs. — But little has been written on the functions of these sense-organs. Although Darwin ('82) did not know of their presence, his physiological experiments may aid in determining their function. He found in Lumbricus the sense of hearing lacking, the sense of smell feeble, the sense of taste v/ell developed, the ability to perceive light

2 26 LANGDON. [Vol. XI.

present on a few anterior metameres only, and the sense of touch highly developed over the whole body. Leydig {'65) seems to have regarded the sense-organs as " Geschmacksknospen." Vejdovsky {'84) objected to this, and thought them " Tastorgane," an opinion which is shared by Ude ("86). Vogt and Jung {'88) stated that Lumbricus perceived light and sojind. It is likely that they mistook the great sensitiveness of this worm to mechanical disturbance for an ability to perceive sound. Retzius, in writing of the probable function of the " Sinnesnervenzellen," says, " Der Regenwurm besitz keine anderen, hoher entwickelten Sinnesorgane ; er ist aber bekanntlich fiir Sinnesreize verschiedener Art empfindlich. Es ware deshalb von ausserordentlich hohern biologischen Interesse zu erfahren, ob dieselben Sinnesnervenzellen diese verschiedenen Eindriicke vermitteln, oder ob einzelne derselben eine besondere physiologische Function haben." Since these sense-organs form the only known sensory apparatus of Lumbricus, and since their structure is not visibly different in different parts of the body, it is likely that they are sense-organs of a general nature capable of reacting to mechanical, chemical, thermal, or luminous stimuli. Those situated in the buccal cavity doubtless serve principally as organs of taste. The presence of a thick layer of pigment in the dorsal epidermis of the head metamere suggests an explanation of the fact that light is perceived only on the anterior metameres. Lenhossek homologized his nerve-cells with ganglion-cells of the dorsal roots of the spinal nerves.

Intraepidermal nei-ve- fibres. — Lenhossek {'92) brought up the question of intraepidermal nerve-endings but to decide against the possibility of their presence. After an examination of a large number of preparations, he stated, "kann ich nun das Vorkommen einer freien Nervenendigung in der Haut des Reerenwurms mit grosser Wahrscheinlichkeit ausschliessen." As these intraepidermal nerve-fibres have appeared in every one of my sections, I can account for the fact that they were not seen by Lenhoss6k only on the assumption that the silver stain produced somewhat different results in his hands and in mine. That there was a difference in our results is


shown by the fact that the gland-cells were stained in his preparations while they were always unstained in mine, and by the fact that Lenhossek could not find a basement-membrane and did not notice the basal cells of the epidermis, while both of these structures showed in my preparations. Retzius ('92) says, " freie Nervenendigungen sah ich ebenso wenig wie v. Lenhossek im Hautepithel des Regenwurms." It is, however, evident that he saw such a nerve-fibre in the buccal cavity, but as he saw this appearance but once, judged himself mistaken. ^

1 After my work was completed and the account of it in the hands of the editor of this Journal, a preliminary paper by Dr. Alexis Smirnow, entitled " Ueber freie Nervenendigungen im Epithel des Regenwurms," appeared in Anat. Anzciger, Bd. 9, No. 18 (June 23, 1894). In this paper Smirnow records his discovery of freeending nerve-fibres in the epidermis of Lumbricus. My own discovery of these fibres was made in the spring of 1893, and briefly mentioned in a preliminary account of my work which was read before the American Morphological Society at the New Haven meeting in December, 1893.

In the main Smirnow's work on these free nerve-endings confirms my own ; but his account differs from mine in the following particulars : —

1. Smirnow overlooks the presence of sense-organs and follows Lenhossek and Retzius in ascribing the sensitiveness of Lumbricus to isolated nerve-cells. These cells I regard as the constituent cells of epidermal sense-organs. I regard Smirnow's statement (I.e., p. 574) concerning the striation and elevation of the cuticular over the "sensibel Nervenzellen " as strong evidence of this. He says : " Die Strichelung der Cuticula ist haufig scharfer ausgebildet an den Stellen, v.o sich die ausseren Enden der Nervenzellen mit der Cuticula beriihren, worauf bereits G. Retzius aufmerksam macht, ebenso wie auf die hiigelartige Verdickung der Cuticula, die an diesen Stellen manchmal vorkommt. Ob diese Strichelung von einer Canahsation der Cuticula abhangt, kann ich ebensowenig entscheiden wie Prof. G. Retzius." My study has convinced me that these striations over the "Nervenzellen" are truly due to canals in the cuticula, and that the hairs borne by these cells pass through these canals to the exterior. The " hiigelartige Verdickung " is, as I interpret it, the elevated summit of a sense-organ.

2. I have never found the protoplasmic processes of the " Nervenzellen " very long or forming a part of the " subepithelialen Plexus." In thick sections, bits of the subepidermal network sometimes appear as if continuous with these processes ; in every case in which I have been able to trace the protoplasmic processes of the sense-cells, I have found that they end at or on the basement-membrare under the sense-organ in which the cell they come from is situated, and that their course is usually sinuous owing to the fact that they have to reach this membrar.e by passing between the small basal cells of the sense-organ, — cells which Smirnow does not mention.

3. I do not find that the sensory fibres form a part of the subepidernial network. I have always found that these fibres pass in bundles, without any connection with other structures in the epidermis, directly to the nearest epidermal nerve.

2 28 LANG DON. [Vol. XI.


From the foregoing study, the following conclusions seem to me to be warranted :

I. The epidermis of Lumbricus agricola, Hoffm. contains a

4. Smiinovv believes that some of the intraepidermal nerve-fibres surround the sense-cells. Although I have searched carefully, I have failed to find any evidence of such connection between these fibres and the " Nervenzellen " — i.e., the sense-cells of a sense-organ. In thick sections, it sometimes appeared as if the intraepidermal nerve-fibres penetrated the sense-organs, but under the oil immersion such fibres were found to be always outside of the sense-organs, among the supporting cells. I have preparations in which the intraepidermal fibres are stained in such numbers in the tissue between sense-organs that it seems to me unlikely that my failure to find fibres surrounding the sense-cells is due to a failure of the stain to act on such fibres.

5. Smirnow describes and figures the " Geschmackszellen " found by Retzius ('92) in the buccal cavity. Smirnow's figures and description of these so-called cells lead me to the conclusion that they are but the outer ends of ducts from the gland-cells which are so numerous in this region. He says: " Der verjlingte Teil der Zelle und der sich an ihn unmittelbar ausschliessende Fortsatz impriigniren sich viel schwacher, als der verdickte Teil der Zelle, wobei die Impragnation mit dem Silbersalz an den Randern des Fortsatzes intensiver ist, als in dem axialen Telle, der heller erscheint und zum Teil von schwiirzlichen Kriimeln und Brocken erfiillt ist. Unter diesen Bedingungen macht der Fortsatz den Eindruck eines hohlen Gebildes, eine Rohre, deren Lumen z. T. von kriimligen Massen erfiillt ist. Dieser eigentiimliche Ban lasst mich vorlaufig zweifeln an der nervosen Natur dieser Gebilde." I have some sections stained by Kleinenberg's haemato.\ylin which were not treated with sodium bicarbonate and which have, therefore, faded. But the glands have retained the stain and stand out distinctly. I find, above the pharynx, glands from which long ducts pass in a sinuous course through the surrounding tissue to the upper wall of the pharynx. Each tube enters the epithelium of this wall and opens exteriorly by a minute pore in the cuticula. Just beneath the cuticula each duct is slightly enlarged. In fact, these tubes appear exactly like the illustrations which Retzius and Smirnow give of the so-called "Geschmackszellen." I find the ducts filled with minute, deeply stained particles of the secretion, which would answer to the "schwiirzlichen Kriimeln " seen by Smirnow ; the secretion in the enlarged apex is often clear, homogeneous, which' would explain the appearance which both Retzius and Smirnow thought might be a nucleus.

Smirnow's discovery was later verified by Retzius in an article entitled " Die Smimow'schen freien Nervenendigungen im Epithel des Regenwurms," which appeared in the Anat. Anzeiger, Bd. 10, Nos. 3 and 4 (Oct. 6, 1894). In new preparations, Retzius succeeds in obtaining the intraepidermal nerve-fibres. lie, himself, now feels doubtful concerning the nervous nature of the " Geschmackszellen " (or, as he calls them in this article, the " Kolbenfasern "). He finds these back of the " Mundhohle," and says : " In eine Reihe von Priiparates habe ich sie nun so massenhaft gefarbt gefunden, dass ich gestehen muss, das mir zuerst ebenfalls ihre nervbse Nature etwas zweifelhaft erschien."


sensory apparatus composed of definite groups of sense-cells whose outer ends pass through the cuticula as sense-hairs and whose inner ends give origin to nerve-fibres which pass directly to the central nervous system and there end freely.

2. The great number of these sense-organs, their existence over the entire body, their great abundance at the two extremities, and the zones of large ones in such positions as easily to be brought in contact with foreign bodies, accounts for the well known and extreme sensitiveness of Lumbricus.

3. These epidermal sense-organs were known to Leydig ('65), Schulze, Mojsisovics ('77), Vejdovsky ('84), Ude ('86), and Kulagin ('88), and their presence can be readily demonstrated.

4. As I have found no isolated nerve-cells in the epidermis of Lumbricus, I am able to account for those described by Lenhossek only on the assumption — which seems to me to be fully warranted by the facts — that he saw the sense-cells of the sense-organs, but failed to recognize the grouping of these cells into organs.

5. The sense-cells are the only cells with which these fibres are connected. They are therefore the nutrient centers of the sensory fibres and true ganglion cells.

6. If there is any differentiation in function between senseorgans of different regions, it is not correlated with any pronounced differences in structure.

7. The efferent nerve-fibres which pass from the central nervous system to the epidermis are not in continuity with any cellular element of the latter. They form a subepidermal network which gives rise to intraepidermal nerve-fibres which end freely between the epidermal cells.

Ann Arbor, Mich., June 18, 1894.

230 LANG DON. [Vol. XI.


'92 Apathy, Dr. St. Erfahrungen in der Behandlung des Nervens)'S tems fiir histologische Zwecke. Mitteilung I. Methylenblau.

Zeitschr. f. wiss. Mikrosk., Bd. ix, pp. 15-37. '90 Cerfontaine, Paul. Recherches sur le systfeme cutane et sur le

systeme musculaire du Lombric terrestre (L. agricola Hoffm.). Arch.

de Biol., tome x, pp. 327-428. Pis. 11-14. '57 Clarke, J. Lockhart. On the Nervous System of Lumbricus ter restris. Afin. and Mag. Hist., Ser. 2, vol. xix. pp. 250-257. '81 Darwin, Charles. Formation of Vegetable Mould through the Action of Worms, with Observations on their Habits. '88 Kulagin, N. Zur Anatomie und Systematik der in Russland vorkom menden Fam. Lumbricidae. Zool. Ans., Jahrg. xi, pp. 231-235. '90 Kulagin, N. Zur Anatomie der in Russland vorkommenden Regen wurmer. Zool. Auz., Jahrg. xiii, pp. 404-406. '65 Lankester, E. Ray. Anatomy of the Earthworm, Part III. Quar.

Jour. Micr. Set., New Sen, vol. v, pp. 99-116. '92 Lenhossek, Michael v. Ursprung, Verlauf, und Endigung der

sensibeln Nervenfasern bei Lumbricus. Arch.f. micros. Anat., Bd.

xxxix, pp. 102-136. Taf. 5. '61 Leydig, Fr. Die Augen und neue Sinnesorgane der Egel. Arch.f.

Anat. raid Physiol., pp. 588-605. '65 Leydig, Fr. Ueber Phreoryctes menkeanus nebst Bemerkungen iiber

den Bau anderen Anneliden. Arch. f. micr. Anat., Bd. i, pp. 249 294. Taf. 16-18. '77 Mojsisovics, Aug. v. Kleine Beitrage zur Kenntniss der Anneliden.

I. Die Lumbricidenhypodermis. Wien.Sitsimgsb. math.-nat. CI., Bd.

Ixxvii, pp. 7-20. Taf. i. '79 Mojsisovics, Aug. v. Zur Lumbricidenhypodermis. Zool. Anz.,

Jahrg. ii, pp. 89-91. '73 Perkier, Edm. Recherches pour servir h. I'histoire des Lombricins

terrestres. Arch. Zool. exper. et gen., tome i, pp. 70-71. '74 Perrier, Edm. Organisation des Lumbricins terrestres. Arch, de

Zool. exper. et geti., tome iii, pp. 331-350. Pis. 12-17. '53 PoNTALLiE. Observations sur le Lombric terrestre. Attn. Sci. Nat.

(^Zool.), Sc^r. 3, tome xix, pp. 18-24. '92 a. Retzius, Dr. Gustaf. Das Nervensystem der Lumbricinen.

Biol. Uiiters., Neue Folge, iii, pp. 1-16. Taf. 1-6. '92 b. Retzius, Dr. Gustaf. Das sensibel Nervensystemder Polychaten,

and Ueber die neuen Principien in der Lehre von der Einrichtung

des sensibeln Nervensystems. Biol. Unters., Neue Folge, iv, pp.

i-io, 49-56.


'86 Ude, Herrmann. Uebcr die Ruckenporen dcr terrecolen Oligochaeten. ZeitscJir. f. luiss. ZooL, Bd. xliii, pp. 87-143. Taf. 4.

'84 Vejdovskv, Dr. Franz. System und Morphologie der Oligochaeten. Taf. I -1 6.

88 VoGT AND Jung. Lehrbuch der praktischen vergleichenden Anatomie. I.


has. c. basal cells.

bas. m. basement-membrane.

bl. V. blood-vessels.

cir. m-as. circular muscle-layer.

cov. c. covering cells.

ad. cuticula.

cut. sp. cuticular spot, surface view of elevation over sense-organ.

cut. sh. cuticular sheath of setae.

ef.f. efferent nerve-fibres going to epidermis.

epth. epithelial cells lining buccal cavity.

i^l. c. gland-cell.

■,'/. /. external opening of the gland-cell.

inters, gr. intersegmental groove.

intra, f. intraepidermal nerve-fibres.

met. metamere.

mus. fib. motor nerve-fibres of muscles.

nepli. p. nsphridial pores.

/. can. pore-canal for sense-hair.

pros, prostomium.

sen. c. sense-cell.

sen. f. sensory nerve-fibres.

sen. //. .sense-hair.

sen. org. sense-organ.

Slip. c. supporting cell.

sub. net. subepidermal network.



All the work was done with Zeiss apochromatic lenses. The figures were drawn to a scale of 9 mm. for .01 mm. and reduced one-half in the plate. Except where noted, they are from camera outlines with compens. ocular No. 8 aiid obj. 4 mm.; the details were filled in with the 2 mm. oil immersion. All the figures except I, 2, and 4 are from silver nitrate preparations.

Fig. I. A sense-organ from a cross section of an anterior metamere, haematoxylin preparation. The sense-hairs are retracted. In the circular muscle-layer is seen an epidermal nerve and a characteristic loop of a blood-vessel. It was impossible to distinguish all structures in lower right side of figure.

Fig. 2. A sense-organ from the buccal cavity, haeni. prep. This shows the depression in the cuticula around the border of the summit of the sense-organ.

Fig. 3. Part of the epidermis from a cross section, showing the appearance of the sense-organs when unstained ; the connection of the intraepidermal nervefibres with the subepidermal network and of this network with the efferent fibres is also shown. Some of the basal processes of supporting cells were made lighter to show the nerve-fibres passing across them.

Fig. 4. A bit of the removed cuticula showing the two layers of cuticular fibres, the gland-pores, and a cuticular spot over a sense-organ. The minute openings in the latter are the outer openings of the pore-canals of the sense-hairs.

Figs. 5 and 6. Intraepidermal nerve-fibres among the gland-cells of a caudal metamere. The subepidermal network obscures the basement-membrane. In Fig. 6, the cuticula was absent and a heavy black deposit of silver took its place.

Fig. 7. Ends of intraepidermal nerve-fibres of the buccal cavity. Cuticula absent in a and b.

Figs. 8 and 9. Surface view of the subepidermal network from sections which passed through the epidermis tangent to the basement-membrane. In Fig. 8 the fibres merely cross each other. In Fig. 9 they anastomose at the part marked n.

Fig. 10. Intraepidermal nerve-fibres and an unstained sense-organ in the epithelium lining the buccal cavity. The cuticula was absent and the basementmembrane could not be distinguished. Several efferent nerve-fibres and two bloodvessels traverse the circular muscles.

Figs. 11-14. Sense-organs of the median zone from cross sections of an anterior metamere. The plump, rounded form and bright brown color of the sensecells could not be represented in the.se illustrations. The section passed somewhat obliquely through the organs, so that the bases of some cells are cut off and the cuticula appears wider than is normal. A silver deposit beneath the cuticula adds to the apparent width of the latter. In Figs. 11 and 12 the connection of a sensory fibre with its sense-cell is shown.

Fig. 15. Bases of sense-cells from oblique sections of the sense-organs of the ventral surface of the prostomium. The summits of the cells have been cut off.

JourMorph. l/o/.X/

PI. Kill

FE L<\ngc/on del.

2 34



A chart showing the distribution of the sense-organs. This chart was prepared by camera drawings of characteristic metameres from the removed cuticula of one worm. The metameres are numbered along the mid-dorsal line. The jagged edge represents the mid-ventral line. The metameres are separated by heavy black lines which represent the intersegmental grooves. In each metamere the small black dots represent the cuticular spots over the sense-organs, the double circles the nephridial pores, and the pairs of black rectangular spots the cuticular sheaths of the setae. The chart was drawn to a scale of i dm. for 2 mm. In the plate it is very much reduced.

Journal of Morphology, Vol. XI.





^Vi^^.. .^^










, - cul. sh.



■-.■• ■■+..»•...•..


' ' - ■ ' -.1 ■.




F. E. Langdon del.



Volione XI. October, iSgS- Number 2.





GEORGE WILTON FIELD, Associate Professor of Cellular Biology in Brown University, Providence, R. I.

The present work, begun at Naples, October 4, 1892, was carried on there until March 10, 1893. It was continued at the Zoologisches Institut of the University of Munich during the following summer, and brought to a close at the Marine Biological Laboratory, Woods Holl, Mass., during the summer of 1894.

At the Naples Zoological Station I occupied the " Smithsonian Table," and my heartiest acknowledgments are due to that institution, and to Professor Dohrn and his corps of assistants at Naples ; also to Professor Dr. Richard Hertwig at Munich and to Professor Whitman, Director of the Marine Biological Laboratory, who have extended to me the privileges of their laboratories.

The adaptability of the cells of echinoderms to cytological studies is practically limited to the ovum ; nearly all the other cells are remarkable for their small size, and although even



[Vol. XI.


this smallness is in certain cases of great advantage to the investigator it has up to the present time prevented a comparative study of the spermatogenesis of the echinoderm groups. By the use of remarkably good apochromatic objectives (3 mm.: apert. 1.30 Zeiss) and compensating oculars, I have been able to study the process in several representatives of each class. The species studied included :

Stichopxis regalis. (Salenka.) ^

Holothuria Poli. (D. Ch.) iHolothurioidea.

Cucumaria ciicumis. (C Planci.') J

Antedon rosacea. (Norman.) } Crinoidea.

Echinus microtuberculatiis. (Blr.)

Sphaerechiims gramilaris. (Ag.)

Strongylocentrotiis lividiis. (Brandt.) J>Echinoidea.

Arbacia ptistulosa. (Gray.)

Ec/ihtocarditwi cordatw/i. (Gray.)

Ophioglypha lacetosa. (Lyman.)

OphioDiyxa pentagona. (Miill., Fr.)

Ophiothrix fragilis. (Miill.)

Ophiodei'ma longicatida.

Chaetaster longipes. (Miill., Fr.)

Astropecten pentacatithus.

Asterias glacialis. (O. F. M.)

Echinasier sepositus.

Asterias Fo7'besii.

Many other species were examined. Those species which show any marked deviation from the typical dioecious condition as do, for example, the hermaphroditic forms such as Synapta, Asteriua, etc., or those which have become viviparous as, e.g., AmpJiiura sqiiamata, have been intentionally avoided. The other species examined were either obtainable only in limited numbers, or at the time the sexual organs were in an unfavorable condition.

As was to be expected, throughout the phylum the general aspects of the process were found to be closely similar.

In case of several species of Echinoidea which were examined the reproductive organs were found to be infested by a sort of "yellow cells," apparently an alga (see PI. XVI), Fig. 13. So far as I am aware, these have not been described. They were found frequently in Arbacia pustulosa, but most abundantly in



those forms which live in the deeper water, e.g.^ most of the spatangoids found at Naples, and particularly in Dorocidaris papillata. In many specimens the entire reproductive gland had become a mass of these "yellow cells." The presence of these yellow cells suggests many interesting questions. It is apparently a pathological condition and not symbiotic in the true sense. It is remarkable that here shut in within the urchin's shell they should find conditions favorable for growth. The shape of the sexual glands in the different classes varies very considerably as a comparison of PI. XV, Figs. 6-1 1, will show. The variations are coordinated with the shape of the body cavity in the different species, but the same general plan is found usually throughout the class. So that the figure given may be regarded as more or less typical for each group.


I have made use of the terminology, introduced by LaValette St. George now most commonly adopted. The term spermatogone = " sperm mother cell " is applied to the original cell which detaches itself from the germinal epithelium. The spermatid is the cell which changes directly into the spermatozoon. The one or more generations of cells which mediate between the spermatogone and the spermatid are the spermatocytes. As constituent parts of the cell the term cytoplasm is used for the protoplasm in distinction from the nucleus : nuclein = chromatin ; caryoplasma or caryolymph ^ nuclear sap.

In view of the fact that the term " Nebenkern " has come to be applied to many sorts of intracellular structures including those whose morphology and physiology is known, it seems best, since the history of the middle piece or " Nebenkern " is now better understood, that this non-committal term " Nebenkern " should be replaced by some term which gives a hint as to the nature of this body. Since the middle piece or " Nebenkern " of the echinoderm spermatozoon is formed from the mitotic spindle, the term "mitosome," introduced by Platner, has been adopted and will be used to designate the middle piece, = Nebenkern = corpuscle accessoire of other writers.

238 FIELD. [Vol. XI.


Owing to the extreme delicacy of the cells involved, and the extreme distortion caused by the great majority of the fixing fluids in common use for other tissues, I have endeavored to record with great care the methods which I have found best adapted. Pictet has already given figures showing the effect upon spermatozoa of the reagents most commonly employed (18). All the points described have been studied {A) on the fresh material, teased in sea-water on the slide. The fresh material teased on the slide was then treated with the various reagents run under the cover glass, and the reaction watched through the microscope. (B) The carefully killed and hardened material has been studied by dissociation methods, and {C) by means of sections cut in paraffin.

A. Fresh Material on the Slide ; for Structure and Development

of the Spermatozoon.

I. Neutral dahlia and methyl green. — To a watchglassful of sea-water add a small quantity of concentrated aqueous solution of dahlia. Filter very carefully, several times. Place the living spermatozoa in a drop of this liquid on the slide. After three to five minutes add a drop of a dilute solution of methyl green prepared in the same way. Examine under cover-glass. The nucleus is stained a delicate green ; the mitosome and centrosome violet. This method gives a minimum distortion ; (Figs. 14 and 27; compare with Figs. $B and 4 D) ; but unfortunately the results are very transitory, for after an hour or so the colors fade and finally the nucleus swells {Fig. 28) and bursts, leaving only the tail and mitosome (Figs. 53, 54) (frequently also the centrosome) visible.

If the dahlia in solution in distilled water is added to the salt water under the cover-glass, a precipitate of granules of dahlia is formed, which destroys the value of the preparation.

The value of this method lies in its delicacy, and for that reason very clean and very dilute stains are necessary. The solutions of dahlia and methyl-green in sea-water must be freshly prepared.


2. Tincture of iodine ; a very weak solution in sea-water (it must be very carefully filtered) preserves the shape and size very well. It stains the mitosome and centrosome a dark yellow or brown, the nucleus and tail a light yellow (Fig. 15),

3. Methyl-green in dilute aqueous solution drawn under the cover-glass by aid of filter-paper, stains nucleus darkly ; but after a few minutes a swelling of the heads of the spermatozoa, probably due to osmosis, begins, and the nucleus finally bursts.

4. Chloride of manganese (10% solution), add concentrated aqueous solution of dahlia ; filter. In a drop of this tease the fresh material on the slide. Examine immediately. For after a short time the heads of the spermatozoa become swollen. This reagent was particularly recommended by Pictet (18). I found it remarkably good for use in studying the mode of formation of the tail of the spermatozoon, and fairly good for the mitosome and centrosome (Fig. 24), but not so good as some other methods, notably i and 2.

5. a. Acetic acid (0.1% to 3.0%) combined with dahlia and methyl-green gave fair results. In many instances, particularly with the stronger solutions, the centrosome very gradually swelled, and slipped itself out of the cap-shaped depression in the anterior wall of the nucleus, into which it had fitted (Fig. 20). In some cases it separated completely (Fig. 22).

b. Acetic acid (33%), while preserving very well the shape of the nucleus, distorted beyond recognition the mitosome and centrosome (Fig. 23).

c. Schneider's acetic carmine. Stain for ten minutes ; decolorize in 33% acetic acid; wash in distilled water. Examine in dilute glycerine. The size and shape of the nucleus wellpreserved. There occurs a great distortion by swelling in case of the mitosome and centrosome, but this swelling is in a large measure reduced by the action of the glycerine. The nucleus is stained carmine ; the chromosomes darker. Mitosome and centrosome unstained.

6. Osmic acid is very useful for demonstrating centrosome and mitosome, particularly the former, making it very refringent.

Fix in osmic vapor, by inverting a drop of sea-water containing the teased material over the mouth of a bottle of osmic

240 FIELD. [Vol. XI.

acid (1% or 2%)). The heads of the spermatozoa become slightly distorted by swelling (Fig. 21). The osmic vapor appears to blacken the parts of the spermatozoon unequally, acting quickest upon the mitosome and centrosome.

a. Add methyl-green in solution in sea-water ; after 2-3 minutes remove the surplus with filter-paper ; add a drop of glycerine diluted with a weak aqueous solution of dahlia : the nucleus stains a delicate green ; mitosome and centrosome violet.

b. If treated with very dilute Delafield's haematoxylin the nucleus stains deeply ; the mitosome and centrosome slightly tinged.

c. Stain with very dilute aqueous solution of gentian-violet, mixed with glycerine : after a short time the mitosome and centrosome are darkly stained, the nucleus but slightly.

B. Prolonged Fixation : Teased Material.

This method is valuable for studying the details of spermatogenesis from the spermatogone to the mature spermatozoon. A piece of the testis is teased in a small quantity of water. Fix in Flemming's chrom-osm-acetic (strong formula), Hermann's fluid, or platinum chloride (0.3%) for 24 hours. Wash for 24 hours or more in distilled water, frequently changed and shaken. Treat with some aqueous stain. Mount in dilute glycerine or in damar. Dissociate cells by tapping cover-glass gently with the point of a needle.

7. Fix in Flemming's chrom-osm-acetic (strong formula) for 24 hours.

a. Stain with aqueous solution of safranin 10-20 minutes : only nucleus stained (Fig. 17). Add very dilute aqueous solution of dahlia : nucleus violet-red, mitosome and centrosome violet,

b. Run methyl-green under cover-glass : nucleus darkly, mitosome and centrosome slightly stained.

c. Stain in dilute aqueous solution of dahlia ; add dilute methyl-green : nucleus green, centrosome and mitosome violet (Fig. 16).


d. One per cent aqueous solution of gentian-violet under cover-glass : nucleus deeply stained, mitosome and centrosome slightly.

e. Stain strongly in gentian-violet ; decolorize in water having a trace of acetic : only the nucleus is stained.

/. Gentian-violet, 24 hours ; decolorize in acid absolute alcohol ; stain 3-10 minutes with eosin in absolute alcohol : nucleus purple, mitosome and centrosome pink.

g. With very dilute Delafield's haematoxylin only the nucleus stains.

8. Fix in platinum chloride 0.3% for 24 hours.

a. Stain in aqueous solution of safranin : only nucleus stains.

b. Run dilute aqueous solution of dahlia under the coverglass : nucleus stains red ; mitosome and centrosome violet.

C. The material for sectioning is best killed in Flemming's or Hermann's fluid. Sections in paraffin of different thickness (i-io/i) were studied.

Historical Summary.

The history of the discovery of spermatozoa by Ham and Leeuwenhoek, and the original belief that they were animalcules — the spermatic animalcules, by some investigators regarded as infusoria, by others as allied to cercaria — need not be entered into here. It was not until about 1841 that the first important generalization upon spermatogenesis was made. Kolliker discovered that the " Spermfadern," as he called them — substituting this term for the hitherto adopted " Samenthierchen " — developed from the internal wall of the testis, and hence were metamorphosed cells (15). Later (1847) he described the spermatozoa as formed not from a single entire cell, but only from its nucleus, or even only a part of its nucleus.

As the result of a third work he decided that the spermatozoon is formed purely and simply of nuclear matter ; that the entire nucleus of the germinal epithelium cell became transformed into the spermatozoon. This nucleus, at first

242 FIELD. [Vol. XI.

spherical, elongates and divides into two parts, a denser anterior and a posterior portion. The anterior forms the head, the posterior the tail, which remains rolled up in the interior of the cell, until the spermatozoon becomes free ; it then unrolls itself.

For many years following, the discussion was warmly waged between the school of Kolliker, who regarded the spermatozoon as a purely nuclear production, and those who led by Henle, Schweigger-Seidel, and LaValette St. George believed that the cytoplasm as well as the nucleus took part in the formation.

LaValette St. George by a long series of brilliant work, begun in 1867, made known the general course of spermatogenesis, and established the nomenclature now in general use. He traced the cells from the germinal epithelium, through the stages called by him the spermatogone, spermatocyte and spermatid to the mature spermatozoon.

In 1 84 1 about the same time that Kolliker discovered that the spermatozoa are formed from the cells of the germinal epithelium, R. Wagner gave the first description of the echinoderm spermatozoon ; describing that of HolotJmria Uibulosa as a "lively-motioned organism with a quite round body, and a delicate tail, similar to the Samenthierchen of the teleosts." Later observers confirmed this for other echinoderms: Quatrefages, 1842, for Syjiapta ijihaerens (19) ; Leydig, 1852 (16), A. Baur, 1864 (i), Hamann, 1883 (10), for Synapta digitata; Semper, 1868 (22), for Anapta gracilis, Chirodota iiico7igriia, and HolotJmria edidis ; Koren and Danielssen, 1882 (5), for Trochostoma Thompsonii; Jourdan, 1883 (14), and Vogt and Jung, 1887 (23), for Holothiiria Uibulosa and Cucumaria Planci.

Practically the first investigator to give an approximately complete account of the structure of the echinoderm spermatozoon was Jourdan (14). In addition to the head and tail (usually the only parts noticed by the earlier writers) he seems from his description to have seen not only that portion, which came to be called the " Nebenkern," " middle piece," " corpuscle accessoire," etc., but even that part which will in this paper be shown to be the centrosome ; he says : " at the point where


the tail begins there is a hyaline 'cupula,' distinctly marked off from the finely granular appearing protoplasm of the head." He also noticed that "after treatment with osmic acid, etc., the hitherto spherical form of the head became more heart-shaped, and in the interior of this, a small shining body becomes apparent."

Others who have more or less considered points in echinoderm spermatogenesis are Selenka (21), Fol (9), Flemming (8), Jensen (13), Carnoy (2), O. Hertwig (12), Hamann (10, 11), Russo (20), Cuenot (3, 4). Recently Pictet (18) studied among that of other invertebrates the spermatogenesis in the Echinoids. But besides restricting his work to this single class, he still further limited it to the manner of the change of the spermatid into the spermatozoon. He showed that the head of the spermatozoon is formed by the entire nucleus, and is consequently composed of two substances, the nuclein (chromatin) and the caryoplasma. Further that the nuclein was no longer in separate bodies (chromosomes) but was dissolved, so to speak, in the caryoplasma to form a single homogeneous mass. That the tail of the spermatozoon is formed by the cytoplasm of the spermatid. That the " ' corpuscle accessoire,' or * Nebenkern,' is a body whose office is to eliminate from the seminal cell those substances which have become useless to the spermatozoon." As to its origin he accepts the results of Platner and Prenant.

The General Mode of Echinoderm Spermatogenesis.

The origin of the primary male sexual cells in the echinoderm group agrees with that which obtains in general in the other animal groups, namely, from the germinal epithelium lining the inner surface of the wall of the testis. I have limited my subject to the history of the germinal cells from the point where, as the spermatogones, they detach themselves from the germinal epithelium, until, as spermatozoa, the descendants of these spermatogones penetrate into the cytoplasm of the ovum in the act of fertilization. In other words, I have attempted to follow the origin and the ultimate fate of the various parts which make up the spermatozoon.

244 FIELD. [Vol. XI.

The general history is in its first and simplest aspects a series of repeated mitotic divisions. The number of these divisions in the case of echinoderms is but two. By the first the spermatogone gives rise to two spermatocytes ; by the second each of these spermatocytes forms two spermatids. No later mitoses occur, and each spermatid then by a series of changes of the constituent parts of the cell become transformed into a spermatozoon. The general facts to be noted are, first, that the number of divisions is but two, instead of a large or even an indefinite number as is the case in certain animals, and as was believed to be the case in echinoderms ; and secondly, that both divisions are by mitosis. The spermatogones measure ii-13/x in diameter; the spermatocytes S-io/l* ; the spermatids 5-7/*, as found in teased preparations. Sections of the alveoli of the testis (Fig. 12) show zones pretty sharply characterized by the various stages in the history of the development of the spermatozoa from the spermatogones. Thus the external series of cells, i.e., those next to the germinal epithelium are spermatogones with the nucleus in a resting condition, having large nucleoli. Next to this is a zone where the nuclei are in the various stages of mitosis ; next a zone made up of spermatocytes ; then come the spermatids, and towards the center of the lumen the spermatids in process of change into spermatozoa ; then the immature, and nearer the center the ripe spermatozoa. In certain cases of starfish studied in January, nucleoli were noted in the spermatocytes : this is probably due to the fact that spermatogenesis was not going on actively at that season. Ordinarily, however, there seemed to be no resting periods intervening between the mitoses.

In the echinoderm phylum there exists a very constant and considerable difference in the shape and size of the spermatozoon of each class, and even within the class a slight difference in the shape and size of the spermatozoon of the different species. (Compare Figs, i, 2, 3, 4 and 5.) The Holothurioidea as a group have the largest spermatozoa (Fig. i); next in order of size come the Ophiuroidea (Fig. 4) ; and then the Asteroidea (Fig. 5). In these three classes the head of the spermatozoon


is approximately spherical. In the Echinoidea (Fig. 3) and in the crinoid studied (Fig. 2) {Antedon rosacea) the spermatozoa are smaller and the head is conical.

In many species, e.g., HolotJuwia Poll, SticJiopiis regalis, OpJnoglypJia lacetosa, A. glacialis, CJiaetaster longipes, and Cticiimaria planci, in the fresh living spermatozoa there is seen a clearer, more highly refringent apex to the head (Figs. I, 4, and 5); this has been noticed by several investigators, but so far as I am aware no one has previously worked out its origin and its significance. As will be shown below, this body is the sperm centrosome (thus confirming Cuenot's conjecture) (4).

With the exception of the cell membrane enclosing the head of the spermatozoon, and the probable presence of a minute quantity of cytoplasm, the general anatomy has already been often described, and can be readily understood from the figures ; hence we may turn immediately to a consideration of the more detailed history of the various parts.

The Constituent Elements of the Spermatozoon, Their


In treating the subject it will perhaps be advantageous to consider first the anatomy of the mature spermatozoon, the number and nature of its constituent parts ; and then to trace the history of each of these constituent parts by following its successive phases from the spermatogone, through the cell generations following (the spermatocytes and spermatids), to the mature spermatozoon; and then finally to consider the fate of each of these parts in the penetration of the spermatozoon into the egg in the fertilization process. For it is to be borne in mind that the study of spermatogenesis has shown that the spermatozoon is only a modified cell, and that each of its constituent parts has previously been a part in a preceding less specialized cell. Each, part, then should be traced from the spermatogone.

The mature spermatozoon consists of the following parts :



[Vol. XI,

A. Head.^

[■ Head proper.

Nucleus. Centrosome.

Mitosome. | The so-called " Nebenkern," " middle piece,"

l^Cell membrane. S " corpuscle accessoire," etc. Probably also surrounding the head, and within the cell membrane a minute quantity of cytoplasm.

B. Tail.

Table of Measurements of Constituent Parts of Spermatozoa of the Species of Echinoderms Studied.




Antedon rosacea

Ophiuroidea .


o.. u T \ longitudinal

Stichopus regalis \ *

( transverse

TT 1 ^1. • Ti 1- i longitudinal . Holothuria Poll •,' ^ ^ I transverse

Cucumaria \ longitudinal

cucumis I transverse

cEchinocardium \ longitudinal

cordatum ! transverse

Echinus microtu- \ longitudinal

berculatus ! transverse

Spaerechinus \ longitudinal

granularis ' transverse

Strongylocentro- ( longitudinal

tus lividus ( transverse

. , . , , ( longitudinal

Arbacia pustulosa ; °

( transverse

\ longitudinal

( transverse

Ophioglypha \ longitudinal

lacetosa ' transverse

Ophiomyxa ( longitudinal

pentagona ( transverse

Ophiothrix fragilis < °^

I transverse

Ophioderma \ longitudinal

longicauda ( transverse

'Chaetaster \

, ■) longitudinal

longipes ' °

Astropecten pen- ( longitudinal

tacanthus ( transverse

, . ,. ( longitudinal Asterias giaciahs < °

( transverse

Echinaster ( longitudinal

sepositus ( transverse

, ^ . „ , .. ( longitudinal Asterias Forbsii < "

I transverse

Diameter of

Nucleus, (

Centrosome, M


4.0 fl 4.0 fJ.


2.0 ;tt 4.0 fJ.

3-3f^ 4.0 fl


1.6 fl


4.0 fl




3-3 M

4.2 fj. 2.2 fl

■ 5°/*

2.2 fl

4.0 fl 2.0 fl

0.33 M


2.0 fJL

3-31^ 2.6 fl

.66 fl


2.0 fJ.

3-3 M 2.0 fl

•33 M


4.0 fl 2.0 fl

•SO At

2.0 fl

3-3 M I-3M

.66 m



4.0 fl


2.0 fl 3-3 M

4.0 fl 47 M


I-3M 3-3 1^

2.6fl 2.8 fl


I-3M 3-31^

3-3 M 4-3 M



3-3 M

2.6 fl

1.2 fl

2.0 fl

2.8 fl 3-3 M


3-3 M



l.IfX 2-2 fi

1-3M 2.0 fl

2.0 fj. 3-3M 3-3 M 4.0 fl

I.2M 1.2fl

1-3/* 1.7 fl

1-3/^ 3-M


The Nucleus. — The nuclei of the spermatogones in the outer zone, as seen in a section of the testis, usually alone of all the cells show nucleoli. The nucleus is large relatively to the amount of cytoplasm. It very soon begins the process by which it will ultimately give rise to the spermatozoa. It divides by mitosis and forms the nuclei of two spermatocytes. The number of chromosomes into which the nuclein of the spermatogone collects seems to be 28-36 : in the spermatocyte 16-18. The attempt to count with exactness so small and so numerous bodies, so closely crowded together, is well nigh fruitless. The dividing nucleus plainly is seen to be made up of several substances, e.g., nuclein, staining deeply with methyl green ; the caryoplasma staining lightly ; and another substance which has the appearance of minute granules; these evidently form the mitotic spindle. A centrosome is also present. These granules and the centrosome take a violet color with dahlia (Fig. 29).

The nucleus of each spermatocyte has the same constituent parts as that of the spermatogone. It divides by mitosis and forms the nuclei of two spermatids. The nucleus of each spermatid contains eight or nine chromosomes, and caryoplasma. Within the nucleus there seems to be no signs of the granules which formed the nuclear spindle ; but these granules and the centrosome are now very distinctly seen to be in the cytoplasm (Figs. 32, 33). It should be noted that each of these mitoses are "reducing divisions."

With the spermatid begin those changes in the shape and constitution of the nucleus which are connected with the specialized form of the spermatozoon, (i) The change in shape. In the case of the Holothurioidea, Asteroidea, and Ophiuroidea the change is insignificant, usually only a slight flattening in the antero-posterior direction. But with the Crinoids and with the Echinoidea the nucleus gradually changes from spherical to conical. (2) There seems to be a change in the constitution of the nucleus, as remarked by Pictet (18). I can confirm his observation. The chromosomes (usually nine, sometimes eight) can be demonstrated in the spermatids and in certain of the spermatozoa, probably the immature ones. But in others the

248 FIELD. [Vol. XI.

nucleus remains homosreneous under the same reasjents and conditions which demonstrate the chromosomes in the others. These spermatozoa with the homogeneous nuclei are the most active and most frequently penetrate into the ovum. In sections of fertilized eggs the nucleus of the spermatozoon when in the outer zone of the cytoplasm of the ovum is small, dense, and homogeneous (Fig. 56) ; on the other hand the one which has traveled some distance toward the female pronucleus is considerably larger, and shows the eight or nine chromosomes surrounded by a lightly staining caryoplasma (Figs. 57, 58). Hence it is probable either that the chromosomes in the nucleus of the spermatozoon dissolve in the caryoplasma and form a denser homogeneous mass preparatory to penetrating the membranes and more compacted outer cytoplasmic layer of the ovum, or else the caryoplasma is extruded and the nucleus of the mature spermatozoon consists very largely or even solely of chromosomes (nuclein) closely packed together. There i^ a very considerable reduction in the size of the nucleus in changing from that of the spermatid to that of the spermatozoon, and correspondingly an increase in size, with a reappearance and wider separation of the same number of chromosomes, after the spermatozoon in the fertilization process has passed the peripheral denser portion of the cytoplasm of the ovum. The question is whether this change in the nucleus is merely one of density, i.e., (i) does the nucleus of the mature spermatozoon consist of the same quantity of nuclein and caryoplasma, but with the nuclein dissolved in the caryoplasma, or (2) does the caryoplasma as a liquid portion pass into the cytoplasm before the spermatozoon becomes fully mature, and after the spermatozoon has entered the cytoplasm of the &g% is this caryoplasma restored from the cytoplasm of the egg .-• On account of the very evident alterations in size of the nucleus, one is inclined towards the latter alternative. It seems to be a pretty good case for proving that an interchange of substance goes on between the nucleus and the cytoplasm, and that in this instance at least the substance which passes through the nuclear membrane is the caryoplasma, a liquid protoplasm. If this is the case, it makes strongly for the view that the nuclein is the


essential part, for in this case most of the male caryoplasma passes out into the cytoplasm of the male cell, and the caryoplasma is replaced by liquid from the cytoplasm of a female cell.

The nucleus varies considerably in size and shape in the different classes, as a comparison of the table, p. 246 and figures (1-5) will show. In Antedon and the Echinoidea it is comparatively small ; while larger in the Asteroidea it is largest in the Holothurioidea and Ophiuroidea. In the anterior surface, usually at the very apex, is a depression into which the centrosome fits. In the Echinoidea, however, it is often not at the very tip of the sharp-pointed conical nucleus, but on the side, usually, however, very near the anterior end (Figs. 24, 25, 26).

Centrosome. — I have never seen a centrosome in the spermatogones in the outer zone, i.e., next to the germinal epithelium. I first succeeded in finding it in the dividing spermatogones. At first it seemed to be within the nuclear membrane. But observation on this point is very difficult, and I am by no means positive in regard to the manner or place of first appearance. I have never seen in the spermatogone the actual division of the centrosomes, but it is probable that it occurs preparatory to the mitotic division. With the disappearance of the nuclear membrane the centrosomes are seen to have the usual position at the poles of the nuclear spindle. This spindle seems to be formed of those violet-staining granules which are present in the nucleus before the disappearance of the nuclear membrane (Figs. 29, 30, and 31). As shown in Figs. 30 and 31, these granules remain in close proximity to the chromosomes, the cytoplasm being entirely free from them, previously to the final division, which gives rise to those cells, the spermatids, which will become the spermatozoa. After the division of the nucleus of the spermatocyte into the two spermatid nuclei, and before the division of the cytoplasm has taken place, those granules which composed the spindle fibres are seen scattered through the cytoplasm (Figs. 32, 33) ; the nuclei themselves, rounded up and with a nuclear membrane, are free from the granules. In the cytoplasm also, close beside each nucleus remains one of the centrosomes, which took part in the division of the cell (Figs. 32, 33, 34, 35).

250 FIELD. [Vol. XI.

The probable history of the centrosomes is as follows : the original centrosome appears first in the spermatogone and is of intranuclear origin (I cannot say that the centrosome is not present in the spermatogones while they are still close to the germinal epithelium, but I have never seen it previous to the disappearance of the nucleolus ; and when first seen it seems to be within the nuclear membrane). In the spermatogone it divides into two, which participate in the mitotic division resulting in two spermatocytes. With each spermatocyte nucleus there is left one centrosome, which is one-half of the original centrosome of the spermatogone. Initiatory to the mitotic division of the spermatocyte, this centrosome divides into two, which come to lie at the poles of the spindle. With the completion of this mitosis each spermatid contains one centrosome. Thus from the original centrosome of the spermatogone four centrosomes have been derived, and each of these four centrosomes comes to be placed at the apex of the head of a spermatozoon, and in the fecundation process enters the ovum with the nucleus of the spermatozoon and takes part in the subsequent fertilization.

When the nuclear membrane is formed around the nucleus of the spermatid the centrosome is not included, but is left outside in the cytoplasm (Figs. 34, 36, 41, 43, 45). This fact either points to the probability that the centrosome is extranuclear in the spermatogone, or else that for some reason the condition in the spermatogone varies from that in the spermatid. If the latter is true, a possible explanation for this difference may be found in the fact that with the spermatid the series of mitotic divisions is completed, and that subsequently the constituent parts of the cell will undergo an extreme modification.

In the light of the work of Fol, Guignard, and Conklin on the part which the centrosome takes in the process of fertilization, it is not difficult to see why the centrosome should be extranuclear in the ovum and in the spermatozoon.

In view of the fact that the centrosome is certainly extranuclear in the spermatid and in the spermatozoon, it is only proper to question whether it is not so also in the


spermatogone. Study of some more favorable forms will decide this.

With the formation of the tail it first begins to be possible to say which part of the spermatid will be the anterior point of the spermatozoon. The centrosome, which hitherto apparently has had no special position with reference to the nucleus (Figs. 34, 39), now comes to lie close beside it, and, except in the Echinoidea, usually directly opposite to the point where the tail is forming (Figs. 35, 36, 40, 42, 43).

