Origin of Vertebrates/Chapter X

The shifting of the orifice of the olfactory passage, which led to the old mouth, from the ventral to the dorsal side, as seen in the transformation of the ventrally situated hypophysial tube of the young Ammocœtes, to the dorsally situated nasal tube of the full-grown Ammocœtes, affords one of the most important clues in the whole of this story of the origin of vertebrates; for, if Ammocœtes is the nearest living representative of the first-formed fishes, then we ought to expect to find that the dorsal head-shield of such fishes is differentiated from that of the contemporary Palæostraca by the presence of a median frontal opening anterior to the eyes. Conversely, if such median nasal orifice is found to be a marked characteristic of the group, in combination with lateral and median eyes, as in Ammocœtes, then we have strong reasons for interpreting these head-shields by reference to the head of Ammocœtes.

The oldest known fishes belong to a large group of strange forms which inhabited the Silurian and Devonian seas, classed together by Smith Woodward under the name of Ostracodermi. These are divided into three orders: (1) the Heterostraci, including one family, the Pteraspidæ, to which Pteraspis and Cyathaspis belong; (2) the Osteostraci, divisible into two families, the Cephalaspidæ and Tremataspidæ, which include Cephalaspis, Eukeraspis, Auchenaspis or Thyestes, and Tremataspis; and (3) the Antiarcha, with one family, the Astrolepidæ, including Astrolepis, Pterichthys, and Bothriolepis. Of these, the first two orders belong to the Upper Silurian, while the third is Devonian.

Of the three orders above-named, the Heterostraci and Osteostraci are the oldest, and among them the Cephalaspidæ have afforded the most numerous and best worked-out specimens. At Rootziküll, in the island of Œsel, the form known as Thyestes (Auchenaspis) verrucosus is especially plentiful, being found thickly present in among the masses of Eurypterid remains, which give the name to the deposit. Of late years this species has been especially worked at by Rohon, and many beautiful specimens have been figured by him, so that a considerable advance has been made in our knowledge since Pander, Eichwald, Huxley, Lankester, and Schmidt studied these most interesting primitive forms.

All observers agree that the head-region of these fishes was covered by a dorsal and ventral head-shield, while the body-region was in most cases unknown, or, as in Eichwald's specimens, and in the specimens figured in Lankester and Smith Woodward's memoirs, was made up of segments which were not vertebral in character, but formed an aponeurotic skeleton, being the hardened aponeuroses between the body-muscles. This body-skeleton, which possesses its exact counterpart in Ammocœtes, will be considered more fully when I discuss the origin of the spinal region of the vertebrate.

Of the two head-shields, ventral and dorsal, the latter is best known and characterizes the group. It consists of a dorsal plate, with characteristic horns, which in Thyestes verrucosus (Fig. 128), as described by Rohon, is composed of two parts, a frontal part and an occipital part (occ.), the occipital part being composed of segments, and possessing a median ridge—the crista occipitalis. In Lankester's memoir and in Smith Woodward's catalogue, a large number of known forms are described and delineated, and we may perhaps say that in some of the forms, such as Eukeraspis pustuliferus (Fig. 127, B), the frontal part of the shield only is capable of preservation as a fossil, while in Cephalaspis (Fig. 127, A) not only the frontal part but a portion of the occipital region is preserved, the latter being small in extent when compared with the occipital region of Auchenaspis (Thyestes). Finally, in Tremataspis and Didymaspis, the whole of both frontal and occipital region is capable of preservation, the line of demarcation between these two regions being well marked in the latter species.



In the best preserved specimens of all this group of fishes a frontal median orifice is always present; it appears in some specimens obscurely partially divided into two parts. Perhaps the best specimen of all was obtained by Rohon at Rootziküll, and is thus described by him:—

The frontal part of the dorsal head-plate carried (Fig. 128) the two orbits for the lateral eyes (l.e.), a marked frontal organ (fro.), and a median depression (gl.), to which he gives the name parietal organ. The occipital part (occ.) was clearly segmented, and carried, he thinks, the branchiæ. I reproduce Rohon's figure of the frontal organ in Thyestes (Fig. 129); he describes it as a deeply sunk pit, divided in the middle by a slit, which leads deeper in, he supposes, towards the central nervous system.

