GASTRULATION IN BIKDS. 145

Gastrulation in Birds.

lticliai'tl Asslicton, IU.A.,Lecturer on Physiology in Guy's Hospital, University of London.

IN the twentieth volume of the'Journal of Morphology 'a p:iper by Mr. J. T. Patterson appeared during the year1909 under the title, "Gastrulation in the Pigeon's Egg:A Morphological and Experimental Study," a preliminarynotice of which was published in 1907 in the ' BiologicalBulletin,' vol. xiii.

In these papers the author gave an entirely novel accountof the process of gustrulation in a bird, which account, iffree from error, described an interesting, albeit perplexing,phenomenon.

The paper is fully illustrated by photographs and diagrams,and. has the appearance of being a careful piece of work,and it has been used by Professor Frank R. Lillie as thebasis of his description of the early stages of bird develop-ment in his recent book, ' The Development of the Chick.'Professor Lillie writes, on page 52, that he " has had theopportunity of following the work step by step, and isconvinced: of its accuracy."

The paper describes so unusual a process that, in spiteof this testimony, it courts a rather close examination.Moreover, if correct the matter should be relieved of allsuspicion, because it would in that case be a highly importantcontribution to the embryology of birds.

Since it seems to me that the description given byVOL. 5 8 , FART 1. NEW SERIES. 10

146 KICHABD ASSHETON.

Patterson is not altogether free from doubt, I venture tooffer the following notes by way of criticism, which maypossibly, and I hope will be, successfully met. Brieflystated Patterson's account is as follows: He denies thatgastrulation iu the pigeon's egg occurs by delamination orany process of ingrowth from the germinal wall or otherlower layer segments. Gastrulation, according to him, takesplace before the egg is laid by a process of involution of theoutermost layer of cells of the segmented blastodisc. At theclose of segmentation this outermost layer of cells, whichforms a continuous membrane, becomes detached from thesub-lyiog cells and yolk along that part of its margin whichis towards the future posterior end, and the detached marginbecoming involuted, grows forward as a thin free edgebeneath all the. loose cells which admittedly exist in thedeeper parts of the segmented blastodisc. Tliis free edgejoins up in front and at the sides with the wall of yolk thatcontains nuclei (i .e. the germinal wall), and forms a con-tinuous sheet of cells—the entoderm or bypoblast. As thesubgerininal cavity expands it excavates the germinal wall,and a sheet of cells derived from the germinal wall is left abovethe cavity. To this sheet the " invaginated " entoderm fuses.Thus the gut entoderm is formed by involution and the yolk sacentoderm by excavation. The loose cells lying beneath theouter layer, now to be termed " epiblast," are said to passinto the outer layer, and so also to form part of the epiblast.This involution process is said to be still further complicatedby the concrescence of the lip thus formed giving rise to alinear seam—the future primitive streak, which is withdrawnlater within the area pellucida by a sweeping round of thegerminal wall in a manner reminiscent of Duval's attempt toprove a process of concrescence at a time subsequent to tlielaying of the egg.

CRITICAL NOTES.

It is claimed that this account of the formation of theentoderm or hypoblast by an infolding of the blastoderm

GA STIMULATION IN BIRDS. 147

edge is supported by experimental observations, such asmarking certain parts of the bhistoderm by injury, andfollowing such marks through several hours of incubation.

Although the account given forms a very complete story,which, if the observations are good, does seem to besupported by a good deal of evidence, yet ' it is very diffi-cult to reconcile it with the process of gastrulation in theother Amniota. In fact Patterson himself hardly mentionsthe reptiles or mammals, but confines his efforts to an at temptto adapt the bird to Amphibians and fishes, and more especiallyto the Teleostean fishes. That is to say, he tries to connectbirds with a group far removed from them in anatomicalfeatures, and ignores the difficulties presented by his theorywhen compared with the most closely allied forms.

In no other group of vertebrates is the dorsal lip of theblastopore, which is in every other case the most bulky andactively proliferating par t of the embryo, known to exist asa thin or f r e e e d g e ! I t is extremely difficult to conceive ofthe mechanism by means of which such an involution couldtake place.

