On p. 945 of the first volume of his ‘Handbuch der Entwickelungslehre ‘Oscar Hertwig refers (in a postscript to his theory of the germinal layers) to a theoretical view concerning the process of gastrulation in mammals (and also in vertebrates generally) which I have attempted to establish in my article on ‘Keimblattbildung und Furchung bei Tarsius spectrum’ (1902, Verhandel. Kon. Akad. v. Wetensch. Amsterdam). He points out that I have “considerably changed my views,’ and that I have “now been induced to look upon matters in a way which differs considerably from what lie (Hertwig) and many other embryologists are ready to uphold.”
This paper is meant to establish that the “divergence” of views which Hertwig accentuates will, in all probability, be only a temporary one. An explanation, or rather a rational interpretation, of what we understand by gastrulation as it has been introduced into science by Haeckel and Ray Lankester can only contribute to eliminate certain difficulties which ever and again present themselves when the different stages of the ontogeny of vertebrates are analysed and compared.
Turning back to the very earliest publications in which Haeckel and Ray Lankester have formulated their ideas, we are immediately struck by a difference in their views which has only too often been undervalued, and which concerns the earliest phylogenetic origin of the two-layered embryonic stage. Haeckel looks upon the process of invagination, Ray Lankester upon that of delamination, as the more primitive. I must confess that to me Ray Lankester’s view is more sympathetic, because 1 find it easier to imagine that a one-layered, hollow, spherical blastula—the starting-point of the Metazoa—has, by division of labour, become two-layered in consequence of the differentiation in each cell of an ectodermal half (principally sensitive and integumentary) and an entodermal portion (principally digestive) than that I could be satisfied with the proposition that the two halves of the hollow sphere have prepared themselves (independently of the natural process of division of labour above noted) to bring about an invagination. The result of this invagination will, then, also be a defensive layer (ectoderm) which encloses a digestive layer (entoderm), both together constituting a two-layered stage that has originated by invagination, with an opening, the primitive mouth (Urmund).
A two-layered gastrula is thus evolved out of a one-layered blastula either by delamination1 or by invagination. In the second case the primitive mouth is the natural consequence of the process of invagination itself: in the first case an opening has to become established at a given point in the course of time, to which Ray Lankester2 has conferred, in 1875, the name of ‘‘blastopore ”(opening, that is, present in the blastoderm).
As a matter of course the names “blastopore” and “primitive mouth” have been looked upon from the beginning as synonyms, the more so as in the phylum of the Coelenterates (Polyps, Medusas, Corals, Sea Anemones, etc.), in which we find duplication of the layers by delamination in one genus, by invagination in others, the primitive mouth or blastopore gradually becomes the so-called mouth of the Medusa or Hydra, or the oral slit of the full-grown Coral or Anemone.
It should at the same time be noticed that the oral slit of the Anemone (Actinia) undergoes a further complication by the formation of a stomodæum. In consequence of the formation of this ectodermal involution, below the level of the oral disk, the primitive mouth is displaced to the lower level at which the stomodæum passes into the entoderm’. It is all the more important to keep this in view as this complication of the full-grown Coelenterate has its echo in the phylogeny of the higher animals. We thus find occasion to insist, already now, on the inadequateness of confusing the oral slit of an Actinia with a primitive mouth.
We shall be able to approach the question of gastrulation, as also that of the formation of the germ-layers in Vertebrates, with a more open mind if we can provisionally disinterest ourselves in the question as to whether delamination or invagination was the more primitive process by which a onelayered blastula became converted into a two-layered gastrula. Those who hold with Haeckel that invagination has been the prototype of all gastrulation are biassed and for them even the definition of what gastrulation really is cannot then be free from prejudice. Such a definition should be applicable both to Invertebrates and to Vertebrates, and should formulate the origin of the two-layered out of the one-layered stage.
When we look round for the definition which the most trustworthy investigators have given of the gastrulation of the Vertebrates we immediately observe how heavy a penalty has been incurred by not leaving the crucial question as to whether delamination or invagination has been the origin of the didermic stage of the vertebrate blastocyst provisionally unanswered. As a rule gastrulation and invagination have been looked upon as synonymous.
All this can be fully appreciated when we open Keibel’s very complete and most conscientious ‘Referai’ on “Gastrulation und Keimblattbildung der Wirbelthiere” in Vol. X’ of the Ergebnisse der Anatomie und Entwickelungsge-schichte (1901). On p. 1111 we read:
“Three definitions of gastrulation stand out conspicuously when among the number that have been given we eliminate certain erroneous elements.