— Thyestes (Auchenaspis) verrucosus. (From .)

Fro., narial opening; l.e., lateral eyes; gl., glabellum or plate over brain; Occ., occipital region.

A similar organ was described by Schmidt in Tremataspis, and considered by him to be a median nose. Such also is the view of Jaekel, who points out that a median pineal eye exists between the two lateral eyes in this animal, as in all other of these ancient fishes, so that this frontal organ does not, as Patten thinks, represent the pineal eye. The whole of this group of fishes, then, is characterized by the following striking characteristics:—

1. Two well-marked lateral eyes near the middle line.

2. Between the lateral eyes, well-marked median eyes, very small.

3. In front of the eye-region a median orifice, single.

In addition, behind the eye-region a median plate is always found, frequently different in structure to the rest of the head-shield, being harder in texture—the so-called post-orbital plate.

— Thyestes Verrucosus. (From .)

What is the structure of this head-shield? It has been spoken of as formed of bone because it possesses cells, being thus unlike the layers of chitin, which are formed by underlying cells but are not themselves cellular. At the same time, it is recognized on all sides that it has no resemblance to bone-structure as seen in fossil remains of higher vertebrates. The latest and best figure of the structure of this so-called bone is given in Rohon's paper already referred to. It is, so he describes, clearly composed of fibrillæ and star-shaped cells, arranged more or less in regular layers, with other sets of similar cells and fibrillæ arranged at right angles to the first set, or at varying angles. The groundwork of this tissue, in which these cells and fibrils are embedded, contained calcium salts, and so the whole tissue was preserved. In places, spaces are found in it, in the deepest layer large medullary spaces; more superficially, ramifying spaces which he considers to be vascular, and calls Haversian canals; the star-like cells, however, are not arranged concentrically around these spaces, as in true Haversian canals.

This structure is therefore a calcareous infiltration of a tissue with cells in it. Where is there anything like it?

As soon as I saw Rohon's picture (Fig. 130), I was astounded at its startling resemblance to the structure of muco-cartilage as is seen in Fig. 131, taken from Ammocœtes. If such muco-cartilage were infiltrated with lime salts, then the muco-cartilaginous skeleton of Ammocœtes would be preserved in the fossil condition, and be comparable with that of Cephalaspis, etc.

The whole structure is clearly remarkably like Rohon's picture of a section of the head-plate of a Cephalaspid (Fig. 130). In the latter case the matrix contains calcium salts, in the former it is composed of the peculiar homogeneous mucoid tissue which stains so characteristically with thionin. With respect to this calcification, it is instructive to recall the calcification in the interior of the branchial cartilages of Limulus, as described in Chapter III., for this example shows how easy it is to obtain a calcification in this chondro-mucoid material. With respect to the medullary spaces and smaller spaces in this tissue, as described by Rohon, I would venture to suggest that they need not all necessarily indicate blood-vessels, for similar spaces would appear in the head-shield of Ammocœtes if its muco-cartilage alone were preserved. Of these, some would indicate the position of blood-vessels, such, for instance, as of the external carotid which traverses this structure; but the largest and most internal spaces, resembling Rohon's medullary spaces, would represent muscles, being filled up with bundles of the upper lip-muscles.

The resemblance between the structure of the head-shield of Thyestes and the muco-cartilage of Ammocœtes, is most valuable, for muco-cartilage is unique, occurs in no other vertebrate, and every trace of it vanishes at transformation; it is essentially a characteristic of the larval form, and must, therefore, in accordance with all that has gone before, be the remnant of an ancestral skeletal tissue. The whole story deduced from the study of Ammocœtes would be incomplete without some idea of the meaning of this tissue. So also, as already mentioned, the skeleton of Ammocœtes is incomplete without taking this tissue into account. It is confined entirely to the head-region; no trace of it exists posteriorly to the branchial basket-work. It consists essentially of dorsal and ventral head-shields, connected together by the tentacular, metastomal, and thyroid bars, as already described. The ventral shield forms the muco-cartilaginous plate of the lower lip and the plate over the thyroid gland, so that the skeleton ventrally is represented by Fig. 118, B, which shows how the cartilaginous bars of the branchial basket-work are separated from each other by this thyroid plate. At transformation, with the disappearance of this muco-cartilaginous plate, the bars come together in the middle line, as in the more posterior portion of the branchial basket-work.

