Fig. 11.—Embrionic Cranium of the Adder. Ventral Aspect. (After Rathke.)

The basioccipital is formed in the posterior part of the basis cranii, and the exoccipitals in the side walls of the trench in continuity with the fundament of the basioccipital (see Fig. 11). The supraoccipital is formed in cartilage above the exoccipitals. The basisphenoid develops, like the basioccipital, in the flat basis cranii, but towards its anterior edge, between the large foramen (h) and the pituitary space (i). It is formed from two centres, each of which is originally a ring round the carotid foramen. The presphenoid develops in isolation between the lateral trabeculæ, just behind the point where they fuse. The side parts of the basisphenoid and presphenoid (forming the alisphenoids and the orbitosphenoids respectively) develop in cartilage separately from the cranial basis, not like the exoccipitals in continuity with it. The hinder parts of the trabeculæ become enclosed by two processes of the basisphenoid; their front parts remain in a vestigial and cartilaginous state alongside the presphenoid. The frontals and parietals show a peculiar mode of origin in the adder, differing from their origin in other Vertebrates. The frontals develop in continuity with the orbitosphenoids, the parietals in continuity with the alisphenoids, and so have much resemblance with the vertebral neural arches which surround the spinal column (p. 195).

Through Rathke's work the real embryonic archetype of the vertebrate skull was for the first time disclosed. Rathke discussed this archetype and its relation to the vertebral theory of the skull in another paper of the same year (1839), but before going on to this paper, we shall quote from the paper on the adder the following passage, remarkable for the clear way in which the idea of the embryological archetype is expressed. "Whatever differences may appear in the development of Vertebrates, there yet exists for the different classes and orders a universally valid idea (plan, schema, or type) ruling the first formation of their separate parts. This idea must first be worked out, though possibly with modifications, before more special ideas can find play. The result of the latter process, however, is that what was formed by the first idea is not so much hidden as partially or wholly destroyed" (p. 135).

Rathke's general paper on the development of the skull in Vertebrates[211] treats the matter on a broader comparative basis than his paper on the adder, and takes into account all the vertebrate classes, in so far as their development was then known. He here makes the interesting suggestion, later entirely confirmed, that the basis cranii or basilar plate is first laid down as two strips, one on each side of the chorda—the structures now known as parachordals (pp. 6, 27). For this supposition, he thinks, speaks the structure of the skull in Ammocoetes, which in this respect is the simplest of all Vertebrates (pp. 6, 22). In Ammocoetes, as Johannes Müller had shown, the foundation of the skull is formed by two long cartilaginous bars, between the hinder portions of which the notochord ends. In these Rathke was inclined to see the homologues of his trabeculæ, and of the parachordals which he was ready to assume from his embryological observations.

Müller was, of course, very ready to accept Rathke's opinions on this subject, for he considered that they supported his own theory of the vertebral nature of the skull. After describing in his Handbuch der Physiologie the cartilaginous bands in Ammocoetes and their highly differentiated homologues in the Myxinoids, he writes in the later editions, "Hence we see that in the cranium, as in the spinal column, there are at first developed at the sides of the chorda dorsalis two symmetrical elements, which subsequently coalesce, and may wholly enclose the chorda. Rathke has recently observed, in the embryos of serpents and other animals, before the formation of the proper cranial vertebræ, two symmetrical bands of cartilage, similar to those which I discovered as a persistent structure in Ammocoetes.... At a later period the basis cranii of vertebrate animals contains three parts analogous to the bodies of vertebræ, the most anterior of which, in the majority of animals, is generally small, and its development frequently abortive, whilst in man and mammiferous animals the three are very distinct. These parts are developed by the formation of three distinct points of ossification, one behind the other, in the basilar cartilage."[212]

Rathke was very cautious about accepting the vertebral theory of the skull; he saw that the facts of development were not altogether favourable to the theory, and he gave his adherence with many reservations and saving clauses. His general attitude may be summed up as follows.[213]

The chorda sheath is the common matrix of the vertebræ and of a large part of the skull. The basilar plate and the trabeculæ, which are developed from the chorda sheath, give origin to three bones, which might possibly be considered equivalent to vertebral centra—the basioccipital, the basisphenoid, and the Riechbein (ethmoid). The Riechbein develops from the fused ends of the trabeculæ. The presphenoid might also be considered as a vertebral body, but it develops independently of the basilar plate and trabeculæ.