“The first would be: Gastrulation is a process during which the cells that will form the intestinal lining find their way into the interior of the embryo.
“The second: Gastrulation is the process during which the material for notochord and mesoderm finds its way into the interior of the embryo.
“The third one, given above, would be; Gastrulation is the process by which the material for entoderm, mesoderm, and notochord finds its way into the interior of the embryo.”
The citation here given shows clearly that the definitions 2 and 3 cannot hold good for the Invertebrata, because the latter do not possess a notochord. The conception of gastrulation is, however, only then of importance when by it we are enabled to hold in one grasp developmental phenomena from Hydra up to man.
Keibel, who rejects 1 and 2 and accepts 3, tries to circumvent this difficulty by emphatically stating(loc. cit., p. 1113) that he refers to gastrulation in Vertebrates, etc. He appears, not to have sufficiently realised that by this limitation to the Vertebrates the important generalisation that in the ontogenesis of all Metazoa a didermic phase occurs, which we term “gastrula,” is mutilated and loses its value as a generalisation.
As to the second definition Keibel rejects it definitely; but with respect to the first he feels inclined to be more lenient. He holds this definition to be quite acceptable for a zoologist who takes into account all the existing Metazoa. This first definition, which was especially patronised by Lwoff (who has given the first energetic impulse towards the clarifying of the conception of gastrulation)1 suffers, however, by the fact that here again Haeckel’s view concerning the primary significance of invagination is tacitly admitted.2 Gastrulation is said to be the process during which the in-testinal entoderm finds its way into the interior of the embryo. This implies that the intestinal epithelium here referred to is originally situated elsewhere, and must now undergo a transportation by which it is transferred to the interior. This petitio principii cannot be allowed to stand, and the definition should be formulated so as to cover delamination as well. It ought, then, to be as follows:
Gastrulation is a process during which an intestinal entoderm is differentiated as against an integumentary ectoderm, and by which a monodermic blastos is changed into a didermic.
This definition is applicable to Invertebrates as well as to Vertebrates.1
We should, however, be very strict in not allowing our conception of gastrulation to be further extended to ulterior processes that give rise to different organs, and during which an undeniable invagination takes place. For that very reason those processes have up to now been erroneously looked upon as gastrulation. By having persisted in considering gastrulation to be necessarily linked to invagination, the actual purpose of gastrulation, viz. bringing about the didermic stage, has been thrust into the background, and undue weight has been attached to the invaginating process.
We would, however, be quite as little justified to look upon the origin of the medullary canal or of the lens and the auditory vesicle by local invaginations as a further continued gastrulation as we are when we assert that the process by which the notochord and the mesoblastic somites are brought about is a gastrulation-process.
During this process true invagination, which has even been cinematographed by Kopsch, undeniably takes place, and is all the more misleading because in Amphioxus it is in direct continuity with the invagination of the gastrula.
However, when we inquire whether the invagination here alluded to contributes towards the origin of the didermic phase of development, viz. towards what we have termed gastrulation, our answer must be emphatically negative. Gastrulation, i. e., the duplication of the germinal layers, is terminated before the commencement of the process of in-yagination, by which the bilateral, symmetrical and metamerical vertebrate is completed. This process of invagination can be looked upon as a kind of budding (notogenesis), which is consecutive upon the formation of the didermic apical portion (cephalogenesis). It gives rise to the trunk.
Moreover the process by which the didermic stage is reached in the higher vertebrates, more especially in mammals, can be traced with such perfect clearness, that we may wonder that in the face of these facts the invagination hypothesis has been so long upheld. The entoderm in mammals makes its appearance by a very striking delamination out of a complex of cells which is comparable to the mono-dermic blastula. And similarly in Sauropsids and Elasmobranchs delamination is the process by which entoderm and ectoderm are respectively differentiated from one another. This takes place in a horizontal plane in. consequence of the accumulation of yolk substance. In Amphibia, Cyclostomes, many Ganoids and Dipnoi, which have been looked upon as important supports for the view that gastrulation is effected by invagination, it has appeared more and more evident to successive investigators (Bellonci, Graham Kerr, Brauer, etc.) that here, too, the primary separation between ectoderm and entoderm is brought about by a process of delamination. As soon as the so-called blastopore (Rusconian anus) appears, which travels a certain distance over the surface of the egg, and which, in many cases, is turned into the definite anus, we have no longer before us the gastrulation process, but a process by which the metamerical and bilaterally symmetrical dorsal organs and the notochord are budded into existence.