The dorsal head-shield of muco-cartilage covers over the upper lip, sends a median prolongation over the median pineal eyes and a lateral prolongation on each side as far as the auditory capsules, giving the shape of the head-shield of muco-cartilage, as in Fig. 118, C.

Not only then is the structure of the head-shield of a Cephalaspid remarkably like the muco-cartilage of Ammocœtes, but also its general distribution strangely resembles that of the Ammocœtes muco-cartilage.

Now, these head-shields in the Cephalaspidæ and Tremataspidæ vary very much in shape, as is seen by the comparison of Tremataspis and Auchenaspis with Cephalaspis and Eukeraspis, and yet, undoubtedly, all these forms belong to a single group, the Osteostraci.

The conception that Ammocœtes is the solitary living form allied to this group affords a clue to the meaning of this variation of shape, which appears to me to be possible, if not indeed probable. There is a certain amount of evidence given in the development of Ammocœtes which indicates that the branchial region of its ancestors was covered with plates of muco-cartilage as well as the prosomatic region.

The evidence is as follows:—

The somatic muscles of Ammocœtes form a continuous longitudinal sheet of muscles along the length of the body, which are divided up by connective tissue bands into a series of imperfect segments or myotomes. This simple muscular sheet can be dissected off along the whole of the head-region of the animal, with the exception of the most anterior part, without interfering with the attachments or arrangements of the splanchnic muscular system in the least. The reason why this separation can be so easily effected is to be found in the fact that the two sets of muscles are not attached to the same fascia. The sheet of fascia to which the somatic muscles are attached is separated from the fascia which encloses the branchial cavity by a space (cf. Figs. 63 and 64) filled with blood-spaces and cells containing fat, in which space is also situated the cartilaginous branchial basket-work. These branchial bars are closely connected with the branchial sheet of fascia, and have no connection with the somatic fascia, their perichondrium forming part of the former sheet. Upon examination, this space is seen to be mainly vascular, the blood-spaces being large and frequently marked with pigment; but it also possesses a tissue of its own, recognized as fat-tissue by all observers. The peculiarity of the cells of this tissue is their arrangement; they are elongated cells arranged at right angles to the plates of fascia, just as the fibres of the muco-cartilage are largely arranged at right angles to their limiting plates of perichondrium. These cells do not necessarily contain fat; and when they do, the fat is found in the centre of each cell, and does not push the protoplasm of the cell to the periphery, as in ordinary fat cells.

In Fig. 132, B, I give a specimen of this tissue stained by osmic acid; in Fig. 132, A, I give a drawing of ordinary muco-cartilage taken from the plate of the lower lip; and in Fig. 133, A, a modification of the muco-cartilage taken from the velum, which shows the formation of a tissue intermediate between ordinary muco-cartilage and this branchial fat-tissue.

Further, in fully-grown specimens of Ammocœtes, in the region of undoubted muco-cartilage, a fatty degeneration of the cells frequently appears, together with an increase in the blood spaces,—the precursor, in fact, of the great change which overtakes this tissue soon afterwards, at the time of transformation, when it is invaded by blood, and swept away, except in those places where new cartilage is formed. I conclude, then, that the tissue of this vascular space was originally muco-cartilage, which has degenerated during the life of the Ammocœtes. The fact that in most cases undoubted muco-cartilage is to be found here and there in this space, is strong confirmation of the truth of this conclusion.

(Mc.); m.ph., muscle of lower lip; m.sm., somatic muscle; Cor., laminated layer of skin. F., fat layer; P., pigment; Bl., blood-space; N., somatic nerve; m.br., branchial muscle; m.sm., somatic muscle.