Now of these bones, the basioccipital is in every way equivalent to a vertebral centrum, for it develops in the basilar plate round the notochord. With the exoccipitals, which arise just like neural arches, it forms a true vertebra. The supraoccipital is an accessory bone developed in relation to bigger brains. The basisphenoid appears in the basilar plate, but in front of the notochord, nor does it arise in exactly the same way as the centrum of a vertebra. The basisphenoid with the alisphenoids, which develop independently in the side walls of the brain, may, however, still be considered as forming a vertebra, though the resemblance is not so great as in the case of the occipital ring. The presphenoid, being long and pointed, is very unlike a vertebral body. The orbitosphenoids develop separately from it. The ethmoid also differs from a vertebra, for it surrounds not the whole nervous axis as the two hinder "vertebræ" do, but only two prolongations of it, the olfactory lobes. In its development and final form it shows no particular resemblance to a vertebra. Its body, the pars perpendicularis (mesethmoid) shows no similarity with a vertebral centrum. Completing the three hinder cranial "vertebræ" and roofing in the brain are the supraoccipital, the parietals and the frontals. The premaxillaries, vomer, and nasals do not belong to the cranial scheme; they are covering bones connected with the ethmoid. So, too, the ear-capsule is not part of the cranial vertebræ, but is rather to be compared to the intercalary bones in the vertebral column of certain fish. Summing up as regards the cranial vertebræ Rathke writes, "We find that the four different groups of bones, consisting of the basioccipital with its intercalary (the supraoccipital), the basisphenoid with its intercalaries (parietals), the presphenoid with its intercalaries (frontals), and the ethmoid with its outgrowths (turbinals and cribriform plate), taking them in order from behind forwards, show an increasing divergence from the plan according to which vertebræ as commonly understood develop, so that the basioccipital shows the greatest resemblance to a vertebra, the ethmoid the least" (p. 30).

In a posthumous volume published in 1861 the same opinion is put forward. "In the head, too," he writes, "some vertebræ can be recognised, although in a more or less modified form. Yet at most only four cranial vertebræ can be assumed, and these differ from ordinary well-developed vertebræ in their manner of formation the more the farther forward they lie."[214]

Rathke was an able and careful critic of the vertebral theory of the skull, but he accepted it in the main. Actual attack on the theory upon embryological grounds was begun by C. Vogt, in his work on the development of Coregonus,[215] and in his paper on the development of Alytes.[216] He described for Coregonus an origin of the skull in the main similar to that established by Rathke for the adder. There was a "nuchal plate" in which the front end of the notochord was imbedded; the notochord ended at the level of the labyrinth; there were two lateral bands, comparable to Rathke's lateral trabeculæ; a "facial plate" was also formed, which seems on the whole equivalent to the plate formed by the fused anterior ends of the trabeculæ. A little later the cranium formed a complete cartilaginous box surrounding the brain, very similar to the adult cranium of a shark.

In his criticism of the vertebral theory of the skull, Vogt started by defining the vertebra as a ring formed round the chorda. Now since only the occipital segment of the skull is formed actually round the notochord, the parts of the skull lying in front of this cannot themselves be vertebræ, though they may be considered as prolongations of the occipital or nuchal vertebra. "We must regard the nuchal plate as a true vertebra, modified, it is true, in its formation and development by its particular functions. Now, since the notochord ends with the nuchal plate we can no longer regard as vertebræ the parts of the skull that lie beyond, such as the lateral processes of the cranium and the facial plate, for they have no relation with the notochord" (p. 123).