Again, on pp. 86-87, the author speaks of the whole thinedge as the dorsal lip of the blastopore, and the yolk as theventral lip. Now this is never the case in any vertebrate,whether we consider the meroblastic eggs of the Blasmo-branch or Teleost or the less heavily yolked eggs of theAmphibia. In all cases the ventral lip of the blastopore, ifformed, is a thickened curved rim which is formed in con-junction with the inflection of the epiblast. If Pat terson isright in calling the inflected, edge of the blastoderm thedorsal lip of the blastopore, then the par t which he calls theventral lip of the blastopore is surely the floor of the gutcorresponding to the yolk-plug in Rana. The yolk is neverthe lip of the blastopore.

Stronger evidence is the table given (p. 90) of meaSure-meuts made upon living blastoderms during the time sup-posed to be taken for the process of gastrulation. If theedge of the ectoderm is inflected, one might expect to find a

148 RICHARD ASSHETON.

diminution of the length of the blastoderm occurring at thatmoment. This is s;iid. to lmve been so in two eggs whichwere kept under observation for 3 | and 3 | hours respectively.

Piittersou objects to some experiments made by myself in1896 ( 'Proa Eoy. Soa,' vol. lx, 1896) with a view to testingDuval's theory of concrescence on chicks, thus: (1) Theywere performed after concrescence had occurred, which is avain] objection if Patterson's contention th;it all this occursbefore laying is correct. (2) Tlie cells may have flowed roundthe bristle held in place by vitelline membrane and yelk.

This objection, which, of course, is irrelevant in this par-ticular case if the first holds good, must be considered ns ageneral objection to the use of bristles for such purposes.The bristle, it may be said, makes a perfectly unmistakablelandmark, which cannot be said of injuries by cauterisation.

The objection is one which has naturally occurred to me,but I am coaviuced- that the objection is groundless for thefollowing reasons :

In the numerous experiments made upon chick and frogs'eggs with sable hairs, I have never seen any evidence thatcells can flow round tlie hair.

The results would not be so constant if there were anyflowing of cells round the bristle.

If cells could move so easily as to avoid a bristle withoutmaking auy visible sign of disLurbauce, they would beaffected by the force of gravity and become displaced wheneggs are not in their normal position. This is not the case.A fully segmented egg of Rana te inporar ia maybe helddown at an angle of 90° to its normal position without affect-ing the normal relation of its cells to one another. Whenthere has been a very severe drag upou the bristle on accountof some excessive stress in an egg, due to some displacementwith reference to tlie vitelline membrane, nnd an attempt liasbeen made by the cells to flow round the bristle, the effect isobvious, and is seen as a. bay or wrinkle which is quite absentfrom properly performed experiments of the kind, and indi-cates only an attempt of a soft tissue to swing round tlie

GASTKUr.ATION JN BIltDS. 149

obstacle, and is not in any way comparable to an actualflowing of a fluid past a fixed and solid object.1

If the segmented egg of a frog, or the segmented blasto-discof a bird, were perfect fluids, then the objection would boa fatal one. Or if the segmented egg were like a heap ofshot, then also such a mass could flow slowly past a fixedobject without producing any visible rippling. Ri.it the seg-mented ovum is not a pile of separated cells. The cells, ormany of them, arc in continuity by means of viscous cyto-plasmic strands, and the whole mass is by no means a perfectfluid.

If a bristle is inserted into the yolk-plug of an egg ofE a n a t empora r i a during the crescent slage of the blasto-porc close up against the advancing dorsal lip of the blasto-poie, there is no tendency either for the advancing lip to bedivided by the bristle, nor for the bristle to be driven throughthe yolk plug-colls, thus proving the absence of anythingapproaching a perfect fluidity of either the ectodermal or theendodermal layer of cells.

I may remark here that when one of my experiments suitsPatterson's purpose he accepts it! (p. 115). If cells can sweeppast to concresce when the bristle is placed to one side, thefact that my bristle, when placed in the area opaca in theposterior margin, did not appear in the embryo, is no proofthat the cells that do form the embryo have not come sweep-ing past the bristle.