With the diminution of the cytoplasm around the nucleus, the centrosome comes to lie in a depression in the nucleus, but remains entirely outside of the nuclear membrane. This condition is probably brought about mechanically by the pressure of the tightly drawn cell-membrane which invests the spermatozoon, pushing the unyielding refringent centrosome into the wall of the nucleus.

In the Asteroidea examined the centrosome consists of two distinct parts or substances ; a clear, transparent, lightly staining part surrounding a dumbbell-shaped body deeply staining with dahlia (Figs. 14, 22, 41, 49, 51). This dumbbell-shaped body reminds one of the figures given by various investigators for the first stage in the division of the centrosome.

As the table (p. 246) shows, the size of the centrosome varies widely in the different classes, and even in the different species within the class. It is largest in the Holothurioidea, and smallest in the Echinoidea. Of the Echinoidea examined it is smallest in Echinus microhibeiriilatiis (Fig. 46), where its diameter is only about equal to the diameter of the tail of the spermatozoon. The very small size of the centrosome in the Echinoidea is probably the reason why it seems to have been overlooked by Pictet and others, who have worked on these forms alone.

By means of those reagents which cause a slight swelling of the nucleus or of the centrosome, notably chloride of manganese, dilute acetic acid, etc., the centrosome can be made to slip out of the little depression in the surface of the nucleus (Fig. 20), though it is still held close to the nucleus by the cell membrane which surrounds the entire spermatozoon. The centrosome preserves its shape very well with most reagents.

252 FIELD. [Vol. XI.

strong acetic acid excepted (Fig. 23); even when the nucleus swells and bursts, the centrosome as well as the mitosome remains uninjured.

In the living spermatozoon in sea-water the centrosome can be seen as a brighter, more refringent spot near the anterior end, in the case of the Holothurioidea, Asteroidea, and Ophiuroidea examined (Figs, i, 4, 5). This fact was noticed by Cuenot (4). It was in 1883 seen by Jensen (13), but he interpreted it as merely a depression at the anterior end of the nucleus caused by the shrinkage due to expulsion of the achromatin from the nucleus.

As to the further history of the centrosome : the sperm centrosome was seen and figured by O. and R. Hertwig, Bovieri, and Fol, in the fertilized echinoderm ^^g. But, so far as I am aware, no one has previously worked out the history of this centrosome. The Hertwig brothers believed that it came from the middle piece, since it showed the same micro-chemical reactions. Fol believed that the anterior portion of the nucleus became fragmented off. Pictet pointed out the erroneousness of Hertwigs' and inclined to Fol's view. But, as shown above, the sperm centrosome is the centrosome of the spermatid, and of the previous cell-generations, and it continues in company with the nucleus of the spermatozoon until the completion of the process of fertilization. As soon as it has passed the denser outer layer of the cytoplasm of the Q.gg it draws farther away from the nucleus, and the characteristic radiations appear in the cytoplasm. It no longer directly precedes the nucleus, but comes to lie at one side (Fig. 58).

The Mitosome. — The very small granules, darkly staining with dahlia, visible in the nucleus before the disappearance of the nuclear membrane (Fig. 29), and which later become the fibres of the mitotic spindle (Fig. 31), have been already referred to (p. 246) ; as well also the fact that after the division of the spermatocyte into the spermatids, these granules, which were the mitotic spindle of the spermatogone and spermatocyte, are no longer included within the nuclear membrane, but are scattered through the cytoplasm (Figs. 32, 33, 34). With the beginning of those changes which mark the transformation of


the spermatid into the spermatozoon, these granules gradually fuse into larger and larger bodies (Figs. 34, 35, 36, 39w), until they come to form in the cytoplasm a small number {2-%) of refringent spheres (Figs. 35, 36, 4gm): these finally fuse into a single spherical mass, the mitosome (Figs. 41, 42, 4pn). The mitosome apparently may at first take any position whatever in the cytoplasm, but with the formation of the tail and the consequent gradual diminution of the cytoplasm in the head of the spermatozoon the mitosome soon becomes pressed upon by the cell membrane, and is gradually drawn into the place of least resistance, i.e., between the nucleus and the beginning of the tail, its normal position in the mature spermatozoon.

LaValette St. George's discovery in 1867 of this body, named by him the " Nebenkern," introduced new questions into cellular morphology and physiology. What is and whence comes this body ? What purpose does it subserve .■* The discoverer at first held that this "Nebenkern" is formed by the condensation of a part of the protoplasm of the spermatid, i.e., that it is of cytoplasmic origin (36). This view was adopted by Metschnikoff, Balbiani, and Biitschli from results obtained from their investigations upon various Arthropods, and by Keferstein upon Molluscs.

According to the more recent researches of LaValette St. George, Platner, Prenant, and myself, the "Nebenkern" is formed from the granules which are the remains of the nuclear spindle, and which after the final mitotic division fuse into a single large spherical mass.

As to what part of the mature spermatozoon is formed by the "Nebenkern," there seems to have been a great disagreement among investigators, not only in the different animal groups, but even in many cases with different species within the same group. According to the view of Keferstein (33), LaValette St. George (36), Metschnikoff (43), and Duval (27, 28), for Molluscs, particularly the Pulmonates ; and of Grobben (30) for the Decapods, the " Nebenkern " forms the head of the spermatozoon. According to the observations of LaValette St. George (38), and Biitschli (25), upon Insects, and of Pictet (18) and myself upon Echinoderms it forms the middle piece.

254 FIELD. [Vol. XI.

Prenant (45) finds that in the Reptiles it forms a cap to the head ("la coiffe cephalique "), but that in the Pulmonates it is dissolved in the cytoplasm which finally forms the tail of the spermatozoon.

From the fact that the " Nebenkern " does not in all species act in exactly the same manner in the building up of the spermatozoon, but apparently sometimes the major part becomes placed anteriorly, sometimes posteriorly to the nucleus proper, the earlier investigators have overlooked the smaller portion and hence thought that the head alone or the middle piece alone was formed from the " Nebenkern," i.e., that the mitosome passed entirely into the head or into the middle piece as the case might be. But Platner's observations upon Pulmonates and these upon Echinoderms may put the question in a clearer light. In many cases, e.g., in the Pulmonates according to Platner, the " Nebenkern " comes to form a considerable part of the spermatozoon anterior to the nucleus proper ; the rest of the "Nebenkern" forms the middle piece. He found that in the spermatid the " Nebenkern " is made up of two portions, a part, the "mitosoma," svirrounding a much smaller body, the " centrosoma." In the change of the spermatid into the spermatozoon these two parts separate ; the centrosome part takes a position anterior, but the mitosome a position posterior to the nucleus {i.e., becomes the middle piece). Now in the Echinoderm spermatid the centrosome is at no time contained within the mitosome, but the ultimate arrangement of these parts in the mature spermatozoon is the same as in the Pulmonates. The fact, then, that not only the middle piece but also a part of the head proper is formed from the " Nebenkern " largely explains the cause of so many apparently disagreeing results.

This comparison of the conditions obtaining in different groups in regard to the "Nebenkern," and the relations of mitosome and centrosome, together with the fact that mitosome and centrosome throughout their history have identical micro-chemical reactions (compare Figs. 14 to 28), has led me to infer that they are differentiations of the same substance, yet so well specialized that they subserve specifically distinct func


tions : that the sperm centrosome is that portion of the mitosome material which must be transmitted with the nuclear material in the act of impregnation in order to initiate the ontogeny of a new individual. The differentiation into mitosome and centrosome must have taken place previous to the division of the spermatogone. The question as to whether the nuclei are or are not the sole participants in fertilization depends to a considerable extent upon the decision of the question of the nuclear or the cytoplasmic origin of the mitosome and centrosome. It is clearly not sufficient to show that the mitosome or centrosome is in the nucleus or in the cytoplasm in any certain cell, but the point must be confirmed by reference to as large a number as possible of successive cell generations.

As to what is the fate of the mitosome in the process of fertilization there has been a wide range of opinions, and if we may judge from the great number of divergent observations, its fate is quite different in the various animal groups, and much careful work is still necessary on this point. Observations upon its fate in the case of the Echinoderms have been made by Selenka (21), Pictet (18), Cuenot, and myself. According to Selenka's account, the mitosome becomes swollen, and moving towards the female pronucleus, finally fuses with it ; while the nucleus and the tail are absorbed.

The view which Pictet brought forward, and which seems to be confirmed by my preparations, is that the mitosome breaks away from the nucleus, and that while the nucleus and the centrosome proceed on towards the female pronucleus, the mitosome and tail are left close to the point where the spermatozoon entered the cytoplasm of the ^g%, and there break down and are absorbed. The observations of Pictet and Cuenot appear to rest largely upon the fact that the mitosome was seen in some cases previously to the penetration of the spermatozoon into the ^^^^ ^^ break off from the spermatozoon, and that such spermatozoa, lacking the mitosome, are quite capable of normal fertilization. My results are based upon sections of artificially fertilized eggs of Asterias glacialis.

At this point it may be well to add that though I have a very great number of times observed carefully the mode of

256 FIELD. [Vol. XI.

penetration of the Echinoderm spermatozoon into the q^^, I have never seen the cytoplasm of the ^g^ send up a little projection to meet the incoming spermatozoon, as described by Fol, and now incorporated into all the text-books. On the contrary one sees a depression in the surface of the cytoplasm caused by the mechanical pressure of the spermatozoon ; exactly the same condition as when one presses a sharp pin into a piece of rubber. The path made by the nucleus as it plows through the cytoplasm is plainly visible (Fig. 57).

The female pronucleus moves towards the advancing male pronucleus in this way : at first spherical, it takes on a pear shape, the narrower apex being directed towards the male pronucleus. This anterior projecting process reminds one of a very large, blunt pseudopodium of an amoeba ; and the pronucleus travels by an almost amoeboid motion ; the shape of the nucleus constantly changing, the projection being sent out and the posterior portion pushing forward and filling it out.

Tail. — The tail, a round flagellum about 0.211, 0.3 /u, in diameter varies in length not only in the different groups, but also in different individuals of the same species. In the species studied the extremes are found among the Holothurioidea ; in SticJwpus regalis it is very short, about 40.0 /*, while in HolotJmria Polo it is about 90.0 \x. The other species studied showed all intermediate lengths.

As Pictet has already shown for the Echinoidea (18) the tail is formed from the cytoplasm of the spermatid. Although I can add but little to the results obtained by him, more than to confirm them for representatives of all the classes of Echinoderms, yet for the sake of more fullness I will go over the subject here. Soon after the mitotic division of a spermatocyte into two spermatids, the cytoplasm of the spermatid begins to form a bulging which increases into a large projection like an enormous blunt pseudopodium (Fig. 40) ; the cytoplasm continues to push or flow into this projection and it becomes elongated and flask-shaped, the body of the flask consisting of a large drop of cytoplasm, which is connected with the cytoplasm proper by a narrow neck (Figs. 42, 43). The continued lengthening of the tail takes place with the elongation and


diminution in the diameter of this neck, together with a diminution in the size of the drop of cytoplasm at the tip, as well also by a diminution of the cytoplasm which has hitherto remained around the nucleus and the mitosome in the cell proper (Figs. 35, 36, also Figs. 40, 42, 43, 44, 45, 50). The tail then is the cytoplasm of the spermatid, which has become modified in a very special way. It is possible that its violent motions in the water are expressions of some molecular change which the sea-water brings about in the protoplasm of the tail, for I have often noticed that spermatozoa when first removed from the testis into sea-water lie motionless ; after a short time a slight motion begins, which after a few minutes increases to the normal rapid motion.

Pictet concluded that the tail is attached to the posterior part of the nucleus, but certain facts seem to point otherwise, for in those cases where the nuclei are caused to swell and finally burst, the tail is almost invariably left attached to the mitosome (Figs. 53, 54). Further in the process of fertilization, the mitosome and the tail together separate from the nucleus (Figs. 56, 57). On the other hand, however, is the observation made by Pictet and Cuenot, referred to above, that in many cases, apparently with one species, Asthenosoma, as the rule, the mitosome becomes detached from the nucleus before the spermatozoon penetrates the tg%. It seems most probable that the tail is in direct continuation with the cell membrane which surrounds the spermatozoon. The cell membrane being morphologically but the external slightly changed cytoplasm, probably differs very little, morphologically not at all, from the tail, and these two, the tail and the cell membrane, probably pass insensibly into one another ; the mode of development would seem to prove this.

Cell memhranc. — A delicate cell membrane surrounds the head of the spermatozoon, inclosing the nucleus, centrosome, and mitosome. This membrane is best seen in cases where from some mechanical cause a slight separation has taken place between the nucleus and the mitosome ; this membrane in that case being stretched but still unbroken (Fig. 55). It can also be seen stretching over the centrosome in cases where the

258 FIELD. [Vol. XI.

centrosome has been crowded out of the socket in the anterior end of the nucleus (Fig. 20). In certain preparations which I made of spermatozoa, killed in osmic vapor, stained in Delafeld's haematoxylin, and mounted in dilute glycerine, after a time the nuclei burst ; a portion of the darkly stained nuclear contents escaped, leaving the cell-membrane distinctly visible, I regard this membrane as the original cell membrane which has persisted from the spermatid. Pictet (18) thinks that this membrane dissolves and with the cytoplasm contributes to the formation of the tail, but in the case of the Siphonophores he found that it persisted, but he adds that " it is probable that it ultimately dissolves at the moment of fecundation."


Many of the facts found in this study of Echinoderm spermatogenesis range themselves with a large mass of facts which are being accumulated from all branches not alone of the animal, but also of the vegetable realm, tending still more to strengthen the theory first advanced by E. L. Mark, and lately confirmed by O. Hertwig (32) and others that the polar bodies are aborted eggs ; that there is a close parallelism between the histories of the cells concerned in the formation of the egg and of the spermatozoon ; that the o.g'g, the spermatozo5n, and the three polar bodies are strictly homologous ; and that any difference apparent is to be regarded as a specialization for specific purposes ; the accumulation of all of the food yolk in one of the ova results in the uselessness of the other three (the polar bodies) ; the modification of the four spermatids, derived from the spermatogone (which is the homologue of the unmaturated ovum) by differentiation of the cytoplasm into a vibratile tail, the separation and subsequent extrusion of the no longer useful material of the nuclear spindle, in the form of the mitosome, are modifications merely of parts of the cell. (It is hardly necessary to call attention to the absolute absence of homology between the extrusion of the "polar bodies" and the extrusion of the "mitosome.")


Cases where the conditions lie so simply are by no means common. But investigation will show that it is much more general than has been supposed, and that cases where the spermatogone gives rise to four spermatozoa have been found in widely different groups, in many Vertebrates, Pulmonates, Lepidoptera and other Arthropods, and Ascaris (similar conditions also are found in the pollen formation in many plants) ; hence we are led to consider the probability that this condition, which is well exemplified in the case of the Echinoderm, is the simple ancestral condition, from which the great variety of types of spermatogenesis have been derived in adaptation to the various modes of life of the different animals and plants ; the necessity for a greater or a lesser number of spermatozoa being met by an alteration in the number of cell divisions between the spermatogone and the spermatozoon. The great majority of the Echinoderms have continued in this primitive condition : the number of spermatozoa being at least four times the number of the eggs, since each spermatogone, the homologue of the unmaturated &g2,, gives rise to four spermatozoa, but further from the fact that the spermatogones being very much smaller than the ova, an enormously greater number can be contained in the testes. This numerical ratio between the ova and the spermatozoa has been sufficient to maintain the race, particularly since the conditions governing the existence of the Echinoderms have undergone remarkably slight changes from the time of their first appearance up to the present day.

The demonstration of the close similarity between the histories of the male and female cells throughout the animal and vegetable kingdoms will render easier the understanding of hermaphroditism and its kindred subjects.

In closing, I will call attention to the following points :

1. The size and shape of the spermatozoa differ in the various classes. In the Holothurioidea, Ophiuroidea, and Asteroidea, the head is spherical; in the Crinoidea and Echinoidea it is conical.

2. A cell membrane completely surrounds the spermatozoon. The tail is in connection with this cell membrane, and not attached directly to the nucleus or to the middle piece.

26o FIELD. [Vol. XL

3. The number of spermatozoa formed from a spermatogone is four, by means of two mitotic divisions.

4. Each mitotic division from the spermatogone to the spermatid is a "reducing division," i.e., the number of chromosomes is reduced one-half by each division.

5. The number of chromosomes in the spermatozoon is nine. This number is characteristic for the Echinoderm phylum.

6. The nucleus of the spermatid contains chromosomes and caryolymph. When this nucleus becomes the nucleus of the mature spermatozoon chromosomes and caryolymph are not distinguishable. They have either mingled to form a homogeneous mass, or else the caryolymph has been extruded. The smaller size of the nucleus of the spermatozoon points to the latter alternative.

7. When the nucleus of the spermatozoon in the fertilization process has passed the outer denser cytoplasmic portion of the ovum it increases in size ; caryolymph and chromosomes appear again. Hence, one would infer that the male pronucleus derived its caryolymph from the cytoplasm of the ovum.

8. The small refringent body seen by various investigators at the apex of the head of the spermatozoon is shown to be the centrosome. It is derived directly from the centrosome of the previous generations of cells. It is extranuclear in position in the spermatid and spermatozoon ; possibly intranuclear in the spermatogone and spermatocyte.

9. The sperm-centrosome, with its definite spherical outline, consists of two sorts of substances, a denser central portion surrounded by a clearer homogeneous mass. I have never succeeded in finding the ^gg centrosome.

10. The sperm-centrosome is very small in Crinoidea and Echinoidea ; much larger in Holothurioidea, Ophiuroidea, and Asteroidea. Since throughout it has the same optical appearance and shows the same micro-chemical reactions as the mitosome (Nebenkern ; Mittelstiick ; middle piece), and also from a comparison with the conditions found elsewhere, particularly in the Pulmonates (where in the spermatid the centrosome is contained within the mitosome, but in the mature spermato


zoon, the centrosome becomes placed anterior, the mitosome posterior to the nucleus [Platner]), it seems probable that centrosome and mitosome are differentiations of one and the same substance. This is th^ same conclusion reached by Watase (46) arguing on another line.

1 1 . The mitosome and centrosome represent the material of the nuclear spindle (cytomicrosomes). The mitosome is that portion which is of no further use, while the centrosome is that part which enters with the nucleus, and is either the necessary complement of the centrosome and spindle of the ovum in fertilization, or else it alone constitutes the mechanism for initiating cleavage. The differentiation of mitosome and centrosome may be a device for overcoming the mechanical difficulties of transferring a large quantity of spindle-forming substance through the ^g^ membranes and denser outer-cytoplasmic portion.

12. The mitosome takes no part in the fertilization process, but together with the tail remains near the periphery of the Qg^, and is rapidly broken down and absorbed ; while the nucleus and centrosome of the spermatozoon push on towards the nucleus of the egg.

Marine Biological Laboratory,

Woods Holl, Mass.,

Aug. 24, 1894.

262 FIELD. [Vol. XL


1. 1864. Baur a. Beitrage zur Naturgeschichte der Synapta digi tata. Dresden, 1864. Nova acta Acad. Leap. Carol. ^ Vol. XXXI.

2. 1884. Carnoy, J. B. La Biologie cellulaire. Lierre, 1884, pp. 225 226.

3. 1 891. CuENOT, L. Etudes morphologiques sur les Echinodermes.

Arch, de Biologie, T. XI, 1891.

4. 1892, CuENOT, L. Notes sur les Echinodermes. Zool.Anz.,V>6i.l^,

1892, p. 121.

5. 1882. Danielssen, D. C. Holothurioidea. (The Norwegian North

Atlantic Expedition, 1876-78. Zoology^ Christiania, 1882.

6. 1892. Field, G. W. The Larva of Asterias. Quar. Journ. Micr.Sc.

London, Vol. 34, 1892.

7. 1893. Field, G. W. Echinoderm Spermatogenesis. Anat. Anz.,

VIIL Jahrg., 1893.

8. 1 88 1. Flemming, W. Beitrage zur Kenntniss der Zelle und ihrer

Lebenserscheinungen. 3. Theil. Arch. mikr. Anat., Bd. XX, pp. 1-86, 1 88 1.

9. 1879. FoL, H. Recherches sur la fecondation, ^/^. Mem.Soc. Phys.

Nat. Geneve, T. XXVI, p. 260.

10. 1883. Hamann, O. Beitrage zur Histologic der Echinodermen,

I. Die Holothurien {Pedata^ und das Nervensystem der Asteriden. Zeitschr. f. wiss. ZooL, Bd. XXXIX, pp. 145190, 1883.

11. 1889. Hamann, O. Anatomie der Ophiuren und Crinoiden. Jen.

Zeitschr. f. Naturwiss., Bd. XXIII, 1889.

12. 1885. Hertwig, O. Das Problem der Befruchtung und der Iso tropic des Eies. Jen. Zeitschr. f. Naturwiss., Bd. XVIII, pp. 276-318.

13. 1883. Jensen, O. S. Recherches sur la Spermatogdn^se. Arch.

Biol., T. IV, 1883.

14. 1883. JouRDAN, Et. Recherches sur I'histologie des Holothuries.

Annates du Mttsee dhist. nat. de Marseille. Zoologie, T. I, No. 6. Marseille, 1883.

15. 1841. KoLLiKER, A. Beitrage zur Kenntniss der Geschlechtsverhalt nisse und der Samenfliissigkeit wirbelloser Thiere. Berlin, 1 841.

16. 1852. Leydig, F. Anatomische Notizen iiber Synapta digitata.

Miillers Archives, 1852, pp. 407-519.

17. 1 889-1 892. LuDWiG, H. Echinodermen. Bromi's Klassen und

Ordnungen des Thierreichs.


18. 1 89 1. PiCTET, C. Recherches sur la Spermatogdnese chez quelques

Invertebrds de la Mediterrande. Mitt. a. d. Zool. Station zu Neapel, Bd. X, 1891, pp. 92-108.

19. 1842. QuATREFAGES, A. DE. M^moire sur la Synapte de Duvernoy

{Synapta Duvernoea. A. de Q.). Ann. des Sciences naf., 2 Ser. Zool., T. XVII, 1842, pp. 19-93.

20. 1891. RUSSO, A. Ricerche citologiche sugli elem. semin. delle Ophi ureae. Internat. Monatssch. f. Anat. u. Phys., Bd. VIII, Heft 8, 1891.

21. 1878. Selenka, E. Zoologische Studien. (i) Befruchtung des Eies

von Toxopnetistes variegatus. Leipsic, pp. 5-7, 1878.

22. 1868. Semper, C. Reisen im Archipel der Philippinen. 2. Theil,

Wissenschaftliche Resultate. I. Band. Holothurien. Leipsic, 1868.

23. 1887. VOGT u. Jung. Lehrbuch der praktischen vergleichenden

Anatomie. Braunschweig, pp. 646-679, 1887.

Among other works referred to are :

24. BoviERi. Befruchtung. Merkel und Bonnet's Ergebnisse, Bd. I,


25. BiJTSCHLi. Nahere Mittheilungen iiber Bau und Entwickelung der

Samenfaden bei Insecten.

26. CONKLiN, E. G. The Fertilization of the Ovum. Biological Lectures

at M. B. L., Woods Holl, Mass., 1893.

27. Duval. Etudes sur la Spermatog^nese chez la Paludine vivipare.

Revue Sc. N. MontpelHer, T. VIII, 1879; ^Iso in Jour. Micr. Paris, 4me Annee, 1880.

28. Duval. Recherches sur la Spermatogdnese etudi^e chez quelques

Gastdropodes pulmonis. Jour. Micr. Paris, y^^ Annde, 1879; also in Revue Sc. N. Montpelher, T. VII, 1878.

29. FoL. Le quadrille des Centres. Arch, des Sc. phys. et nat. de

Genlve, T. XXV, 1891.

30. Grobben, C. Beitrage zur Kenntniss der mannlichen Geschlechts organe der Decapoden. Arb. z. Inst. Wien, Bd. I, 1878, pp. 23-50.

31. GuiGNARD. Nouvelles Etudes sur la Fdcondation. Ann des Sc.

nat., T. XIV, Botanique, 1891.

32. Hertwig, O. Vergleich der Ei- und Samenbildung bei Nematoden.

Arch.f. mikr. Anat., Bd. XXXVI, 1890.

33. Keferstein, W. Spermatogenesis der Pulmonaten. Bronn's Klassen

tmd Ordnungen des Thierreichs, Bd. Ill, p. 121 5.

34. KoLLiKER, A. VON. Die Bildung der Samenfaden im Blaschen.

Nouv. Mdm. Soc. Helvet. Sc. N., Vol. VIII, 1847, pp. 1-82.

35. KoLLiKER, A. VON. Uber die VitaHtat und die Entwickelung der

Samenfaden. Verh. Physik. Med. Ges. Wiirtzburg, Bd. VI, 1856, pp. 80-84.

264 FIELD. [Vol. XI.

36. LaValette, St. George. tJber die Genese der Samenkorper.

2. Mitt. Arch.f. mikr. Anat., Bd. Ill, 1867, pp. 263-273.

37. LaValette, St. George. Spermatologische Beitrage. i. Mitt.

(Bombinator igneus.) Arch. f. mikr. Anat., Bd. XXV, 1885,

pp. 581-593 38. 2. Mitt. (Blatta germanica), idem, Bd. XXVII, 1886, pp. 1-12.

39. 3. Mitt, idem, pp. 285-397.

40. 4. Mitt., idetn, Bd. XXVIII, 1886.

41. 5. Mitt., idefu, Bd. XXX, 1887.

42. LaValette, St. George. Zelltheilung und Samenbildung bei For ficula auricularia. Festsch. zti von Kdllikers 70. Geburtstage, 1887, pp. 409-430.

43. Metschnikoff. Researches upon Spermatogenesis (in Russian).

Art. Versamml. Russ. Nat. Abth. Anat. Phys., 1868, § 56.

44. Platner. Samenbildung und Zelltheilung in Hoden der Schmetter linge. Arch.f. mikr. Anat., Bd. XXXIII, 1889.

45. Prenant. Observations cytologiques sur les elements sdminaux des

Gast^ropodes pulmonis, et des Reptiles. La Cellule, T. IV.

46. Watase, S. Homology of the Centrosome. Journ. of Morphology,

Vol. VIII, No. 2.


















connective tissue.








mitosome (Nebenkern

= middle



piece = corpuscle




All figures except 6, 7, 8, 9, 10, ii, 12, and 13 are magnified 1500 diameters. Camera drawings, with Zeiss 3 mm., 1.30 apert. apochrom. hom. immers. objective, and compensating ocular 18. All the details so far as possible are drawn to scale. The coloring was done directly from the preparations.

Only in Figs, i, 2, 3 A, 4 A, 5 C and D, 35, 36, 40, 42, 43, 44, 45, 48, 50, and 55 is the length of the tail represented.

266 FIELD.


Fig. I. Ripe living spermatozoa of Holothurioidea. K. Holothuria Poll. B. Cucumaria cucumis. C. Stichopus regalts.

Fig. 2. Ripe living spermatozoa of Crinoidea. Antedon rosacea.

Fig. 3. Ripe living spermatozoa of Echinoidea. A. Strongylocentrotus lividus. B. Sphaerechhms granularis. C. Echinus microtuberculatus. D. Echinocardium cor datum (a spatangid). E. Arbacia piistiUosa.

Fig. 4. Ripe living spermatozoa of Ophiuroidea. A. Ophiomyxa pentagona. B. Ophioglypha lacetosa. C. Ophioderma longicauda. D. Ophiothrix fragilis.

Fig. 5. Ripe living spermatozoa of Asteroidea. A. Echinaster sepositus. B. Asterias glacialis. C. Chaetaster longipes. D. Astropectcji pentaca^ithus.

Fig. 6. Diagram of testis of Ophioglypha lacetosa (Ophiurid).

Fig. 7. Diagram of testis of Astropecten pefitacanthus (Asterid).

Fig. 8. Diagram of testis of Cucumaria cucumis (Holothurid).

Fig. 9. Diagram of testis of Asterias glacialis (Asterid).

Fig. 10. Diagram of testis of Stichopus regalis (Holothurid).

Fig. II. Diagram of testis of Brissus unicolor (Echinid).

Fig. 12. A portion of a section of an alveolus of the testis, showing the general mode of spermatogenesis ; the spermatogone {sg) being next to the germinal epithelium, the mature spermatozoa (j-«.) in the center of the lumen of the alveolus. Many of the cells are omitted for the sake of clearness.

.lounial of Morphology. Vol.M.


1 r f y 1

A \ B ' r /) I

■.? £




Sd -■'ij

A ?■


-^s (Si; 9ft^k •••?»

M- iP »>:

Z^X-jluUff/uTOtfrJec r-OTii/i/-! «.« 

268 FIELD. [Vol. XI.


Fig. 13. " Yellow cells " from testis of Dorocidaris papillaia. A, fresh, living. B, stained with methyl green. C, stained with dahlia.

Fig. 14. Spermatozoon. y4i-/^r;a.f ^/a«'a/zj', sea-water + dahlia ; sea-water + methyl green.

Fig. 15. Spermatozoon. Asterias glacialis, dilute tincture of iodine in seawater.

Fig. 16. Spermatozoon oi Asterias glacialis. Flemming (strong formula) 24 hours : water 24 hours, dahlia, methyl green, glycerine.

Fig. 17. Spermatozoon of Astei'ias glacialis. Flemming (strong formula) 24 hours : water 24 hours, safranin 1 5 minutes.

Fig. 18. Spermatozoon of Asterias glacialis. Flemming (strong formula) 24 hours : water 24 hours, Delafield's haematoxylin.

Fig. 19. Spermatozoon of Asterias glacialis. Flemming (strong formula) gentian violet (24 hours); acid absolute alcohol: eosin in abs. alcohol 3 minutes, clove oil.

Fig. 20. Spermatozoon of Astropecten pentacanthtis. Very dilute acetic + dahlia ; after 5 minutes. The centrosome and mitosome swell slightly.

Fig. 21. Spermatozoon of Astropecten pentacaiithiis. Osmic vapor; dahlia; glycerine.

Fig. 22. Spermatozoon of Chaetaster longipes. Acetic 3%; methyl green + dahlia. The centrosome has separated from the depression in the wall of the nucleus, and has become entirely free from the spermatozoon.

Fig. 23. Spermatozoon of Chaetaster longipes. Acetic 33%; methyl green. A great distortion of the centrosome and mitosome results.

Fig. 24. Spermatozoon of Strongyloccntrotiis lividus. Chloride of manganese 10% + dahlia.

Fig. 25. Spermatozoon oi Echinocardium cordatum. Dilute gentian violet in sea-water.

Fig. 26. Spermatozoon of Echinocardiitm cordatum. Chloride of manganese 10% -f- dahlia. The normal shape of the nucleus is somewhat altered by the reagent.

Fig. 27. Spermatozoon of Ophiothrix fragilis. After 25 minutes in sea-water + dahlia.

Fig. 28. Spermatozoon of Ophiothrix fragilis. After one hour in sea-water + dahlia. The nucleus has swollen considerably.

Fig. 29. Spermatogone : Stichopus regalis. (Note. — Figs. 29 to 34 upon treatment with Flemming (24 hours); water (24 hours); methyl green and dahlia; glycerine.) The nucleolus has disappeared. The nuclear membrane still persists. Two centrosomes are present.

Fig. 30. Spermatocyte : Stichop7is regalis.

Fig. 31. Spermatocyte: Stichoptis regalis. In mitosis; all of the chromosomes are not shown in the drawing. (Note. — The line from cy should go to the spindle. No nuclear membrane present.)

Fig. 32. Spermatocyte : Stichopus regalis. After division of the nuclei to form two spermatids. The cytomicrosomes (cy') which formed the spindle, previously inside the nucleus, are now in the cytoplasm. (Note. — In the upper cell the letters, c and w, should be reversed.)


Fig. 33. Spermatocyte : Stichopus regalis. After division of the nuclei to form two spermatids. The cytomicrosomes which formed the spindle, previously inside the nucleus, are now in the cytoplasm.

Fig. 34. A Spermatid of Stichopus regalis. The cytomicrosomes are fusing into larger masses, 7n, which will ultimately fuse and form the mitosome. The centrosome is plainly in the cytoplasm.

Fig. 35. A metamorphosing Spermatid of Stichopus regalis. Sea-water + dahlia and methyl green preparation. Shows a later stage in the formation of the mitosome, and of the tail.

Fig. 36. A metamorphosing Spermatid of Stichopus regalis. Sea-water + dahlia and methyl green preparation. Shows a later stage in the formation of the mitosome, and of the tail.

Fig. 37. Spermatogone: Asterias Forbsii. Flemming (24 hours); distilled water (48 hours) ; Delafield's haematoxylin ; dilute glycerine.

Fig. 38. Spermatocyte: Asterias Forbsii. Flemming (24 hours); distilled water (48 hours) ; Delafield's haematoxylin ; dilute glycerine.

Fig. 39. Spermatid : Ophyoglypha lacetosa. Sea-water -f- dahlia and methyl green.

Fig. 40. Metamorphosing Spermatid of Ophiothrix fragilis. Platinum chloride 0.3% {12 hours): dahlia and methyl green. Shows mode of formation of mitosome, and of the tail.

Fig. 41. Spermatid of Chaetaster longipes. Platinum chloride 0.3% (12 hours); dahlia and methyl green. Shows mode of formation of mitosome, and of the tail.

Fig. 42. Metamorphosing Spermatid : Chaetaster longipes. Platinum chloride 0.3% (12 hours); dahlia and methyl green. Shows mode of formation of mitosome, and of the tail.

Fig. 43. Metamorphosing Spermatid : Chaetaster longipes. Platinum chloride 0.3% (12 hours); dahlia and methyl green. Shows mode of formation of the tail.

Fig. 44. Metamorphosing Spermatid nearly transformed into the spermatozoon. Sphaei-echimis granularis. Sea-water + dahlia and methyl green.

Fig. 45. Metamorphosing Spermatid nearly transformed into the spermatozoon. Arbacia pustulosa. Chloride of manganese 10% + dahlia and methyl green.

Fig. 46. Metamorphosing Spermatid : Echinus microtuberculatus. Sea-water -f- dahlia and methyl green.

Fig. 47. Metamorphosing Spermatid nearly transformed into the spermatozoon. Arbacia pustulosa. Chloride of manganese 10% -f- dahlia and methyl green.

Fig. 48. Spermatid nearly transformed into a spermatozoon. Sphaerechinus granularis. Sea-water -f dahlia and methyl green.

Fig. 49. Spermatid, almost completely transformed Asterias glacialis. Osmic vapor. Centrosome and mitosome seem to be rendered more refringent, and are blackened somewhat more than the rest of the cell. The mitosome has not yet become a single mass.

Fig. 50. Spermatid nearly transformed into a spermatozoon. Arbacia pustulosa. Chloride of manganese 10% -\- dahlia and methyl green.



Fig. 51. Mature spermatozoon : Asterias Forbsii. Sea-water + dahlia, drawn while it was living. Note the structure of the centrosome (c), also compare with centrosome in Figs. 14, 22, 41, 42, 43, 49.

Fig. 52. Mature spermatozoon of Arbacia pustulosa after one hour in chloride of manganese 10% + dahlia.

Fig. 53. Mature spermatozoon of Arbacia pustulosa, but after a drop of methyl green, aqueous solution was run under the cover-glass. The nucleus has burst and disappeared, leaving only the mitosome and tail visible.

Fig. 54. The same result upon similar treatment of a spermatozoon of an Asteroidea.

Fig. 55. Spermatozoon, very nearly mature. Chaetaster longipes. Sea-water + dahlia.

Fig. 56. Spermatozoon of Asterias glacialis very soon after entering the cytoplasm of the ovum. Dahlia, methyl green, glycerine preparation. The mitosome is left behind in the periphery of the cytoplasm of the ovum.

Fig. 57. Spermatozoon of Asterias glacialis, later. Flemming, section in paraffin. Safranin, section in paraffin.

Fig. 58. Spermatozoon of Asterias glacialis. The male pronucleus ; the centrosome now lies at the side of the nucleus of the spermatozoon. The radiations are beginning to appear. The arrow shows the direction which the male pronucleus is taking to meet the female pronucleus. The mitosome and tail have been absorbed. Section in paraffin.

Journal of Morphology Vol. XI.








^ ■;;, t















'"•■■Jmi- ■»









W- ch





':>m TO-'

■n q/-i
















i "




-^ r

CW '"

p " '






1 /



The spermatogenesis of the earthworm is characterized, in its external features, by a number of peculiarities which have not yet been brought into relation with the more usual types. The only published work on the subject is by Bloomfield ('80), who gave a fairly accurate account of the external features, drawn chiefly from living specimens or from glycerine preparations, but he failed to make out the internal changes undergone by the developing cells. Bloomfield' s principal results may be briefly summarized as follows: (i) The early germ cell is not entirely used in the formation of spermatozoa ; a central part remains passive and serves to carry the developing spermatic cells. This central part is called the sperm blastophore and may or may not be nucleated. (2) The sperm blastophores increase by division while in the testis, and disappear, probably by atrophy, after the spermatozoa leave them. (3) The blastophore corresponds to the nucleated supporting cells (Sertoli cells) of the frog and salamander. (4) The large nucleus of the early sperm cell divides many times to form "secondary" nuclei which " stand out around the central mass or blastophore of the generating spheroid with very little protoplasm clothing them." These nuclei become the rod-like heads of the spermatozoa. (5) The protoplasm collects in a small " cap or knob-like mass at the distal end" of the developing cell, and from this grows out the long vibratile tail of the spermatozoon. (This "mass" must be the archoplasm of the spermatid.)

The present investigation was first undertaken in the hope of explaining the significance of the "sperm blastophore" and of identifying the various stages of spermatogenesis in accordance with the usual terminology. It soon became apparent that Liimbriais is an especially favorable form for tracing the origin of the various parts of the spermatozoon, and especially the history of the archoplasm, and to this division of the sub

2 72 CALKINS. [Vol. XI.

ject the following paper is mainly devoted. The history of the nuclear elements proved less easy to follow and the results are less satisfactory, although they present certain new points of interest.

The nomenclature adopted is that of LaValette St. George (•78), according to which the three principal stages of the developing cells are designated as (i) "spermatogonia," (2) "spermatocytes," and (3) "spermatids." It is, however, difficult strictly to apply these terms to the Annelida, for the different stages are not confined here to different zones of the testis. In Lumbricus, for example, the later stages occur in the seminal vesicles, while the testes contain only spermatogonia. Careful study of the vesicles has failed to demonstrate in them any definite arrangement or order comparable to the " Wachsthumszone" or ^'Reifujigssone found in many other forms.

Relative position, therefore, is not a guide to the distinction between spermatogonia, spermatocytes, and spermatids ; nor can dimensions of cells be utilized as a means of recognition, for the same stages are represented by quite diverse sizes and forms. The progressive development of the cell is, however, indicated by differences in the make-up and in the number of chromosomes as many observers have shown, and here, therefore, a trustworthy basis of comparison may be found. According to recent investigations the presence of Vierergnippen indicates the late stages of " spermatocytes of the first order " or cells where there are one half the normal number of chromosomes and where each chromosome is in four distinct parts. "Spermatocytes of the second order" are recognized both by the half number of chromosomes and by the fact that each chromosome is no longer quadruple but double {Zw'eiergriippen, or double chromosomes). Spermatids are recognized, in their early stages, by the presence of single chromosomes of one half the normal number ; in the late stages by the compact and homogeneous chromatin and by the elongating form. The mature spermatozoon consists of head, middle-piece {Mittclsttick), and tail, parts which are represented in the spermatogonium by nucleus, archoplasm, and a part of the cytoplasm, respectively.


The state of differentiation of the sperm blastophore is a superficial guide to differences in age of the cells, but can be depended upon only to mark the latest stages, where spermatids and blastophores are connected by the merest traces of protoplasm.

I desire to express my gratitude to Professor Wilson for his continued encouragement and advice. Also I take this opportunity to express my indebtedness to the American Association for the Advancement of Science for the use of their Investigators' Room at the Marine Biological Laboratory during the season of 1894.

I. Methods,

My best results have been obtained by killing with Hermann's fluid, a solution consisting of fifteen volumes of platinum chloride one per cent, two volumes of osmic acid two per cent, and one volume of glacial acetic acid. Specimens in various stages of sexual development are selected, and cleaned with filter paper in the usual manner. Segments 9 to 13 inclusive are then cut from the worms and left in the Hermann solution for not more than thirty minutes. By this method sections of the testes and seminal vesicles can be obtained {71 situ. When merely the contents of the seminal vesicles are desired a simple and much more satisfactory method is to cut away from the dorsal side the entire alimentary tract together with the calciferous glands, leaving the large seminal vesicles exposed. The sections may then be stained on the slide. For staining, my best results were gained by the use of Heidenhain's iron haematoxylin, and the safranin, gentian violet, and orange triple stain of Flemming. Good results were also obtained by Mayer's haemacalcium and the Biondi-Ehrlich mixture.

The above methods involve the use of heat and imbedding in paraffine, a process which almost invariably shrinks the tissues of Lnmbricus. To avoid this difficulty I found it advantageous to use teased specimens. The method is as follows : the large seminal vesicles of sexually mature worms are selected and teased in a watch crystal containing, usually, the killing agent. The teasing is easily accomplished with fine

2 74 CALKINS. [Vol. XI.

needles, for the groups of spermatic cells are not attached to the trabeculae of the vesicles. After an exposure of not more than ten minutes in Hermann's fluid the acid is drawn off and the cells washed many times with distilled water, allowing them to settle each time. The cells are then passed successively through fifty, seventy, and ninety per cent, and are finally brought into absolute, alcohol. With a fine pipette they are then transferred to a slide coated with equal parts of ^^^ albumen and glycerine. The alcohol evaporates rapidly, leaving the cells firmly attached to the glass by the fixing substance. The only care necessary is to prevent actual drying of the specimens. After this treatment slides can be handled roughly in the alcohols and stains without detaching the cells. This method offers great advantages over the preceding, and gives, I think, truer pictures. Any stain may be used with these preparations. For archoplasm masses Kleinenberg's haematoxylin gives the best result, but curiously enough this stain does not affect the archoplasm in sections prepared either by the paraffine or celloidin method. Chromosomes are most clearly defined by the use of Flemming's triple stain, while the best differentiation of the cytoplasmic and nuclear elements is obtained by the combined use of iron haematoxylin and orange. Good results are also obtained by fixation with corrosive sublimate, and with chromic and picric acid solutions. These acids, although unquestionably distorting the cells, distribute the nuclear elements to a certain extent and render them more conspicuous. Their peculiar effect on the archoplasm will be. described later.

II. The Blastophore,

The blastophore, considered by Bloomfield ('80, '8i) as equivalent to the vertebrate Sertoli cell, is thus briefly described in his paper : " It is in this stage (late spermatogonium) that there is first any indication that as the spermatoblasts are being formed, a slight quantity of protoplasm is being left behind in the center of the generating polyplast, which, as development proceeds, will form the cushion on which the sperm rods may rest. It is best seen in polyplasts which have


been subjected to pressure, when the filament cells or spermatoblasts will be squeezed asunder, but remain connected with the central substance by fine strands of protoplasm. This central mass is the blastophore."

My observations confirm Bloomfield's in regard to the origin of the blastophore, but his description is so meager that a further account will not be superfluous.

These "cushions" with their developing cells (spermatospheres) can best be studied in teased specimens. They are frequently of fantastic shape (Fig. 5), a condition which led Bloomfield to suggest a normal increase by division of the entire blastophore. They become spherical during the later stages of spermatogenesis, especially during the metamorphosis of the spermatid.

A testis of Ltimbriais in section exhibits numerous ellipsoidal and multinucleate cells (Figs, i and 2). The number of nuclei in these cells is inconstant, varying from one or two to as many as twenty, and at this stage they are distributed through all parts of the cell, in the center as well as at the periphery. A study of the various forms shows that the single nucleated cell represents the earliest germ cell, which, by repeated division of its nucleus, gives rise to the multinucleated form. Division figures are frequently seen, and it is a curious fact that in whatever stage of maturation a group may be, the nuclei are all in the same stage of activity at the same time (Figs. 5, 12, and 43).

The multinucleate cells pass from the free edges of the testes into the seminal vesicles, and sections of the latter show that changes in the distribution of the nuclei have taken place. They no longer lie without order in the cell but are arranged around the periphery like the nuclei of an arthropod Qg^ (Fig- 3)- The resemblance to the centrolecithal ^gg is even more striking as development continues, for cytoplasmic cleavage occurs later around each nucleus (Fig. 4), thus differentiating the blastophore from the germ cells. These cleavages deepen until the attachment of the germ cells to the blastophore is reduced to a thin film of cytoplasm. The blastophore remains non-nucleated throughout.

276 CALKINS. [Vol. XI.

The blastophore, therefore, seems merely an excess of cytoplasm, and evidently is not morphologically the equivalent of a vertebrate Sertoli cell as Bloomfield assumes. The Sertoli cell, according to Watas6 ('92), is derived from the primordial germ epithelial cells ; according to von Ebner ('88) it is derived from a spermatogonium which ceases to divide and is devoted to the absorption of nutriment in the shape of fats, etc., for the use of the developing germ cells. These cluster about and upon the Sertoli cell, where, like parasites upon a host, they complete their development. The blastophore, on the other hand, is not a cell, for it has no nucleus. It lives as long as the developing germ cells are connected with it and dies when deserted by the spermatozoa. Nor is there any reason to suppose that it provides nutriment for the spermatic cell.

It is conceivable, however, that the vertebrate Sertoli cell was derived from some such structure as the Annelid blastophore. Both originate from the early spermatogonia or perhaps from the earlier reproductive tissue ; neither of them forms any part of the mature spermatozoon ; the function of the former, however, is clearly defined, while that of the latter is purely conjectural. If, as appears not improbable, the blastophore is merely an excess of cytoplasm, the nuclei of the multinucleate cell upon migrating to the periphery of this cell must have about them that portion of the cytoplasm which is essential to the formation of the mature spermatozoon. If one of the nuclei should remain behind, the result would be, morphologically, a Sertoli cell.

I have been unable to follow the history of the blastophores after they have been deserted by the mature spermatozoa. Their appearance is so altered that we hardly recognize them as the same substance as the cytoplasm of the spermatid. At this time they resemble more closely the protoplasm of protozoa after diffluence. It is probable that their final disappearance is due to atrophy, as suggested by Bloomfield, although many of them serve as food for embryonic parasites [Monocystis) which invariably infest the seminal vesicles of Lumbricus.