To support this view he adduced the fact that the vertebral divisions (primitive vertebræ) visible in the trunk do not extend into the head. He used precisely the same arguments in his paper on Alytes to destroy the vertebral theory of the skull. We quote the following passage translated by Huxley (1864, p. 295) from this paper. "It has therefore become my distinct persuasion that the occipital vertebra is indeed a true vertebra, but that everything which lies before it is not fashioned upon the vertebrate type at all, and that efforts to interpret it in such a way are vain; that, therefore, if we except that vertebra (occipital) which ends the spinal column anteriorly, there are no cranial vertebræ at all."

L. Agassiz, himself a pupil of Döllinger, in the general part (1844) of his Recherches sur les Poissons fossiles (Neuchâtel, 1833-43), repeats in the main his pupil Vogt's criticism of the vertebral theory (vol. i., pp. 125-9).

These arguments of Vogt and Agassiz were not considered by Müller to dispose of the theory,[217] which maintained a firm hold even upon embryologists. It was still upheld by Reichert, and Kölliker in 1849 showed himself convinced of its general validity.

A useful step in the analysis of the concept "vertebra" was taken by Remak,[218] who showed what a complex affair the formation of vertebræ really is, involving as it does a complete resegmentation (Neugliederung) of the vertebral column, whereby the original vertebral bodies were replaced by the secondary definitive bodies (p. 143). Remak showed, as he thought, that the protovertebral segmentation of the dorsal muscle-plates did not extend into the head, and he denied Reichert's assertion (1837) that the cranial basis in mammals showed transverse grooves delimiting three cranial vertebræ (p. 36). The gill-slits, he considered, could not possibly be regarded as marking the limits of head vertebræ.

In 1858 appeared Huxley's well-known Croonian Lecture, On the Theory of the Vertebrate Skull,[219] in which he stated with great clearness and force the case for the embryological method of determining homologies, and criticised with vigour the vertebral theory of the skull. By this time the two rival methods in morphology had become clearly differentiated, and Huxley was able to contrast them, or at least to show how necessary the new embryological method was as a corrective and a supplement to the older anatomical, or, as he calls it, "gradation" method. Applied to the "Theory of the Skull," the gradation method consists in comparing the parts of the skull and vertebral column in adult animals with respect to their form and connections. "Using the other method, the investigator traces back skull and vertebral column to their earliest embryonic states and determines the identity of parts by their developmental relations" (p. 541). This second method is the final and ultimate. "The study of the gradations of structure presented by a series of living beings may have the utmost value in suggesting homologies, but the study of development alone can finally demonstrate them" (p. 541). As an example of the utility and, indeed, the necessity of applying the embryological method Huxley takes the case of the quadrate bone in birds. This bone had been generally regarded by anatomists as the equivalent of the tympanic of mammals, on account of its connection with the tympanum; but Reichert showed (1837) that the same segment of the first visceral arch developed into the incus in mammals, and into the quadrate in birds, and that therefore the quadrate was homologous with the incus. Similarly, on developmental grounds, the malleus or hammer of mammals is the homologue of the articular of birds, since both are developed from a portion of Meckel's cartilage identical in form and connections in the two groups. The homologies of the bones connected with the jaws in bony fishes had long been a subject of contention among comparative anatomists; Huxley shows from his personal observations how the development of the visceral arches throws light upon these difficulties. The mandibular arch in the developing fish is abruptly angled, as in the embryo of Tetrapoda; the upper prong of it ossifies into the palatine and pterygoid; at the angle is formed the quadrate (jugal, Cuvier), and to the quadrate is articulated the lower jaw, which ossifies round the lower prong or Meckel's cartilage. The scheme of development of the jaws is accordingly similar in fish to what it is in other Vertebrates, and this similarity of development enables Huxley to recognise what are the true homologues of the quadrate, the palatine and the pterygoid in adult bony fish, and to prove that the symplectic and the metapterygoid (tympanal, Cuvier) are bones peculiar to fish. In developing Amphibia Huxley found a suspensorium of hyoid and mandibular arches similar to the hyomandibular of fish.