Personally I cannot agree with Patterson's view (p. 109)that concrescence and gastriilation are different phases of thesame process, Gastrulation is the formation of the gutcavity. If this formation is accompanied by the productionof a blastopore (which is by no means always the case, e.g.Hydrozoa, probably all mammals—I would even add all

1 These remarks refer to Rana temporaria only. There is muchvariation in the viscosity of amphibian eggs. I have failed with thesegmented egg of Triton cristata. The eggs of Bufo are also lesssuitable than those of B. tempovaria owing to greater fluidity. Theresults obtained from the chick and Rana temporaria I believe to becjuite reliable.

150 .RICHARD ASSHIOTON.

Amniota), then that blastopore may close by concrescence,but the two processes are entirely different phenomena.

I have myself tried for years to emphasise this difference('94, '96, '08, '09), and the difference is recognised by manycmbryologists such ns Hertwig, Hubrecht , Keibel, MacBride,although they do not use the terms which I humbly protectdo most correctly express the essence of the process, namelyprotogenesis and deuterogenesis. The phenomenon of gas-trulation or the formation of the primitive gut-cavity orarchenteron, whether with or without a blastopore, is proto-genetic, and represents a more ancient phase of evolution.The subsequent phenomenon of deuterogenesis is growth inlength and is post-gastrulaj and in those animals which havea blastopore formed in connection with the first appearanceof gut-cavity it involves all the changes by which the blasto-pore becomes wholly or partially closed, whether by coales-cence, convergence, or concrescence, partial or total. I trepresents a stage in evolution subsequent to that representedby the gastrula stage.

IE there is any concrescence it is concerned with deutero-genesis in the vertebrates and not with gastrolation ; but itis extremely doubtful, in spite of Patterson's work, whetherthere is any such thing as concrescence in the sense whichcan be interpreted as meaning that the embryo of thevertebrate is formed by the fusion of the lips of an elongatedblastopore.

Patterson adheres with patriotic tenacity to the view socommonly held by Americans ns to the formation of tlicvertebrate embryo by concrescence. He writes thus on p.103 : " In other words, in the teleost the entire margin ofthe blastoderm separates from the periblast, and this entiremargin (germ-ring) concresces to form the embryo." He waspresumably unaware of Kopsch's work on the eggs of Salmo,1905, or he could not possibly have written so dogmatically.Kopsch's experiments prove as conclusively (so it seems tome) as anything can be proved that m the trout the maindorsal axis of the embryo is not formed by concrescence.

GA STIMULATION IN BIRDS. 151

From these experiments it is perfectly plain that the germring representing the lips of a posteriorly placed blastoporoprovides the mnterisd for growth in length thus—the mid-dorsal part for the mid-dorsal region of the embryo, thenerve-cord and notochord, the lateral parts for the sides, theventral part for the ventral surfaces. I may refer the readerto some remarks on this in my paper on Teleostean develop-ment, 'Guy's Hospital Reports,' vol. lxi, 1907.

If, therefore, Patterson's account of theformation of the mainaxis of the pigeon by concrescence is correct it is interestingand remarkable, but at any rate it is not like the Teleostean.

Again, where in the animals most closely connected withthe birds in adult characters, the reptiles and mammals, canwe possibly find the slightest hint of any process either of aninvolution of a free edge or a process of concrescence ?

If we turn from such general considerations to his actualexperiments we are not convinced by them.

In the first place there is some, bnt not much, chance ofmistake in the orientation. Patterson says that in thepigeon's egg the embryo lies with its longitudinal axis at anangle of 45° with the longitudinal axis of the egg (" chalazalaxis") in 90 per cent, of eggs. Presumably he discardedexperiments in which on the development of the embryo itwas found to deviate from 45°.

Exp. I. (Operation 83-J hours, examination 37 hours afterthe estimated time of fertilisation.)

The posterior margin of the blastoderm, at this time a freeedge, was injured by cauterisation before it had becomeinvoluted, which injury " ought to be carried down beneaththe blastoderm during the course of further development, thatis, it ought to be found in the entoderm " (p. 88).