If this conclusion is correct, we may expect that it would be confirmed by the embryological history of the tissue, and we ought to find that in much younger stages a homogeneous tissue of the same nature as muco-cartilage fills up the spaces in the branchial region, where in the Ammocœtes only blood and fat-containing cells are present. For this purpose Shipley kindly allowed me to examine his series of sections through the embryo at various ages. These specimens are very instructive, especially those stained by osmic acid, which preserves the natural thickness of this space better than other staining methods. At an age when the branchial cartilages are seen to be formed, when no fat-cells are present, a distinctive tissue (Fig. 133, B) is plainly visible in the velum and at the base of the tentacles, in the very position where in the more advanced Ammocœtes muco-cartilage exists. Taking, then, this tissue as our guide, the specimens show that the space between the skin and the visceral muscles in which the cartilaginous basket-work lies is filled with a similar material. At this stage a sheet of embryonic tissue occupies the position where, later on, blood-spaces and fat-cells are found, and this tissue resembles that seen in the velum and other places where muco-cartilage is afterwards found.



I conclude, therefore, that originally the branchial or mesosomatic region was covered with a dorsal plate of muco-cartilage, which carried on its under surface the dorsal part of the branchial basket-work, and sprang from the central core of skeletogenous tissue around the notochord; this plate was separated from the plate which covered this region ventrally by the lateral grove in which the gill-slits are situated. The ventral plate carried on its under surface the ventral part of the branchial basket-work, and was originally continuous with the plate over the thyroid gland.



Muco-cartilage, red; soft cartilage, blue; hard cartilage, purple. sk$1$, sk$2$, sk$3$, skeletal bars; c.e., position of pineal eye; ''na. cart., nasal cartilage; ped., pedicle; cr., cranium; nc.'', notochord.

In Fig. 134, A, B, C, the cranial skeleton of Ammocœtes is represented from the dorsal, ventral, and lateral aspects. The muco-cartilage is coloured red, the branchial or soft cartilage blue, and the hard cartilage purple. The degenerated muco-cartilage of the branchial region is represented as an uncoloured plate, on which the branchial basket-work stands in relief. If it were restored to its original condition of muco-cartilage, it would represent a uniform plate, on the under surface of which the basket-work would be situated; and if it were calcified and made solid, the branchial basket-work would not show at all in these figures.

Is it possible to find the reason why this skeletal covering has degenerated so early before transformation, and why the thyroid plate remains intact until transformation? We see that all that part which has degenerated is covered over by the somatic muscles,—by, in fact, muscles which, being innervated by the foremost spinal nerves, belong naturally to the region immediately following the branchial. I suggest, therefore, that the original skeletal covering of muco-cartilage has remained intact only where it has not been invaded and covered over by somatic muscles, but has been invaded by blood and undergone the same kind of degenerative change as overtakes the great mass of this tissue at transformation wherever the somatic muscles have overgrown it.

The covering somatic muscles in the branchial region form a dorsal and ventral group, of which the latter is formed in the embryo much later than the former, the line of separation between the two groups being the lateral groove, with its row of branchial openings. This groove ends at the first branchial opening, but the ventral and dorsal somatic muscles continue further headwards. It is instructive to see that, although the lateral groove terminates, the separation between the two groups of muscles is still marked out by a ridge of muco-cartilage, represented in Fig. 134, A, which terminates anteriorly in the opercular bar.

Passing now to the prosomatic region, we find that here, too, the muco-cartilaginous external covering is divisible into a dorsal and a ventral head-plate, the ventral head-plate being the plate of the lower lip, and the dorsal head-plate the plate of muco-cartilage over the front part of the head. The staining reaction with thionin maps out this dorsal head-plate in a most beautiful manner, and shows that the whole of the upper lip-region in front of the nasal orifice is one large plate of muco-cartilage, obscured largely by the invasion of the crossing muscles of the upper lip, but left pure and uninvaded all around the nasal orifice, and where the upper and lower lips come together. In addition to this foremost plate, a median tongue of muco-cartilage covers over the pineal eye and fills up the median depression between the two median dorsal somatic muscles. Also, two lateral cornua pass caudalwards from the main frontal mass of muco-cartilage over the lateral eyes, forming the well-known wedge which separates the dorsal and lateral portions of the dorso-lateral somatic muscle. In fact, similarly to what we find in the branchial region, the muco-cartilaginous covering can be traced with greater or less completeness only in those parts which are not covered by somatic muscles.

In Fig. 134, A, B, C, this striking muco-cartilaginous head-shield, both dorsal and ventral, is shown. Seeing that the upper lip wraps round the lower one on each side, and that this most ventral edge of the upper lip contains muco-cartilage, as is seen in Fig. 117, the dorsal head-shield of muco-cartilage ought, strictly speaking, to extend more ventrally in the drawings. I have curtailed it in order not to interfere with the representation of the lower lip and tentacular muco-cartilages.