Tackling his main problem of the unity of plan of the vertebrate skull, Huxley shows, by a careful discussion of the anatomical relationships of the chief bones in typical examples of all vertebrate classes, that there is on the whole unity of plan as regards the osseous skull. This unity of composition can be established, on the gradation method, by considering the connections of the bones of the skull with one another, their relations to the parts of the brain and to the foramina of the principal cranial nerves. The assistance of the embryological method is, however, necessary in determining many points with regard to the bones developed in relation to the visceral arches. But there is a further step to be taken. "Admitting ... that a general unity of plan pervades the organisation of the ossified skull, the important fact remains that many vertebrated animals—all those fishes, in fact, which are known as Elasmobranchii, Marsipobranchii, Pharyngobranchii and Dipnoi have no bony skull at all, at least in the sense in which the words have hitherto been used" (p. 571). The membranous or cartilaginous skull of these fishes shows a general resemblance in its main features to the ossified skull of other Vertebrates; the relations of the ear to the vagus and trigeminal nerves are, for instance, the same in both; the main regions of the cartilaginous skull can be homologised with definite bones or groups of bones in the bony skull; but discrepancies occur. It is again to development that we must turn to discover the true relationship of the cartilaginous to the ossified skull. "The study of the development of the ossified vertebrate skull ... satisfactorily proves that the adult crania of the lower Vertebrata are but special developments[220] of conditions through which the embryonic crania of the highest members of the sub-kingdom pass" (p. 573). It is with the embryonic cranium of higher Vertebrates that the adult skull of the lower fishes must be compared, and the comparison will show a substantial though not a complete agreement between them. Thus, speaking of the development of the frog's skull, Huxley writes:—"If, bearing in mind the changes which are undergone by the palatosuspensorial apparatus, ... we now compare the stages of development of the frog's skull with the persistent conditions of the skull in the Amphioxus, the lamprey, and the shark, we shall discover the model and type of the latter in the former. The skull of the Amphioxus presents a modification of that plan which is exhibited by the frog's skull when its walls are still membranous and the notochord is not yet embedded in cartilage. The skull of the lamprey is readily reducible to the same plan of structure as that which is exhibited by the tadpole when its gills are still external and its blood colourless. And finally, the skull of the shark is at once intelligible when we have studied the cranium in further advanced larvæ, or its cartilaginous basis in the adult frog" (p. 577). Development, therefore, proves what comparative anatomy could only foreshadow—the unity of plan of all vertebrate skulls, ossified and unossified alike. "We have thus attained to a theory or general expression of the laws of structure of the skull. All vertebrate skulls are originally alike; in all (save Amphioxus?) the base of the primitive cranium undergoes the mesocephalic flexure, behind which the notochord terminates, while immediately in front of it the pituitary body is developed;[221] in all, the cartilaginous cranium has primarily the same structure—a basal plate enveloping the end of the notochord and sending forth three processes, of which one is short and median, while the other two, the lateral trabeculæ, pass on each side of the space on which the pituitary body rests, and unite in front of it; in all, the mandibular arch is primarily attached behind the level of the pituitary space, and the auditory capsules are enveloped by a cartilaginous mass, continuous with the basal plate between them. The amount of further development to which the primary skull may attain varies, and no distinct ossifications at all may take place in it; but when such ossification does occur, the same bones are developed in similar relations to the primitive cartilaginous skull" (p. 578).

In a word, there is a general plan or primordial type which is manifested in the higher forms most clearly in their earliest development—an embryological archetype therefore.