The truth of this contention is supposed to be demonstratedby a photograph (fig. 66). There is nothing to indicatewhich is anterior or posterior end, but I take it that thenumber "66" is close to where the edge of the blastodermshould be, and that the space under the letters " o p " repre-sents the deficiency in the entoderm. We are asked to compare

152 RICfTAftD ASSHETON.

this with a section of " an uninjured blnstpderm at a corre-sponding stage/' and to note that " (lie entoderm in this regionis very thick (see fig. 37). It is clear, therefore, that while suchan operation destroys most of the cells that are to give rise tothe entoderm, yet the posterior margin is still capable offorming a rounded dorsal lip." I venture to submit that itis perfectly impossible to deduce any such conclusion fromthe figures given.

Fig. 66 represents a magnification of 125 diameters, andthe point of injury is about, 3} in. from the dorsal lip. Pig.37 is magnified 245 times, but as the whole section mensuresless than 5 in. it cannot contain the required spot. There is,however, another figure of the same section, fig. 35, which ismagnified 107 times. If we examine the region 2\ in. oreven 2 in. to the left of the edge of the blastoderm, we failto see any greater accumulation of entoderm cells in theuninjured than in the injured one.

Possibly I may have mudo a mistake in my interpretationof his fig. 66, and the number " 66 " is at the anterior endand not the posterior end ns I assumed. In that ense I amat a loss to find either the cells which have been injured orthe deficiency in the entoderm referred to. If the latter isindicated by the clearer spot near a letter " z " (of the figureabove) then the corresponding spot in fig. 35 or 37 is justns devoid of entoderm as in 60. Or if, as he seems tosuggest, we are to contrast fig. 67 with a part still further tothe left in fig. 35, I fail to spe much difference in the con-dition of the "entoderm." On this latter nssuinption, thespot labelled " o p " is presumably the " break" in thevitelline membrane made by the operating needle, from whichthe free edge has curled away forwards. Since there is nota trace of vitelline membrane shown, the photograph failsto strengthen the argument in the text.

I think it must be admitted thnt the author has not beensuccessful here in his attempt at demonstration.

Exp. II . (Operation 35J hours after fertilisation. Subse-quent incubation 49 hours.)

GA.STU.0LAT1ON IN BIRDS. 153

The injury was again on the edge, but now the edge is alip. . In these he finds an injury in e n t o d e r m on ly , there-fore he says there is still an inrolling of entoderm.

In specimens "sl ightly older" such experiments showinjuries in ectoderm and mesoderm but not in the entoderm," showing that the involution has ceased."

The whole series o£ experiments recorded under the head-ing "Exper iment I I " seems to me to be questionable in theextreme. Anyone reading the first two paragraphs of thatsection with a critical mind must perceive how fragile is theevidence upon which such far-reaching results are based.He writes (p. 93), in describing the subsequent effect of aninjury made to the edge of the lip in the middle dorsal line," There is no evidence of an-injury either in the ectoderm orlnesodenn, and hence we must conclude that the affectedcells have been brought to their present position (in theentoderm) by an- inrolling under the posterior margin.Althougli this operation has been repeated.several times withthe above results, yet the position of the injury in the ento-derm may vary in an anterior posterior direction ; but thisvariation is easily accounted for by the fact that one can tellin the living egg only approximately the extent to whichinvagination has progressed."

Thus we see the -results obtained are variable; and he goeson to say that " if an injury be made in the same inp.nner asabove on slightly older blastoderms, the affected region isnot found in the entoderm, but in the ectoderm and meso-derm, showing that the involution has ceased" (p. 93)..

There is little here in the nature of exact or accurateexperiment. There are no times1 or measurements given, butinstead of these he bases results upon operations performedon "sl ightly o lder" blastoderms than those the stage ofdevelopment of which " one can tell only approximately."