III. Phenomena of Reduction.

As previously stated, my results regarding the reducing divisions are not wholly satisfactory, owing to the smallness of the cells and to their lack of consecutive arrangement in the testis and vesicles,

A. Spermatogonia.

The spermatogonia are represented in the testes by the multinucleate cells (Fig. 2). The nuclei are small, yet larger than those of the epithelial cells from which they arose, and as a rule have no nucleoli (Figs. 2 and 3). The chromatin is distinct but not abundant, nor is there any characteristic form. In the cytoplasm by the side of each nucleus is a large and rather indefinite faintly staining body (see Section V). As the nuclei prepare for division, the Kndicel or spirem stage is less conspicuous than in the later sperm cells, its filaments being fine and indefinite. In full karyokinesis neither asters nor astral rays can be seen. Picric acid preparations show small dots, the centrosomes, or archoplasm masses at the poles, and spindle fibres can be traced directly into the chromosomes (Fig. 8) ; the nuclear plate contains thirty-two small and apparently spherical chromosomes (Figs. 7 and 8).

B. Spermatocytes of the First Order.

Spermatocytes of the first order cannot be distinguished by their external appearance. The nuclei have migrated, however, to the periphery of the cell (Fig. 3). I am unable to assert whether karyokinetic divisions of these nuclei intervene between the divisions occurring in the testis and the migration of the nuclei to the periphery of the cells in the seminal vesicles. Such, however, is probably not the case, because of the scarcity of karyokinetic figures in the testis, and because of the average number of nuclei in the multinucleate cell. It seems probable, therefore, that the final division in the testis gives rise to spermatocytes of the first order. Each of these

278 CALKINS. [Vol. XI.

contains thirty-two chromosomes (Fig. 9). The nuclei enlarge after this final division and the entire spermatosphere increases in size (compare Figs. 4, 6, and 12). The archoplasm also is large and distinct (Fig. 4, A).

The resting nuclei of this stage are large and round and exhibit well marked nucleoli. It has been frequently stated that, in the case of developing spermatic cells, the nucleolus is thrown into the cytoplasm ; this certainly does occur in Lu7nbriciis (Figs. 44 and 46), but from the nature of these preparations (these results were seen in only one series of sections) I have not the least hesitancy in asserting that the phenomenon is here an artefact.

After a period of rest (Fig. 10) during which the cytoplasmic cleavages deepen, the nuclei begin their period of greatest activity. The chromatin collects at one portion of the nucleus in a rather small lump (Fig. 11) which afterwards expands and becomes mesh-like, while fine offshoots of chromatin soon appear (Fig. 12), The chromatin elements grow during this period and a thick spirem is formed, nearly filling the nuclear space. The thread is ragged at first (Fig, 12), but later becomes apparently smooth (Figs. 12 and 13), In favorable preparations the spirem is seen to be split longitudinally, and is therefore double (Fig, 12, D). The cell elements are much too small and the spirem much too twisted and interlooped to tell whether it is one long piece. In one case, however (Fig, 13), I was able to see that at most there were only two pieces. Later the spirem becomes transversely segmented into many rod-like bodies (Figs. 14 and 15), which at first appear to be without order, but in later stages are found to be definite both in number and in structure. It is then seen that there are thirty-two of these bodies and that each is double (Fig. 16), From them the " Vic7'ergnippen " are formed by a process differing widely from the method usually described. The thirty-two double chromosomes unite two by two (Figs, 17-20), and are finally arranged in sixteen quadruple groups (Fig. 21). In some cells four quadrivalent groups were seen, together with twentyfour double chromosomes ; in others twelve quadrivalent groups and eight double forms were seen (Fig. 19).


The further history of this stage is as follows : during the ensuing cell division (Figs. 22-27) the " Vierergnippeii" are halved, sixteen double chromosomes going to each daughter cell. Whether this is a reducing division in the Weismann sense cannot be ascertained. It is interesting to note that in no case is the axis of the spindle directed towards the center of the spermatosphere (Figs. 5 and 43). The division planes are radial and the daughter cells thus remain attached to the blastophore.

The reduction in the number of chromosomes has now taken place, not during mitosis, however, but during the antecedent period and through the activity of the chromosomes themselves. The process thus corroborates the view advocated by Boveri (-89) and sustained by Brauer ('93) and others, that the chromatin particles have power to arrange themselves. The formation of the " Vierergricppen, however, differs from that which takes place in Ascaris.

Two main conceptions of the " Vierergnippen" and their mode of origin have been current. One, as advanced by Brauer ('93), is that each quadruple chromosome originates by a double splitting of each chromatic element, i.e., by two longitudinal divisions of each chromosome. The other, advanced as a theory by Weismann, and since demonstrated by Ishikawa ('91), Haecker ('93), vom Rath ('92), Henking ('91), and most ably by Ruckert ('93, '94), is that the " Vierergnippeji originate by two divisions of the chromosomes, the first being longitudinal, the second transverse.

According to my conception the origin of the " Vierergnippen " in Liivibricus is quite different from either of the above. It is as follows : there is indeed a horizontal division (Fig. 12) (shown by the double spirem), and later a transverse division, but this gives thirty-two double instead of sixteen quadruple chromosomes ; the " Vierergr?ippen, or quadruple chromosomes, are formed later by union of these double chromosomes two by two. In the forms examined by Ruckert sixteen double chromosomes are formed ; these are arranged in a nuclear plate and transverse division takes place while in this position, thus forming the " Vierergriippen."



[Vol. XI.

These differences can best be seen, perhaps, by the aid of diagrams. If «, by c, d, etc., represent consecutive chromatin elements in the monospirem, the following series can be obtained, in which I indicates the composition of the Viercrgriippen, II the first division, and III the second division :


CBrauex "93)


© ©



(Haecker '93)



© ©

® ®


a a



tKenking '90


© © ® ®



© ©

According to this schematic arrangement, the Vierergruppeiiy which, in most cases, consist of two contiguous elements (a and b) doubled, in Lumbricus consist of the element a doubled, and united with any other element, x, also doubled. In this case the end result is the same as in Pyrrhocoris or in the copepods, but the method of arriving at it is different.

The process, as in Lumbricus, is carried a step further in Caloptcjius, where, according to Wilcox ('95), there is no preliminary doubling of the spirem ; and, since there is here a union of the spirem segments as in Lumbricus, the elements of

the Vierergruppen would be indicated thus : , x and y representing any other chromatin segment than a and b.

The results obtained in Lumbricus may, therefore, serve as a link in the chain, the extremes of which are represented by the views of Brauer and Wilcox.


C. Spermatocytes of the Secoitd Order.

In spermatocytes of the second order there are but sixteen double chromosomes — one-half as many as in the spermatogonia. After their division the daughter cells, or spermatids, contain sixteen single chromosomes (Figs. 28 and 29).

The following distinctions may now be drawn between spermatocytes of the first and second orders : in the first order occurs reduction in number of the chromosomes due to chromatic activity ; this is followed by a reduction in quantity of chromatin through karyokinetic division ; in the second order the chromosomes are passive, while reduction in mass occurs through karyokinetic division.

D. Spermatids .

The spermatid nucleus is the result of the double reducing division. It contains only half as many chromosomes as in the spermatogonium.

After division of the spermatocytes of the second order the chromatin gives rise to a reticulated mass which gradually becomes more and more compact and homogeneous (Figs. 3033). The spermatid is a small round cell closely attached to the outside of the blastophore (Fig. 6). The archoplasm lies at the extremity furthest removed from the blastophore and is closely applied like a cap to the nuclear membrane (Fig. 35), or lies like an independent cytoplasmic sphere between the membrane and the extremity of the cell (Fig. 6). The cytoplasm has no other distinctive feature ; it is small in amount and apparently only sufficient to cover a comparatively large nucleus. There is no indication of granules, which are so characteristic of the blastophore at this stage (Fig. 36).

The mature spermatozoon is developed from the spermatid by a simple metamorphosis. The nucleus elongates, and the distal extremity becomes drawn out to form the tail.

The first indication of the spermatid metamorphosis is loss of reticulation in the nucleus (Figs. 31, 32). The chromatin becomes dense and homogeneous, forming at first a rounded mass like a ball lying within the nucleus (Fig. 33), but later becoming

282 CALKINS. [Vol. XI.

cylindrical, with rounded ends (Figs. 35 and 36). At the same time the beginning of the tail appears as a minute pointed prominence of cytoplasm immediately above the archoplasm at the distal extremity of the cell (Fig. 6). As the elongation continues the chromatin portion becomes longer and thinner. In chromic acid preparations it still retains its reticulated appearance (Fig. 36), but after Hermann's fluid it is more dense and compact (Fig. 35), and ends distally in the growing tail filament. The archoplasm, which in some cases is drawn out distally, lies in the triangular area at the base of the tail, and, at a later stage, a large vesicle is formed about it (Fig. 37, V) which is typical of the spermatid at this time. The further history is briefly as follows : the nucleus and tail elongate still more ; the vesicle disappears by elongation of the archoplasm mass, and the head of the spermatozoon finally assumes the shape of a long rod with a much longer filamentous tail.

The cytoplasm is apparently stretched to its utmost, for only the most careful observations on favorable preparations will reveal it lying as a mere line around the nucleus and middlepiece (Fig. 37). The young spermatozoon now consists of a filament composed of three parts. First : the nucleus or head, which is directed towards and seems to be attached to the blastophore, but which is in reality separated from it by a small amount of cytoplasm (Figs. 37 and 38 j-). Second : the middlepiece, which appears as a direct continuation of the nucleus so that it can be distinguished only after differential staining (Figs. 38 and 47 m). Third : the tail or flagellum, which is much longer than the nucleus and much finer. The small amount of protoplasm between the head of the spermatozoon and the blastophore becomes drawn out into a sharp spur after the cell breaks away (Fig. 47 s), and this, according to the observations of Foot ('94), acts as a boring-point with which the spermatozoon pierces the Q.^g.

IV. Archoplasm.

The spermatogenesis of Lumbricus offers exceptional advantages for the study of archoplasm, which forms a conspicuous


body in the developing cells, and can be readily demonstrated by the use of most stains.

The position of the archoplasm in the spermatozoa of different animals is interesting because of its bearings on fertilization of the &^g and origin of the male attraction center within the Q.gg. The origin of this body within the ^^^ has been observed in only a few cases. Fick ('93) followed its history in the fertilization of the Axolotl egg and found that its Anlage is the middle-piece of the spermatozoon. Foot ('94) has recently examined the fertilization and maturation processes in Allolobophora foetida {Ltimbriats foetida), and asserts that the attraction center seems to arise from the middle-piece of the spermatozoon. Fol ('91) asserted that in the fertilization of the echinoderm egg, the male attraction sphere originates from the anterior tip of the spermatozoon, but Wilson ('95) has clearly shown that Fol's interpretation of the fertilization process in echinoderms was erroneous. Besides demonstrating the absence of Fol's "Quadrille of the Centers" he shows that the attraction sphere in the fertilized eggs of Toxopneustes variegatits is derived from the middle-piece of the spermatozoon, which rotates through an angle of 180 degrees after entering the Q.gg. Mathews ('95) has observed the same rotation in the fertilized eggs of Asterias and Arbacia. In all of these cases, therefore, where the process of fertilization has been carefully followed, the "sperm center" originates from the middle-piece of the spermatozoon.

What now is the relation between the archoplasm {i.e., sperm center) in the middle-piece of the spermatozoon, and the archoplasm of the early germ cells 1 Is the " sperm center " composed of the same substance as the archoplasm of the spermatocytes and spermatid 1

There are unfortunately too few cases where the fertilization processes have been carefully followed out to enable us to decide this question, and in the cases cited above the results of different investigators are, in a measure, contradictory, while the results obtained by other observers apparently cohere. For example, in the spermatogenesis of the echinoderms. Field ('93) has followed the centrosome from the early spermatic

284 CALKINS. [Vol. XI.

cells to a position at the extreme tip of the spermatozoon, whereas in \\iQ fertilization of the echinoderm Q.gg Wilson ('95) and Mathews ('95) have shown beyond question that the sperm center originates from the middle-piece. In the fertilization of the Q-^Zi and in the spermatogenesis of Limibriais, however, the results of different observers agree. In the fertilization of the Limtbrictis egg Foot ('94) has shown that the sperm center probably arises from the middle-piece, and in the spermatogenesis of Liimbriais I have shown that the archoplasm is a continuous element of the cell and forms ultimately the middlepiece of the spermatozoon ('94).

Many other investigators besides Field have followed the course of the centrosome from the early sperm cells to the anterior end of the spermatozoon, but in most cases the fertilization of the eggs has not been studied. Flemming ('88) in the spermatogenesis of the salamander was in doubt as to the origin of the extreme tip of the spermatozoon, but thought that the middle-piece originated from the chromatin. Platner ('85, '89) showed that the extreme tip of the spermatozoon in butterflies and in some pulmonates {Helix poviatia and Litfiax agrestis) is formed from the centrosome. Benda ('9i) showed that in the spermatozoa of mammals, birds, and reptiles, the ^^ archiplasma'" forms the Spitzenknopf and Kopfkappe, and Moore ('94) asserted that in rat spermatogenesis a portion of the archoplasm (archosome) forms the anterior tip of the spermatozoon, but that the place of the Mittelstiick "is apparently occupied by the spermatid centrosomes and an intermedidr KorpercJien "

A possible explanation of some of these differences may be that in many cases the ^^ Ncbenkcrn" is confounded with the archoplasm, and that the extreme tip of the spermatozoon is formed from this or from a portion of it, instead of from the centrosome and archoplasm. Henking ('9i) showed that in the spermatozoon of PyrrJiocoris aptenis the tip of the spermatozoon is formed from a part (Mitosomd) of the Nebenkern.

I have found two peculiarities in the archoplasm of Ltimbricics due to treatment of the preparations. First in prepara


tions which have been imbedded either in paraffine or in celloidin, the archoplasm does not stain so readily as in teased specimens. Second, different killing agents affect the archoplasm in different ways. After Hermann's fluid it is large and has a more or less reticulated appearance. After picric or chromic acid or corrosive sublimate it is much contracted, and in full karyokinesis appears as a minute and homogeneous dot at the spindle pole. The differences in appearance following the use of these various reagents is quite remarkable and cannot be attributed to individual variation in the several cells.

I have been unable to trace the early history of the archoplasm in spermatogonia, but in the fully developed spindle it may be seen at the two poles (Fig. 8, a specimen killed with picric acid). There is here no indication of radiating cytoplasmic fibres (astral rays).

The archoplasm is a much more favorable object for study in spermatocytes of the first order, since it may be observed in teased specimens which can be more easily stained, whereas in spermatogonia the cells are necessarily studied from sections. In teased preparations from the vesicles the spermatocyte archoplasm can be best demonstrated by staining with Kleinenberg's haematoxylin. An exposure of fifteen minutes is sufficient to turn them a dark blue, the other structures of the cell remaining quite unstained. The archoplasm masses are then so distinct that their history can be easily followed, although preparations which show these changes to the best advantage do not indicate the corresponding changes of the chromatin, and the various parts of the cell must, therefore, be studied independently.

The normal position of the archoplasm is at the distal extremity of the nucleus, although I have frequently seen it at the opposite extremity as well as at all intermediate positions. In some cases it is elongated and flattened and lies closely applied to the exterior of the nuclear membrane. In other cases it is divided into finger-like processes and loops, closely resembling certain stages of the so-called Nebenkern as figured and described by Platner ('85) in the spermatic cells of Liniax ag. and Helix pomatia. Again it is in the form of a sphere ;

286 CALKINS. [Vol. XI.

— but in all forms it is one and the same substance. It divides at the beginning of nuclear activity (Figs. 12, 39 and 45), and, if it occupies a distal position, each half travels around the nucleus through an arc of about 90 degrees (Figs. 39, 40, 41, 42). This makes the axis of the spindle tangential to the periphery of the central sphere, and the subsequent division plane passes radially through nucleus and cytoplasm in such a manner that the resulting daughter cells remain attached to the blastophore. The spindle fibres are formed directly from the archoplasm masses (Fig. 45). In Hermann fluid preparations (teased) the archoplasm can be considered as nearly true to life as is possible with this material. Such is the condition represented in Fig. 43, one of a large number of similar preparations ; the archoplasm extends partly up the spindle fibres, and the latter appear taut, as though they had been pulled from the main mass. In some cases the archoplasm extends up the spindle fibres as far as the chromosomes (Fig. 43). After division of the chromosomes the spindle fibres seem to be in part drawn back into the archoplasmic substance (Figs. 5, and 43 A), although I do not mean to assert that all of the spindle fibres are withdrawn in this manner.

After the reconstruction of the nucleus the archoplasm lies upon the nuclear membrane like a closely fitting cap. It soon becomes spherical, preparatory to redivision, but meanwhile it wanders from its position at the side of the cell to the distal extremity.

I have been unable to follow the division of the archoplasm in spermatocytes of the second order, although it is distinctly seen at each pole of the karyokinetic spindle (Fig. 29). From here it passes directly into the spermatid archoplasm, which appears very distinctly, and is much more susceptible to stains than at any previous period. It is the same in size as at any antecedent resting-nuclear stage (Figs. 4, 12, and 35), and the idea that it is of nuclear origin cannot for a moment be sustained. It lies upon the distal part of the nuclear membrane, and since after karyokinesis it lay at the side of the cell, its position here can be explained only by the supposition that it has moved through an arc of 90 degrees, or that the nucleus


revolves so as to bring the archoplasm into this position. The former is the more reasonable hypothesis.

The further fate of the archoplasm may be described as follows : the spermatozoon is formed from the spermatid by elongation in the direction of its radial axis. The nucleus of the spermatid, which is here compact and homogeneous, elongates, and a tail forms at the distal extremity of the cell (Fig. 6). This elongation of the nucleus continues until the spermatid becomes a long columnar cell, with a filament growing out from one end, the other end remaining attached to the blastophore. The archoplasm mass now lies in a vesicle between the tail and the nucleus (Fig. 37). The nucleus and tail elongate still more ; the vesicle disappears by elongation of the archoplasm mass, and the spermatozoon finally assumes the shape of a long rod with a much longer filamentous tail (Fig. 38). This rod consists of middle-piece and nucleus which can be differentiated only by careful staining. The archoplasm mass of the early sperm cells and the middle-piece of the spermatozoon of Ltimbricus are, therefore, one and the same substance.

The action of acids upon the archoplasm mass deserves further notice. Why should picric acid, for example, reduce the archoplasm from such a structure as that seen in Fig. 43 to. that in Fig. 8 in active nuclei, or from such a structure as that seen in Fig, 4 to that in Fig. 41 in resting periods? The condensation here may give some new light on the nature of centrosome structure. It lends support to the view of recent cytologists, that the centrosome is an uncertain and indefinite element of the cell and composed of granules in a greater or less degree of density.

If, as these observers suppose, the centrosome is but an aggregation of protoplasmic granules, it is easy to conceive that in the archoplasm of Ltimbricus, these granules are not tightly packed, and that a favorable fixing agent will preserve them in this condition. Such a result is given by the use of Hermann's fluid on teased specimens. If a killing agent is used which does not act so quickly as the platino-osmic-acetic mixture, it is conceivable that a condensation of the granules

288 CALKINS. [Vol. XI.

may take place. Certainly in chromic and picric acid preparations the archoplasm masses are greatly reduced, and stain much more intensely with miscellaneous stains than in other cases.

An accessory body {Nebenkeni) is often found in the developing germ cells of Lu^nbricus, usually in the spermatocytes of the first order. This body also appears in the spermatic cells of a great variety of forms where it is often mistaken for archoplasm. It was discovered by LaValette St. George ('67), and has been repeatedly described by subsequent observers. Biitschli ('7l) gave to this body, which he found in the spermatic cells of insects and Crustacea, the name Nebenkern, a name which has clung to it ever since.

The term Nebenkern ("accessory nucleus") is unfortunate, for in it there is nothing to designate any peculiarity or characteristic of the body in question, and it might as well be applied to any unusual structure of the cell. Such various application has, indeed, been made. O. Hertwig ('75) used the term Nebenkern to designate the micronucleus of the ciliates. Nussbaum ('82) applied the same name to bodies constricted off from the nucleus in cells of the hepato-pancreas, and Blochmann ('84) gave the name ^^ Nebenkern " to the "yolk-nucleus," a body found in the early stages of the developing ^g^ cells. Many later writers have used the term in a more or less undefined manner, until to-day the word Nebenkern has no especial significance.

In all of these cases, with the exception of Hertwig's application of the term to infusoria, the name Nebenkern is unsuitable and non-descriptive, for the structure so designated is not a nucleus, although perhaps nuclear in origin. In the case of the infusoria the structure called Nebenkern is much better described by the term " micronucleus " which is now generally adopted. The other bodies in the cell to which the name Nebe?tkern is applied, notably in spermatic cells, in egg cells, and in the glandular cells of the hepato-pancreas, are not homologous with each other ; and the same term therefore should not be used to designate them. In the spermatic cells LaValette ('86), Platner ('86, '89) and Henking ('91) showed that


the Nebenkent originates as the remnant of the interzonal fibres of the karyokinetic spindle, and that it is, therefore, archoplasmic in origin. In pancreas cells, Nussbaum ('82) showed that the Nebenkern is derived by a budding off from the nucleus and is therefore of nuclear origin. The socalled Nebenkern in the &g^ cell originates in the same manner apparently as in the pancreas cells. Valaoritis (-82) described this body as originating by metamorphosis of the germinal vesicle ; Van Bambeke ('93) claimed that it arises by direct transportation of the chromosomes ; and in a recent preliminary paper I ('95) have shown that this body (which Carusin 1850 called the "yolk-nucleus" (Dotterkern) originates from the chromatin network in the nucleus. Here also the socalled Nebe7ikern of Blochmann is nuclear in orisfin. It is apparent therefore that these bodies which differ so greatly in their origin should not have the same name. In the ^g^ cell the so-called Nebenkern is sufficiently well described by the term Dotterkern' or "yolk-nucleus," and a more satisfactory name can be found for the so-called Nebenkern of the pancreas.

The word Nebenkern, if used at all, should be applied to the element of the spermatic cell as originally proposed by Butschli ('71), and I shall adhere to this use of the term in the present paper.

A Nebenkern, then, can be described as an element of the early spermatic cell originating from the interzonal fibres ( Verbindungsfdsern) of the karyokinetic figure. It must, therefore, be archoplasmic in origin, for the Verbindungsfdseni are derived from the archoplasm. Its function, however, as archoplasm, is lost.

That there is need of such an accurate definition of the term Nebenkern in spermatic cells is well shown from the confusion caused by its rather loose application in many instances. For example, Platner ('85, '89) in his several papers on spermatogenesis uses the term Nebenkern as the equivalent to Boveri's ('84) Archoplasnia. This body as described by him, however, seems to be equivalent to both archoplasm and Nebenkern, for according to Platner's observations it consists of two parts, both coming from the archoplasmic spindle. These are first, a

290 CALKINS. [Vol. XI.

central part, derived from the centrosome and which ultimately forms the tip of the spermatozoon, and second, the Alitosoma, derived from the equatorial spindle fibres ( Verbindungsfdsern) which ultimately form the membrane about the axial filament of the tail. The tip of the spermatozoon, therefore, is formed from the Nebenkeim. Moore ('94) confuses the two terms and calls archoplasm a body originating from the spindle fibres after the disappearance of the spindle. Henking ('9i) uses the term Nebciikern in the sense of Butschli, but restricts it to the body formed by the peripheral spindle fibres, while the central spindle forms what he calls the Mitosoma. The Mitosoma in Pyrrhocoris apteriis forms the tip of the spermatozoon, and the Nebenkern a caudal membrane. Hermann ('9i) also gives to the Nebenkern the significance of archoplasm, while Balbiani (93) considers archoplasm, Nebenkern, and yolk-nucleus homologous bodies. Other spermatologists give various meanings to the terms archoplasm and Nebenkern, making it difficult to determine whether a certain element in a spermatozoon (as the tip, for example) comes from the centrosome and archoplasm or from the Verbindungsfdsern of the karyokinetic spindle.

The Nebenkern in Lwnbricus usually lies at the extremity of the cell opposite the archoplasm, and in some cases it persists even during the process of karyokinesis (Fig. 45).

The presence of archoplasm and Nebenkern in the same cell in Lunibricus throws some light on the distinction between these bodies. These differences can be tabulated as follows: (i) the Nebenkern originates as the remnant of the interzonal fibres, while the archoplasm is a constant element of the cell ; (2) the archoplasm plays a certain definite role, while the Nebenkern has no apparent function ; (3) the archoplasm divides and behaves during karyokinesis in the same manner as a so-called centrosome and finally forms the middle-piece of the mature spermatozoon, while the Nebejikem is passive and disappears after a seemingly useless existence. These distinctions are not true, however, for all forms. In many cases the Nebenkern has an important function, such as the formation of enveloping membranes in the salamander spermatozoa.


The archoplasm as described above is apparently equivalent to the central body (centrosome) and its archoplasm. There is no indication that it is of the nature of a Nebejikern. According to Moore ('94) in the spermatogenesis of the rat, the "archoplasm " originates from the spindle fibres of the preceding karyokinesis, probably from the interzonal fibres, or, to quote his own words : "Accordingly in the spermatogenesis of the rat, the sphere seems to be divided into two parts, one made up of nothing but the residual spindle fibres (archoplasm), and another containing nothing but the centrosomes." Again : " The centrosomes divaricate and assume their usual position at the poles of the growing spindle, while the archoplasm remains an inactive structure in the body of the cell." At another place he says : " The archoplasm is reabsorbed into the cytoplasm." In this account of the archoplasm he is simply describing the Nebenkern as defined above ; its origin from the spindle fibres, its lack of connection with the centrosome, its final absorption into the cytoplasm and its inactivity in the cell, — all these are phenomena of the Nebenkern and not of an active cytoplasmic organ like the archoplasm of Ltimbricus. It is not improbable that the same mistake is made by others, and it may be found that the so-called " centrosome" at the tip of the spermatozoon in many forms is in reality a portion of the Nebenkern as Henking has described in Pyrrhocoris apterus.


The results of my investigation on the spermatogenesis of Lumbriciis may be summarized as follows :

1st. A multinucleate cell is formed in the testis. This represents a group of the earliest spermatic cells or spermatogonia. Each spermatogonium gives rise to several spermatozoa.

2d. The nuclei arrange themselves around the periphery of the multinucleate cell. Cytoplasmic cleavages then ensue between the nuclei, as in a centrolecithal ^g^. These cleavages deepen until the nuclei are separated from the central mass of the cytoplasm by mere filaments.

292 CALKINS. [Vol. XI.

3d. The residual mass of cytoplasm thus formed — the blastophore — is not nucleated and cannot be compared with a Sertoli cell in function, form, or mode of origin. It finally disappears. The blastophore furnishes perhaps the chief source of food supply for the parasites — Monocystis — which live in the seminal vesicles. A possible explanation of the function of the blastophore is that it represents superfluous nutritive cytoplasm, the vital protoplasm having gathered around the nuclei.

4th. There is a reducing division in the number of chromosomes and in the quantity of chromatin. In the spermatogonium thirty-two single chromosomes divide, giving thirtytwo to each spermatocyte of the first order. During the succeeding resting stage the chromatin substance, nucleus, and entire spermatosphere become enlarged. The chromatin emerges as a double skein. It divides by transverse division into thirty-two double chromosomes, and these unite two by two to form sixteen quadruple chromosomes. Reduction in number thus takes place without the aid of karyokinesis. In spermatocytes of the second order the nuclei contain sixteen double chromosomes. In the succeeding division these double chromosomes divide, and spermatids result, each with sixteen single chromosomes.

5th. Metamorphosis of the spermatid takes place independently of external conditions. The nucleus or head is formed from the entire chromatin. TJie arcJioplasm of the spermatid forms the middle-piece of the spermatozoon.

6th. The archoplasm mass persists throughout every change of the cell. In the preparatory stages of cell division it divides to form the poles of the karyokinetic spindle, as well as the spindle fibres themselves. All evidence points to the conclusion that, during the anaphase, the spindle fibres are partly withdrawn into the archoplasm. The interzonal fibres probably form the Ncbenkern^ which lies quiescent within the cell and finally disappears.

7th. The archoplasm mass is affected by the method of treatment. After Hermann's fluid it remains large and conspicuous, both at the spindle poles and in the resting cells.


After chromic or picric acid it is small, more dense, and stains more deeply.

8th. There is a distinct difference between a Nebenkern and the archoplasm. The latter gives rise to spindle fibres, retracts parts of them into its substance and finally becomes the middle-piece of the mature spermatozoon. The Nebe7iker7i, on the other hand, is derived from interzonal fibres and disappears after an apparently useless existence.

Columbia College, 1894-95.

294 CALKINS. [Vol. XI.


'93 Balbiani, E. G. Centrosome et " Dotterkern." Joitrn. de VAnat.

et de la Phys. xxix, 1893. '86 Benham, W. B. Studies on Earthworms. Quar. Joicrn. Micr. Sc.

xxvi, 1886. '87 Benda, Carl. Untersuchungen iiber den Bau des funktionirenden

Samenkanalchens einiger Saugethiere. Arch, f.viikr. Anat. 1887. '91 Benda, Carl. Verhandl. der Physiol. Gesellsch. zu BerHn. No. 4

und 5. 1891. '85 BiONDi, D. Die Entwickelung der Spermatozoiden. Arch, f. inikr.

Anat. 1885. '84 Blochmann. Ueber eine Metamorphose der Kerne in den Ovarial eiern und iiber den Beginn der Blastodermbildung bei den Ameisen.

Verh. Nat. Med. Ver. Heidelberg, iii, 1884. '80 Bloomfield, J. E. On the Development of the Spermatozoa.

I. Lumbricus. Quar. Jour n. Micr. Sc. xx, 1880. '81 Bloomfield, J. E. On the Development of the Spermatozoa. II,

Helix pom. and Rana. Quar. Journ. Micr. Sc. xxi, 1881. '89 BovERi, Th. Zellen - Studien. Jena Zeitschr. f. Naturwiss. n.s.

23-24, 1889. '93 Brauer, Aug. Zur Kenntniss der Spermatogenese von Ascaris

megalocephala. Arch.f. mikr. Anat. xlii, 1893. '94 Calkins, G. N. On the History of the Archoplasm Mass in the

Spermatogenesis of Lufnbrictis. Trans, of the New Yoj'k Acad.

of Sciences, xiii, 1894. '95 Calkins, G. N, Observations on the Yolk Nucleus of Lumbricus.

Trans, of the New York Acad, of Sciences. 1895. '88 Ebner, V. VON. Zur Spermatogenese bei den Saugethieren. Arch.

f. mikr. Anat. xxxi, 1888. '94 EiSMOND. Einige Beitrage zur Kenntniss der Attractionspharen und

der Centrosomen. Anat. Anzeiger. x, 1894. '93 Field, G. W. Echinoderm Spermatogenesis. Anat. Anzeiger. ix,

1893. '93 FiCK, R. Ueber die Reifung und Befruchtung des Axolotleies.

Zeitschr. f. wiss. Zool. Ivi, 1893. '88 Flemming, W. Weitere Beobachtungen iiber die Entwickelung der

Spermatosomen bei Salamandra maculosa. Arch, f 7nikr. Anat.

xxxi, 1888. '91 Fol, H. Le Quadrille des Centres, un dpisode nouveau dans I'his toire de la fdcondation. Arch, des Sciences Phys. et Nat. de

Geneve. Troisi&me pdr. xxv, 1891. '94 Foot, Katharine. Preliminary Note on the Maturation and Fertilization of ihc Y.gg oi Allolobophora fcetida. Journ. of M or ph.

ix. No. 3, 1894.


'87 FiJRST, Carl M. Ueber die Entwickelung der Samenkorperchen

bei den Beutelthieren. Arch. f. mikr. Anat. xxx, 1887. '91 GuiGNARD. Nouvelles Etudes sur la Fdcondation. Annal. des

Sc. Naturelles. Botanique. xiv, 1891. '93 Haecker, Val. Das Keimblaschen, seine Elemente und Lage veranderungen. I. Ueber die biologische Bedeutung des Keim blaschenstadiums und iiber die Bildung der Vierergruppen. Arch.

f. mikr. Anat. xli, 1893. '91 Henking, H. Untersuchungen iiber die ersten Entwickelungsvorgange

in den Eiern der Insekten. I und II. Zeitschr. f. wiss. Zool.

li, 1 891. '89 Hermann, F. Beitrage zur Histologie des Hodens. Arch./, mikr.

Anat. xxxiv, 1889. '91 Hermann, F. Beitrage zur Lehre von der Entstehung der karyo kinetischen Spindel. Arch. f. mikr. Anat. xxxvii, 1891. '93 ISHIKAWA. Studies on Reproductive Elements. \\. Noctiluca mili aris. Journ. of Imp. Univ. of Japan, vi, Dec, 1893. '91 ISHiKAWA. Studies on Reproductive Elements. Spermatogenesis,

Ovogenesis, and Fertilization in Diaptomus. Journ. of Coll. of Sc.

Imp. Univ. Japan, v, 1891. '87 Jensen, O. S. Untersuchungen iiber die Samenkorper der Sauge thiere, Vogel und Amphibien. Arch, f mikr. Anat. xxx, 1887. '88 KoROTNEFF. Beitrage zur Spermatologie. Arch, f mikr. Anat.

xxxi, 1888. '87a St. George, LaValette. Spermatologische Beitrage. Arch. f.

mikr. Anat. xxx, 1887. '87b St. George, LaValette. Zelltheilung und Samenbildungbei For f cilia auricularia. K'6llikers Festschrift. 1887. '78 St. George, LaValette. Ueber die Genese der Samenkorper.

Arch.f mikr. A fiat, xv, 1878. '67 St. George, LaValette. Ueber die Genese der Samenkorper. II.

Arch.f fuikr. Anat. iii, 1867. '95 Mathews, A. P. (with E. B. Wilson). Maturation, Fertilization and

Polarity in the Echinoderm Egg. Journ. of Morph. x, 1895. '93a Moore, J. E. S. On the Relationship and Role of the Archoplasm

during Mitosis in the Larval Salamander. Quar. Journ. Micr. Sc.

xxxiv, 1893. '93b Moore, J. E. S. Mammalian Spermatogenesis. Anat. Anzeiger.

1893. '94 Moore, J. E. S. Some Points in the Spermatogenesis of Mammals.

Intern. Mon. f. Anat. u. Physiol, xi. No. 3, 1894. '82 NussBAUM. Ueber den Bau und Thatigkeit der Driisen. Arch.f.

mikr. Anat. xxi, 1892. '85a Platner, Gustaf. Ueber die Spermatogenese bei den Pulmonaten.

Arch.f. mikr. Anat. xxv, 1885.

296 CALKINS. [Vol. XI.

'85b Platner, Gustaf. Ueber die Entstehung des Nebenkerns und seine

Beziehung zur Kerntheilung. Arch.f. mikr. Anat. xxv, 1885. '89 Platner, Gustaf. Samenbildung und Zelltheilung im Hoden der

Schmetterlinge. Arch.f. tnikr. Anat. xxxiii, 1889. '82 Renson, G. De la Spermatogen^se chez des Mammif^res. Arch.

de Biol, iii, 1882. '93 RucKERT. Die Chromatinreduktion bei der Reifung der Sexualzellen.

Ergebnisse, Merkel und Bonnet, iii, 1893. '94 RiJCKERT. Zur Eireifung bei Copepoden. Anat. u. Entwick., Anat.

Hefte, Merkel und Bonnet, xii, 1894. '82 Valaoritis. Die Genesis des Thiereies. Leipzig, 1882. '92 VoM Rath, O. Zur Kenntniss der Spermatogenese von Gryllotalpa

vulgaris. Arch.f. mikr. Anat. xl, 1892. '93 VoM Rath, O. Beitrage zur Kenntnis der Spermatogenese von Sala mandra maculosa. Zeit. f. wiss. Zool. Ivii, Bd. Heft I. '93 Van Bambeke. Contribution a I'histoire de la Constitution de

I'Qiuf. Bull, de VAcad. Roy. de Belgiqne. xxv, 1893. '87 Van Beneden and Neyt. Nouvelles recherches sur la fecondation

et la division mitotique chez VAscaridc megalocephala. Bull, de

VAcad. Roy. de Belgiqne. 1887. '85 Wiedersperg, G. Beitrage zur Entwickelungsgeschichte der Siimen korper. Arch.f. mikr. Anat. 1885. '92 Watase, S. Sertoli Cell. Americatt Naturalist. 1892. '93 Watase, S. Homology of the Centrosome. fourn. of M or ph. viii,

1893. '94 Wilcox, E. V. Spermatogenesis of Caloptenus femur rubrwn.

Anat. Anzeiger. x, 1894. '95 Wilcox, E. V. Spermatogenesis of Caloptenus femur-rubrum and

Cicada tibicen. Bull. Mus. of Comp. Zool. Harv. College. May,

1895. '95 Wilson, E, B. (with A. P. Mathews). Maturation, Fertilization, and

Polarity in the Echinoderm Egg. New Light on the " Quadrille of

the Centers." fourn. of Morph. x, 1895. '92 Weismann. Das Keimplasma. 1892. '87 Weismann and Ishikawa. Ueber die Bildung der Richtungskorper

bei thierischen Eiern. Ber. d. Naturforsch. Gesellsch. zu Freiburg, iii, 1887. '88 Weismann and Ishikawa. Weitere Untersuchungen zum Zahlen gesetz der Richtungskorper. Zool. fahrb.m, Morjfh. Abth. 1888.



All figures are magnified about 1,600 diameters, except Fig. i, Plate XIX. Figs. 1-6, 9-33, 35, 37, 38, 43, 43A, 45, and 47 are from preparations fixed with Hermann's fluid ; Figs. 7 and 8 with picric acid ; Figs. 34, 36, 39-42, 44, and 46 are from preparations fixed with chromic acid.

All figures save i, 7, 8, 44, and 46 are from teased preparations ; the others from sections.

In all figures the following letters have the same meaning :














Double spirem (Fig. 12).




Four-group (Vierergruppen).









Fig. r. A section through the testis, showing position of the latter in relation to the dissepiment. There is no distinct membrane about the testis as there is in the case of the ovary. The nuclei at the extremity of the testis are arranged in groups of varying size. Each group consists of spermatogonia, and each group is one multinucleate cell.

Fig. 2. A multinucleate cell in the testis, showing the general distribution of the nuclei throughout the cell. Optical section.

Fig. 3. Multinucleate cell in a seminal vesicle, showing a peripheral arrangement of the nuclei before the cytoplasmic cleavages have taken place. Optical section.

Fig. 4. A multinucleate cell in the seminal vesicles, showing the beginning of cytoplasmic cleavage very similar to the cleavage of a centrolecithal egg. The interior cytoplasm is finely granular. At the extremity of the nucleus is a large mass of archoplasm which is characteristic of the resting stage.

Fig. 5. An eccentric spermatosphere, showing distortion of the blastophore, increase of spermatocytes of the first order, anaphase of karyokinesis, reconstruction of archoplasm at spindle poles, and beginning of cytoplasmic cleavage with traces of interzonal fibres in some cells.

Fig. 6. A spermatosphere in a late stage, showing elongation of spermatids with beginning of tail formation at the free end of the cell. The archoplasm is at the distal extremity of the homogeneous and compact nucleus.

Fig. 7. A spermatogonium in the testis from a multinucleate cell, showing the nuclear plate with 32 apparently single chromosomes.

Fig. 8. A spermatogonium from the testis in full karyokinesis, showing 24 of the 32 chromosomes and the archoplasm reduced by the picric acid to a mere dot at the poles.

Journal of Morplioloyij \ol.XI.




«'"-l' V.


' — '•*<

^•v^5) ,#:f^*.

.- -, ^VvJ&

^B ■' c

.> .- r

'■ -,^

■^^'.l:• -, ■••

•,■■-- ,/t::^% \.

y:i^' \











■^■■'m I



V;i..\ /

^^ ;V






tf ;if <; .I'Z

,.,: , ,., r.'ier ^ Muter, Frank furnM.



Fig. 9. Spermatogonium in the anaphase of karyokinesis ; chromosomes of the daughter nuclei single and 32 in number (only 19 and 21 are shown in the figure).

Fig. 10. Resting nucleus of spermatocyte of the first order, showing nucleolus and the reticulated chromatin.

Fig. n. Spermatocyte with the chromatin in a lump at one part of the cell; first phase after the resting period.

Fig. 12. Spermatocyte of the first order. The cells of spermatosphere show different stages in the formation of the Knauel or spirem. At D the spirem is double. The archoplasm is in different stages of division.

Fig. 13. Spermatocyte of the first order, showing double spirem in at most one or two pieces.

Fig. 14. Spermatocyte of the first order, showing" the spirem breaking into parts.

Fig. 1 5. Spermatocyte of the first order, showing formation of double chromosomes and the presence of Nebenkern.

Fig. 16. Spermatocyte of the first order, showing 32 double chromosomes.

Figs. 17, 18, 19, 20, and 21, spermatocytes of the first order, showing different stages in the formation of the four-groups, or Vierergruppen.

Figs. 22, 23, 24, 25, 26, and 27. Different stages in karyokinetic division of the spermatocyte of the first order.

Figs. 28, 29, and 30. Different aspects of the spindle of spermatocytes of the second order. Fig. 30 represents the nuclear plate of a spermatocyte of the second order, showing 16 double chromosomes.

Figs. 31, 32, and 33. Spermatids, showing different stages in the concentration of the chromatin.

Journal of Morpliology lot. XL











f^ F







» J »*


























6J^C- i^ci;

litk.Amt. V. IVenier i MKUr, Frank fkrt 'i-M



Fig. 34. Spermatid, showing reticulated nucleus and minute archoplasm mass due to the killing agent.

Fig. 35. Spermatid highly magnified, showing homogeneous nucleus becoming drawn out, tail forming from cytoplasm, and with a clear space around the chromatin. Archoplasm clear and distinct.

Fig. 36. Spermatids on blastophore, chromatin not homogeneous, archoplasm minute (like a centrosome), cytoplasm more or less distorted, all due to the killing agent.

Fig. 11. Spermatid much elongated and detached from the blastophore, showing homogeneous nucleus with a comparatively large vesicle at the extremity which contains the archoplasm mass.

Fig. 38. Spermatid. The vesicle has here disappeared ; the archoplasm now connects the tail and nucleus and is the middle-piece.

Figs. 39, 40, 41, and 42. Spermatocytes of the first order, showing different stages in the history of the archoplasm.

Fig. 43. Spermatosphere in the spermatocyte stage, showing archoplasm masses extending up the spindle fibres. The fibres come from the archoplasm.

Fig. 43A. Spermatid in the anaphase of karyokinesis, showing withdrawal of the spindle fibres into the archoplasm mass.

Figs. 44 and 46. Artefacts, showing the protrusion of nucleoli.

Fig. 45. Spermatocytes of the first order, showing central spindle between the divided archoplasm masses. Nebenkern is also present.

Fig. 47. Late spermatozoon, showing tail, middle-piece, and head, also the spur (S) formed from the point of cytoplasm by which the spermatid had been attached to the blastophore.

Joiinial of Morphology Vol. XI.




A r



58. ■M
















oo .






'w^ \.i>"~






{^Investigation aided by a grant frovi the Elizabeth Thompson Fund of the A.A.A.S.)

No group of mammals offers greater difficulties in the way of arranging all its members in their order of relationship to one another than the Artiodactyla. These difficulties arise largely from the great number of extinct genera which are very imperfectly known from fragmentary remains, and as a consequence of this it is frequently impossible to determine the taxonomic values of resemblances and differences. The first step toward bringing order out of this chaos must consist in completing our knowledge of the structure of the many imperfectly known genera, and wherever it is possible, in carefully tracing out all the cases of parallel or converging development in one or other organ, where it can be shown to have taken place. In this way we may hope eventually to bring together such a body of facts as will clear up these complicated relationships. The Princeton Museum has recently obtained, thanks to the energy and skill of Mr. Hatcher, very extensive and wonderfully preserved collections of Tertiary mammals from the Western States, comprising many nearly complete skeletons of artiodactyls hitherto known only from scattered fragments. As soon as this material can be prepared, it is my purpose to render it available for comparison by careful description and full illustration. The perhaps tedious minuteness of such descriptions cannot well be avoided, if the object in view is to be effectively served.

The extraordinary genus Protoceras (Marsh) is one of those which will best repay exact and minute study, for, although its systematic position is far from clear, for reasons which will appear later, yet it throws much welcome light upon the mode of development among the artiodactyls, and shows how it has

304 SCOTT. [Vol. XI.

happened that the different characters of structure are so variously and puzzlingly combined among the several groups of the order.

Protoceras (Marsh).

The type specimen of the genus is an imperfect skull described by Marsh under the name P. celer (No. 5, p. 81), which is characterized by the presence of a small pair of protuberances, resembling horns, from the parietal bones. Subsequently Osborn and Wortman (No. 8) showed that the type specimen belonged to a female and described the male skull, which is one of the most curious fossils ever found and which displays very remarkable secondary sexual characters. They also described and figured the fore and hind feet, which differ in the most striking way from what the structure of the skull would lead us to expect. In the explorations of the Princeton Expedition of 1893 Mr. Hatcher collected a large number of female skulls and one nearly complete skeleton, the material which forms the subject of the present paper. During the season of 1894 Mr. Hatcher has added considerably to this material and has visited the spot where the type of the genus was collected, in order to establish its exact stratigraphical position, which was previously unknown.

I. The Dentition,

The dental formula is i^, ^ i>/|j ^^^f> which differs from that believed to characterize Gelocns in an additional upper premolar, although it is not certain that Pj_ may not have been present in the latter genus also,

A. Upper Jaw (PI. XXI, Figs, i and 2). — The incisors have entirely disappeared, leaving no traces of alveoli in the premaxillaries. In the female the canine is rudimentary, but in the male it forms a long curved tusk, comparable to that of the tragulines, though rather shorter and of entirely different shape. While in the latter it forms a compressed scimetarlike blade, in Protoceras it is trihedral like that of the oreodonts, but relatively longer, more compressed and slender, and


is quite strongly everted laterally. As in the oreodonts, the canine is abraded upon the posterior face, which renders it probable that in the male the first lower premolar had taken on the form and function of a canine. The first premolar is much the smallest of the series and has a simple compressed and trenchant crown, implanted by two quite widely diverging fangs, and consisting only of the protocone without additions. The tooth stands by itself, being isolated by considerable diastemata both from the canine and from Pj^. The second premolar is remarkably elongate in the antero-posterior direction, resembling in this respect the corresponding tooth of XipJiodon and in a less degree that of Leptonieryx. The crown is very low, compressed and trenchant ; obscurely marked anterior and posterior basal cusps are separated by shallow grooves from the protocone. There is a strong internal cingulum running the whole length of the crown, thickened and elevated at the median point to form an incipient deuterocone, which is supported on a third fang. In some specimens the deuterocone can hardly be said to exist and the cingulum is feebly marked, and then the crown is carried on two fangs only. Seen from the outer side this tooth much resembles /£ in Leptovieryx, except for its greater elongation, but in the latter the deuterocone is much larger and the cingulum absent. The third premolar is very much like the second, but with some modifications ; thus the antero-posterior length is somewhat less, while the transverse width is somewhat greater, an increase which is chiefly due to the greater thickness of the protocone, though the deuterocone is also better developed. This tooth differs from A? in Leptonieryx in the much smaller deuterocone and in the presence of a complete internal cingulum. The fourth premolar is of the usual ruminant pattern, consisting of a single pair of crescents. A fairly developed cingulum is present in the inner crescent, which is very faint on the anterior side, but distinctly marked on the posterior. This cingulum is of course not homologous with that on Pj^ and P3_. A comparison at once shows that the cingulum of P3_ is represented by the horns of the inner crescent on p_4 and that the inner cingulum of the latter is something superadded.