Thus the naked facts force the conclusion upon us that in the Acrania (Amphioxus) the gastrula arises by invagination, in the Craniata (all other Vertebrates) by delamination. Once the didermic gastrula-stage reached, a second phase of ontogenetic development is inaugurated which is also of high phylogenetic importance. In this phase the bilaterally symmetric metameric animal gradually appears which we have to compare with possible phylogenetic transition forms that have connected the Vertebrates with radially symmetrical ancestors. This attempt at a plausible and rational reconstruction of the Vertebrate ancestry is, of course, hampered by the circumstance that no trace of those forms is any longer in existence. Still, an actinia-like, vermiform being, elongated in the direction of the mouth-slit, imposes itself upon our imagination, such as has served for the theoretical speculations of Sedgwick on this same subject, and has once been accepted by van Beneden for the precursors of the Chordata.
We have already above considered that the processes of growth by which a Coelenterate gastrula becomes fixed and gradually changes into a sessile Actinian can hardly be looked upon as protracted phases of gastrulation. This will be more difficult yet when the animal has already acquired a higher degree of complication than that of the Coelenterates, and swims about in the shape of a worm-like, lower chordate animal. We know of Polygordius and of other primitive worm types that to the radial, didermic larval stage—the Trochophora—another developmental phase succeeds, during which we observe proliferation in the anal region, leading to an increase in the distance between the anus and the apex of the metamerical worm, the latter budding off, so to say, from the radial trochophora.
We find similar processes in the Vertebrates, but without a free trochophora larva, and to this latter radial and didermic primitive stage corresponds in the Craniata the rapidly passing-earliest phase in which delamination calls forth two germinal layers. Both in Elasmobranchs and in mammals we notice that the cellular material which is present in those very earliest stages contributes especially—as it does in the trochophora—towards the formation of the anterior part, the head and that, following upon this, a proliferation-process is in-augurated (comparable to the origin of the metamerical worm out of the trochophora larva) by which the notochord and the somites, i. e. the bilaterally symmetrical metameric animal are called into existence.
And so this latter process must necessarily correspond—if we go back into phylogeny only far enough—to the transition of the gastrula larva into the longitudinally stretched Actinialike animal. We have already demonstrated above that it is irrational to continue to use the term “gastrulation ”for this phase of development. The coelomic diverticula of the Actinia that are yet in continuity with the intestine are the predecessors of the somites, the nerve-ring on the oral disk that of the medulla, the stomodæum stands for the notochord and the oval slit of the Actinia (no primitive mouth or blastopore !) is reflected in the primitive groove which is in so natural a continuity with the notochord.
In a former publication1I have distinguished simultaneously with Keibel,2 but independently of him, two phases in the gastrulation process, of which the second was erroneously termed “the palingenetic phase of gastrulation.” It is identical with the second process above described, which has only certain superficial analogies with gastrulation by invagination, and which, nevertheless, is bound up quite as closely to the real gastrulation as is the growth of the elongated Actinia to that of the gastrula-larva.
The primitive mouth of the Actinia-gastrula is gradually elongated into the mouth-slit of the Actinia, which in its turn leads into the stomodæum. The blastoporus of the Erinaceus-gastrula, which very soon closes up again, lengthens out posteriorly in the primitive groove, the floor of which—the primitive streak—produces the material for the notochord.
There is, then, during ontogeny an unbroken continuity between the blastopore of the Actinian and its oral slit, between the blastopore of the Vertebrate (often only potential in mammals1 and not identical with the opening that is called by that name in Sauropsids) and its primitive groove. A phylogenetic continuity has to be statuated between this oral slit of the Actinia and the peculiar spot (behind the so-called anterior lip of the blastopore) which on the Vertebrate embryonic shield gradually moves backwards and establishes in many cases an open communication between a portion of the Vertebrate intestine and the exterior. The primitive streak, however, the solid material that proliferates downwards from the ectoderm, coalesces with the entoderm, and brings forth the notochord from its median (though really paired) portion and the somites from its lateral wings—this primitive streak can never be identified with a blastopore or primitive mouth, neither with the lips of the blastopore. For we have above attempted to demonstrate that in this primitive streak we encounter the material which, also in the Actinia, (1) proliferates downwards from the ectoderm and produces the stomodæum, (2) coalesces with the entoderm, (3) is in direct continuity with those parts which are preparing to give rise to cœlomic pouches but are yet continuous with the primitive enteron.