There is also' the difficulty presented by this hypothesis offormation of a blastoporo lip before the laying of the egg,

1 Some times are given in the explanation of the plates.

154 RTCUAUD ASSHETON.

that there would then be a very str iking difference comparedwith other vertebrates. In all other vertebrates the blasto-pore lip is the growing point for growth in length, andgrowth in length begins at once, therefore showing itselfclearly in the origin of deuterogenetic (peristomial) mesodermfrom the angle of the lip laterally, notochord dorsally, anddenterogenetic epiblast superficially. This condition is wellknown not to occur until some hours later—in the chick aboutthe twelfth to fifteenth hour of incubation. So we shouldhave to account for a very remarkable disappearance andreappearance of this proliferating centre.

Again, Patterson in his second series of experiments saysthat an injury made during the involution process is foundin the entoderm, posterior to the position of the nineteethpair of mesoblastic somites. W e are faced with thefollowing dilemma. We can hardly have a proliferatingblastopore lip formed as in other vertebrates so long as theouter layer is turning in to form entoderm. Therefore thisproliferating lip must come into being after the cessation ofthat process. Any injuries mode to the involuting membranemust surely occur in front of, or beyond, all the tissues,mesoderm included, which are produced by the proliferatinglip when it comes into being.

But fig. 50 shows such an alleged injury far posterior to thenineteenth pair of somites. The injury ought to be in front ofall the primitive streak mesoblasfc, whereas, according to tliofigure, there are many somites of mesoblast in front of theinjury.

Another argument which is difficult to follow is the sngges-tion on p. 99 that certain " c a v i t i e s in t h e d o r s a l l i p "are the homologues of Kupffer's vesicle. I t is surely wellenough established that whatever the physiological meaningmay be of Kupffer's yesicle in Teleostean development, it is,ns a cavity, par t of the gut-cavity. According to Pattersonthe archenteron is the cavity roofed in by the inturning edgeof the blastoderm. Yet here he says thnt these vacuoles abovethis roof are homologous to the Kupffer vesicles, which are

flASTKULATION IN BIltDR. 155

well known to be below tins roof, i . e . they are part of thearchenteron.

Exp. I I I . (Operation 34^- hours after Fertilisation. Subse-quent incubation 34 hours.)

From the plan of his text-fig. 16 one is bound to concludethat the stage does not materially differ from the stage of thepreceding experiment, his text-fig. 10. Each shows ;i similardiameter, a similar forward extension of the endodenn, asimilar width of what he regards as blastopore opening. Theonly difference is that in text-fig. 10 the dorsal lip of the" blastopore " is slightly convex, in text-fig. 16 slightly concavein surface view.

In Exp. I I an injury was made on the edge, and the resultwas a defect in the endoderm at a spot posterior to thenineteenth somite.

In Exp. I l l au injury was made on the surface just withinthe margin. The difference in position of the injury wouldappear to be not more than the diameter of' the needle used.Result, a defect in the region of the head-fold. Therefore thedifference in position of less than a needle's diameter in themarking of a blastoderm corresponds .with a difference inthe embrj'o which includes the greater parb of the body. If.this is so we must despair of getting anything approachingaccurate results by such methods.

Patterson likewise thinks it unlikely that this small area,should give rise to so much embryo directly, and assumes, aswe have seen, that there is a concrescence.

Exp. IV. (Operation 3 4 | hours after fertilisation. Subse-quent incubation for 3 6 | hours.)

On a blastoderm similar to Exp. I l l a spot was marked onthe margin 10° to the right of the middle line so close tothe margin that the outer surface of the needle was levelwith it. Result, a defect "on the right neural fold in the mid-brain region."

If the main axis of the embryo is formed by coalescence ofthe two germ-rings, surely, then, it is in the median plane thatthe injury should be found, i . e . the ventral wall of the neural

156 KICHAUD ASSHtiTON.

tube and notochorJ and, perhaps, gut, yet tlie defect is shownon the upper part of the neural tube only.

Exp. V. (Operation 33^- hours after fertilisation, subsequentincubation for 36J hours.)

A similar blastoderm was injured at the edge of the horn ofthe junction zoue 45° to the right of the middle line, with theresult that a defect is s.'iid to have occurred in the primitivestreak, though one cannot see much of it in his fig. 71.

In none of the cases so far considered do (he figures con-vince one that the spots called defects are really such, orhave any constant relation to tlie spots injured.