From what has been said, it follows that the past history of the skeletal covering of the whole head-region of Ammocœtes, both frontal and occipital, can be conjectured by means of the ontogenetic history of the foremost myomeres.

Dohrn and all other observers are agreed that during the development of this animal a striking forward growth of the foremost somatic myomeres takes place, so that, as Dohrn puts it, the body-musculature has extended forwards over the gill-region, and at the same time the gill-region has extended backwards. It is therefore probable that in the ancestral form the myotomes, innervated by the first spinal nerves, immediately succeeded the branchial region. Judging from Ammocœtes, the forward growth was at first confined to the dorsal region, and therefore invaded the dorsal head-plate, the ventral musculature being distinctly a later growth. With respect to this dorsal part of the myotomes, the first myotome is originally situated some distance behind the auditory capsule, and then grows forward towards the nasal opening; the lateral part, according to Hatschek, grows forward more quickly than the dorsal part, and splits itself above and below the eye into a dorso-lateral part, which extends up to the olfactory capsule, and a ventro-lateral part (m. lateralis capitis anterior, superior, and inferior), thus giving rise to the characteristic appearance of the muco-cartilaginous head-shield of Ammocœtes.

According, then, to the extent of the growth of these somatic muscles, the shape of the muco-cartilaginous head-shield will vary, and if it were calcified and then fossilized we should obtain fossil head-shields of widely differing configuration, although such fossils might be closely allied to each other. This is just what is found in this group. Let the muco-cartilage extend over the whole of the branchial region of Ammocœtes, the resulting head-shield would be as in Fig. 135, A; the branchial bars below the muco-cartilaginous shield might or might not be evident, and the line between the branchial and the trigeminal region might or might not be indicated. Such a head-shield would closely resemble those of Didymaspis and Tremataspis respectively. Now suppose the somatic musculature to encroach slightly on the branchial region and also laterally to the end of the anterior branchial region, then we should obtain a shape resembling that of Thyestes (Fig. 135, B). Continue the same process further, the lateral muscle always encroaching further than the median masses, until the whole or nearly the whole branchial region is invested, and we get the head-shield of Cephalaspis (Fig. 135, C); further still, that of Keraspis, and yet still further, that of Ammocœtes (Fig. 135, D).



So close is this similarity, from the comparative point of view, between the dorsal head-shield of the Osteostraci and the dorsal cephalic region of Ammocœtes that it justifies us in taking Ammocœtes as the nearest living representative of such types; it is justifiable, therefore, to interpret by means of Ammocœtes the position of other organs in these forms. First and foremost is the hard plate known as the post-orbital plate, so invariably found. In Fig. 134, C, I have inserted (cr.) the position of the membranous cranium of Ammocœtes, and it is immediately evident that the primordial cranium of the Osteostraci must occupy the exact position indicated by this median hard plate. For this very reason this median plate would be harder than the rest in order to afford a better protection to the brain underneath. This plate, because of its position, may well receive the same name as the similar plate in the trilobite and various palæostracans and be called the glabellum.

We may thus conceive the position of the nose, lateral eyes, median eyes, and cranium in these old fishes. In addition, other indications of a segmentation in this head-region have been found. The most striking of all the specimens hitherto discovered are some of Thyestes verrucosus, discovered by Rohon, in which the dorsal shield has been removed, and so we are able to see what that dorsal shield covered.

In Fig. 136, I reproduce his drawing of one of his specimens from the dorsal and lateral aspects. These drawings show that the frontal part of the shield covered a markedly segmented part of the animal; five distinct segments are visible apart from the median most anterior region. This segmented region is entirely confined to the prosomatic region, i.e. to the region innervated by the trigeminal nerve. An indication of similar markings is given in Lankester's figure of Eukeraspis pustuliferus (see Fig. 127, B), and, indeed, evidence of a segmentation under the antero-lateral border of the head-shield is recognized at the present time, not only in the Cephalaspidæ, but also in the Pteraspidæ, as was pointed out to me by Smith Woodward in the specimens at the British Museum. Also, in Cyathaspis, Jaekel has drawn attention to markings of a similar segmental nature (Fig. 137).