3o6 SCOTT. [Vol. XI.

The upper premolars as a whole are very much like those of Leptomeiyx, but more elongate and of somewhat simpler construction. In Hypertragulus they are still simpler and more trenchant, as they are in existing tragulines.

The molars are of the extreme brachyodont type, short antero-posteriorly and broad transversely, so as to present a nearly square outline ; ^ is slightly the largest of the series. The external buttresses (proto- and mesostyles) and the median ribs on the outer faces are very prominent and sharply defined ; the internal pillar varies considerably in the different specimens, — in some being quite absent and in others strongly developed ; the inner cingulum also varies much in strength and would appear to be more prominently marked in the males than in the females. The valleys are narrow and shallow and rapidly wear down to mere lines. The shape and proportions of these molars are closely like those of Leptomeiyx, the principal difference between the two genera being that in Protoceras the inner crescents are rather more complete. Doixatherium has molars of which the outer crescents are extremely like those of Protoceras, but the inner ones are thicker and more sloping. In Hypertragidiis the molars are like those of Leptomeryx.

B. Lozver Jaw (PI. XXI, Figs, i and 4). — None of the specimens in the Princeton collection have the lower incisors or canines in position. Osborn and Wortman give the following description of them. " The inferior incisors present delicate spatulate crowns ; the median [and] second incisors are slightly larger than the lateral incisor, which is very delicate. The canine has precisely the same delicate structure as the lateral incisor " (No. 8, p. 359). In the female pi resembles the corresponding tooth in the upper jaw, but is rather smaller and is implanted by a single fang. I have seen no example of this tooth in the male, but as has already been mentioned, the abrasion on Xhe. posterior i2iCQ of the upper canine renders it probable that in that sex pi was caniniform. Even in the female it is considerably in advance of Pj_ and nearly far enough forward to abrade the upper canine, being separated by a diastema from the lower canine and by a longer one from p^. In Gelocus ^ is very small and simple and stands close to p2.


The second premolar is much more elongate antero-posteriorly than the first and is inserted by two widely separated fangs ; the crown is low and compressed, with trenchant edges, and is of very simple construction. The para- and metaconids are but obscurely developed, and a faint ridge on the inner side of the protoconid encloses an incipient valley. The third premolar, like Ai, is the longest (fore and aft) of the series, though the difference is less marked than in the upper jaw. In construction it resembles 'p~2, but all the elements are more distinctly differentiated and the posterior valley is deeper. The fourth premolar is shorter and wider than the third and has its component cusps more clearly demarcated ; the deuteroconid is well shown and the ridge which passes backward from that cusp extends to and fuses with the metaconid, so that the posterior valley forms a "lake."

The inferior premolars of Leptomeryx are in general very similar to those of Protoceras, but p~i, which stands isolated by diastemata, is much more reduced, and on the others the anterior and posterior basal cusps (para- and metaconids) are much more distinct and form sharp elevated points. In Hypertragulus pi is wanting, p~2 has a considerable diastema both in front of and behind it, and /j and p~^ are smaller and simpler than in either of the genera mentioned.

The molars are very low-crowned and have narrow, shallow valleys, which are worn nearly to the bottom at a comparatively early stage of attrition. The crowns are quite broad in proportion to their length, which increases regularly from TfTi to JJTj. The internal crescents are thinner and less distinctly conical than in Geloais. In the smaller and more abundant specimens the external pillar is but feebly developed, especially upon l^i, in the largest-sized jaws (perhaps old males) the pillar is well shown ; in such cases it is largest on wj, smallest on liTi. The talon of wj? is large and consists of two elements, which are more or less distinctly separated from each other. In Leptomeryx the lower molars differ from those described chiefly in their relatively less width and greater height, in the absence of the external pillar, and in the complete separation of the two elements which compose the talon of 'Wj.

3o8 SCOTT. [Vol. XI.

C. The Milk Dentition (PI. XXI, Fig. 5) of the lower jaw is not represented in the collection, and that of the upper jaw only partially so. The second mi,lk molar is very elongate and compressed and consists of three elements, the protocone in the middle, with a basal cusp in front of and behind it, all of which have acute apices and trenchant edges ; the internal cingulum is strongly developed and forms a prominent ridge. This tooth, aside from the cingulum, is strictly in the " triconodont stage," though, of course, no phylogenetic significance can be attributed to this fact. The third milk molar (<£) is very peculiar and not quite like the corresponding tooth of any artiodactyl with which I have compared it, though most resembling that of Gelociis, Leptomeryx, and the tragulines, and thus is quite distinct from the type of d3_ in the existing Pecora. As in the former and in very many ancient genera, such as Dichobime, Dichodon, CcBuotheriiim, etc., the crown bears three external cusps, the protocone and tritocone and in front of the former an anterior basal cusp, for which I have proposed no name, because of the comparative rarity of its occurrence. The tritocone is shaped precisely like the external crescent of p4 , while the protocone is of compressed conical form. The postero-internal cusp, or tetartocone, is likewise crescentic, and thus the hinder half of the tooth is like the entire crown of p4. So far, this description would apply equally well to Leptomeryx, the difference between the two genera consisting in the great prominence of the inner cingulum on the anterior half of the crown in Protoceras, which is thickened and elevated at two points to form two incipient cusps ; of these the posterior, opposite the protocone, is of course the deuterocone, while the anterior one has not been named.

In the Pecora ([3^ has the form of a permanent molar. I have elsewhere shown (No. 10, p. 370) that Rutimeyer has probably attributed too great taxonomic value to this fact, since in the same family both types of f^ may occur, as, for example, in Oreodon and Mejychyus. Yet, nevertheless, the feature is important and it is not without significance that while Protoceras resembles the Pecora in many details of skull structure, the dentition, and especially the milk teeth, arc those of the more ancient genera.


The fourth milk molar is molariform and differs from the permanent ones only in size.

II. The Skull (PI. XXI, Figs. 1-3).

No genus hitherto known from the White River formation has such a modernized type of skull as Protoceras, which, indeed, is in some respects in advance of the existing hornless deer, such as MoscJms and Hydropotes, and in one or two points exceeds all the CervidcB, approaching to the highest type of artiodactyl structure, the Cavicornia. At the same time, remnants of the primitive condition are by no means wanting, and the resulting combination is a very curious one. Another characteristic feature of the skull in Protoceras is the extreme degree to which the sexual differences are carried. Such an amount of sexual difference is altogether unparalleled among the ancient artiodactyls and is not attained even among the modern Ccrvidce. The modernization of the skull, which has been referred to, consists in the following characters : (i) the shortening and rounding of the cranium ; (2) the backward shifting of the orbit, which is removed entirely behind the line of the molar teeth ; (3) the great elongation of the facial region, due not only to the shifting of the orbit, but also to the lengthening of the muzzle ; (4) the bending downward of the face upon the cranio-facial axis, which does not occur in the CervidcB.

The material at command displays three well-marked types of skull-structure, two of which are undoubtedly due to sexual distinctions, while the significance of the third is not yet entirely clear. It will be most convenient to begin the description with an account of the unmistakably female skulls, since these are the less extremely specialized type, and therefore best fitted to display the fundamental characters of structure. Of these females, there are five finely preserved skulls in the Princeton collection, with fragments of several others. The females are much more abundant than the males, as would be expected from the analogy of recent ruminants. Among these skulls there is a considerable degree of variation, and perhaps more than one species is represented by the

3IO SCOTT. [Vol. XI.

specimens, but, as this is not clearly shown, it will be better to treat the variations as individual.

Aside from certain peculiar specializations, the general character of the skull has already been pointed out in the list of modernizations given above. As a whole, the skull is very long and narrow, tapering rapidly toward the anterior end, where the muzzle becomes extremely slender. The face is constricted in front of P4, and again in front of p£, so that the base of the skull has a remarkably llama-like appearance. The upper contour rises at the forehead, the cranium being considarably elevated above the level of the face, though less than in the antlered deer. The occiput is of antique type, high, narrow and with its upper portion deeply concave and projecting backward, very different from the low, broad, and convex occiput of the tragulines. However, as in that group, the upper margin of the foramen magnum forms the hindermost part of the skull. The lambdoidal crest is prominent, the sagittal much longer and higher than in any existing selenodont, except the camels, and the temporal ridges are likewise unusually prominent.

The basioccipital is quite broad and massive, in some specimens faintly keeled, in others with a shallow groove along the median line ; the very small tympanic bullae do not encroach upon it at all. The body of the bone is relatively narrower and deeper in the vertical direction than in the antelope or deer skull, or even than in Moschiis. The condyles are large, especially in the vertical dimension, and are quite widely separated below, though less so than in the deer. The articular surface is continued forward upon the two tubercles of the basioccipital, as in the deer and in some antelopes. Gn the outer side of each condyle there is a curious emargination of the articular surface, which invades the process along the line where the dorsal and ventral surfaces meet. This emargination occurs, but in a much less marked degree, in Moschus, Cariacns and other recent ruminants. The exoccipitals are low and narrow, as compared with their relatively great breadth in the recent Pecora ; in the median line they are quite strongly convex, bulging out here to form a fossa for the


vermis of the cerebellum ; laterally they become concave, so as to enclose a deep fossa on each side. The paroccipital processes are long, slender and laterally compressed ; in shape and relative position, they are quite like those of MoscJms. The inferior surface of the exoccipital, between the paroccipital and the condyle, is not wide, but is unusually long in the fore and aft direction ; this space is not bounded in front by a ridge running from the paroccipital to the basioccipital, as is the case in many deer. The foramen magnum is rather small and of subcircular shape ; its dorsal border is notched in the middle line and on each side of the notch is a more or less prominent and thickened process. The supraoccipital is high, but narrow and very thick, especially in the upper portion, where the diploetic structure is well developed. The posterior surface is concave and small "wings" are formed on the sides. The parietals do not take part in the formation of these wings, because the supraoccipital is reflected over upon the top of the skull and forms an appreciable part of the cranial roof, not only in the middle line, as in MoscJms, but also upon the sides, as in the tragulines. The summit of the lambdoidal crest is formed entirely by the supraoccipital. None of the existing Pecora has such a primitive type of occiput as Protoceras. In all of them the occiput is much lower and broader ; there are no such marked median convexity and deep lateral fossae, and the crest is almost or entirely obliterated. Even the tragulines have this region shaped more like that of the Pecora than has Protoceras. On the other hand, the shape found in the latter occurs also in many ancient artiodactyls, such as the oreodonts, but in this family the occipital wings are much more prominent and the parietals take part in the formation of them. The Deep River genus, Blastovieryx, which is an undoubted member of the Pecora, has an occiput strangely like that of the oreodonts.

The cranium, which is remarkably well-rounded and capacious for a White River animal, though not very large, judged by the modern standard, is roofed in principally by the large parietals, which cover nearly the whole cerebral fossa and extend farther forward than in the existing Pecora, though the parietal

2^12 SCOTT. [Vol. XI.

zone is somewhat shortened as compared with that of the tragulines, and very much so when contrasted with the long parietals of such genera as Oreodon or Ancodus. For most of their length the parietals unite to form a thick and prominent sagittal crest, which is cancellous internally and encloses a small sinus. Anteriorly, near the frontal suture, the crest bifurcates into two temporal ridges, which are recurved and overhanging, thus enclosing a deep grove in which a goosequill may be concealed. These temporal ridges are confined to the parietals, and become thickened and more rugose at the points where in the male skull the large protuberances arise. These thickened areas differ somewhat in shape and in prominence in the various specimens, but in none of the skulls before me do they rise into protuberances at all comparable even with those of the specimen figured by Marsh (No. 6, PL XXI, Fig. i). In most of the female skulls the parietals extend quite far down upon the sides of the cranium, though there is variation in this respect, and in front of the squamosals send down narrow processes to meet the alisphenoids. The latter are antelopine rather than cervine in shape, which is no doubt due to the great amount of backward displacement which the orbit has undergone. The ascending process of the alisphenoid is narrow, but its lower portion is reflected backward, and internally to the glenoid cavity extends almost to the tympanic. The pterygoid process is also narrow and has decidedly less vertical height than in recent Pecora.

The basisphenoid forms a rather slender and almost cylindrical rod, which is distinctly narrower than in the deer. The tympanic is much like that of MoscJms ; the exceedingly small bulla is but seldom well preserved, so that it is not surprising that Marsh should have concluded that " there were apparently no auditory bullae" (No. 5, p. 82). The bulla is produced anteriorly into a long, slender spine, and on its postero-external side is a deep pit for the tympano-hyal. A long auditory meatus occupies the space between the postglenoid and posttympanic processes of the squamosal, but does not form a complete tube, the upper part being covered in by the squamosal. Sir Victor Brooke's account of these structures in the musk deer


may be quoted here with advantage. " In MoscJms the auditory bulla is remarkably small, the petrous portion of the periotic being visible, when viewing the base of the skull, for nearly the entire length of the bulla from before backwards. The tympanic is considerably prolonged to form the inferior floor of the external auditory meatus " (No. i, pp. 522-3). In Protoceras the approximation of the tympanic and basioccipital is too close to allow the periotic to be seen, except in one specimen where the exposure is clearly due to a displacement of the surrounding parts. As in Leptomeryx and in the Pecora generally, the tympanic is hollow and not filled with cancellous tissue, but the bulla of Leptomeryx, though decidedly smaller than in Tragnlus, is relatively much larger than in Protoceras. The mastoid portion of the periotic forms a narrow and more or less rugose strip between the exoccipital and the squamosal, and is continued upon the anterior face of the paroccipital process for some distance below the posttympanic.

The squamosal forms a varying part of the side wall of the cranium, the ascending plate rising higher in some specimens than in others. The parietal suture is regularly arched upward and backward. As in the deer, the bone forms a truncated process at the anterior end of this suture, where the cranial wall turns inward to form the front boundary of the cerebral fossa ; apparently, however, the alisphenoid takes less share in the formation of this process than is the case in Cervics. The root of the zygomatic process is thicker and less concave on its upper side than in the recent Pecora and the process itself is heavier than in those animals, in which it is remarkably weak ; as to length, it is longer than in the cavicorns and shorter than in the deer, the orbit occupying a position intermediate between the rather anterior place it holds in the latter, especially in the hornless forms, and its very posterior position in the former. The glenoid cavity is typically ruminant in character ; in front it is broad and slightly convex, passing behind into a concavity. The postglenoid process is longer and considerably heavier than in most recent ruminants, though very much less so than in such types as the oreodonts or Ancodiis. The process does not, as in the deer, bound the

314 SCOTT. [Vol. XI.

whole width of the glenoid cavity, nor, as in the cavicorns, is it confined to the postero-internal angle, but is placed somewhat internal to the median line. The posttympanic process is very short and slender and is closely applied to the mastoid portion of the periotic. In all the recent Pecora there is a large opening between the auditory meatus and the inferior margin of the squamosal for the passage of nerves and blood-vessels ; in Leptomeryx, Hypertragidiis, and the recent tragulines no trace of this opening is to be found. Protoceras agrees with the latter in this respect, the auditory meatus being in contact with the squamosal in front, behind, and to some extent, above.

The jugal is heavier than is usual in existing ruminants ; its posterior branch extends backward beneath the zygomatic process to the outer edge of the glenoid cavity. Beneath the orbit the jugal projects outward into quite a broad horizontal shelf, corresponding to the prominence of the supraorbital border. The postorbital process is short, and deeply notched to receive the anterior end of the zygomatic process. The orbit, which is completely encircled by bone, is bounded behind more by the postorbital process of the frontal than by that of the jugal. The vertical portion of the jugal, which articulates with the maxillary, is rather small and is not much expanded upon the face ; this is due to the rather low position held by the orbit, which is not so much elevated as in the typical deer. This has an effect also upon the conformation of the posterior nares. The lachrymal, which is quite large, is not depressed to form a pit or fossa, as it is in so many of the recent deer and antelopes, as well as in the oreodonts. It has a limited suture with the nasal, instead of being separated from it by a vacuity, such as occurs in Leptomeryx, Hypertragidtis, and most of the existing Pecora. Short as are the nasals, they are long enough to show indications of such a fontanelle, had it existed. No great taxonomic value can be attached to this character, as within the limits of the same family we may find some genera with and others without it, as, for example, in the oreodonts, Merychyiis and LeptaiicJienia, with its allies, have the vacuity, the other genera are without it.


The frontals are large, though less extended antero-posteriorly than in the modern Pecora and forming less of the roof of the cerebral fossa ; they reach from a little behind the orbits to considerably in front of them, but the temporal ridges do not encroach upon them at all. They are transversely concave and at their outer edges are raised and thickened to form the rugose upper borders of the very prominent orbits. Though far less so than in the males, these projections are much more striking than in the recent Pecora. In the median line of the frontals not far from the parietal suture is a small, rounded and more or less rugose eminence, which varies in size in the different individuals, though it is not prominent in any of the females which I have seen. Judging from Marsh's figures and description, it would appear to be better developed in the type specimen. As in the oreodonts and primitive artiodactyls generally, the supraorbital foramina are placed near to the median line and the vascular grooves which run forward from them are very distinctly marked. There is a small frontal sinus.

The nasals are very remarkable ; they are broad from side to side, but exceedingly short. The median portion is elevated, transversely convex, and projects slightly beyond the maxillary suture, the free portion rapidly tapering to a point. The lateral portion is more depressed and flattened and unites suturally with the maxillary, frontal and lachrymal. Owing to their extreme shortness, they are, of course, very far removed from any contact with the premaxillaries. Though the reduction of the nasals is less extreme than in the saiga antelope, this part of the skull has a striking resemblance to that animal, and as in it, the nasal chamber has no osseous covering for nearly the whole of its length. It can hardly be doubted that Protoceras must have possessed a proboscidiform muzzle much like that of the saiga, and as in that genus, the turbinal bones were doubtless greatly shortened. In the moose {Alces) the nasals and turbinals are considerably reduced in length, though to a much less extreme degree.

The edentulous premaxillaries are very much like those of the modern ruminants, except that they are of unusually small

3i6 SCOTT. [Vol. XL

size ; they project but little in advance of the canines and so do not add much to the length of the muzzle. The alveolar portion forms a thin, depressed, and delicate plate, with regularly rounded free margin, which in shape resembles that of Cariaciis. The spines, on the other hand, are relatively quite broad and very thin, and hence the incisive foramina form narrow crescentic slits, the outer wall of which is formed for nearly half their length by the maxillaries ; these foramina are thus much less widely opened than in most ruminants, which is due to the narrowness of the spines in the latter. The ascending ramus of the premaxillary is short and slender and forms an obtuse angle with the alveolar portion ; it is separated by nearly half the length of the skull from the shortened nasal. In the saiga the premaxillaries are much heavier and wider proportionately than in Protoceras and the ascending portion makes a still more open angle with the alveolar portion ; in Hydropotes the ascending ramus is wider.

The maxillaries have considerable resemblance, when seen in profile, to those of the saiga antelope, but owing to the extremely brachyodont dentition, the alveolar portion is much lower, and as the orbit is considerably less elevated above the level of the palate, the descent of the upper or free margin of the maxillaries toward the front is much more gradual. Another difference between the two genera consists in the more complete reduction of the nasals in the modern type and the larger size of the lachrymals, owing to which the latter form the upper margin of the nasal chamber for some distance. In Protoceras the nasals articulate with the maxillaries and thus cut off the lachrymals from the border of the nasal chamber. The vertical plate of the maxillary above the sinus or antrum is very thin and delicate, but has a round, seam-like free border, which is a little thicker than the rest of the plate. As Osborn and Wortman conjectured, there are no such great and massive protuberances on the maxillaries as occur in the male, but they are faintly indicated by a slight upward arching of the border ; posterior to this the margin is notched and from the notch descends a shallow groove, doubtless marking the course of a blood-vessel. The masseteric ridge is very prominent, though


less so than in the male, and extends far forward, nearly to Af, terminating in a short spine, which again is very much less developed than in the male. Above the masseteric ridge the ascending plate of the maxillary is sharply inflected to form a horizontal surface and toward the median line again assumes a vertical direction, thus giving to the bone a step-like section and greatly narrowing the dorsal part of the nasal chamber. The horizontal surface thus formed ends anteriorly in a deep fossa, to which Marsh has already called attention, but in his specimen the fossa appears to be somewhat differently shaped and more distinctly demarcated. In advance oi P 2 the maxillary is very low, its upper border curving gently downward to the premaxillary suture, and its outer surface is flat, there being no large canine alveolus to cause a swelling. This region of the jaw is very long and it bears the chief share in the characteristic elongation of the muzzle. In one of the specimens, which may possibly represent a distinct species, the portion of the maxillary which is in advance of p£ is shorter than in the usual type of female skull, and the arching of the upper border which represents the great maxillary protuberance of the male, is thicker and has a more rugose margin. From this process the descent of the border to the premaxillary suture is more abrupt. The lower border of the maxillary fossa, which is a continuation forward of the masseteric ridge, though not rugose like the latter, is more prominent and extends farther upward upon the maxillary process. In these respects this skull approximates the characters of the male more closely than do the long-muzzled females which are more commonly found. The teeth of this specimen are remarkable for the relatively feeble development of the internal cingulum upon the upper premolars and its unusual prominence on the molars.

The palatine processes of the maxillaries form nearly the whole of the bony palate, the premaxillaries and palatines taking but little share in it. The palate is widest between the first molars of the opposite sides and gradually narrows in each direction from that point, while in front of p£ it is constricted and becomes very narrow ; it is unusually flat both transversely and antero-posteriorly. The faintly marked rugose ridges

3l8 SCOTT. [Vol. XL

which indicate the limits of the soft palate in front of the premolars, are in many existing ruminants placed quite close to the median line, so that part of the ventral surface of the maxillary is not covered by the soft palate. In Protoceras, on the other hand, these ridges run along the angle formed by the meeting of the horizontal and vertical surfaces of the bones, which thus do not curve into each other so gently as in the modern forms mentioned. The posterior palatine foramina occupy a very unusual position. Ordinarily in ruminants, as in other mammals, they are situated in or near the maxillopalatine suture, but in Protoceras they are placed just internal to pj_ and from them deeply marked vascular grooves run forward to the incisive foramina. These, perhaps, represent the foramina which in Oreodon and Ancodus occur opposite Af.

The palatines are rather small bones and hardly do more than form a border around the posterior nares ; they are separated from the molars by a considerable strip of the maxillaries, whereas in the Pecora the palatines are almost in contact with the molar fangs. The posterior nares are much more ruminant than suilline in character and yet are peculiar in many ways. The canal is very elongate antero-posteriorly, but is very narrow and of considerably less vertical height than is usual in the Pecora, a feature which is correlated with the relatively small elevation of the cranium above the level of the face. The anterior border of the opening is placed between the second pair of molars and has a short thickened median spine. The palatine notches are narrow and shallow, as compared with those of the deer ; they are not formed, as in the Pecora generally, by a constriction of the palatines, but lie between the latter and the maxillaries. The pterygoids are thin, slender plates of bone of no great height, which terminate in slightly thickened and everted hamular processes. There are no pterygoid fossae and no such separation between the distal ends of the pterygoids and alisphenoids as occurs in the Tylopoda.

The mandible is altogether pecoran in character, resembling that of Lcptomeryx but little and that of the oreodonts not at all. The horizonal ramus is very long and quite slender, in accordance with the great proportional length of the molar


premolar series and their very brachyodont condition. In front oip~2 the thin edentulous upper border descends quite abruptly, rising again slightly to form the alveolus of pi. The incisive alveolar border is slightly widened and somewhat spatulate in form, though retaining a greater vertical depth than is usual in the Pecora ; the widening is very limited, as would be inferred from the extreme slenderness of the upper jaw in this region. The symphysis is quite long and rather oblique, and in some aged individuals the two rami of the jaw appear to be coossified. Two rather large mental foramina occur in the symphyseal region, the first underneath and the second slightly behind p~i. The horizontal ramus has the undulating lower margin and anterior taper found in existing ruminants, but is rather stouter than is common in that group and is quite unlike the straight, slender ramus of Leptomeryx. The ascending ramus is rather wider than in the most of the Pecora and does not rise so high as in the cavicorns, which is due to the fact that the glenoid cavity and base of the cranium are not so much elevated above the level of the molars as in those forms. The angle is thin and entirely inconspicuous, not projecting behind the line of the condyle, being shaped very much as in MoscJms, which is a striking difference from Leptomeryx. In the latter the angle projects very far back of the condyle, in a way which recalls that of the ancestral Tylopoda. The masseteric fossa is less profoundly marked than in Leptomeryx and extends farther downward and forward ; it is somewhat more distinctly imprinted upon the jaw than in the musk-deer, but has much the same shape and extension. The coronoid is very low and tapers abruptly to a blunt point. This is the only character in which the mandible of Protoceras resembles that of the oreodonts ; the sigmoid notch is much narrower than in Leptomeryx. The condyle is like that of the Pecora, but in correlation with the better development of the postglenoid process, the articular surface is more reflected upon the posterior face of the condyle. While in the Pecora the latter is slightly concave transversely, convex antero-posteriorly and thus saddle-shaped, in Protoceras, on the other hand, it is plane transversely and more strongly convex in the fore and aft direction.

320 SCOTT. [Vol. XI.

The formnina of the skull are antelopine rather than cervine in character, which is doubtless due to the backward shifting of the orbit, this displacement having proceeded farther than in the deer, especially the hornless genera. The infraorbital foramen is single and occupies rather a posterior position ; it lies above the interval between A? and p_4 (in the deer it is above or in advance of P_£) and is roofed by the tubercle in which the masseteric ridge terminates. The optic foramen is lower than in most antelopes, in correspondence with the less elevated position of the orbit. The foramen lacerum anterius is rather small and rounded, while in the deer it is very large and of irregular shape, especially so in MoscJuls, where it forms a great fissure. As in the Pecora generally {and indeed most artiodactyls) the foramen rotundum is not present, though its position is in some specimens marked by a pair of minute openings. The foramen ovale is quite large and placed internally to the glenoid cavity ; it varies in shape, being in some individuals nearly round and in others narrow and elongate. In spite of the very small size of the auditory bullae, the foramen lacerum medium and posterius are mere slits, much narrower than in Cariaais, and the bulla is not channelled by the carotid canal. A postglenoid foramen is present, though of small size. The condylar foramen occupies a position where it is concealed by the prominence of the inferior portion of the condyle. The squamosal is perforated by two large vascular openings above the root of the zygomatic process. The remarkable position of the posterior palatine foramina has already been described ; it only remains to mention an irregularity in the two sides which sometimes occurs, — thus, in a male skull the left foramen is fully a quarter of an inch behind the right.

The second type of skull is that of the male, the knowledge of which we owe to Osborn and Wortman. Their beautifully preserved specimen is still by far the best that has been discovered, and is one of the most remarkable and curious of mammalian fossils. The striking resemblance of this skull to a miniature Uintatheriimi has been commented on by the authors mentioned, and is an excellent example of a superficial parallelism between two forms which are as widely separated


as two ungulates well can be, since they have no common ancestor later than the most ancient Condylarthra.

The most obvious difference in skull structure between the two sexes lies in the great development of osseous protuberances from various parts of the head in the male, all of which are, however, faintly indicated in the female. The latter is far less bizarre in appearance, and aside from the shortened nasals (a character which is found still more strongly marked in the saiga antelope and to a less degree in A Ices), does not depart in any radical way from the modern hornless Pecora. Indeed, most of its deviations from the skull-structure of those animals is in the direction of the higher members of the same group. Besides the protuberances, the various ridges and processes for muscular attachments are naturally much stronger in the male, as is to be seen in the greater prominence, thickness, and rugosity of the lambdoidal, sagittal, temporal, and masseteric crests. From the temporal crests of the parietals arise a pair of horn-like protuberances, which in the female are mere roughenings of the crests. These processes are laterally compressed, of elongate oval section, and with their long axes running in the direction of the temporal ridges and oblique to the sagittal suture ; their free ends are rounded and roughened. The second pair of protuberances are formed by the great development of the supraorbital borders of the frontals, which are drawn out into a pair of large, depressed, wing-like bodies, completely overhanging the orbits. These processes are relatively better developed in the female than any others of the paired protuberances. The posterior median elevation of the frontals is not much more conspicuous in the males than in some females. A third pair of conical, hornshaped processes arise from the antero-external angle of the frontals, where they unite with the nasals and lachrymals. Feebly-marked indications of these processes are to be seen in the female, more decidedly in some individuals than in others. A fourth pair of protuberances are formed by the short, thick, conical processes which are given off from the anterior ends of the masseteric ridges of the maxillaries. These are likewise to be found in the female, though very much less strongly

322 SCOTT. [Vol. XI.

developed. From the upper margin of the maxillaries arise the most anterior and the largest of the protuberances. " The whole conformation of the maxillaries is, so far as we know, unique among the mammalia ; the superior borders curve sharply upward into two powerful plates of bone, concave on the outer side, convex on the inner, and rising to the level of the parietal processes, with a concave posterior and convex anterior border" (Osborn and Wortman, No. 8, p. 357). The concavity of the outer surface is interrupted by two convex ridges, one of which is the continuation upward and forward of the masseteric ridge, though its smooth surface shows that it is not a part of that ridge, and the other is the bulging caused by the alveolus of the large canine. These maxillary protuberances are, as we have already seen, shown in the female skull by a slight upward arching of the borders of the maxillaries ; in some specimens they are quite distinct, and have slightly thickened and rugose margins, though not in any degree approximating their great size in the male. In UintatJiei'iiini the maxillary protuberances are rounded and horn-like, while in Protoceras they are plates. Sus larvatus, a recent species from Africa, has maxillary protuberances which are thick and very rugose. The infraorbital foramen is slightly more anterior in position in the male, and in that sex the premaxillary spines are rather narrower, so that the incisive foramina are somewhat more widely open. All the bones of the skull are thicker and more massive, as a result of which the male skulls are seldom so distorted by pressure as the female specimens very generally are.

Still a third kind of skull-structure is presented by the type specimen of Protoceras celer, as described and figured by Marsh (Nos. 5, 6). This specimen is imperfect and some points of importance cannot be determined from it. On the whole, it agrees best with the females which have already been described, but in none of the latter which I have seen are the parietal protuberances anything more than mere roughenings of the temporal crests, while in the type specimen they are fairly well developed, conical eminences, which are described as " a pair of small horn-cores, situated not on the frontals.


but on the parietals, immediately behind the frontal suture. . . . The horn-cores are well separated from each other, and point upward, outward and backward, overhanging somewhat the temporal fossae. They are conical in form, with obtuse summits" (No. 5, p. 81). Osborn and Wortman, who have compared the type with their own specimens, report that these parietal "horn-cores" are about equal in size to the anterior frontal protuberances of the male skull figured by them, and are therefore very much smaller than the parietal protuberances of that animal and of quite a different shape. In Marsh's specimen, so far as can be judged from the figure (No. 6, PI. XXII, Fig. i), the median frontal eminence would appear to be more prominent than in the females which have been described above and the maxillary fossa is of a different shape and more distinctly bounded behind. The thinness of the maxillaries precludes the supposition that there can have been any great protuberances upon those bones.

Osborn and Wortman conclude that the type specimen was the skull of a female, and this result is altogether probable. At the same time, however, it is remarkable that among the numerous undoubted female skulls which have been collected, not one should show the conical, horn-like protuberances on the parietals which characterize the type. More extensive material will be required to determine the significance of this feature. It may be only an individual variation, in which the female has approximated the characters of the other sex, as we have already seen to occur sometimes in the case of the maxillary protuberances. Or, in the second place, the type specimen may represent the female of a species distinct from that to which the individuals described in this paper belong. I was at one time tempted to believe that the type might be the male of a more ancient genus than the specimens which Osborn and Wortman referred to Protoceras, and that it would be found to have come from the Oreodou beds. Mr. Hatcher has, however, lately visited the spot whence the type was taken, and writes me that it is in the Protoceras beds, well up toward the top.



[Vol. XL


Male. Length of skull from occipital condyles . 0.215 Breadth of skull at supraorbital margins . .111

Length of face from anterior border of

orbit .130

Length of cranium from occipital crest to

front margin of orbit .100

Occiput, height

Occiput, breadth at base

Sagittal crest, length .030

Frontals, length in median line .... .050

Nasals, length in median line .030

Palate, breadth at ?«/ .028

Maxillary, length on alveolar border . . .120

Premaxillary, length in front of canine . . .018

Diastema between canine and iJ., length . .012

Diastema between ^ and tl •oi8

Mandible, length

Mandible, height of condyle above lower


Mandible, breadth of angle

Mandible, depth below JTj

Mandible, depth below /~i

Diastema between J~i and /I, length . . .

N.B. — The measurements of the male are taken Wortman.

Female Female No. I. No. 2. MoscHus.

0.223 0.225

.^072 .oSo



0-137 .052

•07 s




















.018 .015















in part from Osborn and

III. TJie Brain.

No specimen of the brain-cast is preserved in the Princeton collection. Osborn and Wortman's figure shows that it was decidedly more advanced than in the case of any other White River genus so far known. Their account is as follows : "The brain is deeply convoluted. We observe upon "each hemisphere four longitudinal gyri ; these, according to Owen's nomenclature, would be the median, medilateral, suprasylvian, and sylvian" (No. 8, p. 355). A remarkably modern feature of this brain is in the width of the cerebral hemispheres, as compared with their length, and in this respect, as well as in its richness of convoliition, it is superior to the brains of the small modern deer, such as MoscJms, Hydropotes, Cervus hiwiilis, etc.


The sulci give more precise means of comparison than do the gyri. Krueg describes the fissures of the traguline brain as follows : " Die Figur zeigt, dass auf der oberen Seite nicht nur die dahin gehorigen Hauptfurchen, sondern auch ein Theil der auf der Medianseite gelegenen Fissura splenialis sichtbar wird. Die Fissura coronalis ist bei alien Zeichnungen mit dem oberen Fortsatz der Fissura suprasylvia verbunden. . . . Die Fissura suprasylvia zeigt keine Marke zwischen Korper und hinterem Fortsatz. Ausser der sehr kurzen Fissura lateralis findet sich nach hinten und aussen von ihr noch eine accessorische Furche, . . . Die Hauptfurchen an der Lateralseite scheinen alle vorhanden zu sein. . . . Der Gesammthabitus der Furchen sowohl als der ausseren Umrisse ist aitsserordentlich dJinlicJi dem der ElapJiier [i.e., Cervidae.] In the smaller deer, such as MoscJius, ElapJiodiis, Cej'vits Jniinilis, the splenial fissure (sulcus calloso-marginalis) extends over upon the dorsal surface of the hemisphere, as in the tragulines, the condition which Krueg calls " supination," while in the larger members of the group it is confined to the median side ("pronation "). "Was die iibrigen Furchen anlangt, so ist der Processus acuminis fissurae Sylvii gewohnlich lang ausgezogen ; so dass er fast die Kuppel der Fissura suprasylvia erreicht ; es pflegt keine accessorische Furche zwischen beide eingeschoben zu sein. Die Fissura diagonalis ist mit ihrem Hinterende durch einen nach oben gerichteten Fortsatz der Fissura suprasylvia verbunden" (No. 4, pp. 315-317).

Osborn and Wortman's figure shows that the principal sulci in Protoceras correspond closely with those of the larger deer. There is no " supination " and the splenial fissure does not appear on the dorsal surface ; the lateral fissure is long and a well marked accessory furrow lies between this and the suprasylvian. The latter is connected anteriorly with the coronal sulcus. The diagonal sulcus is plainly visible in the view from above, but does not quite reach the suprasylvian, and the acuminate process of the sylvian fissure is long. The convolutions are thus distinctly of an advanced and modern type. The cerebellum is not known, but as Marsh has suggested, its small size is indicated by the narrowness of the occiput.

326 SCOTT. [Vol. XI.

IV. The Vertebral Colimui.

The skeleton which forms the principal subject of this paper retains twenty-one of the presacral vertebrae. We may conjecture with considerable confidence that the entire number of such vertebrae was twenty-six, and of the five missing ones several are represented in other specimens. The neck is relatively longer than in the tragulines or Leptomeryx, shorter than in Mosc/ms, and is rather slender. As its actual length is almost the same as in the musk-deer, its shortness as compared with the length of the skull is very marked. In general characters the atlas (PL XXI, Fig. 6) is like that of the Pecora. It is long in proportion to its breadth, which is due to the comparatively small lateral extension of the transverse processes. The anterior cotyles for the occipital condyles are broad, of considerable height, and widely separated both above and below. Their external borders are quite deeply notched, corresponding to the similar emarginations on the outer sides of the occipital condyles, to which attention has already been called. The ventral incision between the anterior cotyles is more deeply cut than in Cervus, but the articular surfaces for the accessory facets developed on the tubercles of the basioccipital in front of the condyles are not reflected so far over upon the inferior arch of the atlas as in that genus. Owing to the depth of this anterior incision and the position of the surfaces for the axis, the inferior arch is not so elongate fore and aft as in the deer, nor is it keeled ventrally, and the hypapophysis is hardly more than a rudiment. The arch is strongly convex transversely and has two shallow fossae upon each of its sides. The neural arch is likewise short from before backward ; for though the incision between the anterior cotyles is not so deep as on the ventral side, that between the posterior pair is much deeper. Like the inferior arch, the neural is strongly convex from side to side and the two together form a nearly circular ring. The neural spine is remarkably well developed and conspicuous ; it is much compressed, but quite high, and occupies about one half the length of the arch. On each side is an elevated ridge, enclosing a lyrate area, in the middle


of which the spine rises. As usual in the Pecora, the arch is perforated by foramina for the first pair of spinal nerves. The posterior cotyles for the centrum of the axis are more oblique and less directly transverse than in the Pecora ; they are quite high, but of no great width, and the articular surface is reflected within the ring of the atlas, where it forms a broad, continuous facet for the odontoid process. There is, however, no facet upon the hinder face of the inferior arch for the centrum of the axis below the odontoid, such as occurs in the Pecora. The transverse processes have no great width, but they extend out more widely than in Cervits or in Moschus, especially toward the posterior end ; they are pierced by foramina for the inferior branches of the first pair of spinal nerves, but there is no vertebrarterial canal, as there is in AgriochceruSf the swine, etc.

The axis (PI. XXI, Fig. 7) is peculiar in many ways. The centrum is very long, much more so than in any other vertebra of the neck, and is broad and much depressed in front, contracted in the middle, and subcylindrical behind. The posterior surface is obliquely placed and deeply concave, forming a hemispherical pit for the centrum of the succeeding vertebra. The ventral surface of the entire centrum is marked by a prominent but very thin hypapophysial keel. The articular surfaces for the atlas are narrow and high, though extending less upward along the sides of the neural canal than in the Pecora. In the latter the two surfaces are fused together beneath the odontoid process, forming an uninterrupted facet, and their dorsal ends pass into the pedicels of the neural arch. In Protoceras the articular surfaces do not extend beneath the odontoid, but are completely separated by a wide and deep emargination, and a still more extensive notch divides the dorsal end of each surface from the neural arch. The odontoid process is neither conical nor spout-shaped, but rather semicylindrical, with bluntly rounded point. The upper surface bears a median ridge with a shallow fossa on each side of it, and the ventral surface is gently convex from side to side. This shape of odontoid is the half-way stage in the conversion of the conical into the spout-like form.

328 SCOTT. [Vol. XI.

It is remarkable how many of the approximately contemporaneous Oligocene genera have reached the same stage in the transformation of the odontoid, and as I have elsewhere shown, this transformation is one of the most convincing cases of parallelism, it having been accomplished many times independently and always in the same way. Protoceras is in the same stage of the change as Oreodon, Gelociis among the Pecora, Poebrotheriiun among the camels and llamas and MesoJiippjts among the horses. The resemblance is particularly close in the cases of Gelociis and Mesohippiis, while in Poebrothermin the process is wider and shorter, and in Oreodon it is broader and more flattened than in any of the genera named. The significance of this oft repeated change in the character of the odontoid process is to be sought for in the relations between the axis of the skull and the line of the neck. In the short-necked forms with conical odontoid, it will be found that the craniofacial axis is continuous with the line of the neck, or, at most, that they form a very open angle, while in the long-necked forms with spout-shaped odontoid, the two lines meet at a more or less acute angle. It is, of course, of the highest importance to give the spinal cord a channel without sharp bends, and this is accomplished in long-necked ungulates by making the odontoid concave.

In Protoceras the neural spine of the axis is different, not only from that of the Pecora, but also from that of most ungulates in general, in forming a great hatchet-shaped plate ; its upper margin descends in a gentle, regular curve, from behind downward and forward, describing an arc of about 45°. The free border is but slightly thickened and rugose and the whole plate is very thin and delicate, though in some cases the hinder border is much thickened. The posterior zygapophyses are prominent, present obliquely downward and outward and have facets which are slightly concave transversely. The transverse processes are short and very slender, and their bases have no great antero-posterior extension along the sides of the centrum. The foramina of the axis are quite different from those found in the Pecora. In the latter the pedicels of the neural arch are perforated for the exit of the second pair of spinal nerves.


and, as the vertebral artery " usually enters the neural canal between the arches of the second and third vertebras," the transverse process is generally not perforated by the vertebrarterial canal, but is pierced anteriorly by the inferior branch of the second spinal nerve. In Protoceras, on the other hand, the pedicel is imperforate, the nerve passing out through the deep notch which separates the atlanteal facet from the neural arch. As the transverse process arises behind this point, it does not traverse the path of the inferior branch of the nerve, and hence there is no foramen for that branch. The process is perforated, however, by the vertebrarterial canal, both openings of which are present, and the artery probably entered the neural canal between the arches of the atlas and axis.

None of the specimens contain representatives of the third or fourth vertebra.

The fifth cervical is relatively short, both actually and proportionately shorter than in Mosc/ms, an animal which agrees well with this species in stature. The centrum is of depressed cylindrical shape, is strongly opisthocoelous and the faces are markedly oblique to the long axis of the vertebra. On the ventral surface is a prominent, but very thin and fragile keel, which terminates posteriorly in a hypapophysial tubercle. The neural arch is broad and nearly plane on the dorsal side and carries a spine which is much higher than in Moschus and which tapers and rapidly becomes very slender, though its base occupies nearly the whole length of the neural arch. The zygapophyses are large and widely separated on the two sides ; the anterior pair are the narrower and present obliquely inward as well as upward. As the obliquity of the centrum is more decided than that of the neural arch, the pedicels of the arch are very low in front, and hence the ventral sides of the prezygapophyses are separated from the centrum only by narrow notches. The postzygapophyses are larger and more prominent than the anterior pair and project directly downward, almost without obliquity. The neural canal is broad and low, especially in front ; posteriorly the increased height of the pedicels and obliquity of the centrum give it greater height. The transverse process is so mutilated on both sides that its

330 SCOTT. [Vol. XI.

exact shape cannot be determined. Enough remains, however, to show that its origin on the centrum is less extended in the fore and aft direction than in Mosclms and in consequence the posterior orifice of the vertebrarterial canal occupies a more advanced position than in that genus.

The sixth cervical has a centrum of nearly the same length as the fifth and is of a similar depressed cylindrical shape. The anterior face is strongly convex and hemispherical in shape, with a large shallow pit, or sulcus, for the attachment of the intervertebral cartilage ; the posterior face is as strongly concave, but both faces are much less oblique with reference to the long axis of the centrum than on the fifth vertebra. The ventral keel is very feebly developed and does not terminate behind in a tubercle, but simply dies away. The neural arch is very short antero-posteriorly ; this is due to the manner in which the arch is cut away between the prezygapophyses. The axis of the neck undergoes a marked change in direction between the fifth and sixth vertebrae and this is provided for by the projection of the anterior zygapophyses on the latter. These are broader, slightly more elevated and less oblique than on the fifth, while the postzygapophyses are very much the same on both, except that on No. 6 the articular surface is continued down upon the neural arch, forming an accessory facet ; the pedicels of the neural arch are higher and hence the neural canal is more rounded and less depressed. The spine does not appear to be much higher, but is decidedly thicker and heavier and inclines forward to about the same degree. As is always the case on the sixth vertebra, the transverse process is very clearly divided into diapophysial and pleurapophysial elements. The former is a stout, depresse'd rod, which stands out nearly at right angles with the centrum and having the distal end expanded in the fore and aft direction ; this process is decidedly more prominent than in Moschus. The pleurapophysis is a very large plate, slightly convex on the inner side and concave on the outer, with thickened free border, especially at the angles. This element is developed very much as in the musk-deer, but is of greater vertical height and the inferior face of the centrum is more clearly demarcated from the origin of the plates.


The seventh cervical is the shortest of the series and has a centrum of a shape differing from that of all the others ; it is narrow in front and the anterior portion is so sharply constricted that it appears to stand upon a short neck ; behind this it immediately broadens and the posterior face is wide and depressed and displays facets for the heads of the first pair of ribs. The faces are still somewhat oblique in position. The neural arch is very short, being cut away between the prezygapophyses, as is also the case in the sixth vertebra, another marked change in the direction taken by the spinal column occurring at this point. The prezygapophyses are very large and widely separated ; for most of their extent they face upward, with but little obliquity, but the articular surface is reflected downward upon the inner side of the process and presents inward. This accessory facet articulates with the corresponding one on the outer side of the neural arch of the sixth vertebra, which has already been described. The postzygapophyses are very small, hardly a third as large as the anterior pair. The neural spine is no higher than that of the sixth, but is much larger in every other dimension and looks like a truncated thoracic spine, which stands erect, instead of pointing forward as in the other cervicals. The transverse process is quite a long, depressed bar, which, as usual, is imperforate. This vertebra differs from the corresponding one of MoscJuis in the heavier neural spine, the much more decided lateral projection of its zygapophyses and the longer transverse processes.

The Thoracic Vertebrce (PI. XXI, Fig. 8) were probably fourteen in number. Their number cannot have been less than twelve, since so many are preserved in a single specimen and do not form an entirely unbroken series, and can hardly have exceeded fourteen, because the same specimen has preserved five lumbars, apparently the entire series. It is not impossible, of course, that the typical artiodactyl formula of nineteen thoraco-lumbar vertebrae was exceeded in this genus, but there is no reason to assume that such was the case.