Huxley now goes on to consider the relation of this general plan or type of the skull to the structure and development of the vertebral column. Does the skull in its development show any signs of a composition out of several vertebræ? The vertebral column develops as a segmented structure round the notochord; the skull develops first as an unsegmented plate extending far beyond the notochord. The processes of this basilar plate, the trabeculæ, are quite unlike anything in the vertebral column. It is true that when the process of ossification begins, separate bones are differentiated in the basilar plate one in front of the other, giving an appearance of segmentation. The hindmost of these bones, the basioccipital, ossifies round the notochord, quite like a vertebral centrum, and its side parts which form the occipital arch develop in a "remotely similar" way to the neural arches of the vertebræ. The next bone, however, the basisphenoid, develops in front of the notochord, and shows very little analogy with a vertebral body. The analogy is even more far-fetched when applied to the axial bones in front of the basisphenoid. The cranium might indeed be divided upon ossification into a series of segments bearing a more or less remote analogy with vertebræ. "In the process of ossification there is a certain analogy between the spinal column and the cranium, but that analogy becomes weaker and weaker as we proceed towards the anterior end of the skull" (p. 585). The best way to state the facts is to say that both skull and vertebral column start in their development from the same point, but immediately begin to diverge. The clear indications of segmentation which fully ossified adult skulls undoubtedly show are, therefore, secondary, and the vertebral theory of the skull, which was originally based upon the appearance of such fully ossified crania, is on the whole negatived by embryology.

We have now to turn back a few years in order to follow up another line of discovery which had an important bearing upon the theory of the vertebrate skull—the working out of the distinction between membrane and cartilage bones.

As early as 1731, R. Nesbitt,[222] in two lectures delivered to the Royal College of Surgeons, demonstrated that in the human fœtus some bones were formed not in cartilage but directly in fibrous tissue, and this observation was confirmed by other human anatomists, particularly by Sharpey at a considerably later date. In 1822 Arendt[223] focussed attention upon the remarkable structure of the skull of the Pike, with its cartilaginous brain-box studded all over with bony plaques, an arrangement which had already attracted the interest of Cuvier and Meckel. K. E. von Baer[224] in 1826 discussed at some length the relation between the bony and the cartilaginous skull in fishes, with particular reference to the sturgeon, coming to the following just conclusion:—"If we consider the fibrous skeleton of Ammocoetes as the first foundation of the skeleton of Vertebrates, we can form a series among the cartilaginous fishes, according as a cartilaginous skeleton penetrates more and more into this fibrous foundation. In the same way the process of ossification supplants the cartilaginous skeleton. So long as the ossifications lie in the skin, as in the sturgeon, they form corneous bones (Hornknochen), but when they lie under the skin, they form true bones, e.g., the bones of the skull in the pike" (p. 374).

Embryologists soon become aware that a similar distinction between a primitive cartilaginous foundation and a secondary overlying ossification of the skull showed itself in the development of all Vertebrates. Dugès, in his Recherches sur l'ostéologie et la myologie des Batraciens (1834), distinguished between such bones as are formed by direct ossification of the cartilaginous groundwork of the skull, and such as are developed in the periosteal fibrous tissue.

Reichert in 1838[225] noted that several of the skull bones in Amphibia are formed without the intermediary of cartilage, such as the nasals, the maxillaries and the lacrymals. So, too, the frontals and parietals of Teleosts developed independently of the cartilaginous skull, and belonged to the skeletal system of the skin, not to the true vertebral axial skeleton (pp. 215-6). Even more interesting was his discovery, afterwards confirmed by Hertwig,[226] that in the newt several bones connected with the palate were formed in the mucous membrane of the mouth by the fusion of a number of little conical teeth (p. 97). Certain of these bones he considered to be the substitutes, not the equivalents, of the palatine and pterygoid of other Vertebrates, which are formed from the upper part of the first visceral arch, a part missing in the newt (p. 100). Owing to the difference of development he would not homologise these bones in the newt with the palatine and pterygoid of other Vertebrates. He recognised also that the bone now known as the parasphenoid was developed in the frog in the mucous membrane of the mouth, and had originally no connection with the cranial basis (p. 34). Rathke in 1839 also allowed the distinction between cartilage and membrane bone, but laid no stress upon it (Entw. d. Natter., p. 197).