Certain prudent changes of nomenclature would perhaps recommend, themselves. For that portion of the vertebrate embryonic disk which I have proposed to compare to the Coelenterate month-slit and stomodæum, the name of dorsal mouth might be chosen. The difference between the phyla of Annelids and Molluscs as compared to Vertebrates is thereby all the better marked.
Provisionally we are not yet enabled to enter into a detailed comparison between ontogenetic and phylogenetic processes. The possibility of instituting such comparisons will probably have disappeared in the graves of the numerous transition forms that now rest among the fossils. But the chief outlines of the evolutionary process of notochord and somites stand out boldly enough and correspond to what Sedgwick has first hinted at and what van Beneden has later extended but never fully worked out. At the same time not only the Diplo- and Hemichordata but also the Cephalochordata are relegated to a more modest lateral situation in the pedigree of Vertebrates.
The distinction between the head-segment of Polygordius and the trunk of the same animal is thus of the same order and belongs to the same category as that between the very foremost portion of the body of a Vertebrate and the segments that extend behind it. In an earlier publication1 I have accentuated this distinction by the use of the terms “cephalogenesis” and “notogenesis.” I wish to adhere to this, and yet to observe that the distinction here intended between “kephale” and“notos” is not identical with that between head and trunk (trunk-segments having been ascertained to enter into the composition of the head), but that on one side should be ranged the very foremost portion of the head to which the ophthalmic and optic nerves belong, whereas on the other we place the further subdivisions of the brain with their cephalic nerves, as also the basis of the skull with the remains of the notochord it contains, the visceral arches and the whole of the trunk.
We are in no way justified in including the whole process of the formation of the notochord in the gastrulation phenomenon, nor either that of the cœlom and the somites. In Echinoderms the formation of enterocœl, hydrocœl, etc., is only inaugurated after gastrulation has been completed; in worms the appearance of the somites is synchronic with the increase in length which has been noted above and which follows upon gastrulation. How the process of cœloniogenesis in the higher vertebrates is derived from that of their invertebrate ancestors (which are unknown to us) cannot for the present be said to be sufficiently elucidated. Certain peculiarities in the development of Balanoglossus should here be considered,and may to some extent be comparable with what occurs in Vertebrates, but up to now we have not sufficient data. Bateson has shown that Balanoglossus possesses an unpaired anterior cœlomic sac (cavity of the gland), that paired sacs of collar-cœlom follow upon this and that an unpaired hinder-most cœlom provides the trunk and soon becomes paired.1
To explain the fact that the mutual relation between the earlier gastrulation phases of vertebrates and their later processes of growth and organisation have been so hopelessly mixed up and misunderstood, we have to keep in view that an exaggerated importance has been ascribed to Amphioxus.
I wish to insist upon this somewhat more fully. It is quite natural that, in the literature of the preceding century, a most particular and isolated position has been allotted to Amphioxus, as the very lowest fish-like being in the system of the Vertebrates, and we can neither wonder that in the second half of that century, in which the theory of evolution made its way, this position has considerably increased in importance, and that there arose a tendency to look upon Amphioxus as the real ancestral Vertebrate, out of which all the other fishes and Vertebrates had originated. Haeckel has from the first accepted this view in his phylogenetic papers, and when then Kowalewsky had elucidated in such a masterly manner the ontogenesis of Amphioxus and of the Ascidians, the theory of the descent of the Vertebrates out of Invertebrates viâ the Ascidians and Amphioxus seemed to be definitely settled. This, however, was not the case, but then the beautiful and detailed researches of Hatschek on the development of the organs of Amphioxus have shown this animal to be such a perfect model of clear histogenesis in its early development, that it was only natural that all subsequent observers, who occupied themselves with the embryonic development of higher Vertebrates, should have attempted to start from the data furnished by Amphioxus.
This position as a central and archaic form in Vertebrate phylogeny, confirmed as it seemed by embryology, is decidedly a usurped position.
For as soon as we also consider the anatomy of Amphioxus it becomes evident that the Craniota have not sprung from this acraniate animal; that the relation between the anterior extremity of the notochord and the brain must have been different in the real ancestral Vertebrates; that eyes and static organs (otoliths) cannot have been absent in those ancestors as they are in Amphioxus—in short, that Dohrn was quite right when he proclaimed Amphioxus to be a distantly related, partly degenerated, relict, which could not furnish us with important conclusions concerning the phylogeny of Vertebrates.