The first thoracic has a short, broad, depressed and distinctly opisthocoelous centrum. The prezygapophyses are like those

332 SCOTT. [Vol. XI.

of the seventh cervical, but smaller, more concave and without the internal accessory facet. The neural arch is narrower and more strongly convex and the spine is thicker, though of less antero-posterior extent, and inclined backward instead of being erect ; the postzygapophyses are on the inferior face of the overhanging neural arch. The pedicels of the arch are very narrow at their point of origin, being deeply notched behind for the passage of the spinal nerves. The transverse processes are long, heavy and prominent and end in large concave facets for the tubercles of the first pair of ribs. Compared with the corresponding vertebra of MoscJius, the chief differences to be observed consist in the greater breadth of the neural arch and the longer transverse process in the fossil.

The first and second thoracic vertebrae are somewhat different from the others, since they form a transition from the structure of the cervicals. The second has a rather more trihedral and less depressed centrum than the first, a somewhat heavier and longer spine and shorter transverse processes. The prezygapophyses are on the anterior face of the neural arch, but are widely separated. From the second to the sixth thoracic the centra gradually become shorter, less depressed, and more distinctly trihedral in section; the transverse processes shorten and terminate in smaller flat or slightly convex facets for the rib tubercles, instead of concave surfaces; the spines increase in length and the zygapophyses draw near to the median line. The sixth appears to have the longest spine of the series, though this point cannot be determined with certainty. Back of this the spines gradually decrease in height, but still incline strongly backward until the (supposed) eleventh thoracic is reached, which is the anticlinal vertebra. The transverse processes continue long and prominent and still have relatively large rib-facets, but on the eleventh (?) they are greatly reduced and disappear on the twelfth Q). The latter has postzygapophyses of the cylindrical pattern and corresponding anterior processes appear on the thirteenth Q.). The centra of the posterior thoracic vertebrae have become considerably elongate ; they have subcircular faces, but are constricted and of trihedral section in the middle.


\J \J \J

The thoracic region is of about the same length as in Moschus and of similar general appearance, but a comparison of the two genera brings to light a number of differences in the details of structure. Thus, in Protoceras the transverse processes are decidedly longer, especially in the hinder part of the region. In MoscJms the spines of the posterior vertebrae are lower, but much more extended from before backward, especially at the tips, which from the tenth to the thirteenth are thickened and project beyond the spine both in front and behind. In this genus also the ninth, tenth and eleventh vertebrae have metapophyses, which are particularly prominent on the tenth and eleventh. In the fossil these arise near the ends of the transverse processes.

The Lumbar VertcbrcB (PI. XXI, Fig. 9) numbered at least five. In the anterior region the centra are shaped like those of the posterior thoracic series, but as we pass backward, they become more and more broadened and depressed ; the centrum is longest in the second, third and fourth, slightly shorter in the first and considerably so in the fifth. The zygapophyses are of the usual cylindrical, interlocking, artiodactyl type, but no episphenial processes are developed. Large and conspicuous metapophyses are present, which in MoscJuls are hardly more than rudiments. The neural arches and spines are more traguline in shape and appearance than pecoran ; the arches are short from before backward, being deeply cleft between the postzygapophyses, which is not the case in MoscJuis, and the pedicels of the arches are not perforated by the spinal nerves (as they are in the modern genus) which pass out through the notches below the postzygapophyses. The spines are less extended antero-posteriorly than in the musk-deer and resemble rather those of Tragidus, being slender and curved forward. The transverse processes are likewise quite different from those of MoscJnis. In the latter they are slender and depressed, extending downward as well as outward, and are not at all like the broad, plate-like processes of the higher Pecora. In Pi'otoceras, on the other hand, they are long, very broad and thin, rounded at the tip and but slightly decurved, and are thus of the typical pecoran shape. The process is quite short on

334 SCOTT. [Vol. XI.

the first lumbar, much longer on the second and reaches its maximum length on the third, while the overlapping ilium prevents its attaining any great length on the fifth.

No Sacral and but one Caudal Vertebra is preserved in any of the specimens. The latter belongs to the proximal part of the tail and has a relatively large subcylindrical centrum, which indicates that Protoceras had a much better developed tail than Moschus or than the modern deer generally, and which was comparable in proportionate length to that of the giraffe. The prezygapophyses are very long, but do not appear to articulate with any other vertebra. The specimen belongs to a younger and somewhat smaller animal than the skeleton here described and both epiphyses are lost. This fact must be allowed for in using the measurements given below.

The Ribs. — In many primitive artiodactyls, such as Oreodoji, the ribs are notably slender and subcylindrical, resembling those of the carnivores. In Protoceras they are entirely pecoran and resemble those of MoscJms in general, but their somewhat greater length and curvature and the increased distance between the head and tubercle point to a wider and more capacious thorax. The anterior ribs are short, subcylindrical proximally, expanded and plate-like distally ; the maximum length is attained about the middle of the thorax ; posteriorly the ribs become continually more slender and rounded, as well as shorter.


Protoceras. Moschus.

Atlas, greatest length 0.050 m. 0.029 m.

Atlas, greatest breadth 072 .036

Axis, length of centrum 041 .035

Axis, breadth of anterior face 032 .024

Axis, length of odontoid process 013

Sixth cervical, length of centrum 025 .025

Sixth cervical, breadth of anterior face 013

Seventh cervical, length of centrum 019 .016

Seventh cervical, breadth of anterior face 013

First thoracic, length of centrum 015

First thoracic, breadth of anterior face 015

Last thoracic, length of centrum 020 .022

Last thoracic, breadth of anterior face 018

First lumbar, length of centrum .022 .024

Second lumbar, length of centrum 024 .024


Protoceras. Moschus.

Third lumbar, length of centrum 0.024 m. 0.024 m.

Fourth lumbar, length of centrum 024 .024

Fifth lumbar, length of centrum 019 .024

Third lumbar, width of transverse process at base . . . .017 .011

Third lumbar, fore and aft diameter of spine on .021

First caudal, length of centrum 015 .012

First caudal, breadth of anterior face 014 .005

First caudal, breadth of posterior face 013 .004

V. The Fore Limb.

The Scapula (PI. XXI, Fig. 10) is entirely ruminant in character and quite different from that of the primitive artiodactyls. In outline it forms a high, narrow triangle, with very slender neck. The glenoid cavity is shallow and nearly circular in shape and the coracoid forms a prominent, recurved hook, which is entirely like that of Moschus. The spine is placed near to the anterior margin, so that the prescapular fossa is very much smaller than the postscapular. The spine itself rises rapidly from the suprascapular border and becomes unusually high and prominent ; its free border is not, as in most ruminants, an erect edge, but almost from the start is curved over toward the posterior side, the curvature increasing distally. Thus, the hinder surface of the spine is deeply concave and its front convex in a manner resembling the scapular spine of Mesoreodoti, but unlike that genus, there is no distinct metacromion. The acromion is longer, broader and thicker than in the existing Pecora. The coracoid border is not quite straight, but rises from the neck with a slight curvature, which is concave below and convex above. The other two borders are straight. Both the coracoid and glenoid borders, and especially the latter, are thickened and elevated, making both the preand postscapular fossae somewhat concave. The subscapular fossa is also rendered slightly concave by the elevation of its borders.

The Hwneviis is short and stout. The head is large, strongly convex in the antero-posterior direction and projecting back much beyond the line of the shaft ; it is much larger in proportion to the length of the bone than in MoscJms. The external tuberosity is very large, forming a high, massive ridge.


36 SCOTT. [Vol. XI.

which extends across the whole anterior face of the bone, is elevated much above the level of the head, and rises toward the inner side, where it forms a blunt, recurved hook, overhanging the bicipital groove. The internal tuberosity is small and laterally compressed, while the bicipital groove is deep and rather narrow. The tuberosities are much more conspicuous than in Moschus and (apparently) than in Gelociis. The deltoid ridge is but little better developed than in the former genus. The shaft is short and heavy ; its promixal portion is laterally compressed, but very thick antero-posteriorly ; the middle part is thicker and more rounded in section, and the distal portion is but moderately widened and has a short but quite prominent and rugose supinator ridge. The anconeal fossa is small and of triangular shape ; it is considerably lower and broader than in Gelociis. The supratrochlear fossa is quite deep, but there is no perforation of the bone at this point.

The trochlea is placed obliquely to the long axis of the shaft, descending somewhat toward the outer side. The intercondylar ridge is a rounded tuberosity, which is much broader and more median in position than in Gelocns, though less so than in OreodoUy to the humeral trochlea of which that of Protocet'as bears but little resemblance. The internal epicondyle is much less prominent than in the latter genus, though decidedly more so than in Gelocns or the recent ruminants.

The Ulna and Radius (PL XXI, Figs, ii and 12) are in old individuals coossified at the distal end, but separate for most of their length. The two bones are, however, closely applied together throughout and the radio-cubital arcade is both short and narrow. The ulna is less reduced than in the necent Pecora and the olecranon is higher, straighter, and projects less backward ; its free end is thickened and rounded, almost club-shaped, instead of being squarely truncate, and the sulcus along the summit is shallow and obscurely marked. The sigmoid notch has a salient beak and shows a fully differentiated humeral articulation ; on the proximal side of the notch the humeral facet extends across the whole width, but distally the facet abruptly contracts and is confined to a narrow strip along the internal border. The shaft of the ulna is much


reduced as compared with that of the oreodonts, suillines, etc., but much less so than in the existing Pecora, or even than in Leptomeryx and the tragulines ; in its proximal portion, it is convex on the inner, concave on the outer side ; distally this arrangement is reversed. The shaft tapers inferiorly, but shows a considerable expansion in the antero-posterior direction about three-fourths of an inch from the distal end, from which it rapidly tapers again. The distal end is relatively much larger than in the recent ruminants and is occupied by a saddle-shaped facet for the cuneiform.

The radius has much the same proportions as in MoschttSy but is somewhat longer. The head is transversely expanded, particularly toward the ulnar side, and antero-posteriorly compressed ; it occupies the entire distal surface of the humerus, the ulna having but a minute facet for this portion of the latter. The proximal surface is divided into three nearly equal facets for the corresponding divisions of the humeral trochlea. The shape of the intercondylar pit of the radius in Protoceras is, so far as it goes, a point of resemblance to the oreodonts, but the other humeral facets and the whole shape of the head are entirely different. As these characters are retained with remarkable persistency throughout the whole history of the oreodonts and even in Agriochoerns, the difference is not unimportant.

The shaft of the radius is of nearly uniform size throughout, and is strongly arched forward ; it is rather slender, but yet of the typical modern ruminant shape, i.e., of transversely oval section, and contrasts strongly with the subcylindrical and remarkably slender radial shaft of Oreodon. The distal end is moderately expanded and thickened and bears two roughened, elevated ridges which enclose a sulcus for the extensor tendons. The distal facets for the carpus seem to show some variations which may be more or less due to age. Of the young specimen described by Osborn and Wortman these writers say : " The process of bone which bears this facet (i.e., for the scaphoid) is not produced backward, as it is in Traguliis, nor has it the marked obliquity seen in Leptomeryx and Cariacus and, to a less degree, in Tragidiis. The scaphoid

338 SCOTT. [Vol. XI.

facet is not sharply defined by a prominent ridge from that of the lunar, as it is in Cariactis, Leptovieryx, and Tragulus, the two articular surfaces being quite continuous in front " (No. 8, p. 360). In the adult female skeleton which we have been considering, the process of bone bearing the scaphoid facet is produced backward more than in Tragidus and almost as much as in Moschus, though the lunar does not abut against the ulnar side of the process so extensively as in that genus. The scaphoid facet is distinctly divided from that for the lunar by a ridge which is prominent even in front, though much less so than in the modern ruminants ; it is concave in front and convex behind. The lunar facet is rather narrower than the scaphoidal and is of similar shape, but the posterior convexity is much less extended toward the palmar side. The straight course taken by the facets, to which Osborn and Wortman have called attention, is a difference from all the modern selenodonts, and even in such ancient forms as Anoplotheriimi and Lcptomeryx the facets run obliquely from before backward and inward. In the camels this obliquity is not very marked. In another respect Protoceras differs from all the existing Pecora, Tylopoda, and Tragulina, viz., in the absence of any facet on the radius for the cuneiform.

The manus has already been well described by Osborn and Wortman, but for our present purpose it will be necessary to examine it somewhat more in detail.

The Carpus (PI. XXII, Fig. 17) is remarkably primitive, as compared with the great degree of modernization displayed in the skull. Schlosser (No. 9) long ago called attention to the fact that a constant difference between the more ancient and the later artiodactyls was to be found in the much greater relative height (vertically) of the carpus in the former, especially with regard to the distal elements. This ancient feature is as decidedly marked in Protoceras as in the oreodonts. In the shapes of the individual bones, on the other hand, may be seen some approximation to modern conditions. The scaphoid is high, but not much extended transversely or antero-posteriorly. Its shape is thus entirely different from the almost cubical scaphoid of Oreodoti, and though considerably higher, it


is shaped more as in Moschus. The radial facet consists of an anterior convexity, which is lower than in Oreodon, and not reflected so far over upon the dorsal face of the bone as it is in that genus, and of a posterior concavity. In the tragulines the scaphoid is relatively much narrower and the anterior convexity for the radius much higher and rising steeply toward the ulnar side, while the inner side is excavated for a descending process of the radius. The proximal surface of the scaphoid is thus much more like that of Mosclms than of the tragulines and differs markedly from that of Oreodon in the sudden narrowing or excavation which invades the posterior concavity from the ulnar side of the bone. The distal end is very different from that seen in any of the groups mentioned above ; it is occupied by two distinctly separated facets, for the trapezoid and magnum respectively. The former is rather the smaller of the two and is simply concave and of irregularly oval shape. The magnum facet stands at a somewhat lower level than that for the trapezoid and is nearly flat in front, becoming concave posteriorly to receive the head of the magnum. There is no distinct facet for the trapezium, though Osborn and Wortman state that the two bones are in contact. On the ulnar side are two facets for the lunar, one proximal and one distal, and both nearer to the dorsal than to the palmar border. The distal facets on the scaphoid of Oreodon are entirely different from those of Protoceras both in shape and in position, which is due to the fact that in the former the magnum has shifted almost entirely beneath the scaphoid, while the lunar has gone over upon the unciform. In the Pecora and Tragulina these facets are changed by the coalescence of the magnum and trapezoid, and in the latter group, by a displacement similar to that which has occurred among the oreodonts. In PoebrotJieriimi, on the other hand, we find a condition very like that of Protoceras, with the addition of a minute facet for the trapezium.

The lunar is remarkably high and narrow, narrower even than the scaphoid, which is the reverse of the proportions found in Mosclms and Tragnliis, though in DorcatJieriiim {Hyccmoschus) the lunar is the narrower of the two. The proximal

340 SCOTT. [Vol. XI.

end is but little expanded and of nearly uniform width on its dorsal face, though it is somewhat constricted in the middle of the radial side. The radial facet is broader and convex in front, narrower and concave behind. The distal end is described and figured by Osborn and Wortman as having "its articular surface divided almost equally between the unciform and magnum." This description does not quite apply to the adult female skeleton which is here described. While the two facets are not far from being of equal width, a comparison with DorcatJierium and Mosc/ms shows a tendency toward the traguline method of displacement, in that the facet for the magnum is more lateral and that for the unciform more distal. Consequently, the salient beak formed by the meeting of the two surfaces is not in the vertical median line, as it is in the Pecora, but is shifted toward the radial side. The magnum surface is the narrower of the two, but it extends farther toward the palmar side ; it is slightly convex in front and concave behind, while the unciform facet is concave throughout. The scaphoid facets correspond to those already described on that bone. On the palmar side the proximal end of the lunar rises somewhat above the level of the scaphoid and has an accessory lateral facet for the radius, though the contact is less extensive than in recent Pecora. Another accessory radial facet of the proximal end lies behind the principal one. On the ulnar side of the bone are two facets for the cuneiform, the proximal one of which is fiat and confined to the dorsal half of the lunar, while the distal one is concave and extends through nearly the entire depth.

The cuneiform is relatively broad, but of proportionately small dorso-palmar depth ; its principal difference from that of existing tragulines and Pecora consists in the absence of any facet for the radius. In these groups the displacement of the ulna is to some extent compensated by an extension of the ulnar facet posteriorly and down upon the outer side, but in Protoceras the ulnar facet remains a narrow groove. As in the Pecora, the infero-external angle of the bone is slightly incurved to form a blunt hook. The pisiform facet is broad and flat above, narrow and concave below, where it extends


down upon the hook. The distal end is occupied by the wide and shallow (fore and aft) facet for the unciform.

The pisiform is neither traguline nor cervine in character, but differs from that of the musk-deer much as the latter does from the larger species of Cervtis.

As compared with the pisiform of Moschus, it is somewhat longer and more slender and is but slightly incurved at the tip. On the other hand, it differs much more radically from the straight, slender, and elongate pisiform of the oreodonts, and tends distinctly toward the shape taken in the Pecora. It is short, deep, and compressed, rugose and somewhat incurved distally. The facets for the ulna and cuneiform are continuous ; the former is small, triangular, and plane, the latter is larger, of irregular outline, and saddle-shaped.

The trapezium is not preserved in any of the specimens in the Princeton collection, but, fortunately, Osborn and Wortman are able to assure us of its presence.

A significant feature in the carpus of Protoceras, pointed out by the authors just mentioned, is the complete separation of the magnum and trapezoid, as in P cebrothermni and the Tylopoda generally. The trapezoid is quite high and deep in the dorsopalmar direction, but very narrow ; it is somewhat wedgeshaped, broad behind and thinning to an edge antero-externally. Its proximal surface forms a rounded, convex head for the scaphoid, the articular surface of which is continued down upon the palmar side, doubtless for the trapezium. The contact with the magnum is by means of two facets which meet in the middle of the ulnar side. Distally the trapezoid bears a triangular, nearly plane surface for the second metacarpal, but has no contact with the third.

The magnum is relatively high and narrow, as compared with that of the recent Pecora. The dorsal face is subquadrate in outline, the palmar very much narrower, and the posterior hook, though stout and rugose, is very short. The proximal surface is unequally divided between the facets for the scaphoid and lunar. Anteriorly the scaphoid facet takes up quite two-thirds of the width of the magnum and is here slightly concave. Upon the " head " the surface is convex, and,

342 SCOTT. [Vol. XL

as the dividing ridge crosses the head obliquely, this region is more equally shared between the two facets. The lunar facet is much narrower than that for the scaphoid ; in front it is more decidedly concave and is lateral rather than proximal in position ; on the head, however, the surface which supports the lunar is truly proximal. On the radial side the small facets for the trapezoid and second metacarpal meet at an open angle and form a salient edge. The distal surface is occupied by a large saddle-shaped facet for the third metacarpal, and rising from this a small one for the second.

The unciform is a large bone, though smaller than the scaphoid, which exceeds it in every dimension save that of thickness ; it is, however, notably high in proportion to its width. The posterior hook, which in the Pecora, including even Moschus, is rudimentary, is still well developed, though smaller than in many ancient forms of the Artiodactyla, e.g.^ AncodiLS. The proximal surface has two facets, a narrower one for the lunar and a much broader one for the cuneiform. The former is low and slightly concave in front, rising behind into a convex head like that of the magnum. The cuneiform facet is continued down over the ulnar side of the bone, almost reaching the surface for the fifth metacarpal. The magnum surface is confined to a minute strip near the dorsal side, above the facet for the third metacarpal. The latter facet is much more extensive than in those Pecora in which the metacarpals have coalesced to form a cannon-bone and is even larger than in Gelociis. The distal surface is taken up by the large facet for the fourth metacarpal ; that for the fifth is very much smaller and rather lateral than distal.

The Metacarpus (PI. XXII, Fig. 17) contains four elements. The median pair (III and IV) have attained about the same stage of development, as regards length and slenderness, as in Gelocus, and are therefore relatively longer than in the more ancient artiodactyls, XipJiodo7t excepted. The laterals (II and V) have at the same time retained a size and importance which is utterly unknown in the Pecora and far exceeds what is found in the tragulines. All of the digits remain free throughout life. The mode of connection between the carpus and


metacarpus might with almost equal propriety be called either "unreduced" or " inadaptive," to use Kowalevsky's terms, because this manus is not reduced farther than is involved in the loss of the pollex and enlargement of the median digits. As Osborn and Wortman have suggested, a rudimentary first metacarpal may have been present. This suggestion is made more probable by the fact that the proximal end of metacarpal II has its internal angle truncated and slightly hollowed for the space of about half an inch, as if for the reception of a short splint-bone. This surface is much too large to have been occupied entirely by the trapezium, and nothing of the kind is apparent on mc. V.

The second metacarpal has a narrow, triangular head, with a plane surface for the trapezoid, and on the ulnar side a minute facet for the magnum, which is oblique and rises above the top of mc. III. The small size of this facet as compared with the same in such a genus as Ancodtts, for example, is an indication that the connection of the magnum with mc. II was undergoing reduction. The shaft is slender, compressed, strongly curved, and of trihedral section, the apex of the triangle being at the contact with mc. III. The distal end is slightly thickened and expanded, but the trochlea is narrow and rounded.

The third metacarpal is the longest of the series, extending both above and below mc. IV, as is also the case in Oreodon and Ancodus. The head is but little expanded, the longest diameter being the dorso-palmar one, and the head not being extended toward the radial side, as it is in Gelocus, in which mc. Ill covers nearly or quite all the distal surface of the coossified magnum and trapezoid. The proximal surface is taken up by the large, slightly saddle-shaped facet for the magnum, while a stout, oblique process overlaps the head of mc. IV, and abuts against the unciform ; this unciform process is longer and more prominent than in Gelocus. There is no facet for the trapezoid, mc. Ill being cut off from that bone by the connection of mc. II with the magnum. On the palmar side of the head is a small circular facet, which projects strongly toward the ulnar side and articulates with a corresponding facet on mc. IV. The shaft is of nearly uniform diameter

344 SCOTT. [Vol. XL

throughout, thickening somewhat toward the distal end ; in section it is an irregular quadrate, with nearly equal transverse and dorso-palmar diameters, but the palmar surface is much narrower than the dorsal. The ulnar side of the shaft of mc. Ill and the radial side of mc. IV are flattened, so that the two bones are closely approximated. The distal trochlea is very different from that of Oreodon and resembles the pattern found in Ge locus and Pcebrothet'hiin. In Oreodon this structure is low and wide, very strongly convex, and separated from the dorsal surface of the shaft by a deep, narrow pit ; while in Protoceras, as in the other genera mentioned, it is higher and narrower, the convexity is not so marked, and the articular surface is separated from the face of the shaft only by an almost imperceptible ridge. As in all the genera named, the carina is confined to the palmar side. The radial side of the shaft is flattened for about two-thirds of its length by the contact of mc. II.

Compared with the third metacarpal of Gelocits that of Protoceras is straighter, with less broadened and thickened head, it has an excavation on its radial side for the head of mc. II. and a larger unciform process ; otherwise the two are much alike.

The fourth metacarpal is the counterpart of the third, except that it is shorter and rather more slender proximally, while the distal portion of the shaft is somewhat broader and more compressed antero-posteriorly. The head is no broader than the shaft, but has a greater fore-and-aft diameter, owing to the presence of a palmar projection. The unciform facet is slightly concave transversely. Both of the median metacarpals are proportionately short ; while the scapula and the forearm bones are considerably longer than in MoscJms, the median metacarpals are noticeably shorter, and even in the musk-deer the anterior cannon-bone is by no means elongate. The relative length of the metacarpals in Protoceras is more nearly that seen in Dorcathermm.

Metacarpal V is shorter and distinctly more slender than mc. II, but otherwise shaped like it; the head has a rugosity on the ulnar side for ligamentous attachment. The unciform facet is small, of triangular shape and, but for a small con


vexity in the middle, is plane ; it rises steeply toward the ulnar side, so as to present internally almost as much as proximally.

The Phalanges (PI. XXII, Figs. 17 and 19) are short, both in proportion to the size of the animal and the length of the metacarpus, but are for the most part of typically ruminant pattern, nevertheless. The proximal phalanx is slender and relatively longer than in Dorcatlierium, though shorter than in MoscJius. It is nearly straight, and of quite a different shape from the curved, depressed phalanges of Oreodon. At the proximal end the dorso-palmar and transverse diameters are nearly equal ; the metacarpal facet is not concave and only the palmar border is notched for the metacarpal carina. The distal end is depressed, wider than deep, and its trochlea is obscurely divided by a shallow notch into two condyles.

The second phalanx is notably short and rather broad and depressed, the transverse diameter exceeding the dorso-palmar. This phalanx is thus of quite a different shape from that which characterizes the recent chevrotains and Pecora. The distal trochlea is neither deeply grooved nor reflected far upon the dorsal side of the bone.

The ungual phalanx is altogether ruminant in character, its shape being trihedral, slender and pointed ; it is not so slender and elongate as in Mosc/ms, nor so curved and depressed as in the tragulines, and is less symmetrical than in Posbrotheiium, indicating that the median digits were more closely approxi

mated than in that genus.


Protoceras Protoceras

No. I. No. i}- MoscHus.

Scapula, height 0.131 0.106

Scapula, greatest breadth 069 .059

Scapula, breadth of neck 015 .014

Scapula, fore-and-aft diameter of glenoid cavity . .022 .018

Scapula, transverse diameter of glenoid cavity . . .018 .015

Humerus, length (fr. ext. tub.) -149 -^49

Humerus, thickness of proximal end (ant.-post.) . .045 .032

Humerus, breadth of distal trochlea .025 .019

Radius, length I47 -UO -127

1 This individual is part of the type specimen of P. celer Marsh; it was found in 1894 by Mr. Hatcher.

346 SCOTT. [Vol. XI.

Protoceras Protoceras

No. I. No. 3. MoscHus.

Radius, breadth of proximal end 025 .026 .019

Radius, breadth of distal end 015 .016 .019

Radius, breadth of shaft in middle 015 .011

Ulna, length 182 .154

Ulna, height of olecranon 033

Ulna, breadth of distal end 009 .010 .007

Carpus, height 021 .014

Carpus, breadth 023 .018

Magnum, height of dorsal face 008 .C04

Metacarpal II, length o8r .027

Metacarpal II, breadth of proximal end 007

Metacarpal II, breadth of distal end 008 .005

Metacarpal III, length 089 .102

Metacarpal III, breadth of proximal end on .012 .010

Metacarpal III, breadth of distal end . . . . . .010 .010

Metacarpal IV, length 085 .100

Metacarpal IV, breadth of proximal end 008 .010 .010

Metacarpal IV, breadth of distal end 010 .010 .010

Metacarpal V, length 077 .034

Metacarpal V, breadth of proximal end 006

Metacarpal V, breadth of distal end 006 .005

First phalanx. III digit, length 024 .022 .028

First phalanx. III digit, breadth of proximal end . .010 .010 .010

First phalanx, III digit, thickness of distal end . .006 .007 .006

Second phalanx. III digit, length 012 .021

Second phalanx. III digit, breadth of proximal end .009 .007

Second phalanx. III digit, thickness of distal end .007 .007

Third phalanx, III digit, length 017 .021

Third phalanx. III digit, breadth of proximal end .009 .008

Third phalanx. Ill digit, thickness of proximal end .009 .008

VI. TJie Hind Limb.

The hind leg is much longer and heavier than the fore leg in all its parts. The disproportion is, however, by no means so great as in Tragulus, but rather resembles the condition'found in MoscJms. In the living animal the increased length of the hind leg was compensated by the greater habitual flexure of the knee and hock joints.

The Pelvis is pecoran rather than traguline in character. The ilium has a thin and compressed, but rather short and deep, plate-like neck, which expands gradually into the large and everted anterior portion. The outline of the anterior or supra-iliac border is like that of Cerviis more than of Moschus


in having distinct upper and lower projections. The sacral articulation is placed unusually far back, nearly the whole of the anterior expansion being in front of it. The acetabular and pubic borders are quite widely separated at their points of origin, but approximate forward; and the anterior part of the iliac surface is a narrow groove. The ischial border, which in Traguhis is nearly straight, is arched upward above the acetabulum into a crest, as in the Pecora. The fossa in front of the acetabulum and above the acetabular border is not so large or so deep as in Moschus ; in Tragiilus it is wanting. The acetabulum is small, nearly circular in outline, and deep ; the fossa for the ligamentum teres encroaching but little upon the articular surface. The ischium is long, deep vertically, laterally compressed and plate-like in form. At the posterior end it is considerably expanded, the dorsal border rising more gradually and not forming an overhanging hook, as in Moschus. The tuberosity is rather small, but prominent and rugose and situated higher up than in the recent genus. The pubis at its point of origin is slender and of a rounded, depressed section ; but it soon expands and becomes plate-like, as in the Pecora, not having the rod-like character found in the Tragulina. The obturator foramen is a long, narrow oval, considerably more elongate than in Moschus. As a whole, the pelvis is thus distinctly pecoran in type.

The Feimir (PI. XXII, Figs. 13 and 14) is much like that of Moschus, though not without some considerable differences. Thus, the head is less distinctly set upon a neck and is ovoidal rather than hemispherical in shape. The great trochanter is broader, more massive, and rises higher above the level of the head; the deep digital fossa does not extend so far behind the head and the ridge connecting the great trochanter with the second is much more prominent. The second trochanter forms a large, rugose, pyramidal protuberance. These differences, it is obvious, are mere matters of detail; and the proximal end of the femur is much more like that of Moschus than that of Traguhis. A resemblance to the latter is seen in the shaft, which is heavier than in Moschus, as well as longer, but has a similar rounded shape and strong anterior curvature,

348 SCOTT. [Vol. XI.

and the external linea aspera is much better marked and longer. The pit for the attachment of the plantaris muscle is larger and deeper, but has not such a rugose bottom and is not so conspicuous when viewed from the side, which is due to the fact that in the musk-deer the outer wall of the pit is cut away. The rotular groove is very different from that of MoscJius. In the latter the groove is narrow, symmetrical, with borders of equal height and thickness, and continued well up upon the anterior face of the shaft, while in Protoceras we find a condition more like that of the larger deer, though not in the same degree of development. In this genus the trochlea is broad and shallow ; the inner border is considerably higher and thicker than the outer and the articular surface is reflected over upon the mesial face of the bone ; but there is no such vertical prolongation of the groove upon the shaft. The characteristic rotular trochlea of the tragulines, which is also found in Leptomeryx, is quite different from that of Protoceras, which in this respect approximates more to the higher Pecora. The condyles are narrow and project less behind the plane of the shaft than in MoscJius, more than in Tragiihis.

The Patella (PI. XXII, Fig. 15) is remarkably large in proportion and somewhat peculiar. It is of the same general shape as in the musk-deer, broad above and tapering below to a blunt point, but is more massive and rugose and larger in every dimension. The fossa for the inner border of the femoral rotular groove is more deeply impressed than that for the outer border, and its internal edge forms a prominent projection, which extends along the mesial face of the femur. There is considerable individual or perhaps specific variation in the shape of the patella. One specimen is broader and less tapering than the one above described, and its inner prominence is not so much produced.

Both specimens of the Tibia (PI. XXII, Fig. 16) in the collection have lost the distal end, but fortunately this is preserved in Osborn and Wortman's material. Though longer and heavier than the tibia of Moschus, that of Protoceras is of essentially the same character. The condyles for the femur are rather narrow, but well extended from before backward ;


the spine is bifid and unusually high, but this may in part be due to the crushing which the specimens have undergone. The cnemial crest is very prominent, ends above in a massive rugosity and the sulcus for the peroneal .tendon is deeply incised. The shaft has the characteristic double curvature, both anteriorly and laterally, and the general shape found in the smaller Pecora. According to Osborn and Wortman, the astragalar surface is constituted very much as in the deer {Cariacus).

The proximal portion of the Fibula is coossified with the tibia, as in MoscJms, Tragulus, etc., and as this portion is considerably thicker proportionately than in the genera named, it is likely that a good length of the slender, filiform shaft was preserved in the living animal. Osborn and Wortman state that the distal end forms a distinct malleolar bone, which is shaped as in the Pecora and wedged in between the calcaneum and the distal end of the tibia. They find reason to believe, however, that in the fully adult animal the two bones may coalesce, as in the tragulines. In the latter group the malleolar bone, even when separate from the tibia, as according to Flower it sometimes is in Dorcatheriiim, has quite a different shape from that of the Pecora. The tibia and fibula of Leptomeryx are remarkably like those of Protoceras in almost every respect.

The hind foot has already been described by Osborn and Wortman, but it will be necessary to go over the same ground with somewhat more fullness, in order to display the range of variability in this structure, and to bring together all the material for comparison in attempting to estimate the systematic position of this extraordinary genus.

The Tarstis (PL XXII, Fig. 18) unites a condition of primitiveness with highly advanced characters. The calcaneum is thoroughly pecoran in shape ; the tuber is quite elongate (more so than in Moschus, less than in Cerviis or Gelocus), compressed, with nearly parallel dorsal and plantar borders, tapering less to the distal end than in the musk-deer, and with less definitely marked tendinal sulcus on the free end. The sustentaculum is very prominent and at once distinguishes this calcaneum from that of any of the oreodonts. The fibular facet is not so

350 SCOTT. [Vol. XI.

much elevated as in Mosckns or Cervns, but is longer anteroposteriorly ; the inner surface of this prominence bears a facet for the astragalus. The distal astragalar facet forms a broad, flat band, which is connected with the cuboid facet. The latter surface is relatively broader than in the Pecora, but has a much greater dorso-plantar extent than in MoscJuis, the calcaneum not being so suddenly constricted distally as in that genus.

The astragalus is higher and narrower than in recent Pecora, a primitive character which is repeated in such genera as Gelocus, PcebrotJieruim, etc. The proximal trochlea is widely and deeply grooved, and the external condyle is slightly higher and thicker than the internal, and is separated from the corresponding distal surface by a much wider interval than in the modern deer ; the sustentacular facet is also relatively narrower than in those animals. The distal astragalar trochlea is not only higher and narrower than in the existing Pecora, but is also differently proportioned. The cuboidal surface, in the first place, is distinctly narrower in relation to that for the navicular. In the second place, the junction of the two facets is marked by quite a prominent angulation, while in the modern forms this is a low, rounded swelling. Poebrotheriiim again agrees with Protoceras in this respect, though the cuboidal facet is broader in the former.

There would appear to be considerable variation with regard to the coossification of the various tarsal elements. Osborn and Wortman say : " In our young specimen of Protoceras the cuboid and navicular are perfectly free, but in the adult specimen there is some bony union. The line of junction, however, is clearly indicated by a more or less open suture. What is here said of the cuboid and navicular also applies to the cuboid and ectocuneiform, so far at least as the union of the latter with the cuboid is concerned. There appears to be no tendency to bony union of the ectocuneiform with the navicular." In the Princeton collection are feet belonging to individuals which are not only adult but aged, and in none of them is there any ankylosis of the cuboid with the navicular or of the ectocuneiform with either.


Like the astragalus, the cuboid is higher than in the recent Pecora ; the calcaneal surface is relatively wider, the astragalar facet narrower than in those animals. The calcaneal facet is wider than the corresponding distal portion of the bone and hence forms an overhanging ledge. The astragalar surface is narrow and simply concave in the dorso-plantar direction ; shortly behind the points where the two facets join they are narrowed by a circular sulcus and another invades the astragalar facets of both cuboid and navicular. The principal diameter of the cuboid is the dorso-plantar, due partly to the large size of the posterior hook, which is more or less rudimentary in the modern Pecora. Of the inner or tibial side of the cuboid about one half the height is occupied by the navicular, which is supported upon two narrow, ledge-like projections, the posterior one of which is considerably more prominent than the anterior. Below the ledge on the dorsal side is a flat facet for the ectocuneiform and there is an additional facet for this bone in the middle of the cuboid. The distal surface is occupied by the large and somewhat irregular facet for the fourth metatarsal, which is convex in front and concave behind. The facet for the fifth metatarsal is very small and entirely lateral in position, much as in Pcsbrotherium. An additional facet for the plantar projection of mt. IV is formed on the inferointernal side of the cuboid hook.

The navicular is higher than in the recent Pecora, but not otherwise notably different. The astragalar surface is concave, but has a low, rounded ridge near the fibular border for the groove on the distal trochlea of the astragalus. On the proximal dorsal margin this ridge forms an elevation which is less distinctly marked than in most of the existing Pecora. The portion of the circular sulcus above mentioned as invading the approximate astragalar surfaces of the cuboid and navicular, which affects the latter, forms a channel or groove along the whole fibular side of the bone. On the same side are two well defined facets for the cuboid, of which that on the plantar side is the larger and presents more inferiorly. On the distal side are two very distinctly separated facets for the cuneiforms. The anterior one, which is for the coalesced ecto- and meso

352 SCOTT. [Vol. XI.

cuneiforms, is L-shapcd, occupying the dorsal and tibial sides of the bone. The surface for the entocuneiform is relatively very large in the dorso-plantar direction, but very narrow transversely ; it is placed on a downward projection from the distal face of the navicular and stands at a considerably lower level than the facet for the compound bone. The posterior hook from the plantar side of the navicular, which is so conspicuous in the oreodonts, is absent in Protoceras.

As in nearly all known selenodonts, recent or fossil, the mesoand ectocuneiforms are indistinguishably fused together. This occurs even in the oldest known American artiodactyl, Trigonolestes, of the Wasatch. The compound element thus formed is nearly as broad as the cuboid, and of much greater vertical height relatively than in the recent Pecora ; it articulates with the second and third metatarsals.

The entocuneiform is much larger proportionately than in the recent Pecora or Tragulina ; its principal dimension is the vertical one and it forms a high, deep, and compressed plate, which has numerous connections. Proximally it articulates with the navicular and distally with the head of the second metatarsal, also with the projection from the plantar side of the third. In the Pecora this latter articulation does not exist.

The Metatarstis (PI. XXII, Fig. i8) consists of four elements, two of which (mt. II and V) are splint bones and two (III and IV) are large functional digits. In one of the specimens described by Osborn and Wortman the second metatarsal is more than one-third the length of the functional pair, but ordinarily it is much shorter and forms a narrow compressed splint, tapering to a point inferiorly. It articulates with the mesocuneiform portion of the compound element and by a long truncated surface with the entocuneiform.

The third metatarsal is very much larger, longer, and heavier than the corresponding metacarpal. The head bears a large, somewhat concave facet for the compound cuneiform ; the posterior hook-like projection, which in existing Pecora has become rudimentary, is very large and prominent and bears an oblique facet for the entocuneiform, as is also the case in Oreodofi. On the tibial side of the head is a deep fossa in


which the proximal part of mt. II lies and on the fibular side is a narrow, concave facet into which a projection from mt. IV is received. The shaft is stout, slightly contracted in the middle ; proximally it is laterally compressed and has its greatest diameter in the dorso-plantar direction, while distally it is broadened and flattened and its longest diameter is transverse. This is in accordance with the shape of the metatarsal in the Pecora, in which the hind cannon-bone is laterally compressed in its proximal portion, while the fore cannon-bone is compressed from before backward, but in Protoceras the difference is not so distinctly marked. The contact surfaces of the median metatarsals are flattened, allowing the two to be very closely approximated ; the plantar side is also flattened, but the dorsal and apposite surfaces form one continuous curve. The distal trochlea is strongly convex and is demarcated from the shaft by a shallow pit ; the carina, as in the metacarpal, is confined to the plantar side, but is a little more strongly developed than in the manus.

The fourth metatarsal is the counterpart of the third and exceeds it but very slightly in length ; it projects somewhat below the distal end of mt. Ill, but this is nearly compensated by the fact that the head of the latter rises a little above that of mt. IV. The plantar hook is very prominent and bears a facet for the one upon the hook of the cuboid. The fibular side is excavated for the head of mt. V, and there is a minute articular surface for that bone ; the fossa, however, is not nearly so deep and wide as that on mt. Ill, which receives mt. II. The articulation with mt. Ill is by means of a small but prominent process on the dorsal margin of the proximal end, which fits into a corresponding depression on mt. Ill, and also by flat surfaces on the approximate sides of the plantar hooks. According to Osborn and Wortman, some specimens show a tendency to the ankylosis of the median metatarsals; but none of those in the Princeton collection, even of aged individuals, exhibit any traces of such a process, and it would appear to be very exceptional.

The fifth metatarsal is a short, tapering splint ; it articulates with a small facet on the fibular side of the cuboid and with

354 SCOTT. [Vol. XI.

another and still smaller one on the head of mt. IV. No traces of any distal portions of the metatarsals of the lateral digits, such as Kowalevsky figures for Gelocus, have yet been found, nor is there any reason to believe that such existed. Indeed, Kowalevsky does not explicitly state whether they are figured in the pes on any better evidence than the analogy of the manus.

The Phalanges (PI. XXII, Figs. i8, 20), so far as is known, are confined to the median digits, and in all probability, there were no dew-claws. The phalanges of the pes are very much larger than those of the manus and differ from them in several details of structure, indeed, to quite an unusual degree. Aside from its great increase in every dimension, the proximal phalanx is like the corresponding bone in the manus ; the distal trochlea is somewhat more deeply notched in the median line and has a relatively greater dorso-plantar diameter, while on the plantar side the trochlea is more prominent and continued farther proximally upon the shaft.

The second phalanx is not only actually, but also proportionately, much longer than the corresponding anterior phalanx, and is of quite a different shape, being laterally compressed and having its principal diameter the dorso-plantar instead of the transverse. The proximal articular surface is more deeply concave, more distinctly divided into two cotyles by a median ridge, and the median elevation of the dorsal margin is more pronounced. The rugose prominences for ligamentous attachment are larger and more unsymmetrical, that on the external side (tibial side of digit III and fibular side of digit IV) being decidedly the more prominent. The distal trochlea describes an almost exact semicircle and is extended more proximally upon the dorsal side than is the case in the manus.

The ungual phalanx is longer and straightcr than the anterior one, its outer border being less curved, and the distal end is more obtusely pointed ; the plantar face is more concave, with more elevated borders, and the rugosity on the plantar side beneath the articular facet is much more prominent. The facet for the second phalanx is higher, narrower, and more symmetrical, in that its two divisions are more nearly of


the same size. The dorsal border of the facet is prolonged in the median line into a large overhanging hook, which is but slightly indicated in the anterior ungual. This hook corresponds to the greater dorso-plantar diameter of the distal trochlea of the second phalanx and its greater prolongation up the dorsal face of the shaft in the pes than in the manus.

As will be seen on comparison of the tables of measurements, there is not in MoscJms such a difference between the anterior and posterior phalanges as obtains in Protoceras. Those of the pes are of nearly the same size in both genera, though slightly larger in Protoceras, while those of the manus are considerably larger in MoscJms. In structure also the posterior phalanges agree better in the two animals, the differences being merely in matters of minute detail, except, of course, the grooves for the metatarsal keels.


Protoceras. Moschus.

Ilium, length of ventral border (fr. acetabulum) . . o.ii6 0.069

Ilium, depth of neck 023 .011

Acetabulum, fore and aft diameter 024 .020

Acetabulum, vertical diameter 020 .019

Femur, length 177 .164

Femur, breadth of proximal end 042 .035

Femur, breadth of distal end 041 .032

Patella, height 036 .025

Patella, breadth 024 .013

Tibia, breadth of proximal end 039 .035

Tibia, thickness of proximal end 046 .037

Calcaneum, length 070 .051

Astragalus, length 030 .022

Astragalus, breadth of proximal trochlea 016 .015

Cuboid, height 016 .008

Cuboid, breadth of distal end on .008

Metatarsal III, length 107 .131

Metatarsal III, breadth of proximal end 012 .007

Metatarsal III, breadth of distal end 013 .009

Metatarsal IV, length 107 .134

Metatarsal IV, breadth of proximal end on .008

Metatarsal IV, breadth of distal end 012 .009

First phalanx, III digit, length 035 .031

First phalanx. III digit, breadth of proximal end . . .013 .010

First phalanx. III digit, thickness of distal end . . .009 .009

Second phalanx, III digit, length 022 .023









.U^ j





356 SCOTT. [Vol. XI.

Second phalanx, III digit, breadth of proximal end Second phalanx, III digit, thickness of distal end

Third phalanx, III digit, length

Third phalanx, III digit, breadth of proximal end Third phalanx. III digit, thickness of proximal end

VII. Restoration (PI. XX).

The general aspect of the skeleton, as a whole, resembles that of the musk-deer. The head is proportionately much longer, a character frequently found in the ancient mammals, as compared with their recent representatives, though in this case the elongation principally affects the facial region, which is not an ancient but a modern character. In the living animal the appearance of the head must have been entirely different from that of MoscJms; even the female doubtless had the proboscidiform muzzle, which among the recent ruminants is found only in the saiga antelope and to a less extent in the moose {Alces). The difference is, of course, exaggerated in the case of the male, whose bizarre skull is not to be compared with that of any existing mammal whatever.

The spinal column, in length, weight, curvature, and in the proportions of the various regions, is very similar indeed to that of MoscJms ; the neck is heavier and actually longer, though much shorter as compared with the length of the skull ; the trunk is of about the same length and the thorax deeper and more capacious. The lumbar region has broader transverse processes, but more delicate spines, which curve more decidedly forward, and more prominent metapophyses, all of which are traguline features and perhaps indicate a more pronounced curvature of this region of the back than in the musks. There can hardly be any doubt that Pi'otoceras had a distinctly longer and better developed tail than most recent deer.

The inequality in the length of the fore and hind limbs is very nearly the same as that which is to be observed in Mosckus, but the proportions of the different limb-segments are not similar and the individual bones are heavier and stronger. Thus, the scapula and the bones of the fore-arm are considerably longer than in the existing animal and the carpus is much


higher ; but, on the other hand, the metacarpus and anterior phalanges are very much shorter and the humerus is of similar length in the two genera. In the hind limb similar facts are observable ; the pelvis, femur, tibia, and tarsus are all decidedly longer than in the musk-deer, while the metatarsals are much shorter. The phalanges of the pes are of nearly the same length in both genera. As a whole, the limbs are longer in Protoceras, the shortness of the feet not compensating for the greater length of the other segments of the limbs.

The skeleton of Leptomeryx follows that of the tragulines more closely than it does that of Protoceras, both in regard to the actual size of the body and in the relative length of the limbs and consequent curvature of the back. The inequality of the limbs and curvature of the spinal column are, however, decidedly less than in Tragiihis. Aside from the great difference of stature and the still greater divergence in the appearance and character of the skull, there is an undeniable resemblance between the skeletons of Leptomeryx and Protoceras, though that between the latter and MoscJms is still closer.