Jacobson in 1842[227] introduced the useful term, "primordial cranium," for the primitive cartilaginous foundation of the skull, and drew a sharp distinction between cartilage bones and membrane bones.

In his Recherches sur les Poissons fossiles,[228] L. Agassiz used Vogt's work on the development of Coregonus to establish a classification of the bones of the skull in fish, a classification which had the merit of drawing a sharp distinction between the cartilaginous groundwork and the "protective plates" of the fish's skull. He recognised that the protective plates developed in a different way from the other bones of the skull. "We must distinguish," he writes, "two kinds of ossification; one which tends to transform the primitive parts of the embryonic cranium directly into bone, and another which leads to the deposition of protective plates round this core, which develop not only upon the upper surface, as has hitherto been supposed, but also on the lateral walls and on the lower surface of the cranium" (p. 112). In the skull of all fish there are three elements—(1) the cartilaginous base, including the nuchal plate, the trabeculæ and the facial plate, together with the auditory capsules; (2) the cartilaginous cerebral envelope; (3) the bony protective plates (absent in Elasmobranchs). The bones developed in relation to these cranial elements can be classified as follows:—(1) the basioccipital, exoccipitals (paroccipitals?), supraoccipital and "petrous" (rocher), developed from the nuchal plate; the ali- and orbito-sphenoids developed from the trabeculæ; the "cranial ethmoid"[229] developed from the facial plate; (2) the parietals, frontals and nasals formed from the "superior" protective plate; the "anterior" and "posterior" frontals and the temporal, from the "lateral" plates; the body of the sphenoid and the vomer from the "inferior" plates. The other element, the cartilaginous brain-box, does not ossify, and tends to become absorbed (p. 124).

In 1849 Kölliker published a paper[230] dealing with the morphological significance of the distinction between membrane and cartilage bones, and in 1850[231] he defended his views against the criticisms of Reichert[232] in a further note entitled Die Theorie des Primordialschädels festgehalten. It is convenient to consider these papers together. Kölliker held that there was (1) a histological and (2) a morphological difference between the two categories of bones. The histological development of the two kinds was different, but this difference was not sufficient to establish a morphological distinction between them, a distinction in their anatomical Bedeutung. The true morphological distinction between them was their development in different skeleton-forming layers. Membrane bones were developed in fibrous tissue lying between the skin and the deep layer which formed the primordial cranium, and it was this formation in a separate layer that gave them a different morphological significance from the bones formed directly in the deep layer. Kölliker's distinction, therefore, was between the bones formed in the primordial cartilaginous cranium on the one hand, and the superficial ossifications in fibrous tissue on the other hand. The cartilaginous cranium in Kölliker's opinion was formed upon the vertebral type, and the membrane bones were accessory. This, at least, was his opinion in 1849. In 1850, after Stannius had shown that membrane bones occurred as integral parts of the vertebræ in certain fish, he modified his view of the membrane bones, and admitted them, at least in some cases, as constituents of the cranial vertebræ.

On this morphological distinction of membrane and cartilage bones future comparative osteology was to be based:—

"My sole aim is to state again the principle upon which comparative osteology is to be based and extended, and this is that first place should be assigned to anatomical considerations, and among these to the manner of origin of the whole bone in relation to the skeleton-forming layers" (1850, p. 290).

The homologies established by this new principle might run counter to the homologies indicated by the study of adult structure. "Thus, for instance, although the lower jaw in position, function, form and shape, appears to be the same bone throughout, yet it must be admitted that it shows a difference in the different classes. In Mammals and Man it is an entirely secondary bone (an extremity according to Reichert), in Birds, Amphibia and Fishes only partially so, for its articular belongs to Meckel's cartilage and is accordingly analogous to a rib; indeed, in the Plagiostomes, etc., the whole lower jaw along with the articular is a persistent Meckel's cartilage" (p. 290, 1850).