We find in Amphioxus when we try to compare it with the higher Vertebrates the same difficulties which we encounter when comparing the Cyclostomes with the other fishes or the Ornithodelphia with the higher mammals. All three are representatives of old and early stems that have branched off sideways at a very early stage, and that thus represent lateral lines of development. They have no considerable significance for comparative phylogeny, and may on no account be looked upon as transitional forms that occupy a place on the line of descent between the Coelenterates and the Primates.
Once we have fixed upon this position for Amphioxus in the system, there will be no difficulty in looking upon the subdivisions Acrania and Craniota as being equivalent to a new subdivision, In vagin at a and Delaminata. The process of invagination, which is noticed in so many Vertebrates as accompanying the origin of the notochord and the somites, will not, then, any longer be indicated by the name gastrulation.
Together with this the significance of the primitive streak as coalesced lips of the blastopore will have to be abandoned; the streak and the groove represent the oral slit of the Actinia with its dependent stomodæum, which, when compressed into the notochord, would thus be homologous to that organ in Vertebrates.
I feel very hopeful that on this basis a mutual understanding will soon be arrived at, by which many embryological results which up to now seem hopelessly confused will fall into line.
The numerous and important contributions to Vertebrate embryology which we owe to so many contemporary investigators, who for the present yet adhere to earlier theoretical propositions, retain their full importance as far as the actual observations are concerned. The incorporation of those observations in the new theoretical interpretations will come about by itself.
In this case the blastococl gives rise quite naturally to the archenteron.
“On the Invaginate Planilla,” etc., ‘Quart. Journ. of Mier. Sci.,’ vol. xv, 1875, p. 163.
Lwoif, “Die Bildung der primaren Keimblâtter und die Eutstehung der Chorda,” etc., ‘Bull Soc. imp. der naturalistes de Moscou,’ 1S94.
I must emphatically assert that the reproach of having included the invagination process in the definition of gastrulation does not apply to Keibel himself, however much this might seem to be the case if we consider his definition No. 1. He has, however, on pp. 1109-1110 of the ‘Referat’ above alluded to, most distinctly stated that he wished this definition to apply to a delamination-gastrula as well.
In Haeckel’s ‘Anthropogenie ‘(4th ed., 1891, p. 156), which may also be considered as decisive on this question, we read: “[The cleavage cells] arrange themselves in two separate layers, the two primary germinal layers. These surround a digestive cavity, the ‘Urdarm’ (primitive enteron), with an opening, the ‘Urmund ‘(primitive mouth). The important embryonic stage which possesses these oldest primitive organs we call the gastrula, the process by which it originates gastrulation.”
‘Anat. Anzeiger,’ iii, 1888, p. 911; ‘Quart. Journ. of Mier. Science,’ vol. xxxi, 1890, p. 552.
“Zur Entwickelungsgeschichte der Chorda, etc.,” ‘Archiv f. Anat. u. Physiol., Anat. Abth.,’ 1889, p. 376.
An open blastopore has up to now only rarely been noticed in Mammals. In ErinaceusI Lave observed it quite clearly (‘Furchung und Keimblattbildung bei Tarsius,’ Pl. XII, figs. 8, 9). Keibel (l.c. Taf. 24, figs. 46, 47) is less positive as far as the rabbit goes, so are Selenka (Taf. 17 and 18) for the opossum, and Bonnet (Anat. Hefle, Bd. 9, Taf. 32) for the dog.
* Furchung and Keimblattbildung bei Tarsius,’ Amsterdam, 1902, K. Akademie v. Wetensch.
Whether these arrangements are repeated in higher Vertebrates has not yet been decisively shown. It should, however, be noticed that the mode of formation of the pericardium which was two years ago described by me for Tarsius (1. c., p. 3, El. VIII, fig. 70 a, b), is in no way isolated, butlias since been also observed by me in Sciurus and Tupaja. For Tupaja the gradual transition of an évagination of the entoderm that becomes constricted off into the anterior median portion of the pericardium is demonstrated in more than one preparation. In Scinrus it has not been followed out in full, but an early stage was noticed. Bats, too, seem promising in this respect. At all events, if this should be further confirmed, our whole interpretation of the vertebrate cœlom would have to be recast. That in such a case Balanoglossus among Invertebrates would have to be considered as an object of coinparison would not be astonishing if we remember how a certain comparability between the branchial arrangement of Balanoglossus and Amphioxus has long been known. The so-called notochord of Balanoglossus I would be inclined with Spengel to regard as a delusion. Gegenbaur in the latest edition of his comparative anatomy (Bd. 1, p. 185) has allotted to Rhabdopleura, which is related to Balanoglossus, a decided significance in the pedigree of Vertebrates.