VIII. TJie Systematic Position of Protoceras.

The osteology of this genus is now almost as completely known as that of any living mammal, and yet the determination of its affinities is a very obscure and difficult problem, and it is therefore hardly a matter of surprise that there should be much difference of opinion with regard to it. Marsh cautiously infers from an examination of the skull alone that it was connected with the giraffe. "The characters now known suggest affinities with the giraffes, but indicate a distinct family." (No. 5, p. 82.) Zittel makes the genus the type of a subdivision of his family "Cervicornia," and says of it : "Die Gattung Protoceras bildet ein hochst merkwurdiges Bindeglied zwischen Tragulina und Cervicornia. Gebiss und Extremitaten stimmen mehr mit den ersteren iiberein, wahrend sich der Schadel am besten mit den Giraffen und Sivatheriden vergleichen lasst." (No. II, p. 407.) Osborn and Wortman express no very decided opinion as to the relationships of Protoceras, leaving the

358 SCOTT. [Vol. XI.

question rather an open one. " If now we compare Protoceras with any family of the Pecora, there are so many striking differences at once apparent that we are compelled to conclude that there are no marked affinities in the direction of any of these families. In the possession of bony protuberances on the parietals, which are probably processes of this bone, and not developed separately as in the Giraffe, in the general architecture of the skull, together with so many primitive characters of the feet, this genus apparently occupies a distinct position and cannot be consistently referred to either the Tragulina or the Pecora as at present constituted and defined. The possession of multiple horns suggests the possible relationship of this family to the Sivatheriidse, but the likeness does not extend to other features of the skull." "That it \i.e., Protocej-as] represents a distinct family there can be little doubt. Of its successors we know nothing whatever, and our ignorance is equally great in the matter of its ancestry." (No. 8, p. 369.)

Flower has well stated the difficulty of determining the relationships of artiodactyl groups, the ancestry of which can only be conjectured. " The Pecora or true Ruminants form, as has often been remarked, an extremely homogeneous group, one of the best defined and closely united of any of the Mammalia. But though the original or common type has never been departed from in essentials, variation has been very active among them within certain limits, and the great difficulty of subdividing them into natural groups (*the despair of zoologists,' as Pucheron calls it) arises from the fact that the changes in different organs (feet, skull, frontal appendages, teeth, cutaneous glands, etc.) have proceeded with such apparent irregularity and absence of correlation, that the various modifications of these parts are most variously combined in different members of the group." (No. 2, p. 181.) All this applies almost equally well to the Artiodactyla as a whole, and the mutual relationships of the various subdivisions which compose that order. This difficulty proceeds from the frequent impossibility of determining what points of resemblance between the groups to be compared are due to inheritance from a common ancestor, and what are cases of parallel development. Such determination


can be made with certainty only when the phyletic series has been worked out, and here, as elsewhere, any classification which is made without knowledge of the various phyletic steps, can be only temporary and tentative.

In the case of Protoceras we may hope, with the complete information as to its structure which we possess, to establish its relationship to the Pecora as a whole with reasonable probability, the main outlines of the latter's phylogeny having been fairly well determined. We shall need to remember Osborn's dictum in this discussion, that " no case of exact parallelism in both teeth and feet between two unrelated types has yet been found, or is likely to be." (No. 7, p. 383.) If the European palaeontologists, beginning with Kowalevsky, are right, then the ancestral form of the modern Pecora is the Oligocene genus, Gelocus, and no fact is known which in any way impugns the probability of this conclusion. The first step in our inquiry must therefore be to institute a comparison between Geloc7is and Protoceras. If the latter be referable to the Pecora at all, it must be derived from Gelocus, which there is good reason to regard as ancestral to that entire group.

So far as the dentition is concerned, there is no very essential difference between the two genera, but the European form has molars in a less advanced stage of development, as is to be seen in the thicker, more conical, and less completely crescentic form of the lobes. The premolars are likewise less differentiated. The upper canine is compressed and bladelike, while in Protoceras it is trihedral and opposes the first lower premolar. The skull of the American genus is in many respects much more modernized than that of Gelocus, though the latter is only imperfectly known. Kowalevsky says of it : " Leider habe ich in alien untersuchten Sammlungen keinen completen Schadel finden konnen. Indess aus verschiedenen Bruchstucken des Schadels geht unzweifelhaft hervor, dass Gelocus weder geweihartige Auswiichse, noch Horner auf den Stirnbeinen besass. Dieselben Bruchstiicke haben gezeigt, dass die eigentliche Hirnkapsel nicht so weit nach hinten verdrangt war, wie es bei den heutigen Wiederkauern der Fall ist, sondern eine mehr normale Stellung einnahm, in der Weise,

360 SCOTT. [Vol. XI.

class der vordere Orbitalrand sich genau dem erstcn Molar Sfecfeniiber befand, wahrend bei dem grossten Theil der recenten Wiederkauer die Hirnkapsel, in Folge der starken Entwickelung des Gesichtstheils, so weit nach hinten verschoben erscheint, dass der vordere Orbitalrand dem letzten Molar gegenliber, Oder selbst hinter diesen zu stehen kommt. Der Schadel hatte eine gewisse Aehnlichkeit mit dem unserer heutigen Traguliden, mit denen Gelocus uberhaupt viele gemeinsame Merkmale besitzt." (No. 3, p. 147.)

In sharp contrast with the primitive traguline skull of Gelocus is the advanced and highly differentiated skull of Protoceras, which in some respects is more modernized than that of the deer. The short, capacious, and rounded cranium, the posterior position of the orbits, the great elongation of the face and its depression upon the basicranial axis are characters which at present are confined to the higher Pecora. At the same time, the shape of the occiput, the arrangement of the supraoccipital, the prominence of the sagittal and lambdoidal crests are primitive features, such as are now not to be found in even the lowest Pecora, e.g., MoscJms. The specializations peculiar to this skull form still a third class of characters, such as the extreme shortening of the nasals and the numerous protuberances which are developed in the male, especially those which arise from the parietals and maxillaries. The skull of Protoceras presents, then, a remarkable assemblage of characters, some few even more primitive than are found in the tragulines, some altogether peculiar to itself, but most of them extremely modernized and advanced, and elsewhere among selenodonts combined only in the higher Pecora. Though an examination of the skull alone might lead one to refer this genus to the Pecora, yet there would be much reason to hesitate in doing so. The very combination of such primitive and modern characters is not what we should expect in a transition form ; in such a form we should find an association of features like those of Traguhis and Gelocus, on the one hand, and of the lower Pecora, such as Moschus, on the other. Then, too, the existence of horn-like protuberances on the parietals is not suggestive of relationship with the Pecora, since no member


of that group has any such protuberances, their horns and antlers always rising from the frontals.

The other parts of the skeleton bear testimony of similar import. The atlas and axis do not differ in any important respect from those of Geiocus, so far as the latter are known. In particular, the odontoid process has attained the same stage in the development of the spout-like shape. The remaining cervical and the thoracic vertebrae are very similar to those of the musk-deer, while the lumbars are more traguline.

The limbs are in almost every res'pect far more primitive than those of Geiocus. The scapula, however, has a more modernized form than in that genus, and except for its more prominent and recurved spine, is a copy of that of the muskdeer ; while in Geiocus it is wider than in either the recent tragulines or true ruminants. The fore-arm bones are decidedly less advanced than in Geiocus. In the latter the groove on the proximal end of the radius for the intercondylar ridge of the humerus is much narrower, the distal facets for the carpus are much more oblique in position, and there is an articulation with the cuneiform. The shaft of the ulna is much more reduced, and its distal end no longer covers the entire cuneiform. The carpus of Geiocus shows a great advance over that of Pj'otoceras, both in its reduced vertical height and in the ankylosis of the trapezoid and magnum. In the metacarpus the same advance is visible in the reduction of the lateral digits to splint-bones, which are interrupted in the middle, in the exclusion of the second metacarpal from contact with the magnum, and the articulation of the third with the trapezoid. In all these respects Protoceras is very much more primitive. Filhol adds another modernization of Gelociis, viz., the frequent coossification of the proximal parts of the lateral metacarpals with the median pair, although the latter are not themselves ankylosed to form a cannon-bone.

Femur, tibia, and fibula show no important differences in the two genera, but the pes of Geiocus is much more modernized. This advance consists in the coossification of the cuboid and navicular and the coalescence of the third and fourth metatarsals into a cannon-bone, as well as in the reduced

362 SCOTT. [Vol. XI.

height of the distal tarsals. If Kowalevsky's figure of the pes be correct, Geloctis probably is behind Protoceras in the retention of the distal portion of the lateral metatarsals. However, neither this writer nor Filhol speaks of finding these portions in situ, and, if their association with the pes is conjectural, no great stress can be laid upon this character.

The result of this comparison, then, is that in regard to the structure of the skull, and to a less degree the dentition, Protoceras is far in advance of Gelociis, while the differentiation of the limbs, and especially of the feet, lags as far behind. This association of characters confronts us with very clearly defined alternatives, when we attempt to solve the problem of the relationship of Protoceras to the Pecora. Assuming, as we have every justification in doing, that Gelociis is an ancestral form of the Pecora, then, either Protoceras is not descended from Gelociis at all, and its likenesses to the Pecora are not due to genetic affinity, but have been independently acquired ; or, on the other hand, Protoceras is so descended, its resemblances to the true ruminants are the expression of real relationship and the primitive structure of the limbs and feet has been reacquired, whether by reversion or otherwise. Of these alternatives the former is by far the more probable, (i) Without at all denying the possibility of such reacquisition of primitive characters, yet no plausible reason can be assigned for assuming it, and no case is known, among mammals at least, in which such a mode of development has been rendered in the smallest degree likely. (2) On the other hand, very many cases of the independent acquisition of similar structures have been pretty clearly demonstrated. It will suffice to mention the many different groups which have independently developed the spout-shaped odontoid process of the axis, or the tetraselenodont molar, or the humerus of the horse and camel. It may, perhaps, be objected that these are single characters, whereas the skull of Protoceras displays a whole series of such resemblances to the Pecora, but even such cases of parallelism are by no means unknown. There are many close correspondences in the skull-structure (and limb-structure as well) between the camels and the true ruminants, and yet the sue


cessive genera of these two series show that these resemblances are not due to a common ancestry, since there can hardly have been an ancestor common to both series later than Dichobnne, or some similar form. Even more remote is the connection between the true ruminants and the true swine {Suid(z), yet the latter have acquired several of those very characters which give such a strong pecoran stamp to the skull of ProtoceraSy for example, the backward shifting of the orbits and the depression of the face upon the basicranial axis. In the loss of the sagittal crest the swine-skull is more modernized than that of Protoceras. Among the oreodonts, Merycochcerus has the face bent down upon the cranial axis, and the orbits are shifted much farther back than in any other member of that family. Such resemblances are obviously, therefore, not sufficient in themselves to create any strong presumption of affinity. The foot-structure, by keeping on such a primitive plane, reveals the true nature of these skullcharacters. As Osborn has said, we know of no case where teeth, skull, and feet have converged to a common type from different starting-points, but that one of these may display such convergence or parallelism in several different respects is not at all uncommon.

(3) Even should we go so far as altogether to exclude Gelocus from the pecoran ancestry, the difficulty of accounting for the peculiar assemblage of characters found in the skull of Protoceras on any other hypothesis than that of the independent acquisition of the pecoran features, is not at all diminished. For it must be remembered that these likenesses are to the higher members of the group, and on the same grounds such typical members of the series as Prodremotherium, Dreniotheriiim, AmpJiitragidits and Palcsomeryx would have to be excluded from the line.

It is altogether probable, then, that Protoceras has but a remote connection with the Pecora, and consequently that the affinities with different families in that group which have been suggested, are illusory. Nor can it properly be called a connecting link between the tragulines and the deer, for, having more primitive feet than the former group, it cannot well be

364 SCOTT. [Vol. XI.

descended from it, and the skull being in some respects more advanced than that of the deer, Protoceras can hardly be ancestral to that family.

It is not at all likely that the Tragulina are descended from Gelociis, the latter having a more advanced type of foot-structure and an odontoid process of the axis which has already lost the conical shape. The connection with the true ruminants was, therefore, in all probability by means of some form as yet unknown, which was rather more generalized than Gelociis. With the line terminating in the existing tragulines, which, so far as we know, have always been confined to the Old World, Protoceras can have but a remote connection, but just when and how this connection was established cannot at present be determined. There are, however, certain American genera which are usually referred to the tragulines and which probably are more or less distantly related to that group. These genera are Leptoineryx and Hypertragulus ; they are much alike, and yet with such significant differences as show that they are to be regarded as divergent branches of the same stock. One is tempted on zoogeographical grounds to assume a relationship to these genera on the part of Protoceras. Leptomeryx has an entirely different type of skull from that of Protoceras and one which closely resembles the structure of the traguline skull, differing merely in the concavity or flatness of the occiput, the small size of the auditory bullae and their freedom from cancellous tissue, and in the presence of a fontanelle or vacuity between the frontal, lachrymal, and nasal. The odontoid process of the axis is conical. In the dentition, however, even in details, there is much likeness between this genus and Protoceras. In the character of the limbs Leptomeryx is in many respects in advance of Protoceras ; thus, in the manus the magnum and trapezoid are ankylosed and the second metacarpal is excluded from the magnum. But the scapula, ulna and radius, metacarpals and phalanges, are very much alike in the two genera. Pelvis, femur, tibia, and fibula do not differ in any important respect, but Leptomeryx has a posterior cannon-bone and coossified cuboid and navicular. In spite of the many differences, there is an undeniable likeness of habit between Leptomeryx and Protoceras.


In Hypertragnlus we find certain characteristics peculiar to itself, such as the retention by the lower canine of its original form and function, the loss of the first and isolation by diastemata of the second lower premolar, and the coossification of the ulna and radius, but the general resemblance to Leptomeryx is close. The structure of the cranium and position of the orbits are the same in both, but the elongated, constricted, and slender muzzle, the large and irregular fontanelles which encroach upon the nasals, the character of the palate, the shape and position of the posterior nares, the aspect of the base of the skull, with auditory bullae and glenoid cavities, are all suggestively like what we find in Pr-otoceras. The premolars have the same simple structure found in the latter, but are not so much elongated antero-posteriorly. The pes is like that of Protoceras, except for the coossification of cuboid and navicular. In size Hypertraguhis somewhat exceeds Leptomeryx, but is much inferior to Protocet'as. Still another genus of apparently this same group is the curious little Hypisodus, from the White River, with its hypsodont molars and ten functional lower incisors, made up of the incisors proper, the canines, and first premolars. The distal end of the fibula is coossified with the tibia, and the feet, so far as known, resemble those of Lepto7Jteryx.

This family represents a group of White River selenodonts, each of whose genera has become more or less specialized in a way peculiar to itself, and with a tendency to simulate the Pecora in some respect or other, yet always retaining a number of primitive features. I cannot but believe that Protoceras represents a divergent offshoot of the same stock which, retaining in most respects the foot-structure belonging to the common ancestor of all these genera, has, at the same time, wonderfully paralleled the higher Pecora in many features of the skull. We have yet to find the forerunners of this genus in the two lower divisions of the White River formation, the Oreodon and TitanotJierumi beds, but one of these forerunners may prove to be the problematical genus Stibarits. The Uinta formation may be expected to yield the ancestor common to the entire group, and when it is found we shall prob



[Vol. XI.

ably discover in it the starting-point both of the Pecora and of the Tragulina. The manus of Protoceras appears to have been but Httle modified from that of this hypothetical form, and some such type of manus might easily give rise to those of all the later representatives of these various selenodont lines. The hypothetical genus will doubtless also be found in








the Old World, for nothing seems clearer than that both Pecora and Tragulina originated in that region, and the latter group, as narrowly defined, has always been restricted to it. From this supposed Uinta genus lead no less than four divergent lines in the White River. Leptomeryx has kept nearly the primitive type of skull, but has developed somewhat complex premolars, with some advance in the structure of the manus and still more in that of the pes. Hypertragulus has a somewhat modified skull, but very primitive dentition, with little change in the extremities beyond the coossification of the ulna and radius, and of the cuboid and navicular. Hypis


odus has developed a hypsodont type of molar and curious lower incisors, while in Protoceras the principal modifications have been those of the skull.

Until the forerunners of Protoceras have been found, these conclusions can be only conjectural, but such a solution of the problem offers fewer difficulties than any other now available. The table on the preceding page displays the position of the genus according to this view.

Geological Museum, Princeton, N. J., July 26, 1894.

368 SCOTT.


Brooke, Sir Victor. On Hydropotes inertnis and its cranial

characters as compared with those of Moschus moschiferus. Proc.

Zoological Society of Lofidoti, 1872, p. 522. Flower, Sir W. H. On the structure and affinities of tlie Musk Deer. Ibid. 1875, p. 159. KowALEVSKY, W. Osteologic des Gelocus Aymardi, Palaeonto graphica, Bd. XXIV, p. 145. Krueg, J. Ueber die Furchen der Grosshirnrinde der Ungulaten.

Zeitschr.f. wiss. Zool.^ Bd. XXXI, p. 297. Marsh, O. C. Horned Artiodactyle {Protoceras celer) from the

Miocene. Ainer. Joiirn. Sci. and Arts, 3d ser., Vol. XLI, p. 81. Marsh, O. C. Description of Miocene Mammalia. Ibid., Vol.

XLVI, p. 407. Osborn, H. F. Rise of the Mammalia in North America. Ibid.,

p. 379Osborn, H. F., and Wortman, J. L. Characters of Protoceras

(Marsh), etc. Bull. American Miiseiim of Natural History,

Vol. IV, p. 351.

9. SCHLOSSER, M. Beitrage zur Kenntniss der Stammesgeschichte der

Hufthiere. Morphologisches Jahrbuch, Bd. XII, p. i.

10. ScoTT, W. B. Beitrage zur Kenntniss der Oreodontidse. Ibid.^

Bd. XVI, p. 319.

11. ZiTTEL, K. V. Handbuch der Palseontologie. I Abth., Bd. IV.

Munich, 1891-1893.




Restoration of Protoceras celer, one- third natural size. The skull is taken from Osborn and Wortman's figure of the male, drawn to the proportionate length of the female skull which belongs to the skeleton. The secondary sexual characters being principally confined to the head, there is nothing misleading in this association of the sexes in one figure, especially in one of such small scale. The remainder of the skeleton belongs to a single individual, with the exception of the humerus, atlas, 12th thoracic and ist caudal vertebrae, which are supplied from other specimens.

Joiinial of Morpholoqii^ Vol XL


lilhAnst yW'rrKr tlVinUr: FrankfarlfM

372 SCOTT.


(All the figures, except Nos. 5 and 12, are two-thirds natural size.)

Fig. I. Skull of immature female, from the side, /'a, parietal ; /'a', parietal protuberance ; F, frontal ; Z, lachrymal ; Na, nasal ; P^, third upper premolar ; D 4, fourth upper milk molar ; Mi, first upper molar.

Fig. 2. Skull of adult female, base-view. Bs, basisphenoid ; Ty, tympanic ; C, Pi, alveoli of canine and first premolar.

Fig. 3. Skull of female, rear view ; same specimen as Fig. i. Poc, paroccipital process ; Pa', parietal protuberance.

Fig. 4. Inferior dentition, crown view. C, Pi, alveoli of canine and first premolar; Mi, first molar.

Fig. 5. Superior milk dentition, crown view. D 2, D3, D 4., second, third, and fourth milk molars ; Mi, first true molar. Natural size.

Fig. 6. Atlas, from above. Sn, foramen for first spinal nerve.

Fig. 7. Axis, side view.

Fig. 8. Sixth (?) thoracic vertebra, from the side.

Fig. 9. Third lumbar vertebra, from above. The centrum is concealed in the matrix, met, metapophysis ; tr, transverse process ; pzy, postzygapophysis.

Fig. 10. Scapula of right side.

Fig. i I . Right ulna and radius, seen from the external side. R, radius ; U, ulna.

Fig. 12. Outline of distal end of ulna and radius. Natural size. ^, radius ; U, ulna ; S', U, C, facets for the scaphoid, lunar, and cuneiform.

JournM of Morphology. Vol. XI.


IJh Ami yWerntrik^inter, Frankfar0M

374 SCOTT.


(All the figures, except Nos. 14, 19, and 20, are two-thirds natural size.)

Fig. 13. Right femur, front view.

Fig. 14. Right femur, proximal end. Natural size. TV 2, second trochanter.

Fig. 15. Left patella, from behind. 7«. Z';', process for mesial face of femur.

Fig. 16. Right tibia and fibula. F, fibula.

Fig. 17. Left manus, //, ///, IV, V, second, third, fourth, and fifth digits. S, scaphoid ; Z, lunar ; Cn, cuneiform ; Td, trapezoid ; M, magnum ; U, unciform.

Fig. 18. Right pes, ///, IV, third and fourth digits. The divergence of the distal ends of the metatarsals is not normal, but due to a distortion of the specimen.

Fig. 19. Phalanges of third digit of left manus, seen from the radial side. Natural size.

Fig. 20. Phalanges of third digit of left pes, seen from the tibial side. Natural size.

J our Tied of Morphology. Vol. XL

Pi xxir.

R. Weber dd

liik-Arist. vWerntr S^Winter, Frankfarf/M.


ALVIN DAVISON, M.A. Fellow in Biology, Princeton College, N. J., U.S.A.

The anatomy of the adult skull of this eel-like Amphibian has been well described by Wiedersheim (i). Osborn (8) has given a most excellent description of the central nervous system. Hay (2) has given an interesting account of the finding of its eggs in an Arkansas swamp in September of 1887, and subsequently published the results of his study of the embryos contained in these eggs, and also of the skull of a small specimen six inches long. Cope has furnished a general description of the species, and endeavored to trace out its phylogeny by means of the insufficient data at hand. Kingsley (5), through a study of Hay's embryos, for the most part confirmeyd the latter's account, and also added to our sparse knowledge some important points, especially along the line of phylogeny. The writer has published a description of the conical arrangement of the muscles (6), and in a later article the manner of the fertilization of the eggs (7). No other investigations of any importance than the above mentioned have been made on this -peculiar Amphibian, owing undoubtedly to the fact that the embryological material is so difficult to secure. The animals will not breed in captivity, and the batches of eggs laid are so few and so well concealed as to escape the sharp eye of the naturalists. Since the eggs are fertilized internally, a comparatively short period of external incubation is necessary, thereby limiting the probability of being discovered. Previous to February, 1894, no one had secured a specimen of the young less than six inches in length. In that month I was so

1 I was very materially aided in making these investigations in the Princeton Biological Laboratory by Mr. C. W. F. McClure who placed at my disposal a large number of Amphibians and gave me numerous valuable suggestions.

276 DAVISON. [Vol. XI.

fortunate as to have sent me, by a North Carolina dealer in embryological supplies, several very young specimens varying in size from seventy-eight to ninety millimetres in length.

In this paper it is my intention to give a detailed account of the anatomy of these very young specimens, and by comparisons with the adult structures as well as with other Amphibians to deduce a few new points in phylogeny. Since there has never been published any complete and reliable account of the anatomy of the adult, it will be necessary for me to begin at this point.

External Features of the Adult.

The general form is serpentine, having the same proportions as the body of a teleost eel. The circumference of the largest specimen I have seen was 1 50 mm., and its length almost one metre. The tail occupies about one fourth the length of the animal, and is laterally compressed in such a manner that the area of a cross-section is triangular with the apex at the dorsum. The body is slightly contracted just anterior to the fore limbs forming the so-called neck. The head is twice as long as it is broad, and vertically compressed. The snout is obtusely rounded, but is more pointed than in other urodeles, and extends from six to eight millimetres beyond the lower jaw. The lips of the upper jaw overhang for the most part those of the lower, thus preventing the mud from entering its mouth during its subterranean excursions. The eyes are one and a half millimetres in diameter, have no lids, utilize the epidermis as a cornea, and are situated along a transverse line cutting off the anterior third of the head. The anterior limbs are from fifteen to twenty millimetres in length, and support two or three diminutive digits. The posterior limbs, situated just anterior to the vent, are from fifteen to twenty-five millimetres long, and also have two or three digits. Cope (4) says that specimens have been found having two digits on the anterior limb and three on the posterior.

A few millimetres anterior to the fore limb is the branchial fissure, securely guarded by two membranous flaps. The skin is smooth, silky, and of a dark brown color dorsally, and of a


slate gray ventrally. The head is covered with mucous pores, arranged in several rows which unite in the region of the neck so that only two distinct rows are seen on each lateral area of the body. Cope (4) errs in saying none are present on the body.

Bones of the Head.

The head of Amphiuma is long and narrow, the general outline being somewhat like that of the skull of the Proteus except that the snout is not so pointed. While in the majority of Amphibia the skull is as broad as long, in Amphiuma it is twice as long as it is broad. It is composed of twenty-eight distinct bones : two maxillaries, a premaxillary, two nasals, two frontals, two prefontals, two orbitosphenoids, two vomeropalatines, two parietals, two exoccipitals, two prootics, two pterygoids, a parasphenoid, two stapes, two quadrates, and two squamosals. The maxillary is an irregular, oblong bone with a triangular cavity beneath, along one side of which are attached twenty-two conical teeth. This bone is not curved as in the majority of amphibians, but presents a straight alveolar edge. The foramina (Fig. i, c, b) are seen on its dorsal surface, the anterior of which gives passage to small blood vessels and nerves. Cope (4) believes that the larger foramen is prophetic of the tentacular canal of the Gymnophiona. The maxillary articulates dorso-laterally with the prefrontal and nasal, ventrointernally with the vomero-palatine, and anteriorly with the premaxillary, which in Amphiuma is a single bone. This very irregular bone is composed of three parts : the alveolar portion (Fig. II, P) bearing ten teeth, the dorsal wedge extending backwards between the nasals and frontals to a line joining the orbits, and a ventral wedge lying in the roof of the mouth between the vomero-palatines and parasphenoids. The greatest width of either of these wedges is one millimetre, and the length of each is about one third that of the skull. The nasals are small bones very much pitted, and serve to roof in the anterior part of the nasal chamber. The frontals are the longest bones of the skull, and bound the fore brain both dorsally and anteriorly. They extend underneath the posterior edge of the

^yS DAVISON. [Vol. XI.

dorsal wedge of the premaxillary, and the union of the descending plates at this point functions as a nasal septum, as seen in Fig. III. Through this portion of the bone is the canal for the passage of the olfactory nerve. Hay (2) does not believe that such a canal exists, though Wiedersheim (i) has correctly described it. Cope (4) has called the descending union of the frontals the ethmoid. In this he is wrong, as embryonic investigations (3) have clearly shown, I think the Sarasins (9) have followed Cope in this error, and together with him have sought to show relationships which do not exist. The frontals are in contact with the nasals and premaxillaries anteriorly, and laterally join the prefrontals and orbitosphenoids. Upon reaching the frontal bone the sagittal crest of the parietals divides, each fork running to the outer side of the former bone. In the groove thus formed lies the temporal muscle. The prefrontals take part in roofing the nasal chambers, have a rough surface and an irregularly oblong outline. They join the maxillaries laterally and form part of the orbit. The orbitosphenoids are small, taking part in the lateral boundaries of the brain cavity. They articulate with the parietals, parasphenoid, and vomero-palatines. They are almost separated from the frontals by the interposition of a narrow wedge of the parasphenoids. The vomero-palatines bear the inner concentric row of teeth, which number about forty-four. TJie 7iuviber of premaxillo-maxillary teeth is never less than fifty. The number is wrongly stated by Cope as thirty-one. The vomero-palatines, together with the ventral wedge of the premaxillary, form the roof of the mouth. This pair of bones unite anteriorly, and are nowhere separated more than two millimetres from each other. Their backward extent ceases slightly posteriorly to the beginning of the parasphenoid. The parietals are the largest bones of the skull, and form the roof of the greater part of the brain cavity. Their median juncture is the sagittal crest. Their external borders are deflected upwards to form the temporal crests, thus giving rise to a broad groove for the reception of the powerful temporal and masseter muscles. The prootics and squamosals lie lateralwards of these bones, and posteriorly are joined by the exoccipitals which bear the two

No. 2.] AMPHIUMA MEANS. 2)79

pedestals for articulation with the axis. The exoccipitals join each other only in a very small part of their length, being separated by a V-shaped opening which is the foramen magnum. Immediately within this aperture, on either side, is seen a small facet for articulating with the prezygapophyses of the axis. The large foramina seen at the external sides of the bases of the pedestals give passage to the vagus and hypoglossal nerves. The exoccipitals are in apposition laterally with the squamosals and stapes. They do not reach the prootics, which lie anterior to the squamosals and external to the parletals, with which they join in such a manner as to form the anterior parts of the temporal crests.

The pterygoids are wing-like bones extending from the quadrates forward along the parasphenoid to near where the vomero-palatines arise, at which point the bone gives way to cartilage. The parasphenoid is the broad basal bone of the skull, extending throughout more than half its length. It is broadest in the otic region, and narrows in either direction. Its posterior part is bounded on three sides by the exoccipitals. Anteriorly, it extends beneath the ventral wedge of the premaxillary. The stapes is an orbicular bone, scarcely three millimetres in diameter, articulating with the parasphenoid, the exoccipital, and quadrate. It does not form a part of the suspensorium. The quadrate is a comparatively small bone, lying on the inner side of the descending squamosal, and joins the pterygoid, stapes, and parasphenoid. This bone, with the squamosal, forms the suspensory apparatus. It bears the facet for articulation with the mandible. The squamosal is another peculiar bone in the Amphiuma skull. Wheiras in most of the Urodeles it is directed forivards and slightly otit wards, in Amphitinia it is directed outwards and dowjiwards, and bnt very slightly forwards. It is an exceedingly strong bone, and is firmly adherent to the exoccipitals, parietal, and prootic above, and joins the quadrate and stapes beneath. The possession of this bone, according to Cope (4), allies the families Amphiumidae and Coecilidae. The bone which Cope has called squamosal in the Coecilians is quite differently located, being directed forwards and inwards in such a manner as to form

3S0 I>A VISON. [Vol. XI.

part of the orbit, and therefore deserves the name quadratojugal, as some authors have already called it.

The mandible is a simple structure, each ramus being composed of these bones, viz., the dentary, the angular, and articulatory. The dentary supports twenty-two teeth, and forms the whole external portion of the ramus, and appears for some distance in front on the inner side. The external surface presents a foramen for the mandibular nerve. The angular and articular are so well coossified for the most part that their boundaries cannot be clearly defined. Anterior to the facet for articulation with the quadrate, the articular abruptly rises into a prominence, to which the temporal muscle is attached. The angular lies beneath the articular, and forms a projection behind it, in which the digastric muscle is inserted. Immediately anterior to the condyle is a notch for the insertion of the masseter muscle. The rami are not anchylosed in front, but are held together by cartilage.

The hyoid apparatus oi AmpJiiunia is quite unlike that of any other Amphibian (Fig. XV). There is but one basibranchial to which are joined two ceratobranchials bound by cartilage. There are four epibranchials. The basihyal is cartilaginous, as are also all the epibranchials except the first. The hypohyals are very short, being scarcely more than one-half as long as the basibranchial. The ceratohyals are longer than the ceratobranchials. The few cartilaginous formations of the skull not already described will be discussed later.

The Limbs.

The limb bones of AmpJiiuma are characteristic of reptilia in some respects. The fore limb consists of a humerus, ulna, and radius, carpus, metacarpus, and phalanges. These bones are proportioned with reference to each other, as in mammalia. The humerus of a specimen one metre long is about one centimetre long. Its head is of cartilage. Immediately below the head, on the anterior side, is a prominence for the insertion of the biceps muscle. The distal third of the shaft is slightly flattened to afford a more advantageous surface for the articu

No. 2.] AMPHIUMA MEANS. 38 1

lation of the two bones of the forearm. These bones are approximately of the same length, but the radius is the stronger. The carpals are not ossified. They are five in number. The ulna articulates with the ulnare only, but the radius articulates with both the ulnare and radiale. There may be either two or three metacarpals. Formerly this difference in number served as basis for the specie classification, didactyla and tridactyla. Professor Ryder has since demonstrated the identity of the two forms. The second and third digits have two phalanges each. The fourth digit has only one.

The hind limb of Ajiiphiiima is fully one-third longer than the fore one. The femur has a well developed, cartilaginous head and a prominent trochanter. It broadens gradually from the middle to the distal end. The tibia and fibula are a trifle over one-half the length of the femur, and are approximately equal to each other in strength. The tibia is largest at its proximal end, but the two ends of the fibula have equal surfaces. These bones articulate with the tibiale and fibulare of the tarsals. The third tarsal supports the third and fourth metatarsals, and the tibiale supports the second metatarsal. The second and third metatarsals have each two phalanges, but the fourth has only one. All the phalanges and metatarsals are well ossified, but the tarsals are cartilaginous. The girdle bones are less perfectly formed than the limb bones. The shoulder girdle consists of a cartilaginous coracoid, a bony scapula, and a cartilaginous suprascapula. There is no evidence of true sternal elements. The pelvic girdle is more complete, having an ischium, ilium, and pubes. The ischium and pubes are united to each other and also to their fellows of the opposite sides, so as to form a shield-like plate, which is composed of cartilage, with the exception of two discoid ossifications in the posterior parts. The acetabulum is entirely cartilaginous. The ilium proper is well ossified, very slender, and surmounted by a cartilaginous style which is attached to the sacral rib of the sixty-third vertebrae on the right side and the sixty-fourth on the left. I think Mr. F. A. Lucas of Washington was the first one to observe this asymmetrical disposition of the iliac bones.

38 2 DAVISON. [Vol. XI.

The Vertebrae.

Three divisions of the vertebrae may be recognized : cervical, trunk, and caudal. There is only one bone in the first division known as the axis, the atlas being anchylosed with the skull. The anterior face of the axis presents two concavities for articulation with the occipital condyles. There are also two slight projections between the concavities, which may be called prezygapophyses, as they are applied to the internal facets of the condyles. On the posterior aspect are seen the postzygapophyses descending from the backward extension of the neural arch. The neural spine is only slightly developed, and there are no transverse processes. This vertebra as well as all the others is amphicoelous. The trunk vertebrae number sixty-two. All have prominent transverse processes and high neural spines. The transverse processes of the first seven or eight vertebrae are laterally sulcated in their distal regions, and have short ribs attached. The neural spines bifurcate posteriorly and send their prongs outward on the postzygapophyses. The course of each prong is V-shaped, with the apex directed anteriorly. From this apex a small diapophysial spine extends forward to near the anterior base of the neural spine. This yorocess serves a special purpose in Amphiuma, as I shall show later. The faces of the zygapophyses are in a symmetrical plane, extending in an axial direction. All the trunk vertebrae except the first two have small hypapophyses attached to the anterior part of the body, which project anteriorly. The middle two-thirds of the body of each vertebra is so constricted laterally as to form a rather sharp spine viscerally.

There are two sacral vertebrae. Their processes are the same as the trunk vertebrae. The caudal vertebrae number thirty-seven, making a total of one hundred and one bones in the spinal column. All the caudal elements except the first two have prominent chevron bones. The neural spine, which is so high in the trunk region, is very much depressed, and the posterior bifurcations of this spine are more extensive. The transverse processes are declivous and decrease in length posteriorly as far as the mid-tail region, where they entirely dis


appear. The parapophysial spine remains constant for the important muscular attachments. The many processes and depressions characterizing the bones of AmpJiiiima present but slight genealogical significance until we have made a careful study of the muscular system.

MusctUar System..

During the past six months I have searched carefully for a description, or even a few words of introduction, to the muscular system of this strange animal, but have been able to find only a very terse discussion of the subject. This is given by Dr. Bronn (10), and consists of a few words concerning the muscles of the head. A brief account of the dorsal muscles was published by the writer (6) in April, 1 894. A satisfactory dissection of the muscles requires considerable and careful preparation of the tissues, owing to the fact that the muscular arrangement is so complex and many of the muscles are so minute and massed together. After much experimenting I found the following fluid a most admirable agent for the maceration and differentiation of the muscular elements : one part of one-fourth per cent chromic acid, two parts of ten per cent nitric acid, two parts of seventy per cent alcohol, and three parts of water. The specimen may be left in this fluid a week, at the end of which time it must be thoroughly washed in running water for several hours. Then the muscle-fibres will be found stained a bright red, while the fascial envelopes will remain uncolored and the tendinous origins and insertions will be swollen so as to be readily seen.

Great difficulty is experienced in neatly separating the skin from the underlying muscles, since the two are indissolubly connected by an exceedingly tough fascia. This fascia consists of a dense sheath of tissue arising from the neural spines in two plates, which, scarcely separated at their origin, diverge gradually as they rise to the dorsal surface, thereby bounding laterally an area whose cross-section is triangular. This area is filled with a loose connective tissue and fatty substance. Each plate of fascia is reflected over the external surface of

384 -^^ VISON. [Vol. XI.

its respective side. At a distance of one-JiftJi of the circumference fro77i the dorsal line, a cleavage into two -membranes takes place, one of which descends almost vertically through the bodywall to the cavity where it gives rise to the transversalis abdominis muscle. This muscle continues ventralward until within about one centimetre of the mid-ventral line, where it becomes fascia. The other reflected plate of fascia extends subcutaneously around the body to the mid-ventral line, where it comes in contact with the internal plate, since no muscle takes part in the formation of the body-wall in the mid-ventral region. This tough fascial sheath also envelopes the head, being strongly attached to the median keel in the posterior region, and broadly adhering to the anterior portion of the frontal, the prefrontal, nasal, and premaxillary bones.

Muscles of the Head.

Amphiiuna has four dorsal head-muscles : pterygo-maxillaris, masseter, temporal, and cervico-parietalis. The pterygo-maxillaris arises mainly from the median juncture of the parietalsand the fascia covering the horizontal surface of the frontal. A small portion of the muscle is a continuation of the cervico-parietalis. Its insertion is on the dorsal side of the pterygoid bone and cartilage. It is clear to be seen that this muscle in Amphiumds ancestors must have been the anterior part of the cervicoparietalis. The tendons of the temporal, through enlargement and continual activity have usurped almost the entire space of the parietal groove, thereby causing the unused muscle to dwindle. The masseter is an exceedingly strong muscle, and arises in two parts. One part originates from the lateral area of the prootic, the other from the anterior curved keel of this bone. The two unite almost at their origin, and extend as a thick, muscular mass to its insertion on the mandible external to the coronoid process. The temporal is the long and strong elevator of the lower jaw. It arises from the neural spines of the fifth, fourth, third, second, and first vertebrae. It is inseparably joined with its fellow as far as the parietal bone. At a distance forward of this equal to the interorbital space, the


muscle is transformed into two tendons which, passing along the parietal groove, descend anterior to the prootic, and are inserted together in the coronoid process. The arrangement of this muscle is such as to give great strength and yet preserve the flat attenuate condition of the head in the prootic region. The cervico-parietalis muscle arises from the second and first vertebrae, and is attached to the posterior part of the parietal and the dorsal area of the exoccipital bone. The lateral head-region presents five muscles : cucularis, digastricus-maxillae, interbranchiales constrictores arcuum branchiarum, levatores arcuum, and adductores arcuum. The cucularis arises from the fascia of the transverse processes and descends a narrow band of muscle anterior to the forelimb to its insertion in the walls of the oesophagus. The digastricus maxillae is a large flat muscle arising in three portions. The first portion is attached to the neural spines in the shoulder region and is a continuation of the superior dorsal muscle. The second portion arises from the summit of the first epibranchial and mingles inseparably with the first. The third portion is the strongest, and arises from the posterior otic region, joining the other two immediately, whence the entire mass passes downwards and forwards to a firm insertion in the posterior angle of the lower jaw. Bronn's Their-Reichs (10) describes only two portions as origins of this muscle. The writer has detected several errors in this work in the descriptions of the muscles of the head of Amphiunia. The interbranchialis constrictores arcuum branchiarum exist as thin oblique bands of muscular fibre between the first, second, and third epibranchials, but no fibre joins the third and fourth, between which the gill slit persists in the adult. The levatores arcuum arises from the inferior side of the posterior portion of the digastricus and descends as a flat band of fibres to its insertion on the summits of the epibranchials. The adductores arcuum consist of a tendinous band connecting the summits of the four epibranchials, whence it extends downwards and backwards to a point above the forelimb, where its course becomes transverse, forming the third inscriptio tendinea. The ventral head-region presents eight

386 DAVISON. [Vol. XI.

muscles : thoracico-hyoideus, omo-humero-maxillaris, geniohyoideus, mylo-hyoideus, stylo-hyoideus, genio-glossus, ceratohyoideus externus, and trachealis arcuum. The thoracicohyoideus is a large muscle extending from the median extremity of the cerato-branchial backwards until it is inseparably mingled with the rectus abdominis. Its fibres are interrupted by several inscriptiones tendineae, which are present as far forward as the gill slit. The omo-humero-maxillaris is well developed, arising from the fascia ventralward of the fore limb, and increasing in strength as it runs forward to its insertion on the angle of the maxillary. The genio-hyoideus is a thin band of muscle, arising from the symphysial region of the mandible, and is inserted in the fascial sheath of the thoracicohyoideus. The mylo-hyoideus forms a thin sheet of muscular fibre extending transversely between the rami. The stylohyoideus lies posterior to and deeper than the former muscle, and extends from the cerato-branchial to the cerato-hyal and hypohyal bones and basi-hyal cartilage. The genio-glossus lies in the floor of the mouth parallel with the ramus connecting it with the hypohyal. The cerato-hyoideus-externus lies immediately beneath the stylo-hyoideus. The trachealis arcuum is composed of a transverse band of fibres extending from the fascia of the tracheal region to the tendon joining the summits of the epibranchials. Its function I believe to have been the retraction of these arches. Fischer, Duges, Humphry, Schmidt, Goddard, and Van der Hoeven disagree to such an extent upon the names of the muscles of the Amphibian head that I have not adopted any one man's nomenclature, but have retained the name which seemed most proper for the muscles of Amphmma.

Muscles of the Limbs.

So far as I have been able to learn no one has yet attempted to describe the muscles of AmpJiimna s limbs. The minuteness and massing together of the muscles render it a most difficult undertaking. On the ventral aspect of the fore limb are seen four muscles. The largest one, representing the


pectoralis major, arises from the fibres of the omo-humeromaxillaris posterior to the limb, and extending distalward as a radiate muscle, is inserted in the fascia of the muscles of the arm. Immediately beneath this muscle, which covers the entire coracoidal region, is found the supracoracoideus, a radiate muscle arising from the ventral surface of the coracoidal cartilage and extending to its insertion in the head of the humerus. Its function is that of depressing the fore limb. The slender fascia-like deltoideus arises insensibly from the fibres of the omo-humero-maxillaris, and runs along the anterior side of the arm, being slightly inserted on the distal end of the humerus, but continuing as a flexor carpi radialiis to its final insertion in the carpal cartilages. The coraco-humeralis is a mere branch of the omo-humero-maxillaris, and is strongly inserted in the anterior proximal region of the humerus. It draws the limb cephalad. The flexor digitorum communis is a greatly degenerated muscle arising from the middle part of the humerus and extending downwards to the carpal region. The dorsal aspect of the fore limb presents four muscles very closely bound together by dense fascia. A slender muscle representing the triceps brachii arises from the fascia posterior to the branchial arches, and appears to be attached slightly along its entire course down the arm to the phalanges. From the obliquus externus a band of fibres runs forward to its insertion in the upper part of the humerus, serving to draw the arm backward. This muscle corresponds to the latissimus dorsi. Another muscle arising in common with the last mentioned is inserted along the middle portion of the humerus, sending fibres onward to the phalanges, and is probably the atrophied remains of the infraspinatus. Owing to the fact that AmpJiiinna seldom bends its arm at the elbow, the muscles arising from the shoulder region in many instances continue to the forearm and hand. This is the primitive condition of limb muscles. In fact, I do not think this animal is capable of flexing the forearm or the arm, as the muscles are so bound together by dense fascia and continnons at the elboiv joint. My anatojnical inference on this point was confirmed by observing a large specimen moving across the floor. The limbs did not touch tJie floor, but


88 DA VISON. [Vol. XI.

they were moved quite vigorously backward and forzvard, and were not bent at the elbow or knee joijits.

The muscles of the hind limbs are larger and more distinct than the foregoing. On the ventral aspect are seen three muscles. The large muscular mass arising from the ischiopubic symphysis and taking its course down the posterior side of the limb to the phalanges appears in the reptilia as the adductor and gracilis muscles. Immediately beneath this mass a radiate muscle arises from the ischio-pubic plate, and is strongly inserted in the greater trochanter. The femorocaudal arises from the third and fourth caudal vertebrae, and descends forward in two parts, one of which is inserted in the upper part of the femur; the other joins with the semimembranosus extending down the posterior side of the leg to the insertion into the phalanges. The ischio-caudal is a well developed muscle originating on the posterior margin of the ischium and extending posteriorly to an insertion on the vertebrae of the anterior third of the tail. The pubo-tibialis is a strong adductor arising from the coelomic aspect of the ischio-pubic plate and extending across the middle part of the femur down the front side of the tibia to an insertion in the phalanges.

The dorsal aspect of the hind limb presents two muscles. The rectus femoris is a heavy muscle arising from the fascia in the region of the ilium and extending to the distal part of the femur, where it is attached, thence continuing to the aponeurosis of the foot. The ilio-peroneal arises from the ilium, and extends to the distal bones of the leg. Thus it will be seen that many of the muscles of this limb pass over two joints, thereby indicating very restricted movements, if any, in the knee joint. The phylogenetic significance of these facts will be discussed later.

Muscles of the Trnnk.

The muscles of this region furnish a most intricate as well as a most interesting study. This portion of Ajnphitwias muscular system had not been described prior to my paper in the Anatomischcr Atizcigcr of April, 1894. As was stated in


the early part of this communication, the trunk muscles are separated into two regions, viz., the dorsal and abdominal, by the fascial lamina split off from the external dorsal sheath. This lamina extends through the wall to the body cavity. A similar disposition of the fascia occurs in Crypt obranchus japoniciis, van der Hoeven, as described by Humphrey (11). The dorsal mass of AvipJimma is not differentiated into separate muscles, but for the sake of convenience may be considered as composed of two parts : the superior, lying above the transverse processes, and the inferior, lying beneath these processes. The former corresponds to the erector spinae of some authors. The latter is called rectus trunci internus by Schmidt, Goddard and van der Hoeven in a description of other Amphibia of this order. Mivart (12) speaks of a similar muscle in Menopoma as being a part of the retrahens costarum. The anterior portion of this muscle in Amphhnna undoubtedly functions as a retrahens costarum, being attached to the minute ribs found on the first seven or eight vertebrae of the trunk. The skin having been carefully removed from the back, and the muscles well stained and macerated by the fluid mentioned previously, there will be seen lying along the axis longitudinally-disposed rows of cones, the enveloping fascia of which appears, at first sight to form a kind of network.

In the superior dorsal mass there are three rows of cones lying side by side. The apices of the row adjacent to the axis are directed posteriorly, those of the next row anteriorly, and those of the third row posteriorly. TJius it is seen that the apical direction of the cones varies alternately in the different rows. Each cone is introduced into the preceding one about one-third of its length, as shown in Fig. 10.