For instance, in the last cnse itis quite impossible to satisfyoneself that there is any injury at all from fig. 7] , and text-fig. 17, which is a transverse section through the allegedinjury, shows si perfectly normal primitive streak section witha mass of cells or yolk, or both, lying on the top in no wayconnected with it. This, in fact, is an "extra-ovate" inRoux's sense that may have travelled from anywhere. Myown experiences with such experiments have taught me howdeceptive an extra-ovate may be.

The results are very different to the defects figured byKopsch in his Salmo embryo experiments.

Exp. VJ. At a rather later stage—"late gastrular s tage"—in which the entoderin had advanced a little furl her aninjury was made ;it the posterior margin in the median line.

The result was a defect in the middle line at tlie level ofthe tenth pair of somites affecting ectoderm only.

Exp. I I was supposed to demonstrate the involution of theedge of the blastodisc to form the entoderin, because aninjury to the edge made at 35-f hours appeared only in theentoderin somewhat posterior to the nineteenth pair ofsomites. In Esp. YI an injury also touching the edgealthough made three quarters of an hour earlier appeared inthe ectoderm only. How can this discrepancy be explainedaway on Patterson's hypothesis?

Now this one seems open to another explanation. There isclearly an extra-o\ate consisting of " a mass of dead cells"

GASTRUrATION 11ST BIRDS. 157

lying between the separated halves of the neural tube. Thenotochord is perfect and to ouo side, the entodei'in is uninjured.May not the delect in the neural tube be simply mechanical,due to the pressure of an extra-ovate which became separatedoff from the edge of the blastoderm as a result of thecauterising, and which, passing into the ai*ea pellucida betweenthe blastoderm and vitelline membrane, caused the injuryseen ? Text-fig. 19 strougly suggests this solution.

Bxp. V I I and V I I I . The figures do not enable one toappreciate the churacter of the defects. Sections are notgiven.

The remaining experiments, IX—XIII, were made upon theblastoderm after the eggs were laid, and therefore aFterPat terson's supposed concrescence of the lip must have beencompleted, because by now, according to him, the main axialline of the embryo produced by this concrescence is entirelyenclosed within the blastoderm margin, and there are no longerany free blastoporiclips that could come together.

Although Patterson still speaks of concrescence, Exp .XI , p. 115, it clearly cannot be a phenomenon similar to thatwhich, as he alleges, occurs during gnstrulation. One isinclined in this particular connection to say with ProfessorMacBride (re " Amphioxus," 'Quar t . Journ. Micr. Sci.,' vol.liv, p. 302) : "OE course in every structure there is aniinagiuary middle line, and if anyone chooses to say thatthis band of dividing cells consists of r ight and left halveswhich unite together as quickly as they grow, I shall notWiiste time iu arguing against such a metaphysical conception,which is capable neither of proof nor disproof/'

From Patterson's final discussion it is clear tha t he quitefails to appreciate the distinction between gastrulation andsubsequent growth in length. I t is not really true that " a l lof the chorda and mesoderin are derived from the primaryinvaginated layer " in Amphioxus. The anterior par t is soderived, but the posterior part is derived from the proliferatinglips of the blastopore, which can be described neither as ecto-derm nor einloderm.

158 IUCHA11D ASSHJ5T0N.

To me it seems that there is a very great difference betweengastral and peristomal mesoblast: the one is protogenetic, theother deuterogenetic.

Nor, again, is it true that in the case of birds the wholemesoblast is formed from the primitive streak. It is altogetherdifficult to understand whj' if the primitive streak is formedby the fusion of thickened rims, that thickening should dis-appear, only to reappear a little later as primitive streak. Itis very remarkable and significant of the narrowness of tin'swork that in dealing with avine early stages as comparedwith other vertebrates the word " reptile " should occur onlytwice and the " mammal " is mentioned but a single time !

It is pretty evident that the author is utterly unable toreconcile his description (which is an attempt to fit the birdson to fishes) with the facts of reptilian or mammalian embryo-logy, the two groups of animals most nearly connected withthe birds.