There seems, then, little doubt but that these primitive fishes possessed something in this region which was of a segmental character, and indicated at least five segments, probably more.

Rohon entitles his discovery 'the segmentation of the primordial cranium.' It would, I think, be better to call it the segmentation of the anterior region of the head, for that is in reality what his figures show, not the segmentation of the primordial cranium, which, to judge from Ammocœtes, was confined to the region of the glabellum.

What is the interpretation of this appearance?

Any segmentation in the head-region must be indicative of segments belonging to the trigeminal or prosomatic region, or of segments belonging to the vagus or mesosomatic region. Many palæontologists, looking upon segmentation as indicative of gills and gill-slits, have attempted to interpret such markings as branchial segments, regardless of their position. As the figures show, they extend in front of the eyes and reach round to the front middle line, a position which is simply impossible for gills, but points directly to a segmentation connected with the trigeminal nerve. Comparison with Ammocœtes makes it plain enough that the markings in question are prosomatic in position, and that the gill-region must be sought for in the place where Schmidt and Rohon located it in Thyestes, viz. the so-called occipital region.

This discovery of Rohon's is, in my opinion, of immense importance, for it indicates that, in these early fishes, the prosomatic segmentation, associated with the trigeminal nerve, was much more well-marked than in any fishes living in the present day. Why should it be more well-marked? Turning to the palæostracan, it is very suggestive to compare the markings on their prosomatic carapace with these markings. Again and again we find indications of segmentation in these fossils similar to those seen in the ancient fishes. Thus in Fig. 138 I have put side by side the palæostracan Bunodes and the fish Thyestes, both life size. In the latter I have indicated Rohon's segments; in the former the markings usually seen.

From the evidence of Phrynus, Mygale, etc., as already pointed out, such markings in the palæostracan fossils would indicate the position of the tergo-coxal muscles of the prosomatic appendages, even though such appendages have not yet been discovered, and it is significant that in all these cases there is a distinct indication of a median plate or glabellum in addition to the segmental markings. Especially instructive is the evidence of Phrynus, as is seen by a comparison of Figs. 107 and 108, which shows clearly that this median plate (glab.) covered the brain-region, a brain-region which is isolated and protected from the tergo-coxal muscles by the growth dorsalwards of the flanges of the plastron. In this way an incipient cranium of a membranous character is formed, which helps to give attachment to these tergo-coxal muscles. As such cranium is derived directly from the plastron, it is natural that it should ultimately become cartilaginous, just as occurs when Ammocœtes becomes Petromyzon and the cartilaginous cranium of the latter arises from the membranous cranium of the former. In Galeodes also the growth dorsalwards of the lateral flanges of the plastron to form an incipient cranium in which the brain lies is very apparent.

— Thyestes Verrucosus Bunodes Lunula.

I venture, then, to suggest that in the Osteostraci the median hard plate or glabellum protected a brain which was enclosed in a membranous cranium, very probably not yet complete in the dorsal region—certainly not complete if the median pineal eyes so universally found in these ancient fishes were functional—a cranium derived from the basal trabeculæ, in precisely the same manner as we see it already in its commencement in Phrynus and other scorpions. With the completion of this cranium and its conversion into cartilage, and subsequently into bone, an efficient protection was afforded to the most vital part of the animal, and thus the hard head-shield of the Palæostraca and of the earliest fishes was gradually supplanted by the protecting bony cranium of the higher vertebrates.

Step by step it is easy to follow in the mind's eye the evolution of the vertebrate cranium, and because it was evolved direct from the plastron, the impossibility of resolving it into segments is at once manifest; for although the plastron was probably originally segmented, as Schimkéwitsch thinks, all sign of such segmentation had in all probability ceased, before ever the vertebrates first made their appearance on the earth.

It follows further, from the comparison here made, that those antero-lateral markings indicative of segments, found so frequently in these primitive fishes, must be interpreted as due not to gills but to aponeuroses, due to the presence of muscles which moved prosomatic appendages, muscles which arose from the dorsal region in very much the same position as do the muscles of the lower lip in Ammocœtes; the latter, as already argued, represent the tergo-coxal muscles of the last pair of prosomatic appendages—the chilaria or metastoma. Such an interpretation of these markings signifies that the first-formed fishes must have possessed prosomatic appendages of a more definite character than the tentacles of Ammocœtes, something intermediate between those of the palæostracan and Ammocœtes.