From the exterior apex of each cone in the two distal rows a tendinous cord extends to the interior apex of the following cone, thus serving to hold the apices in position. The row most distant from the axis has the deep part of the base of each cone firmly attached to the outer half of a transverse process. That part of the base distal from the axis is reflected to form an inscriptio tendinea extending transversely to the mid-ventral line. The superficial base of the cone blends with

390 DA VISON. [Vol. XI.

the fascial body investment. That side of the base proximal to the axis is continued forward as the distal side of a cone in the adjacent row. Therefore it is seen that a transverse line through the apex of a cone in one row will pass through the base of a cone in the adjacent row.

In the middle row the deep sides of the bases are attached to the post-zygapophyses and their spines. The distal and proximal sides of the bases are continued as the lateral boundaries of cones in the adjacent rows. The superficial sides of the bases have the same insertions as those in the row previously described. The cones in the row adjacent to the axis are somewhat flattened laterally by their close apposition to the neural spines. The deep side of the base of each cone is securely inserted on the postero-lateral division of the neural spine. The distal side of each base takes the same course as the corresponding side in the adjacent row. The proximal sides are fastened to the neural spine and also to the fascia arising from the neural spines to serve as the body investment. The superficial sides of the bases and also one-half of the superficial lateral boundaries of the cones are blended with the external fascial envelope. The apices of the cones in this row give off ribbon-like tendons which extend to the interior of the following apices. Such is the general arrangement of the cones in the superior dorsal mass.

The size of these cones varies. Those of the distal row are all of the same size, and are somewhat larger than those of the other two rows, the length being fully three centimetres, and the diameter of a base about one and a half centimetres. The length of a cone in the proximal row is scarcely two centimetres, and its base is about one-half a centimetre. The preceding measurements were made on an animal almost one metre long.

Since the arrangement of these cones is so regular, it is easy to estimate their number, which I have calculated to be three hundred and seventy-two in the superior dorsal mass.

A view of the inferior mass from within the body cavity reveals no evidence of a conical arrangement, but instead are seen, very prominently marked, the transverse septa at regular

No. 2.] AMPHIUMA MEANS. 39 1

intervals, corresponding to the lengtlis of the vertebrae. It will be noticed, however, that the septa appear to cease very abruptly at a distance of two-thirds of a centimetre from the axis. A careful dissection of a well stained specimen along this line brought to view the same conical arrangement observed in the superior mass. The cones in the distal and middle rows are quite perfectly developed, but those of the proximal row are very imperfectly formed, being too closely apposed to the spinal axis. The direction of the apices in these rows is exactly opposite to those in the superior mass ; that is, the proximal row of cones has its bases pointing anteriorly, whereas in the corresponding row of the superior mass the apices pointed posteriorly. The cones are much smaller, being scarcely half as large as the overlying ones. The superficial sides of the bases, as well as a large part of the superficial lateral area, are inseparably united to the dense fascia lining the body cavity. The outer sides of the bases in the distal row are reflected to form the transverse septa, while the deeper sides of the bases are firmly attached to the lower side of the outer half of the transverse processes. The inner sides of these bases are continued to form the lateral boundary of a cone in the adjacent row. The attachments of the middle row are so similar to those of the same row in the superior mass that I will not give them. The apices of these two rows are connected with the interior part of the apices of the cones following by a ribbon-like tendon.

In the row adjacent to the spinal axis the deep sides of the bases adhere to the hypapophyses of one vertebra, and the apices are inserted on the hypapophyses of the vertebra following, so that each hypapophysis serves for the attachment of an apex and the deep side of a base. From this brief description it can be readily seen that the general plan of the cones is the same in both dorsal masses.

The conical arrangement of the muscles prevails not only in the dorsal portion of the tail of Amphiuma, but also in the ventral portion. The disposition and attachments of the cones here are so very similar to those of the trunk region that it would be unprofitable to describe them. The number of cones

392 DAVISON. [Vol. XI.

in this region is approximately four hundred and sixty-eight, though some of them near the extremity are rather imperfectly formed. The whole number in the trunk, counting twelve to each vertebra, equals seven hundred and forty-four, which, added to the four hundred and sixty-eight in the caudal region, gives a total of one thousand two hundred and twelve cones. Ill other words, the dorsal and caicdal muscles of this animal have over one tJionsand strong fascial attachnents.

Having discussed the general structure of these muscles, it is now in order to determine the direction of the fibres. .These are not parallel to a line through the apices of the cones, as we should expect, but are so directed as to form an angle of about ten desrrees with that line. Since there are no cones found in the outer half of the dorsal muscle, the direction of the fibres there is exactly parallel with the axis, being, however, completely interrupted by the inscriptiones tendineae.

So far as I have been informed, this peculiar conical disposition of the fibres of the dorsal muscle has been observed in only two other vertebrates and in no other Amphibians. Dr. Hair {13) describes similar cones in the alligator, and I have noticed the same structure prevalent in SpJioenodon (Hatteria), the peculiar New Zealand lizard. The genealogical significance of this muscular arrangement will be discussed later.

The ventral trunk muscles are, as in other Urodeles, composed of four sheets of fibres : the transversalis abdominalis, the obliquus internus, obliquus externus, and rectus abdominis. The first named is a mere continuation of the descending lamina of the external dorsal fascia. It may therefore be said to arise from the neural spines, and with its fellow, form a tube inclosing the viscera and dorsal muscles. In other words, the lamina may be considered as an aponeurosis, the muscle-fibres originating just before the aponeurosis emerges into the body cavity. The muscle passes transversely around the ventrum, dwindling again to fascia about one centimetre before joining its fellow along the mid-line. The striking feature in this mnscle is that it is unaffected by the inscriptiones tendineae, a condition not present in any other Amphibian. The obhqnns internns and also the obliquus externus are tJiicker on the left


side of AmpJiiitma than on the right. The left dorsal mass is also considerably stronger than the right. This asymmetry is probably due to the manner in which the animal lies coiled, though I have not had opportunity to demonstrate that fact. The obliquus internus muscle is composed of fibres taking origin from the tendons of the transverse processes, and extending obliquely anteriorly between the inscriptiones tendineae. External to this muscular plate lies the obliquus externus, readily recognized, as the fibres extend obliquely posteriorly between the inscriptiones. As the fibres of these two oblique muscles approach the ventral line they gradually change their direction, becoming finally parallel with the axis of the animal, and thus form the rectus abdominis. In all other Amphibians except AmpJiiimia this muscle is continuous over the ventral line. Its fibres are completely interrupted by the inscriptiones. Anteriorly it is continuous with the thoracico-hyoideus and omohumero-maxillaris. Posteriorly it is attached to the pelvic elements, but continues as the ventral caudal muscle. Thus it will be noticed that the trunk and head muscles of Amphiuma are more highly specialized than those of other Urodeles, while the limb muscles are less specialized.

Digestive System.

The food of this Urodele consists of crayfish, small teleosts, and other similar aquatic life. The lining membrane of the buccal cavity is tough and smooth, but its continuation into the pharyngeal and oesophageal region is loose and somewhat corrugated longitudinally. On the ventral side are numerous ciliated columnar cells. The stomach is a slight enlargement of the oesophagus, beginning at a distance behind the shoulder girdle, equal to the distance of that girdle from the tip of the nose. The mucous lining of the stomach is thrown into small longitudinal ridges in the anterior portion of the organ, these ridges increasing in prominence as they extend posteriorly. The walls of the stomach in its anterior parts are but little thicker than those of the oesophagus, but posteriorly they become nearly twice as heavy. The length of the stomach is

394 DA VISON. [Vol. XI.

equal to almost one-third of the distance between the fore and hind limbs. The transition from stomach to duodenum is readily recognized by the vast difference in the thickness of the wall, that of the latter being very thin. The membranous vascular folds increase in prominence and continue throughout the entire intestine to the rectum. The intestine is for one or two centimetres folded upon itself at two or three points in its course. The pancreatic gland lies dorsalward of the posterior half of the stomach, and for an equal distance along the duodenum. The liver extends from the region of the tenth to the thirty-eighth vertebra, being almost twice as long as the stomach. It is entire. The gall bladder lies near the caudal termination of the liver. The rectum consists of an abrupt expansion of the digestive canal at a distance of four or five centimetres from the cloaca. The internal wall of the rectum is quite smooth. This portion of the canal passes gradually into the cloaca, recognized only by its location and smaller diameter. The cloacal region will be described in the discussion of the urogenital apparatus.

The Circulatory System.

This system in Amphiiima is very much the same as in the salamanders. The heart is surrounded by a very large sac, through which may be seen the ductus cuvieri entering the large auricle. The bulbus arteriosus is long, giving off an aorta bow on each side. The carotids are exceedingly large. The iliac arteries and veins are very prominent. The venae revehentes are clearly visible in the kidneys. The portal vein lies along the dorsal side of the liver, receiving its numerous tributaries from the intestines, pancreas, and liver. The pulmonary vein is distinctly visible, passing to the apex of the lung on the external wall. Other features of the vascular system are in common with the order of Urodeles.

I shall not attempt a description of the nervous system, as that has been admirably discussed by Dr. Osborn (8).


TJlc Gcnito-7irinary Systein.

The kidneys are leguminoid bodies located immediately anterior to the hind limbs. In my largest specimens they measured four centimetres in length. The ureter is very short. The urinary bladder is exceedingly elongated, extending from the termination of the liver to the cloaca. Cope has assented tlmt AmpJiiinna has only one testis, but I fijid pahrd testes extending from the liver half ivay to the vent. In alcoholic specimens they are of a brownish spongy texture. They are attached to the body wall by folds of the peritoneum. A mesonephric or Wolffian duct is present as a thick-walled tube running in a straight course from a point near the gall bladder to the urogenital sinus. The vesicula seminalis is a thin-walled tube ending blindly in front near the gall bladder and extending in a convoluted course to a common opening with the mesonephric duct into the urino-genital sinus. The vasa efferentia are present as delicate tubes arising from the testis and emptying into the Wolffian duct, which acts as a vas deferens. The ovaries are very slender bodies lying a short distance posterior to the liver. The oviducts extend along the dorsal side of the body wall, terminating in the urogenital sinus. It has for a long time been a question whether the eggs of Aniphinma are fertilized internally or externally. I published a short article relating to this subject last April (7). I am now fully convinced that internal fertilization takes place.

In the early part of the spring of 1893 I secured a male specimen of Amphinma a trifle over a metre in length in Northern Tennessee. This animal was the largest of its kind on record, and was found farther north than any previously discovered. On examining the vent I found exuding a viscid substance which, when placed under the microscope, revealed numerous spermatozoa. The inner walls of the vent were covered with dense papillae on their posterior parts. These papillae under the microscope proved to be the orifices of numerous glands which secreted the almost colorless waxy substance in which the spermatozoa were lodged. The ajtterior parts of the internal ve?it xvalls are furnished ivith fro7n fftcn

396 DA VISON. [Vol. XI.

to twenty membranous laminae extending obliquely fro7n witJiin outwards and backivards in stick a manjier as to transfer the generative products slowly from the cloaca to the external lips of the vent. Wlien these lips are placed in apposition to the lips of the female vent, the reproductive agents are induced within the cloaca of the latter by means of a series of capillary tubes (Fig. XI, C) arranged oti the inner walls of the vent and extending from without inwards and forwards. I do not see how these different features in the vent structure of the two sexes can serve any other purpose than that which I have described. Furthermore, the fact that the male was so filled with spermatozoa as to cause them to exude indicates that that month, May, was the natural time for their evacuation; and, inasmuch as the eggs are not deposited until August or September, fertilization must occur within the body of the parent. This theory of internal fertilization is further strengthened by the fact that IcJitJiyophis glutinosus, the blind-worm of Ceylon, an animal closely related to the family Amphiumidae, is reported by the Sarasins (9) to have its eggs fertilized internally.

TJie Respiratory System.

The anatomy of the respiratory apparatus in this animal is very simple. The external apertures of the nostrils are exceedingly small, being situated near together on the forward aspect of the fascial region. The internal nares appear lateralwards of the posterior limit of the vomero-palatine series of teeth. The trachea is very long, being i^i my largest specimen nearly six centimetres. The glottis is a small longitudinal slit on the ventral side of the pharyngeal region. There is no epiglottis. The trachea is thin-walled, without any cartilaginous formations. The lungs are annexed to the trachea dorsalward of the heart. Cope has greatly erred in saying the lungs are subequal. TJie left lung is coextensive posteriorly tvitJi the liver, but the right one extends within tJiree or four centimetres of the vent. The diameter of the left is much smaller than the right also. The walls of these respiratory sacs are quite thick anteriorly, and grow thinner as they pass caudalward.


The transverse trabeculae in the cardiac region are ten millimetres apart, and are thrice as heavy as the longitudinal trabeculae. Immediately anterior to the front limbs is seen a spiracle guarded by membranous flaps to exclude the mud. This aperture is in communication with the pharynx, and is supplied with special muscles for closing it, as previously described.

Having now called attention to the important features of the adult, it is in order to make a brief examination of the young.

The Young Amp/mima.

For a long time embryologists have been seeking a good supply of specimens representing the development of this peculiar Amphibian. Hay (2) was successful in finding the eggs of Amp/muna in an Arkansas swamp in 1887. This material furnished important information. Last February I was so fortunate as to secure a number of very young specimens from a North Carolina dealer. They ranged in length from sixtyeight to ninety millimetres. They were found in a damp locality under some large rocks. Hay's embryos were 45 mm, long. It is probable that my specimens were hatched in November or December. The general shape of the young (Fig. 12) is very much the same as the adult. There are no signs of gills, and only one gill-opening persists on each side. The eyes are rendered useless by the heavy shields of epidermis, as in the adult. The dennal glands aj'c more prominent than in the old. The caudal fin is less atrophied. The lower jaw is correspondingly shorter than the adult's. The projecting dermal folds of the upper jaw are little developed. TJie legs are relatively longer tJian in tJie adult, but the digits are imperfectly fortned, those Oft the hind limb being qjnte distinct, while there are no signs of any on the fore limb. Hay (3) states the reverse of this to be true in his embryos, viz., that the digits of the fore limbs are better differentiated than those of the hind. A gross dissection of this small specimen showed that the internal structure accorded with that of the adult, except in the case of the left lung, which did not extend to the caudal

398 D^ VISON. [Vol. XI.

termination of the liver. The most important information was gained from the study of the heads of several specimens stained in toto in borax-carmine and cut serially into sections one-fiftieth of a millimetre in thickness. Horizontal as well as sagittal and transverse sections were made.

TJie Skeletal Anatomy of the Head.

The skeleton of the head is shorter and broader than in the adult. The ossifications are, of course, vastly different. There are but sixteen fully ossified elements in the cranium: premaxillary, nasals, frontals, prefrontals, parietals, squamosals, maxillaries, vomero-palatines, and parasphenoid. The following elements are partially ossified: orbitosphenoids, prootics, and exoccipitals. The pterygoids, quadrate and stapes are wholly cartilaginous. The premaxillary is quite the same as in the adult, and needs no further description than to state that the interosseus septum is not derived from the nasal roofing cartilage, as Wiedersheim (i) has suggested. The two elements, though in contact, present no transition from one to the other. The enveloping cartilage of the nasal sac is incomplete where the osseous septum is present. The union of the cartilaginous floors of the nasal sacs medially extends beneath the anterior part of the brain for a short distance, when the middle portion of the cartilage passes into connective tissue, leaving two lateral bars. Hay (3) speaks of an unpaired piece of cartilage lying in the roof of the mouth of the adult between the anterior ends of the vomero-palatines, and states that it is not found in his embryos. Wiedersheim speaks concerning the same as follows : " Da wo die Vorderenden der Vofnero palatine in der Mittellinie zusammenstossen, ragt ein conisch gestalteter Knorpelzapfen vom Boden der Nasenhohle in die Schleimhaut des Mundes herab, von welcher er einen Ueberzug erhalt."

Hay believes that this nodule of cartilage has been cut off from that forming the floor of the nasal sacs by the union of the bones in the roof of the mouth. Fortunately my specimen is of just the proper stage to settle the question. The roofing


bones of the mouth are fully ossified before the nodule is formed. In a specimen eighty-eigJit millimetres long the cells of the dejise connective tissue are concentrically arranged preparatory to the formation of the true cartilage^ which in a specimen ninety niillimctres long has made its appearance. The frontal bone, though completely ossified, differs from the adult in the manner of giving exit to the olfactory nerve. Hay implies that Wiedersheim has erred in the following description : " Es handelt sich namlich, wie am besten aus der Figur 20 F ersichtlich ist, um einc an der Unterfiache der vorderen Stirnbeingegend auftretende Knochenzwinge, deren mediale Circumferenz vorn und einwarts, und deren laterale mehr nach hinten auswarts gelagert ist. Beide stehen parallel zum Medianebeine und sind unten gegen die Schadelbasis zu durch eine schmale knocherne Commisur in Verbindung." This description is exactly correct for the adult (Fig. Ill h), biLt in the young tJiere exists only an unmodified aperture in the frontal for the exit of this nerve. Cope (14) attaches great importance to the free margin of the frontal bone in the adult. The frontal in the young has no such margin, the surface being slightly depressed in the middle and regularly convex laterally.

The frontals overlap the parietals to a considerable extent posteriorly. A cross-section through the posterior part of the parietal of the adult presents the curve of a quarter circumference, the depression being external for the temporo-cervical tendon. In the young this bone slopes from the median line outwards at an angle of thirty degrees, and is but slightly depressed in the middle of the distal half.

The orbito-sphenoid confirms Hay's account, being higher in front than behind. Its ossification is almost complete. The exoccipitals remain almost entirely cartilaginous, being invested with a thin parostosis. The ossification of the condyles is beginning. The pterygoid is represented by a bar of cartilage, free anteriorly but attached to the quadrate posteriorly. The otic capsule is quite surrounded by cartilage. The otolithic deposit is extensive. The prootic is cartilaginous to a considerable extent. The stapes shows no evidence of ossification, as is likewise the case with the columella. The quadrate

400 DA VI SON. [Vol. XI.

is perfectly formed in cartilage. The squamosal is the only otic element entirely ossified. A thick plate of basioccipital cartilage lies beneath the hinder portion of the brain. The parasphenoid forming the floor of the brain-case is somewhat convex in its anterior part, but markedly concave in the posterior region. The lateral walls of the brain-case in the pituitary region posterior to the orbitosphenoid are of cartilage, as in the adult. The other features of the cranium are so similar to the descriptions of Hay and Kingsley that I deem it useless to give them.

The Visceral Skeleton.

The mandible is quite the same as in the adult. The rami are united anteriorly by strong connective tissue. Meckel's cartilage lies in the groove of the ramus as far forward as the point just below the eye, and extends backwards so far as to be in contact with the quadrate. The dentary and angular are well ossified. The teeth are fully developed. The hyoid apparatus is unlike either the adult or Hay's embryo. It is for the most part cartilaginous. The middle third of the basibranchial appears quite well ossified. Thin parostoses invest the ceratohyal and ceratobranchial. The following elements compose the apparatus : one basihyal, two hypohyals, two ceratohyals, one basibranchial, two ceratobranchials, and eight epibranchials. Hay found no basihyal in his embryos. A distinct nodule of cartilage is present at the juncture of the hypohyal and ceratohyal on the external side. Its significance is unknown to me. The ceratohyals extend outwards rather than backwards, as in the adult. The larynx is an exceedingly simple structure, consisting of a fibrous connective tissue tube, strengthened by two lateral bars of cartilage. The external orifice is a longitudinal slit. The trachea has no rings or partial rings of cartilage.

Soft Parts of the Head.

The disposition of the muscles is approximately the same as in the adult, with the exception that the place of the temporocervical tendon appears to be filled by muscular tissue. Ven

No. 2.] AMPHIUMA MEANS. 40 1

trally the first inscriptio tendinea is seen just below the basihyal. The remnants of some nasal glands are present along the olfactory sac. Hay found these glands better developed in his specimens. The most important and interesting structure is found below and external to the eye in my smallest specimen, seventy-eight millimetres in length. There appears in tJiis region a canal (Fig. 15) one-tentJi of a millimetre long, which is availed by columnar epithelial cells extremely regular in outline. External to the epithelial wall there is seen a thick layer inferiorly of degenerated tissue, which is bounded by a thin layer of fibrous connective tissue. In three other specimens, eighty-eight, ninety, atid ninety-two millimetres respectively, no trace of this degenerate canal could be discovered, and in the smallest specimen I was able to detect it on the right-hand side ojtly. Here it was very clearly seen, as shown in Fig. 15. The significance of this atrophied element will be discussed later. Hay (3), Kingsley (5), and Osborn (8) have to a limited extent described the nervous system of Amphiuma. Owing to the fact that my tissue was not prepared for the demonstration of the nervous elements, I can add only one or two points of interest. The brain (Fig. 14) is much shorter than in the adult, caused mainly by the wedging in of the middle brain between the hemispheres of the great brain. The brain viewed dorsally, excluding metencephalon, presents the outline of a longitudinal section of a hen's ^g^, the anterior end corresponding to the smaller end of the Q-^^. The olfactory lobes are not marked off as in the adult. The brain of Siphonops annulatus, as figured by Wiedersheim, resembles to a considerable extent the brain of young Amphiuma. The latter, however, is not so elongated as the former. The pineal gland and pituitary body so prominent in the adult are scarcely distinguishable. Hay and Kingsley have described the origin and distribution of the cranial nerves sufficiently for our purpose. My sections clearly corroborate the statement by Hay that the facial nerve passes beneath the columella. Kingsley failed to find the roots of the fifth nerve. My sections shoiv but one root. The dorsal ganglion in connection with the twelfth nerve is quite large, being at least onethird the size of the gasserian ganglion.

402 I>A VISON. [Vol. XI.

Axial Skeleton and Appettdages.

The vertebrae are only partially ossified. The transverse processes are wholly cartilaginous, and a portion of the internal part of the body of the vertebrae is unossified. The cartilaginous posterior projection of the roof of the spinal cord is invested with a thin parostosis. The cartilaginous intercentra are present. The ribs are present as cartilage. The neural and diapophysial spines are imperfectly developed. The hypapophyses are well marked. The shoulder girdle is wholly cartilaginous, and presents elements of scapula, coracoid and precoracoid. The scapula is exceedingly thin, being only two cells in thickness. The other elements are correspondingly slender. No sternum is present. The humerus is covered with a thin layer of bone, except in the regions of the extremities. The radius and ulna are also ossifying externally. The carpus is present as pure cartilage, and the phalanges remain as hyaline tissue also. The pelvic girdle is entirely cartilaginous. All the elements are present as in the adult, but there is no evidence of the posterior bony disc. The femur is invested with a very delicate ectostosis in the shaft region. The future prominent trochanter process is indicated by a slight flexion of the cartilage at that point. The two abductor, the adductor and two rotator muscles have the same locations as in the adult. The tibia, fibula, tarsus, metatarsus, and phalanges show no signs of ossification. This affords weighty evidence that the formation of the anterior limbs is in advance of the posterior.

I have not been able to discover any conical arrangement in the fibres of the dorsal muscles. The abdominal muscles have the same relative dispositions as in the adult. The transversalis is unaffected by the inscriptiones tendineae, and arises from the internal plate of fascia, originating in common with the external plate on the neural spine. The digestive, respiratory and excretory systems correspond with those of the adult. Having described the anatomy of these different individuals, it remains to determine the genealogical status of Atnphiimia among vertebrates.

No. 2.]



Phylogenetic Conclusions.

Amphinma has always been considered a degenerate form. Cope (4) says that Amphiiivia is the annectant type which Wiedersheim sought for in tracing the ancestry of the Coeciliidae to the Stegocephali of the Carboniferous period, and then adds that he derives the Coeciliidae from the Urodela direct through the AmpJimmidae, and adds the following table of affinity:




Amphiumidae Desmognathidae Amblystomidae Cryptobranchidae



It is evident to all phylogenists that this table presents an absurdity, since representatives of each of the five families in the direct line of descent are existing at the present time. That these families are closely related cannot be denied. Cope bases his strongest point of relationship between Coeciliidae and Amphiumidae on the common possession of an ethmoid, when in fact the latter family does not possess an ethmoid. Sections of my young specimens clearly demonstrate this. What Cope has called the ethmoid are inerely the descending processes of tJie frontals . Kingsley (5) believes that the many peculiar resemblances of Gymnophiona and Amphinma are those of homoplassy. The recent information gained by the examination of the young specimens in my possession, enables me to prove that these resemblances are due, in part at least, to relationship. The TchthyopJiis glntinosjis of Ceylon, as described by the Sarasins (9), is undoubtedly closely allied to Amphinma. It is now known that the eggs of the former are fertilized within the body of the parent. In my description of the reproductive organs of Amphinma I have demonstrated that in this genus also internal fertilization takes place. Both deposit eggs of about the same size, which are united by a

404 DA VISON. [Vol. XI.

twisted cord. Both incubate the eggs by lying in a coil about them. In their larval life the young of IchtJiyophis possess gills and dwell in the water. Hay's embryos of AmpJihmia, evidently near the period of hatching, had well developed gills. The young specimens I secured last February were found under rocks near the water, indicating that their transformation from aquatic to land habits had lately been accomplished. All these superficial features common to the two genera indicate affinity; but by far the stronger evidence of their affinity is based on the structure of the soft parts as well as the skeletal elements. The peculiar disposition of the fascial investment in AmpJiumia is also seen in IcJitJiyopliis. The dorsal lamina arising from the neural spines splits into two plates before reaching the lateral line, and one enters the body cavity to give rise to the transversus abdominis, which is unaffected by the inscriptiones tendineae probably on account of its late formation in the embryo. The omo-humero-maxillaris is absent in all urodels except the Coeciliidae and Amphiumidae. The lungs of Amphimna are very unequal in length, a condition characteristic of the Coecilians, according to MacAlister. There is also a striking similarity in the trachea of the two families. Wiedersheim speaks thus of the Gymnophiona: "Die Luftrohre ist entsprechend den weit nach hinten geriickten Lungen fur ein Amphibiem von sehr bedeutender Lange und componirt sich aus zahlreichen hyalinknorpeligen Ringen, welche dorsalwarts nicht geschlossen sind, sondern hier durch Bindgewebe ersetzt werden." As I have already shown, the trachea of Amphimna is comprehended in the above description of a Coecilian. The brain of the young AmpJiiuma is much more unlike the brain of the adult than the latter is unlike the brain of Siphonops annnlatus, as figured by Wiedersheim. The distribution of the cranial nei-ves in the two species is almost identical.

In Siphonops annnlatus, Coccilia rostrata, Coccilia oxyura and IcJithyophis glntinosus Weidersheim has figured and described an orbital gland and peculiar tentacular apparatus of the nasal region. This apparatus consists of a canal beginning posteriorly to the eye, whence it extends forward to its external orifice


near the narial aperture, a tentacle or feeler, and a muscle bymeans of which the tentacle is retracted or protruded to guide the animal in its dark underground expeditions. As I have already shown, there exists in my youngest specimen of Amphiuma the atrophied remnants of this tentacular apparatus. The columnar epithelial linijig of the canal is very distinct in about one dozen transverse sections taken throiigJi the orbits. In some of the sections I have discerned what I believe to be the degenerated retractor muscle. This apparatus in Amphi?ima has precisely the same relative location as in the Coecilians. For some unexplainable reason, neither Hay nor Kingsley found this organ in the young embryo. Hay speaks of nasal glands, which, from his description, I conclude to be identical with the glands I noticed in conjunction with the olfactory cavities. The tentacular canal of Amphiuma cannot be mistaken for the duct of a nasal gland, as it lies too far lateralward and posterior to the nasal region. It will not answer for the duct of the orbital gland, as it is too far inferior to the orbit, although it is possible that its relation to the surrounding, parts may have become somewhat distorted. However that may be, the occurrence of this degenerated structure in the young Amphiuma and its complete disappearance in the adult gives unmistakable evidence of the relationship of the Coeciliidae and AmpJiitimidae. The disappearance of the ethmoid in the latter family can be accounted for by the fact that the descending processes of the frontals have displaced the cartilage for the ethmoid so that it lies beneath the forebrain in my young specimens. The nodule of cartilage between the anterior ends of the vomeropalatines of the adult has no genealogical significance, so far as I can discover. As before stated, the dense connective tissue is being transformed into cartilage in my specimen of 6S mm. The cause of this formation is found in the fact that the teetJi. of the lozver faiv in biting do not meet the corresponding ones in front, but pass inside of them, pressing against the roof membrane of the mouth, thereby exciting the grozvth of the cartilage in the same manner as the horns originated in the Cavicornia and Cervidae, according to Eimer (17). That the teeth in the adult Amphiuma bite anterior to the summit of the car

4o6 DA VI SON. [Vol. XI.

tilage, is explained by the anatomy of the young, in which the lower jaw is relatively shorter than in the adult. A similar abbreviation of the lower jaw is exhibited by the majority of the Gymnophiona, as shown by Wiedersheim (Taf. V, Figs. 58 and 59).

The vertebrae of AmpJiiiuna are highly specialized, having definite processes for the attachments of its complexly constructed trunk musculature. As Cope has already suggested, the prominent anterior hypapophyses are peculiar to the Coccilians and A^nphiiima. Thus far I have pointed out the features in these two families which give evidence of genealogical affinity. The proof of relationship furnished is to my mind conclusive ; but the gravest question — what that relationship is — remains to be answered. If Huxley's dictum, " It is more important that similarities should not be neglected than that differences should be overlooked " were maintainable, near afhnity of these two families must be admitted. Before such affinity can be asserted, important contrasts in skull structure must be explained. Thus Kingsley says : " The presence of an ethmoid in the GymnopJiioma (and its absence from AvipJiijtma and other Urodeles), the existence of a turbinal, the absence of a parasphenoid, and the presence of a basisphenoid are all points of importance." Another striking contrast is seen in the structure of the orbit which is only partially encircled by the bony elements in Amphiiuna, there being no jugal or quadrato-jugal bone present. Gervais gives a concise description of the coecilian orbit : " Cependant I'orbite des Cecilies n'est percee ni dans le maxillaire seul ni dans le corps de I'os jugal ; c'est ce que Ton voit tr^s-bien sur la tete d'un jeune animal de ce genre; et avec quelque attention, surtout en se servant d'une loupe, on retrouve meme chez I'adulte des traces de la suture des deux OS entre lesquels I'oeil est ici place, et qui concourent, comme chez beaucoup d'autres animaux, a former un cercle orbitaire." These differences in skull structure make it patent that AmpJiijima cannot be the connecting link between the leg-bearing Urodeles and the Coecilians, as Cope has asserted. The elongated cranium, the double series of teeth, the tentacular apparatus, the degenerated optic sense, the manner of

No. 2.] A MP HI UM A MEANS. 407

fructification and incubation of the eggs, the habits of the young, the degenerate hmbs, the unusual disposition of the transversalis-abdominis, the inequality of the length of the lungs, the anterior hypapophyses and the amphicoelous vertebrae, all of which these two families possess in common, point to a common parent form of the Coeciliidae and Amphimnidae. The numerous differences in the skull structure of the two families make it manifest that the common ancestor is a form far back in Geologic time ; a fact which tends to verify Wiedersheim's statement that the origin of the Coeciliidae is to be sought in the Stegocephalans of the Carboniferous. The well-developed columella auris of AmpJmivia is very probably a character retained from the GanocepJiala and RacJiitomi. In the light of present paleontological and embryological knowledge any detailed phylogeny of Amphibia must be very uncertain. However, the facts at hand seem to me of such significance as to warrant the following table : —

Salientia Amphiumidae

Gymnophiona Urodela Trachystomata

\ /



Embolomeri Stegocephali




Although AmpJiimna is considered a degenerate form, yet certain parts of its structure are highly specialized. The general shape of the cranium presents a marked contrast in comparison with other Amphibian skulls. The great length of the face and the pointed snout are of no phylogenetic significance, as they have been developed by the habits of the animal. The vertebrae with their numerous processes cannot be accounted for on any other ground than that of adaptation to the mode of life which rendered it necessary that a complex trunk

4o8 DA VI SON. [Vol. XI.

musculature should exist, and the required processes for the many attachments. The conical arrangement of fibres in the dorsal muscle reveals a condition quite the opposite of degeneration. The fact that similar muscular cones are found in the alligator (13) and SpJiaenodon does not imply that these three forms are in any way related. The existence of an unusually long and strong temporo-cervical tendon in Amphiuma and DcsviognatJms does not furnish sufficient evidence that they are closely allied, as Cope has tabulated them. These are merely cases of parallelism, as is plainly shown when we take into consideration the marked contrast of the more important structural elements of the two families. Scott (19) has demonstrated this condition of parallelism in numerous mammalian families. " The prismatic, cement-covered molar has been independently developed in many forms : e.g., several of the ruminants, certain pigs, the horses, one of the rhinoceroses (Elasmotherium), the elephants, many rodents, etc. The selenodont molar pattern has been several times independently evolved ; (i) in the true ruminants ; (2) in the camels; (3) in the oreodonts. The spout-shaped odontoid process of the axis has arisen in the true ruminants, the horses, the camels, and, to a certain degree, in the later oreodonts, such as Merychyus." Gegenbaur (20) [p. 669, Fig. 232] has described and figured incorrectly the muscular arrangement in the tail of the fish. The wall of the cone is incomplete adjacent to the spinal column, in all the fish which I have examined. This tendency toward conical arrangement of fibres in the tail of the fish has been evolved by the same mechanical principles that obtain in Amphiuma. Thus it may be noticed that in many respects AmpJdtima is not a degenerate form, but on the contrary possesses highly developed structures; and were it not for the fact that the brain exhibits such primitive characters [Osborn (8), p. 178], I would consider this type the result of progressive rather than retrogressive evolution.



1. WiEDERSHEiM, ROBERT. Das Kopfskelet der Urodelen. Morph.

Jahrb., Ill, 1877.

2. Hay, O. p. American A^atiiralist, Vol. XXII.

3. Hay, O. p. Jotirnal of Morphology, Vol. IV.

4. Cope, E. D. The Batrachia of North America.

5. KiNGSLEY, J. S. American Naturalist, August, 1892.

6. Davison, Alvin, Anat. Anzeiger, April, 1894.

7. Davison, Alvin. Princeton Quarterly Bulletin, April, 1894.

8. OSBORN, H. F. Proc. Acad. Nat. Sc, Phila., 1883.

9. Sarasin, p. and F. Forschungen auf Ceylon. Bd. II, Entwick.

Anat. der Ichthyophis glutinosus.

10. Bronn, H. G. Dr. Klassen und Ordnungen der Amphibien. Bd. V

undVI. Abtheil II.

11. Humphry, G. M. J otirnal of Anatomy and Physiology, Vol. VI.

12. Mivart, St. George. Proc. Zo'dl. Soc, April 22, 1862.

13. Hair, Philip, fournal of Anatomy and Physiology, Yo\. \l.

14. Cope, E. D. fournal of the Academy of Philadelphia, 1866.

15. MacAlister. Morphology of Vertebrate Animals.

16. Wiedersheim, Robert. Die Anatomie der Gj^mnophionen.

17. Eimer, G. H. Theodor. Organic Evolution.

18. Gervais, M. Paul. Zoologie et Paleontologie.

19. Scott, W. B. On the Osteology of Mesohippus and Leptomeryx,

with Observations on the Modes and Factors of Evolution in the Mammalia.

20. Gegenbaur, C. Manuel d' Anatomie Comparde.




Fig. I. Dorsal aspect of Amphiuma's skull, X 2. P, premaxillary ; «, nasal; tn, maxillary; /, frontal; //, parietal; // prefontal; q, quadrate.

Fig. 2. /, premaxillary; vi, maxillary ; /^Z, parasphenoid ; /i^, pterygoid; q, quadrate ; st, stapes ; a, exit for ninth and tenth nerves.

Fig. 3. Lateral view of premaxillary with nasal, maxillary, prefrontal, and portion of frontal removed, X 2. a, brain cavity ; g, frontal bone ; //, canal for exit of olfactory nerve ; /, nasal septum of premaxillary ; /', a break in the septum.

Fig. 4. Dorsal muscles of head, f, pterygo-maxillaris ; h, masseter ; i, cervico-parietalis ; d, e, temporo-cervical tendons ; a, frontal bone ; g, prootic.

Fig. 5. Muscles on dorsal aspect of posterior limb, c, femur ; a, rectus femoris; e, ilio-peroneal.

Fig. 6. Muscles on ventral aspect of posterior limb, c, adductor magnus; e, gracilis ; a, inscriptiones tendineae.

Fig. 7. Great adductor muscle of posterior limb, d, femur; h, tibia ; a, cartilaginous plate ; e, trochanter ; g, pectineus ; l>, ossified disc ; c, pubo-tibialis.

Fig. 8. Transverse section of adult at the fortieth vertebra, a, left side ; i, transversalis-abdominis ; h, obliquus internus ; g, obliquus externus ; k, fascia ; e, fascia split from external envelope.

Fig. 9. Incision through right body wall and both walls reflected. /, lateral fascia line ; c, muscular cones.

Fig. id. The three rows of cones seen when the skin and fascia are removed from right side of Amphiuma. b, triangular rod of fat lying above the vertebrae ; c, tendinous cord connecting apices of cones ; a, fascia of cone reflected to form an inscriptio tendinea.

Fig. II. Vent of the female longitudinally split open, f, capillary tubes ; b, folds of membrane ; a, entrance to the oviduct; d, lip of vent.

P'ig. 12. Young Amphiuma, 78 mm. X ij^. a, gill opening.

Fig. 13. Ilyoid apparatus of adult. ^, epibranchialis ; ^, cerato-branchial ; c, basibranchial ; d, basihyal.

Fig. 14. Brain of young Amphiuma seventy-two millimetres long, a, b, prosencephalon ; c, mesencephalon ; d, metencephalon ; e, spinal chord.

Fig. 1 5. Transverse section of the optic region of young Amphiuma sixtyeight millimetres long, d, eye ; r, retina ; o, orbitosphenoid ; /, parasphenoid ; /, atrophied tentacular canal; e, degenerated gland; a, brain; m, ramus; «, Meckel's cartilage ; h, i, k, hyoid apparatus.

Jotmud of Morphology VoL.M.


ItihUKtr-yAmtTiWimr. FraltUfuTf^M.

Journal of Morphologif Vol. XI.



LilkArsL rWtrntriWinUt, Fr<m]tfurt *M,





Historical Sketch 411

Material and its Preparation 416

acipenser 417

Scaphirhynchops 422

polyodon 424

Lepidosteus 426

Amia 430

Summary 432

References 435

Explanation of Plates 440

Historical Sketch.

The history of the successive discovery of facts relative to the morphology of the enteron in the vertebrates may be said to have undergone a process of evolution comparable to that through which the forms with which it deals have passed.

The first stage may be comprised between the years 1836 and 1869. "In 1836, Dr. Sprott Boyd (9),^ in attempting to discover whether or not there existed an internal membrane of the stomach below that of the mucosa, failed to find what he sought, but saw the mucosa covered uniformly with small tubes, which he described and figured in several species of mammals, birds, and reptiles. The openings of these glands had been seen previous to the year 1836, and also isolated follicular or glandular masses had been noted, but Boyd was the first author who saw the mucosa of the stomach entirely covered with tubular glands from the cardia to the pylorus."

In 1838, Henle and Purkinje (51) each found that these glandular tubes were lined with cells; a year later, Wasmann

1 See References.

412 HOPKINS. [Vol. XI.

(64) discovered in the pig the existence of two kinds of glands, — the gastric or peptic, and pyloric or mucous glands. During this period there was noticed in the gastric glands onlyone kind of element, large rounded and very granular cells, which Frerichs called pepsin cells. Of all the authors, including Kolliker (1856), Milne-Edwards (1859), Leydig (1866), and Klein (1870), who at this time gave attention to the structure of the glands in the fundus of the stomach, no one mentions any other kind of cell.

The second period in the growth of our knowledge concerning the fine anatomy of the enteron dates from the appearance of three articles by Rollet (55), Heidenhain (24), and Ebstein (17) in 1870, and continues to the year 1880.

Rollet distinguished in the glands of the stomach two kinds of cells, which he called " delomorphous " and " adelomorphous " cells. Heidenhain established the same distinction under the names of " Hauptzellen " and " Belegzellen " (principal and border or parietal cells), terms which were also employed by Ebstein, and which have come into general use. Ebstein devoted himself especially to the study of the pyloric glands, and so thorough was he in this that his descriptions and figures have been but little improved upon to the present day. " It was established from this time that in the mammals the base of the cardiac glands of the stomach was occupied by a covering formed of cylindrical cells more or less granular, the principal cells; that the upper portion was covered by very large cells charged with granules, the pepsin cells of Frerichs; and that in the middle of the tube the prismatic cells occupy the axis of the gland tube; the border cells were crowded to the outside, and determined by their exterior projection that bulging aspect of the isolated gastric glands, which one meets in all the previous accurately made drawings." (Pielliet.)

In the second edition of his Tr'aite d' Anatomic, 1873, Sappey gave a morphological description of the glandular tubes in different kinds of vertebrates, but he failed to discern more than one form of cell in the gastric tubule; he says that they are formed of two tunics. The ectal, which is amorphous and homogeneous, presents no apparent difference in the two kinds of


glands. The ental tunic of the peptic glands is composed of large rounded cells which impart an irregular bulging form to the surrounding tunic; the pyloric or mucous glands are lined by small prismatic cells which do not reach to the center of their cavity.

One of the most important contributions that appeared during this period was an article by Edinger (18), in 1877, upon the mucous membrane of fishes. He says that the gastric glands appear phylogenetically first in the class fishes — the Selachians. Consequently the older vertebrates, like the invertebrates, have no specially differentiated portion of the alimentary tract for the purpose of digesting fixed bodies, albumen substances, etc. As to the presence of a stomach in the Dipnoans he is in considerable doubt, but seems to think it rather doubtful if they possess one. Hyrtl, he says, could find no trace of gland openings in the stomach of Lepidosiren paradoxa. In Protopterus, Parker (47) says: "The whole of the mucous membrane of the stomach and intestine is perfectly smooth, and there is no indication of any differentiated gastric or intestinal glands. Cilia are present on the epithelium throughout the stomach and intestine." Among the Teleosts, according to Edinger (18), there are several that have no stomach glands. They are absent in Cobitis fossilis, Gasterosteus pungitius. Tinea vulgaris, Abranus barbio; according to Rathke (53), they are wanting in Blennius ocellatus and Sanguinalentus, Gobius melanostomus, Cyprinus chrysophrasius, and Atherina Boyeri. Probably they fail also in Balistes.

In all these there is found a single somewhat granular cylindrical epithelium without beaker-cells. In such animals digestion must be performed, in part at least, by an intestinal secretion. The gastric glands have probably developed from the ordinary insinking of the alimentary epithelium. Their ontogenesis in the higher vertebrates as well as in those of lower rank indicates the above mode of formation, as shown by KoUiker, Barth, and Laskowsky. In Mammals, Birds, and the Batrachians it has been proved that the epithelium of the stomach glands, at an early stage of their development, is uniform throughout, and only at a later period are those cells

414 HOPKINS. [Vol. XI.

situated in the fundus of the crypts transformed into gastric cells.

The differentiation of cells into principal and border cells does not obtain in fishes; whether it occurs in Amphibia is doubtful. Heidenhain and Trinkler (6i) found in the frog only one kind of cell; Edinger, however, found in the frog cells which gave as sharp color differentiation as one usually finds between the principal and border cells in Mammals and Birds. In Necturus maculatus the examination of several specimens by myself failed to show border cells. In an investigation on the enteron of Necturus by Dr. B. F. Kingsbury ^ no parietal cells were found.

The last stage of advance in our knowledge of this subject comprises the period from 1880 to the present time. During this epoch investigations have been directed principally to the determination of the evolution and mode of regeneration of the glandular epithelial cells and to their physiological significance.

No one, perhaps, has done more towards the solution of these problems than Dr. Nichola Trinkler. His paper (61), " Ueber den Bau der Magenschleimhaut," is devoted almost exclusively to the consideration of the epithelial cells and their transformations. As the result of various experiments and investigations, he concludes that the parietal cells increase in number during digestion, and that the young cells which arise from them by fission move gradually towards the lumen of the gland, and are there transformed into principal cells, and in this manner serve to replace the destroyed principal cells.

From experiments with artificial digestion he shows that the parietal cells of higher vertebrates and also the gland cells of lower forms (Frog, Pike) secrete pepsin, but that the presence of the parietal cells accelerates two or threefold the rapidity of digestion of fibrin and egg-albumen.

Langley (32), in his paper, " On the Histology and Physiology of the Pepsin-forming Glands," confined himself principally to the granules of the gland cells and to changes which

1 The Histological Structure of the Enteron of Necturus maculatus. Proc. Amer. Micro. Soc, Vol. XVI, pp. 18-64, 1894.


occur in them during digestion and hunger. His observations were made upon certain Amphibia and Reptiles.

From the study of mitosis of the epithelial cells, Bizzozero (6) concluded that the cells did not live and die in the place where they originally arose, but that by degrees the deeperlying cells of the epithelium reached the free surface in a manner precisely comparable to that which takes place in the epidermal cells of the skin. In certain glands which had attained their full development he found numerous nuclear figures, premonitors of active cell multiplication. These newlyformed cells, he says, gradually replace the epithelial cells of the free surface of the stomach, which, in certain animals, is in the highest degree desquamous. In a more recent paper (7), in which he examined the enteric glands and epithelium of the mouse, he arrived at the same conclusion as above, namely, that in the mouse, as in the rabbit, the gland epithelium is gradually transformed into the surface epithelium of the mucous membrane.

Pilliet (49), in a paper on the evolution of the glandular cells of the stomach, says that the gastric glands may assume very different appearances, and that by noting the form and structure of the glandular cells at various points of the gastric follicles he has tried to form a general idea, by studying the cells from their appearance to their death, where each of the different conditions were to be fixed in chronological order, and to indicate the particular age to which it corresponds in the life of the cell. He reaches the same conclusion as Bizzozero in regard to the gradual metamorphosis of the glandular into surface epithelial cells.

Other writers on this subject might be cited, but enough has already been said to indicate something of the nature of the work already done, and the forms which have been studied. Comparatively little has been done on the enteron of some of the lower or more generalized forms. Among the fishes or fish-like vertebrates very little attention has been given to that old and interesting group, the Ganoids. Indeed, in the literature that I have been able to examine, only the most meager references are made touching the morphology of the enteric

4 1 6 . HOPKINS. [Vol. XI.

glands and epithelium of this group, and those by only a few observers. To certain of the Ganoids no reference whatsoever was found on this subject.

Material and its Preparation. The forms upon which this paper is based are as follows :

Family Acipenseridae.

Acipenser rubicundus. (Common Lake Sturgeon.) Scaphirhynchops platyrhynchiis. (Shovel-nosed Sturgeon.)

Family Polyodontidae. Polyodon folium ? (Spoon-billed Sturgeon, Duck-billed Cat.)

Family Lepidosteidae. Lepidosteus osseus. (Gar-Pike, Bony Gar, Bill- Fish.)

Family Amiidae. Amia calva. (Bowfin, Dog-Fish, Mud-Fish, Lawyer, Johnny Grindle.)