So, too, the supraoccipital in man cannot be fully homologised with the supraoccipital of many mammals, for its upper half arises at first in isolation as a secondary bone (p. 290).

Reichert objected to the distinction drawn by Kölliker, and denied that there was either a histological or a morphological difference between membrane and cartilage bones. It was shown a few years later by H. Müller[233] that there was in truth no essential difference in histological development between the two categories of bone, that the cartilage cells were replaced by bone cells identical with those taking part in the formation of membrane bones. The morphological distinction continued however to be recognised, particularly by the embryologists. Rathke in his volume of 1861[234] classified the bones of the skull according to their origin from the primordial cranium or from the overlying fibrous layer, distinguishing as membrane bones, the parietals, frontals, nasals, lachrymals, maxillaries and premaxillaries, jugals, tympanic, parts of the "temporal," vomer, part of the supraoccipitals in some mammals, and the mandible (with the exception of the articular in such as have a quadrate bone). Huxley was also inclined in 1864[235] to recognise the distinction, but he writes with some reserve:—"Is there a clear line of demarcation between membrane bones and cartilage bones? Are certain bones always developed primarily from cartilage, while certain others as constantly originate in membrane? And further, if a membrane bone is found in the position ordinarily occupied by a cartilage bone, is it to be regarded merely as the analogue and not as the homologue of the latter?" (p. 296).

We may note here that many comparative anatomists of the period were quite ready to decide Huxley's last question in a sense favourable to the older, purely anatomical, view of homology. Owen, for instance, held that difference of development did not disturb homologies established by form and connections. "Parts are homologous," he writes, "in the sense in which the term is used in this work, which are not always similarly developed: thus the 'pars occipitalis stricte dicta,' etc., of Soemmering is the special homologue of the supraoccipital bone of the cod, although it is developed out of pre-existing cartilage in the fish and out of aponeurotic membrane in the human subject."[236] Similarly he pointed to the diversities of development of the vertebral centrum in the different vertebrate classes as proof that development could not always be relied upon in deciding homologies (p. 89). But he could not deny that the archetype was better shown in the embryo than in the adult (supra, p. 108).

J. V. Carus[237] likewise stood firm for the older method of determining homologies by comparison of adult structure. "We can regard as homologous," he writes, "only those parts which in the fully formed animal possess a like position and show the same topographical relations to the neighbouring parts" (p. 389). Parts homologous in this sense might develop in different ways, but no great importance was to be attached to such a circumstance. Membrane and cartilage bones developed in practically the same way, from the same skeleton-forming layer, and no morphological significance attached to their distinction (pp. 227, 457). Embryology was of considerable value in helping to determine homologies, but the evidence that it supplied was contributory, not conclusive. Perhaps the greatest service which the study of development rendered was to disentangle, by a comparison of the earliest embryos, the generalised type (p. 389).

We have now traced, by our historical study of the theory of the skull, the gradual evolution of the tendency to find in development the surest guide to determining homologies. We have seen how the embryological "type" came to be substituted, in whole or in part, for the anatomical "type" derived from the study of adult structure. But we have had to do only with a modification, not with a transformation, of the criterion of homology recognised by the anatomists. Homology is still determined by position, by connections, in the embryo as in the adult. "Similarity of development" has become the criterion of homology in the eyes of the embryologist, but "similarity of development" means, not identity of histological differentiation, but similarity of connections throughout the course of development. For the purposes of morphology, development has to be considered as an orderly sequence of successive forms, not in its real nature as a process essentially continuous. Morphology has to replace the living continuity by a kinematographic succession of stages. Since it is the earliest of these stages that manifest the simplest and most generalised structural relations of the parts, it is in the earlier stages that homologies can be most easily determined. But these homologies are still determined solely by the relative positions and connections of the parts, just as homologies are determined in the last of all the stages of development, the adult state. And since the generalised type is shown most clearly in the earliest stages and tends to become obscured by later differentiation, homologies observed in embryonic life are to be upheld even if the relations in adult life seem to indicate different interpretations.