The material was obtained and placed in the hardening fluid immediately on the spot where the various forms were taken. Specimens of the common sturgeon, Polyodon, and Scaphirhynchops were obtained in the month of February (or the very first of March) from Knoxville, Tennessee. Other specimens of Polyodon and those of Lepidosteus were taken in the latter part of August from the Mississippi River at Ft. Madison, Iowa.

The material was preserved by hardening in picric alcohol (95/0 alcohol, I part; water, i part; picric acid, \f) from- one to two days; then it was placed in G'j'jo alcohol for a day, after which it was kept in 82% alcohol. When needed for use, small pieces from various regions of the enteron were dehydrated in 95% alcohol for one day, then soaked in chloroform twelve to twenty-four hours, after which they were infiltrated and embedded in paraffin. Mercuric chlorid was also used as a hardening agent with very satisfactory results. The fresh tissue was placed in the solution of mercury (HgCb, 5 grams; NaCl, Yi gram; H2O, 100 cc.) for one-half to two hours — a


longer time seems to do no harm. From this the tissue was transferred to 70% alcohol to which a little tincture of iodine had been added. The tissue is washed in this till the alcohol ceases to lose color; then it is transferred to 75% alcohol for a day, after which it is dehydrated and embedded either in paraffin or collodion.

The cilia of ciliated epithelium are perfectly preserved, so far as can be judged, by each of the above-named hardening agents.

Various stains were tried, but the most satisfactory combination was haematoxylin and eosin. For separating the muscular coats 20/^ nitric acid was used. The tissue was left in the acid until it had sufficiently dissolved the connective substance to permit the ready separation of the fibers. This is effected, usually, in from one-half to two days, depending largely upon the temperature, the acid acting more rapidly in a warm atmosphere. When the tissue has been sufficiently macerated, further action of the acid is prevented by placing the tissue in a saturated aqueous solution of alum, plus 2^0 chloral hydrate, in which it may be kept for an indefinite period. This is also a good method for isolating the gastric glands. For isolating the epithelial cells Midler's fluid, picric alcohol, 30-35%, and alcohol, 30-35%, were used; of these Muller's fluid was the most satisfactory.

It is a source of regret that the material could be examined in a fresh condition only in the cases of Amia and Lepidosteus.

The outlines of the figures were made by aid of an Abbe camera lucida; the details were put in freehand.

Common Lake Sturgeon.

General Form of Enteron. — The conformation of this portion of the abdominal viscera is somewhat complex. From the mouth the enteron extends nearly one-half the length of the abdominal cavity, where it doubles upon itself and ascends several centimeters in front of the opening of the pneumatic duct. The ascending portion is somewhat to the left of the descending part; the intervening space is filled by a lobe of the liver. At the point above indicated the ascend

4l8 HOPKhWS. [Vol. XI.

ing part curves upon itself to the right, and extends backward about two-thirds the length of the abdominal cavity, where it changes its direction and ascends to a point a little cephalad of the middle of the cavity ; here it again forms a loop, and thence extends directly to the vent. From this description and the diagram (Fig. i) it will be seen that there are three descending and two ascending parts to the enteron. The oesophagus and stomach are almost wholly included by the two first portions, and it is with these that this paper has chiefly to do. At the pyloric end of the stomach, in an adult sturgeon, the walls attain a thickness of from two to four centimeters. Immediately following this thickened part are the pyloric caeca. These appear like a single organ, but from the occurrence of three distinct openings leading into the mass, as Ryder (56) says, it must be regarded as a system of pyloric caeca three in number"; the tubes soon subdivide into a number of small branches analogous to a racemose gland. The remainder of the enteron will be passed over by simply remarking that the walls appear somewhat thin, and that at the terminal end of the intestine the spiral valve makes seven complete turns, the last one extending nearly to the vent. The peritoneal coat has a uniformly dark, almost black, appearance, due to the deposition of pigment (Fig. 5). Within the peritoneum is the thin longitudinal muscular coat; the great bulk of the muscular tissue, however, is comprised within the circular layer. Along the oesophagus this layer is very thick, and the striated fibers are grouped into large fascicles. In the gastric region the striated muscular fibers are superseded by the unstriated fibers; the circular layer is thinner, and the fascicles are smaller and more compact. No oblique muscular coat was observed.

The various regions of the enteron are not sharply differentiated from each other. In a general way the subdivisions can be recognized, but the boundary line between oesophagus and stomach, for instance, cannot be determined by the gross appearance of this region. A part, at least, of that portion which most observers have called oesophagus ought, it is believed, to be regarded as really a part of the stomach.


Oesophagus. — In his work on the " Fishes of France," Moreau (41) says of Acipenser: "The oesophagus is covered or armed with papillae more or less conical and directed backwards. The stomach is scarcely larger than the oesophagus; it would be difficult to establish the line of separation if the gastric mucosa did not present a different appearance, it being entirely smooth. It is at the commencement of the stomach, a little back of certain small pad-like structures formed by the termination of the oesophagus, that the canal of the air-bladder opens." Ryder recognizes in the alimentary canal three very clearly defined regions. Of these the first, or oesophageal, " extends as far back as the opening into the air-bladder. The gullet proper, . . . upon being laid open, is found to be covered for some distance with backwardly-directed soft fleshy processes, into which its mucous membrane is elevated. At some distance in its course farther back, its lining membranes again become smooth, but slightly folded longitudinally." With but one or two exceptions all the authors mentioned in this paper, who have expressed themselves on the point, state that the pneumatic duct opens into the oesophagus. The epithelium of the latter, according to Leydig (33) and several others, is like that of the buccal cavity, a stratified pavement epithelium. So far as noted, all authors say that there are no glands in the oesophagus. From the above quotations it will be seen that the stratified pavement epithelium extends backward to the opening of the pneumatic duct, and that this region is non-glandular. The correlation of the parts, as found by the writer, is somewhat at variance with the above statements. In my specimen, an adult sturgeon about two metres (six feet) in length, the stratified epithelium, together with the fleshy papillae, disappear at a point about 5 cm. in front of the pneumatic duct opening. Succeeding the stratified is a columnar epithelium which extends uninterruptedly to the pylorus. Considered from a mechanical standpoint, almost every one would say that the stratified should overlap the columnar epithelium, but the exact reverse is the case. At the transition point the stratified epithelium becomes obtusely wedge-shaped ; the deeper layers are overlaid by cylindrical-shaped cells whose

420 HOPKINS. . [Vol. XL

length varies in conformity to the inchned surface covered by them (Fig. 7).

In the writer's opinion this mode of transition may be regarded as a type to which, it is believed, most if not all of the various groups of vertebrates will be found to conform.

Edinger, whose investigations included Cyclostomes, Selachians, one Ganoid (Lepidosteus), and many Teleosts, makes this statement : " Where pavement epithelium is present, it becomes thinner and thinner towards the stomach, the interspersed beaker-cells increase in number and soon form a continuous stratum which extends over one or two layers of the fiat cells, but at the border of the stomach lies directly upon the connective tissue of the mucosa."

In certain mammals (dog, cat) and in one reptile (soft-shelled turtle) the transition from stratified to columnar epithelium has been found to correspond to the above type, Gage (21). Extending from this point nearly to the pylorus, the epithelium is ciliated (Fig. 12). For a short distance immediately beyond the point of transition the mucous membrane appears very similar to that at the pyloric end of the stomach; the epithelium is infolded, forming deep crypts, or follicles, which closely resemble the pyloric glands; in the former, however, the epithelial cells are ciliated, while in the latter they are not. The first few crypts are lined by columnar ciliated cells only, but the true glandular cells make their appearance before we reach the opening of the pneumatic duct. In the latter case the tubes are lined by two kinds of cells, the glandular cells occupying a short segment at the base, and the ciliated cells the remaining portion (Fig. 6). In many cases two or more glands open into a common outlet. From the above facts the writer concludes that the caudal portion of that segment of enteron which most writers have called oesophagus is in reality a portion of the stomach, a conclusion substantiated by Gegenbaur (22) and Milne-Edwards (39).

Stomach. — Although the stomach of certain of the sturgeons has been investigated to some extent, no reference has been found in the literature on the forms relative to an interesting morphological feature found in the present species.


Ciliated epithelial cells have been found in the oesophagus of sturgeons, but no one mentions their presence in the stomach, and several deny their ever existing there.

Concerning the glands in the stomach of the sturgeon Leydig says : " They are short cylindrical sacs . . . lined with great regularity by a clear and delicate cylindrical epithelium, which is continuous at the edge of the gland orifice with that of the surface epithelium; the cells of the two cylindrical epitheliums are distinguished in this, that in the surface epithelium they are larger and distended towards their free extremity by a molecular mass." If the above statement applies to all the glands of the stomach, Leydig's specimen differed greatly from the one studied by me. It has already been noted that in my specimen the greater part of the gastric epithelium is ciliated. Among the ciliated cells are numerous beaker-cells, many of which are open at the free end. The glands of the stomach are of the two ordinary kinds, cardiac and pyloric; the latter occupy but a small area compared to the former. The cardiac glands are differentiated into two very clearly recognizable portions, a superficial or mouth part and a deeper or body portion. The relative length of the two parts through the middle portion of the stomach is as i to 3 nearly (Fig. 9).

On both sides of this area the glandular cells are gradually replaced by those of a more nearly cylindrical form, till finally the tubules are lined by the latter form of cells only. In all the tubules, so far as observed, except those in the pyloric region, the mouth portion is lined by ciliated columnar cells (Figs. 6, 8, 9). No difference was detected between these and the cells of the surface epithelium except that the latter are longer and their attached ends do not end so abruptly. Beakercells are found in the mouths of the glands as well as on the free surface. The nuclei of the epithelial cells stain deeply with haematoxylin ; the cell-body stains sparingly with eosin unless the stain is left on for some time; this applies equally well to the gland cells proper. The part of the cell next to the so-called basement membrane stains more deeply than the part next to the lumen of the gland. All the cells in the glandular portion of the tubule have approximately the same

422 HOPKINS. [Vol. XI.

form, — irregularly cylindrical or cubical-shaped; those next to the ciliated cells of the mouth may perhaps be somewhat more nearly cylindrical.

The attached ends of most of the glandular cells are continued out into a sort of sheet-like prolongation which appears to anchor the cells in place. Along the middle of the gland the cells overlap each other in a manner very similar to the scales of a pine cone, the apex of the cone corresponding to the fundus of the gland. From numerous instances in which the cells were traced directly into the mucosa it is believed by the writer that in the Ganoids, at least, a basement membrane does not exist. In Edinger's description of the stomach glands in fishes he says that they possess no membrana propria or basement membrane, but that the epithelium borders directl}^ upon the connective tissue of the mucosa. The cell threads lie over each other like tile, so that they surround the upper part of each gland with a kind of membrane which is composed of innumerable fine threads, and sharply fixes the boundaries of the gland. This appearance is obtained only in the cardiac glands. In the pyloric region the glands are lined throughout by cylindrical cells; in these glands no cilia were found. "The differentiation of parietal cells, so far as is known, does not occur in the class fishes, but is a phylogenetic occurrence which appears much later in the vertebrate series." (Edinger.)


Owing to lack of material, almost nothing can be said regarding the form and appearance of the enteron in this genus. The stomach is recurved upon itself in a manner similar to that of the common sturgeon. At the pyloric end the muscular walls are thickened, but not to such a marked degree as in the preceding. To all appearances the pyloric caeca are like the common sturgeon's. The peritoneal coat is unpigmented. In the oesophageal portion of the enteron are numerous large fleshy papillae; these disappear some distance in front of the pneumatic duct opening. In a specimen whose enteron, when straightened, measured about 1 5 cm. from the air-duct to the caeca, the papillae extended only to within about 3 cm. of the


pneumatic opening. The epithelium of this region is stratified; the surface layer is composed chiefly of large beaker-cells almost bladder-like in form; the nuclei are quite small and crowded down close to the attached ends of the cells. Near the place where the papillae disappear the stratified is succeeded by a ciliated columnar epithelium. The arrangement of the cells at the point of transition of the two epitheliums is as in the preceding form, i.e., the ciliated cylindrical cells overlie the deeper layers of the stratified oesophageal epithelium as in Fig. 7.

The segment of enteron between this point and the opening of the pneumatic duct is, from the presence of glandular cells, regarded as the cephalic part of the gastrium. The stomach glands of the individual examined were not so long as those in the sturgeon; the comparative length of the mouth and glandular portion was, however, about the same proportion, i to 3 nearly (Fig. 13).

The mouths of the glands, except the pyloric, are lined by ciliated cells resembling those of the surface epithelium (Fig. 16). Among the ciliated epithelial cells are many greatly distended beaker-cells whose contents are coarsely granular and very sharply differentiated from the neighboring cells (Fig. 16). The granulation extends into the cell to the level of the nucleus. In some instances the granular mass could not be seen projecting beyond the level of the free ends of the cells, but in others this was noticed. Probably in the former the theca had not yet ruptured and the contents exuded as in the latter. The nuclei of the great majority of these cells were situated at a higher level than the others, as if the swelling caused by the granular accumulation within had started the cells somewhat from their normal position. Although these two kinds of cells are so different in appearance, it is not easy to determine in precisely what this difference consists.

Trinkler thinks that the only difference between beaker-cells and the ordinary cylindrical cells is, that in the former the metamorphosis of the protoplasm is more complete; he considers the beaker-cells as simply a later condition of the cylindrical cells.

424 HOPKINS. [Vol. XL

The bodies of the gastric glands are made up of irregularly cubical-shaped cells which overlap each other as in the sturgeon. The cells are granular in appearance; the nuclei are large, and contain a distinct nucleolus. That portion of the cell next to the lumen of the gland stains much less deeply than the basal half of the cell. Usually more than one gland opens in a single mouth ; the usual number was two, but frequently three or more were noted. The glands lie in close proximity to each other, there being but little intervening connective tissue.

In surface sections, i.e., sections at right angles to the long axis of the gland, it was noticed that frequently several glands were united, as it were, into a bundle; the connective tissue surrounding these forming a thicker layer than around the individual glands. As seen in cross-section, the number of epithelial cells of a gland varies from eight to twelve in most instances (Fig. 14).


The appearance of the enteron in this genus differs somewhat from either of the preceding. Its general outline is shown in Fig. 3. The papillae at the cephalic end of the oesophagus are small and numerous; they extend to within about one or one and one-half centimeters of the pneumatic duct. The walls of the stomach are considerably thicker than in either of the preceding specimens. At the pyloric valve the muscular walls are thickened as in Scaphirhynchops. The pyloric caeca are relatively much larger and more deeply subdivided (Fig. 3, c). The caecal cavity communicates freely with that of the intestine; it subdivides into four main branches, corresponding to the four main lobes into which the gland is divided. The intestine is short, morphologically speaking, but the spiral valve, with its six complete turns, really forms a long intestine, in a physiological sense. The last turn of the valve is about two centimeters from the vent. The peritoneal coat is almost entirely unpigmented, there being only a few small pigment patches.


The mode of epithelium transition and the presence of follicles, or crypts, in front of the pneumatic duct opening corresponds to those forms already noted, except that the follicular area is considerably shorter than in either of the preceding. The epithelium of the cardiac portion of the stomach is ciliated (Fig. 21). Owing to the shortness of the cilia and the presence of a thin layer of extraneous matter on the surface of the epithelium, considerable difficulty was experienced in detecting the cilia in this specimen. Among the ciliated epithelium cells were a great many beaker cells; the peripheral end of these takes scarcely any stain, and appears to be open, or without a membrane over the free end. The nuclei of the cells are very large, some oval and others circular in outline, and many of them contained several darkly stained granules. The cells lining the mouths of the glands are much shorter than those on the free surface, but nevertheless they possess fully as large nuclei as the latter; in other respects no differences between the cells in the two situations were observed. In this genus the mucous membrane presents certain features not found in any of the others. The gastric glands are so convoluted that it is almost impossible to get sections in which the whole length of the gland appears; usually the glands are cut at various angles to their long axis, so that perhaps two or three successive sections must be examined in order to study the gland throughout its whole extent. The relative lengths of the mouth and body of the gland are nearly equal (Fig. 18). The lumen of the glands was specially large in the specimen examined, and appeared to be of about the same size at all points except the mouth, which, of course, was somewhat larger. In several instances it was noted that the diameter of the lumen measured at least one-quarter the whole width of the gland (Fig. 20). Several glands open into a single mouth. The gastric cells which line the body of the glands have, at their attached ends, a very slight thread or sheet-like continuation for attaching them to the mucosa. The tile-like overlapping of the cells is very slight ; often the dividing line between two cells extends directly across at right angles to the long axis of the gland. The cells are finely granular and stain lightly, especially the

426 HOPKIXS. [Vol. XI.

part next to the lumen of the gland; the nuclei are large, either oval or circular in section, and situated close to the attached ends of the cells. In the fundus of the glands the nuclei, especially those of oval outline, appear to be undergoing division (Fig. 20). In many of the cells there may be seen extending across the short diameter of the nucleus a very deeply stained band of chromatin; in these cells no nucleoli can be seen. In the cells whose nucleus is circular in outline but one nucleolus is present, as a rule, but sometimes two or more were seen. In several instances nuclei were found somewhat constricted in the middle, and still others where two distinct nuclei were in direct apposition, each being somewhat smaller than the original parent nucleus from which, to all appearances, the two nuclei had been formed.

This appearance of nuclear division in the fundi of the glands is in perfect accord with the statements of several authors who have shown that in the adult of certain mammals the centers of growth of the enteric epithelium are situated in the fundus of the glands. If this is true of fishes, it would be interesting to know the exact concomitant changes which the cell must undergo in the process of its transformation from a non-ciliated glandular cell into a ciliated epithelial cell. The glands are quite widely separated from each other by the intervening mucosa, more so than in any of the other forms. The diameter of the glands is also less than in the two preceding forms, but this may be simply an individual variation. The pyloric glands are lined by narrow cylindrical cells, among which occasionally were seen greatly distended, coarsely granular beaker-cells. No cilia were found in this region, either in the glands or on the surface. The thickness of the stomach walls was previously alluded to; it is due in great part to the excessive thickness of the submucosa, which forms a layer as thick if not thicker than that formed by the muscular coats.


In a specimen measuring 55 centimeters from the tip of the snout to the tip of the tail the entcron extends in a direct line


backward for a distance of 18 centimeters, where the first refolding occurs: the other flexures are indicated in Fig. 2.

There is no line of demarcation between oesophagus and stomach unless, as stated by Balfour and Parker (3), " a glandular posterior region be regarded as the stomach, a non-glandular anterior region forming the oesophagus." The macroscopic appearance of this region gives no indication of the position of the boundary line between the two parts. The intestine is of about the same size in all parts. The pyloric caeca are so small and numerous that the caecal mass formed by them presents an almost brush-like appearance. The caecal cavity extends into the finest subdivisions of the gland. The peritoneal coat is unpigmented. The short spiral valve makes only two or two and one-half turns, and ends about two centimeters from the vent. In connection with this part of the intestine is a structure referred to by Balfour and Parker as follows: "The posterior part of the intestine, from the beginning of the spiral valve to the anus, is connected with the ventral wall of the abdomen by a mesentery. . . . This mesentery, which together with the dorsal mesentery divides the caudal section of the body into two lateral compartments, is, we believe, a persisting portion of the ventral mesentery which, as pointed out by one of us,i is primitively present for the whole length of the body-cavity. The persistence of such a large section of it as that found in the adult Lepidosteus is, so far as we know, quite exceptional. . . . The small vessel in it appears to be the remnant of the subintestinal vein." My specimen agrees perfectly with the above statement; the ventral edge of the mesentery in this specimen measures at least 5 cm. in length. The blood-vessel, supported by the mesentery, divides into two nearly equal branches, one of which extends forward and the other backward, to be distributed to the abdominal parietes. No reference to a ventral mesentery in any of the other ganoids has been noticed, and the writer has not had opportunity of examining other forms than Lepidosteus and Amia. No ventral mesentery was found in Amia.

The papillary structures found in the oesophagus of the

1 Comparative Embryology, Vol. II.

428 HOPKINS. [Vol. XI.

three preceding forms do not exist in Lepidosteus, owing, possibly to the great development of the dental armament, which practically precludes the escape of prey, when once fairly within the mouth, without any intervention of subsidiary structures.

In Lepidosteus^ the non-glandular cephalic portion of the enteron extends from the pneumatic duct caudad a distance of 8 to 10 cm. before the gland structures of the stomach appear. This part, which will be called oesophagus, is covered by a ciliated columnar epithelium interspersed with numerous beakercells (Fig. 17). In front of the pneumatic opening the epithelium is stratified, being composed principally of large mucous cells, but among these are a considerable number of cylindrical and fusiform cells. Owing to the rounded form of the beaker or mucous cells, the nuclei, situated close to the attached end of the cells, presents a disk-like or saucer-shaped appearance. Of the epithelium of the stomach of fishes Edinger says (Ueber die Schleimhaut des Fischdarmes nebst Bemerkungen zur Phylogenesis der Driisen des Darmrohres, p. 666), " Cilia, where such cover the mucous membrane of the oesophagus, disappear in the stomach. . . . The epithelium of the stomach is a cylindrical epithelium which never bears cilia." Possibly these statements are true for all the specimens he examined, but the probabilities are that he overlooked the cilia in the stomach of Lepidosteus. The grounds for this belief are, first, he himself found ciliated epithelial cells in the oesophagus of Lepidosteus ; and second, as already stated, the oesophagus in the present specimen is ciliated, but ciliated cells were found in the stomach as well. Possibly the reason for Edinger not finding them was due to defective preservation of the cilia in this region. Cilia were not found over the whole extent of the cardiac region, but were confined to its cephalic part. The ciliated cells did not form a continuous layer but were more or less scattered among the nonciliated epithelial cells of this region.

The epithelial cells are long and slender; the thread-like

1 Size of specimen not known, but the hardened specimen of oesophagus and stomach measured somewhat over twenty centimeters.


ends can be traced down among the connective-tissue corpuscles of the mucosa; not the slightest indication of a basement membrane was noticed. From many of the beaker-cells of the oesophagus can be seen conical or rounded projections of mucus, but, unlike Scaphirhynchops, these mucous masses are stained very little. The transition from oesophagus to stomach is gradual. Towards the gastric end of the oesophagus the epithelium gradually infolds, forming short, broad follicles. These are soon replaced by the true gastric glands of the stomach. The latter are lined by large granular, irregularly cubicalshaped cells, except the part near the exit of the gland upon the surface, where the cells are of a broad cylindrical shape, but there is no sharp line of demarcation between the two. The glands of Lepidosteus, unlike those of the other forms, have, for the most part, no portion lined by cells of the same size and appearance as those of the surface epithelium. The part corresponding in position to the mouth is lined by cells two or three times as large as those of the surface. The relations of the cells at this point are such as to give the appearance of the gland having been forced up through the surface epithelium like a wedge (Figs. 23 and 26).

As a rule the glands open singly upon the surface, but sometimes two or more have a common outlet. The cells along the upper half of the glands show clearly the thread-like continuations extending from their basal end ; in the deeper part of the gland they are somewhat shorter. The nuclei of the gland cells are circular, those of the surface epithelium oval in outline. No indications of cell division were noted.

As we approach the pylorus the glands grow shorter; at the same time the upper part becomes lined with cells like those of the surface epithelium, till finally the glands are lined with columnar cells only; these glands are considerably shorter than the cardiac glands. In the cardiac region the glands are close together, and of about the same diameter at all parts (Fig. 26). The lumen is distinct and enlarged somewhat just before it opens upon the surface. Cilia were not found in any part of the lumen of the gland in any portion of the stomach.

430 HOPKINS. [Vol. XI.


In this, the most teleostoid in appearance of Ganoids, one might, perhaps, expect to find morphological features unlike those of the other members of the group. Macroscopically, the enteron of Amia does differ to a certain extent from the others, but microscopically the resemblances are very close. But little, if any, taxonomic value can be ascribed to this organ so far as could be ascertained from the individuals examined. The general form of the enteron in Amia is shown in Fig. 4. The chief differences between this and the preceding are that in Amia the gastric portion of the enteron is, comparatively, very much enlarged and of somewhat different shape, and there are no pyloric caeca. The spiral valve makes four or four and one-half turns, the last ending a little more than a centimeter from the vent.

In the adult specimens examined there was no distinct ventral mesentery connecting this part of the intestine with the ventral abdominal wall as in Lepidosteus. Papillae were not found back of the pharyngeal dental pads. The fibers of the circular muscular coat are striated as far as to the place where the glands of the stomach first appear, just caudad of the pneumatic duct opening; elsewhere unstriated fibers were found. The pneumatic duct opens near the junction of the oesophagus and stomach. Between oesophagus and stomach is a short region occupied by rather short, broad follicles lined by columnar ciliated cells; the true gland cells are first seen at a point a little beyond the opening of the pneumatic duct.

According to Schulze (58) the epithelial cells of the stomach in all vertebrates are open, i.e., the free ends of the cells are not covered by a cell wall.

The mucus which these open cells secrete, Edinger thinks, is for the purpose of protecting the cells themselves from the digestive action of the secreted fluids. Brinton (i i) also seems to hold the same opinion. He says: "The protection of the stomach from its own secretion is effected mainly by the salivary and other secretions which enter it from the oesophagus and the duodenum. . , . For units of mucous membrane.


fishes seem to have the most powerful gastric digestion." To the writer these opinions appear unsatisfactory from the fact that in the American Ganoids, at least, the ciliated character of the gastric epithelium would tend to prevent the formation of a distinct mucous coat over the surface of the stomach. But apart from this, it is believed that the vital properties of the cells are sufficiently potent to withstand any deleterious effects which the gastric secretions might possibly have upon them. Edinger thinks that the functions of the mucus are to thin the chyme and to form a protective covering over the hard, indigestible bodies, as sand, shells, etc., which find their way into the stomach. He says that such foreign bodies, surrounded by a tough mass of mucus, are frequently found in the intestine. This explanation seems more reasonable than the preceding.

Ebstein found open as well as closed cells, and is of the opinion that during digestion the membrane of the closed cells is ruptured. In all the specimens examined by me both open and closed cells were found. The epithelial cells of Amia's stomach are very slender (Fig. 28), and the attached ends are continued into long thread-like processes which intertwine with the subadjacent mucosa. Ciliated cells are found uninterruptedly from the cephalic end of the oesophagus to within two or three centimeters of the pylorus. Scattered among these were many open beaker-cells, — the two kinds of cells being in about equal numbers. From the open end of many of the beakercells a mucous mass of varying size was seen projecting a variable distance beyond the free end of the cells (Fig. 27). At the cardiac end of the stomach the gastric glands appear as short tubes at the base of the follicles above mentioned. They, however, rapidly increase in length, and over the middle portion of the stomach make up the greater part of the tubule (Fig. 25). As the pyloric region is approached, the glandular part decreases in length, and about two centimeters from the pylorus disappear. From this point on, the glands are lined with cells like those forming the surface epithelium, only shorter. In the cardiac region the mouths of the glands are short, and are lined by ciliated cells (Fig. 24). The cells of

432 HOPKINS. [Vol. XI.

the body of the gland are, for the most part, cubical in longisection of the gland, but for a short distance below the mouth the cells are more nearly cylindrical in outline. In Fig. 24 it will be noticed that the cells lining the mouth of the gland are placed obliquely to its long axis. In all the forms examined this arrangement holds true, only to a more marked degree than there represented. Frequently cells were seen so bent that the angle formed equaled at least a right angle. In all cases the convexity of the cells projected towards the exit of the gland; the attached end of the cells reached a much lower level than the opposite end.

In the pyloric region the glands are quite widely separated from each other; the lining cells of these are situated at nearly right angles to the long axis of the gland. Towards the pyloric valve the glands become shorter and gradually disappear, or pass into the crypts of the intestine. Cilia were not found in these glands or on the surface epithelium.


The salient points of this paper may be summarized briefly as follows :

1 . The gastric glands were first discovered by Sprott Boyd in 1836. Two years later Henle and Purkinje each discovered that the glands were lined with cells.

2. In 1870 Rollet distinguished in the glands of the stomach two kinds of cells, which he termed " delomorphous " and "adelomorphous" cells. Heidenhain, in the same year, noted the same distinction under the names of "principal cells" and "parietal cells."

3. Phylogenetically, the gastric glands first appear in the Selachians.

4. The ontogenetic development of the glands indicates that they have developed from simple insinkings of the alimentary epithelium.

5. The differentiation of parietal cells does not occur in the class fishes, and it is doubtful if it does in the Amphibia.

6. The centers of multiplication of the enteric epithelial cells appear to be at the fundus of the glands.


7. The boundary line between oesophagus and stomach can be determined only by microscopical examination.

8. In the sturgeon, Scaphirhynchops, Polyodon, and Amia deep follicles are found cephalad of the pneumatic duct opening. In Lepidosteus the pneumatic duct opens very far in front of the place where the follicles first make their appearance. In sturgeon and Scaphirhynchops true gastric cells (as inferred from their form and appearance) are present cephalad of the pneumatic duct; that is, to all appearances, in these two last named forms the pneumatic duct opens into the stomach.

9. The typical form of transition from the stratified epithelium of the oesophagus to the columnar epithelium of the stomach is illustrated in Fig. 7. The columnar and stratified epitheliums are wedge-shaped at the place of transition, the columnar overlying the stratified epithelium.

10. A basement membrane could not be demonstrated in any of the individuals examined.

1 1 . The ontogenetic development of the glands shows that primitively they were lined throughout their whole length by cells like those of the surface epithelium. The true glandular cells are a later specialization; hence those glands with a comparatively long mouth and short glandular portion are regarded as more primitive than those with a short mouth and long glandular portion.

12. From the above statement we conclude that the gastric glands of Lepidosteus are somewhat more highly specialized than those of Amia, and that both these are more highly developed than in any other members of the order.

13. Pyloric caeca are present in all except Amia; a spiral valve is present in all, being most rudimentary in Lepidosteus.

14. Ciliated cells were found in the oesophagus and stomach of all the members of the group examined.

15. Ciliated cells were not found in the pyloric region of the stomach, either on the free surface or in any portion of the glands.

16. The significance of cilia in the enteron is doubtful. Trinkler thinks they have not much meaning; he regards them

434 HOPKINS. [Vol. XI.

as residual structures of the embryonic period. Doubtless they are of little importance. Possibly they facilitate the dissemination of the gastric juice and other fluids secreted by the stomach.


The writer is indebted to Professors Wilder and Gage of the Anatomical Department for kindly assistance and encouragement, for material, and for the loan of numerous papers bearing upon the subject; for all of which most grateful thanks are extended.

To Professor H. E. Summers, for material sent by him from Knoxville, Tenn.; to Mr. M. G. Schlapp, special student in the University, for procuring material from the Mississippi River; and finally to the Buffalo Fish Co., of Buffalo, N. Y., for the generous supply of sturgeon material.



I. Agassiz, L. Living Specimens of young Gar-pikes from Lake Ontario. Proc. Boston Soc. Nat. Hist., Vol. VI, 1859.

1. Agassiz, A. The Development of Lepidosteus. Proc. Amer. Acad.

Arts and Set., Vol. XIV, No. 4, Plates I-V, 1879.

2. Agassiz and VoGT. Anatomiedes Salmones. 1845 (cited by Valatour).

3. Balfour, F. M., and Parker, W. N. On the Structure and Devel. of

Lepidosteus. Phil. Trans. Roy. Soc. London, Part II, pp. 359442, PI. XXI-XXIX, 1882.

4. BiEDERMANN. Untersuchungen iiber das Magenepithel. Sitzungs berichte der Kais. Akad. der Wissenschaften zu Wien. Aprilheft. Jahrgang 1875, Bd. LI (Ref. from A. f. M. A., Bd. XIV, p. 189).

5. BiscHOFF. Ueber den Bau der Magenschleimhaut. Miillers Archiv

fiir A/iat., Phys. ttnd wiss. Medicin, pp. 503-525, Taf. XIV u. XV, 1838.

6. BizzozERO, G. Ueber die Regeneration der Elemente der schlauch formigen Driisen und des Epithels des Magendarmkanals. Anat. Anz., Bd. Ill, pp. 781-784, 1888.

7. BizzozERO, G. Ueber die schlauchformigen Driisen des Magen darmkanals und die Beziehung ihres Epithels zu dem Oberflachenepithel der Schleimhaut. Arch./, viikr. Anat., Bd. XL, pp. 325375. October, 1892.

8. Blanchard, R. Sur la presence de I'epithelium vibratile dans I'intes tine. Zool. Atiz., Bd. IV, p. 637, 1880.

9. Boyd, Sprott. On the Structure of the Mucous Membrane of the

Stomach. Edinbiif-gh Med. and Surg. Journ., XLVI, pp. 382-404, 1836 (cited by Valatour).

10. Braun, M. Zum Vorkommen von Flimmerepithel im Magen. Zool.

Anz., Bd. Ill, p. 568, 1880.

11. Brinton, W. Experiments and Observations on the Structure and

Function of the Stomach in the Vertebral Class. Proc. Roy. Soc. London, Vol. XI, pp. 357-359, 1860-62.

12. Cajetan, J. Ein Beitrag zur Lehre von der Anatomie and Physio logic des Tractus intestinalis der Fische. (^Inaugural-Dissertation zur ErlanguHg der Doctorwicrde bei der medicinischen Facultdt.') Bonn, 1883.

13. Cattaneo, G. Ulterior! ricerche sulla struttura delle glandule pepti che dei selaci, ganoidi e teleostei. Risposta al dot. C. Bergonzini. Boll, scient., VIII, 90-99, Pavia, 1886.

14. CuviER, G. Leqons d'Anatomie Comparde. Tome IV (2d. edition),


436 HOPKINS. [Vol. XL

15. CuviER, G. Le R^gne Animal. Tome VII (Poissons), Paris.

16. DoxiTZ, W. Ueber die Schleimhaut des Darmkanals. Mailer's

Arch.f. Anat., pp. 367-406, Taf. X, 1864.

17. Ebstein, W. Beitrage zur Lehre von Bau und den physiologischen

Funktionen der sogenannten Magenschleimdriisen. Arch./, mikr. Anat., Bd. VI, pp. 515-538, Taf. XXVIII, 1870.

18. Edinger, L. Zur Kenntniss der Driisenzellen des Magens, besonders

beim Menschen. Arch.f. mikr. Anat., Bd. XVII, pp. 193-21 1, Taf. XVI, 1880.

19. Edinger, L, Ueber die Schleimhaut des Fischdarmes, nebst Be merkungen zur Phylogenese der Driisen des Darmrohres. Archiv fiir mikr. Anat., Bd. XIII, pp. 651-692, Taf. XL u. XLI, 1877.

20. Franque, H. Amiae calvae Anatomiam. Berlin, 1847 (i plate).

21. Gage, S. H. The Transition from stratified to columnar Epithelium.

Abstract in " The Association of American Anatomists : History, Constitution, Membership, and Abstracts of Papers for the years 1888-1889, 1890." Washington, D. C, 1891.

22. Gegenbaur, C. Elements of Comparative Anatomy. (Translated

by F. Jeffrey Bell.) 1878.

23. Hamburger, E. Beitrage zur Kenntniss der Zellen in den Magen driisen. TaL XIII. Arch.f. mikr. Anat., Bd. XXXIV, pp. 225235, 1890.

24. Heidenhain, R. Untersuchungen iiber den Bau der Labdriisen.

Arch. f. jnikr. Anat., Bd. VI, pp. 368-406, TaL XX and XXI, 1870.

25. Hopkins, G. S. On the Digestive Tract of some North American

Ganoids. Abstract in Proc. Am. Assoc. Adv. Sci., Vol. XLI, pp. 197-198, 1892.

26. Hopkins, G. S. Structure of the Stomach of Amia calva. Proc.

Amer. Soc. of Microscopists. Thirteenth Annual Meeting, pp. 165169, 1890.

27. Huxley, T. H. The Anatomy of Vertebrated Animals. 1872.

28. Jordan, D. S. Manual of Vertebrated Animals. 1890.

29. Klein, E. On the Ciliated Epithelium of the Oesophagus. Quart.

foiirn. Micr. Sci., Vol. XX, p. 476.

30. Klein and Verson. The Intestinal Canal. Strieker s Manual of

Histology, pp. 342-396, 1872.

31. Kowalevsky. Die Entwickelungsgeschichte der Store. Vorlaufige

Mittheilung. Bull. Acad. Imp. Sci. St. Petersbourg. Tom. XIV,

Col. 3i7-325> 7 figg-> 1870 (cited).

32. Langley, J. N. On the Histology and Physiology of Pepsin-forming

Glands. Phil. Trans., Vol. CLXXII, Pt. Ill, pp. 663-711, 1881.

33. Leydig, F. Traitd d'Histologie. 1866.

34. Leydig, F. Einige histologische Beobachtungen iiber den Schlamm peitzger. Miillcr's Arch, f Anat., pp. 3-8, 1853.


35. List, J. H. Ueber Becherzellen und Leydig'sche Zellen (Schleim zellen). Arch. f. vtikr. Anat., Bd. XXVI, pp. 543-552, Taf. I, 1886.

36. List, J. H. Ueber den Ban, die Sekretion uad den Untergang von

Driisenzellen. Biol. Centralb., Bd. V, pp. 698-704, 1885-6.

37. Macallum, a. B. Alimentary Canal, Liver, Pancreas, and Air bladder of Amiurus catus. In Contributions to the Anat. of Amiurus. Pj'oc. of the Canad. Inst. Toronto, N.S., Vol. II, No. 3, 1884.

38. Marchi, p. Beobachtungen fiber Wimperepithel. Arch. f. mikr.

Anat., Bd. II, Taf. 1, pp. 467-72, 1866.

39. Milne-Edwards. Lemons sur la Physiologie. Tome II et VI.

40. MiNOT, C. S. Human Embryology. 1892.

41. MoREAU, E. Histoire Naturelle des Poissons de la France. Tome I,

Paris, 1 88 1.

42. MuLLER, M. J. Memoire sur les Ganoides et sur la Classification

Naturelle des Poissons. Annals des Sci. Nat., 3d series. Tome IV,


43. MiJLLER, J. Ueber den Bau und die Grenzen der Ganoiden. Abh.

d. Berl. Akad. d. Wisscnsch., pp. 11 7-2 16, 1844.

44. Neuman, E. Flimmerepithel im Oesophagus menschlicher Embry onen. Arch.f. mikr. Anat., Bd. XII, pp. 570-574, 1876.

45. Nussbaum, M. Ueber den Bau und die Thatigkeit der Driisen.

Arch. f. mikr. Anat., Bd. XIII, pp. 721-755, Taf. XLIII, 1877. Ibid., Bd. XXI, pp. 296-351, Taf. XV-XVIII, 1882.

46. Ogneff. Einige Bemerkungen fiber das Magenepithel. Biol.

Centralb., Bd. XII, pp. 689-692, 1892.

47. Parker, W. N. Anatomy and Physiology of Protopterus annectens.

Proc. Roy. Soc, Vol. XLIX (Abstract, p. 551), 1891.

48. Partsch, C. Beitrage zur Kenntniss des Vorderdarmes einiger Am phibien und Reptilien. Arch. f. mikr. Anat., Bd. XIV, pp. 179203, Taf. XII, 1877.

49. PiLLiET, A. Sur revolution des cellules glandulaires de I'estomac

chez I'homme et les vertebres. Jonrn. de VAnat. et Phys., etc., Tome XXIII, pp. 463-497, PI. I, 1887.

50. PiLLlET, A. and Talat, M. Sur les diffdrents stades evolutifs des

cellules de I'estomac cardiaque. Comptes Rendiis des stances et tnhnoires de la societe de Biologic, Tome III, 8e s^rie, 1886.

51. Purkinje. Ueber den Bau der Magendriisen. In Bericht iiber die

Versammlung deutscher Natnrforscher und Aerzte in Prag, 1838 (cited by Valatour).

52. Rabl-Ruckhard. Einiges iiber Flimmerepithel und Becherzellen.

Miillers Arch.f. Anat., pp. 72-87, 1868.

53. Rathke, H. Zur Anatomie der Fische. (Tv/o papers.) Miiller's

Arch.f. Anat., pp. 170-186, 1836, and pp. 335-356, 1837.



54. REGificzY, E. N. Ueber die Epithelzellen des Magens. Arch. f.

mikr. Anat., Bd. XVIII, one wood-cut, pp. 408-411, 1880.

55. RoLLET, A. Bemerkungen zur Kenntniss der Labdriisen und der

Magenschleimhaut. Untersuchungen aus dem histitute fiir Physiologic U7id Histologic in Gratz, II. Heft, 1871 (cited by Nussbaum).

56. Ryder, J. A. The Sturgeons and Sturgeon Industries of the Eastern

Coast of the United States. Extract from the Btilktin of the U. S. Fish Commission, Vol. VIII, PI. XXXVII-LIX, 1888. t)"]. Sappey. Splanchnologie-Embryologie. Traite d' Anatomic descriptive. 36 ^d., Tome IV, 1879.

58. SCHULZE, F. E. Epithel- und Driisen-Zellen. Schjcltzes Arch,, Bd.

Ill, pp. 137-203, Taf. VI-XII, 1867.

59. Sertoli, E. and Negrini. Contribuzioni all' anatomia della mucosa

gastrica. Arch, di med. vet., Bd. Ill, pp. 168-180, PI. I, Milano, 1878.

60. Sewall, H. The Development and Regeneration of the Gastric

Gland Epithelium during Foetal Life and after Birth, fourn. Physiol., Vol. I, pp. 321-334, PI. I, London, 1878.

61. Trinkler, N. Ueber den Bau der Magenschleimhaut. Arch. f.

mikr. Anat., Bd. XXIV, pp. 174-210, Taf. X u. XL 1S85.

62. Valatour, M. M. Recherches sur les Glandes gastriques et les

Tuniques musculaires du Tube Digestif dans les Poissons osseux et les Batraciens. Annates des Sciences Naturelles, 4e serie, Tome XVI, pp. 218-284, 1861.

63. Van Gehuchten, A. Le Mechanisme de la Sdcr^tion. Anat. Anz.,

VI. Jahrgang (1891), No. i, pp. 12-25 (7 Fig.).

64. Wasmann. De digestione nonnulla. 1839 (cited by Valatour).

65. Wiedersheim. Comparative Anatomy of Vertebrates. (Translated

by W. N. Parker.)

66. Wilder, B. G. Notes on the North American Ganoids, Amia, Lepi dosteus, Acipenser, and Polyodon. Proc. Amer. Assoc. Adv. Sci., Vol. XXIV, pp. 151-196, 3 plates, 1876.

67. Wilder, B. G. Gar-Pikes, Old and Young. Popular Sci. Monthly,

Vol. XI, Nos. 61, 62, pp. 1-12 ; 186-19S (10 Fig.), 1877.




All the figures of this plate except 2, 3, and 4 are from the common sturgeon.

Figs. 1-4. Diagrams of the enteron :

Fig. I. Common sturgeon. A pyloric caeca.

Fig. 2. Lepidosteus. B pyloric caeca.

Fig. 3. Polyodon. C pyloric caeca.

Fig. 4. Amia.

Fig. 5. Pigment cells in the peritoneum of the enteron.

Fig. 6. Gland from near the cephalic end of the stomach showing the long upper portion of the tubule lined by columnar ciliated cells and the short glandular portion at the base.

Fig. 7. This shows the mode of transition from the stratified epithelium of the oesophagus (or pharynx) to the ciliated columnar epithelium of the succeeding part.

Fig. 8. Enlarged view of Figs. 6 and 9 at points A and B.

Fig. 9. Gastric gland from near the middle of the stomach.

Fig. 10. Longisection of gastric gland.

Fig. IX. Transection of the same.

Fig. 12. Ciliated epitheUum of the stomach.


I'l. .\'.VT

Hopkins. del.



Fig. 13. Two gastric glands opening into a single mouth. (Scaphirhynchops.)

Figs. 14 and 15. Transection and longisection of gastric gland of Scaphirhynchops.

Fig. 16. Enlarged view of Fig. 13 showing the ciliated character of the cells in the mouth portion of the gland. A, greatly distended beaker-cell.

Fig. 17. Epithelium from the oesophagus of Lepidosteus showing ciliated cells and beaker-cells.

Fig. 18. Gastric glands of Polyodon.

Figs. 19 and 20. Transection and longisection of gland (Polyodon). A, nucleus undergoing division.

Fig. 21. Epithelium from the stomach of Polyodon showing ciliated cells and beaker-cells.

Fig. 22. Transection and longisection of gastric gland (Amia).

Fig. 23. Mouth of gastric gland, of Lepidosteus, showing the relation of the cells of the gland to the ciliated epithelial cells on the surface of the stomach.

Fig. 24. Mouth of gastric gland of Amia, showing the ciliated cells lining the mouth and also a few beaker-cells.

Fig. 25. Gastric gland (Amia).

Fig. 26. Gastric gland (Lepidosteus).

Fig. 27. Two beaker-cells from the stomach of Amia.

Fig. 28. Ciliated epithelial cells from stomach of Amia.

(Figures 18, 25, and 26 are on the same scale ; all the others of this plate, except Fig. 13, are on the same scale as Fig. 15.)

Jour. Morfifi Vol. XI

11. .xxyi'

Hopkins del.



In a short paper published in January, 1895,^ I showed that the fertilization of the egg in the sea-urchin Toxopiieiistes variegaUts Kg. differs radically from Fol's account of the "Quadrille of Centers" in Strongylocentrottis lividus, and at the same time A. P. Mathews announced results nearly identical with mine in the case of the ?,Q?i-urch\n Aj'bacia piuictulata, and the star-fish Asterias Forbesii. Since the appearance of that paper Boveri has published the results of an independent investigation of the fertilization of Echinus inicyotiibeixiilatiis^ which, as far as the essential phenomena of fertilization are concerned, agree "fast Punkt fiir Punkt" with those of Mathews and myself, — a confirmation which seems to render the evidence against the " quadrille " nearly conclusive as far as the echinoderms are concerned.

On two points only is there an apparent conflict between Boveri's results and my own. One of these, as Boveri himself points out, is merely a difference of interpretation, or rather of terminology, since the large reticulated spherical body forming the central mass of the asters of the karyokinetic figure (" astrosph^re" of Fol, centrosphere of Strasburger) is called a centrosome by Boveri, whereas I termed it the " attractionsphere " or "archoplasm-sphere," and asserted the absence of a "differentiated centrosome" (/. c, p. 326).

The second point appears at first sight to be a difference of fact, since Boveri describes the sperm-aster as containing from the first a minute deeply staining centrosome of the usual type, whose division into two precedes the fission of the sperm-aster.

1 Journal of Morphology, X, i, p. 319.

2 Ueber das Verhalten der Centrosomen bei der Befruchtung des Seeigel-Eies, nebst allgemeinen Bemerkungen iiber Centrosomen und Verwandtes. Verh. der Phys.-Med. Ges. Wurzburg