For my part I should not be in the least surprised were I to hear that something of the nature of appendages in this region had been found, especially in view of the well-known existence of the pair of appendages in the members of the Asterolepidæ—large, oar-like appendages which may well represent the ectognaths.

Of the three groups of fishes—the Heterostraci, the Osteostraci, and the Antiarcha—the last is Devonian, and therefore the latest in time of the three, while the earliest is the first group, as both Pteraspis and Cyathaspis have been found in lower levels of the Silurian age than any of the Osteostraci, and, indeed, Cyathaspis has been discovered in Sweden in the lower Silurian. This, the earliest of all groups of fishes, is confined to two forms only—Pteraspis and Cyathaspis,—for Scaphaspis is now recognized to be the ventral shield of Pteraspis.

Hitherto a strong tendency has existed in the minds both of the comparative anatomist and the palæontologist to look on the elasmobranchs as the earliest fishes, and to force, therefore, these strange forms of fish into the elasmobranch ranks. For this purpose the same device is often used as has been utilized in order to account for the existence of the Cyclostomata, viz. that of degeneration. The evidence I have put forward is very strongly in favour of a connection between the cyclostomes and the cephalaspids, and agrees therefore with all the rest of the evidence that the jawless fishes are more ancient than those which bore jaws—the Gnathostomata.

This is no new view. It was urged by Cope, who classified the Heterostraci, Osteostraci, and Antiarcha under one big group—the Agnatha—from which subsequently the Gnathostomata arose. Cope's arguments have not prevailed up to the present time, as is seen in the writings of Traquair, one of the chief authorities on the subject in Great Britain. He is still an advocate of the elasmobranch origin of all these earliest fishes, and claims that the latest discoveries of the Silurian deposits (Thelodus Pagei) and other members of the Cœlolepidæ confirm this view of the question.

This view may be summed up somewhat as follows:—

Cartilaginous jaws would not fossilize, and the Ostracoderms may have possessed them.

They may have degenerated from elasmobranchs just as the cyclostomes are supposed to have degenerated.

Seeing that bone succeeds cartilage, the presence of bony shields in Cephalaspis, etc., indicates that their precursors were cartilaginous, presumably elasmobranch fishes.

Of these arguments the strongest is based on the supposed bony covering of the Osteostraci, with the consequent supposition that their ancestors possessed a cartilaginous covering. This argument is entirely upset, if, as I have pointed out, the structure of the cephalaspid shield is that of muco-cartilage and not of bone. If these plates are a calcified muco-cartilage, then the whole argument for their ancestry from animals with a cartilaginous skeleton falls to the ground, for muco-cartilage is the precursor not only of bone, but also of cartilage itself.

The evidence, then, points strongly in favour of Cope's view that the most primitive fishes were Agnatha, after the fashion of cyclostomes, as is also believed by Smith Woodward, Bashford Dean, and Jaekel.

Among living animals, as I have shown, the Limulus is the sole survivor of the palæostracan type, and Ammocœtes alone gives a clue to the nature of the cephalaspid, i.e. the osteostracan fish. Older than the latter is the heterostracan, Pteraspis, and Cyathaspis. Is it possible from their structure to obtain any clue as to the actual passage from the palæostracan to the vertebrate?

Here again, as in the case of the Osteostraci, a relationship to the elasmobranch has been supposed, for the following reasons:—

The latest discoveries in the Silurian and Devonian deposits have brought to light strange forms such as Thelodus and Drepanaspis, of which the latter from the Devonian must, according to Traquair, be included in the Heterostraci. It possessed, as seen in Fig. 139, large plates, after the fashion of Pteraspis, and also many smaller ones.

The former, from the upper Silurian, belongs to the Cœlolepidæ, and was covered over with shagreen composed of small scutes, after the fashion of an elasmobranch. Traquair suggests that Thelodus arose from the original elasmobranch stock; that by the fusion of scutes such a form as Drepanaspis occurred, and, with still further fusion, Pteraspis.

There are always two ways of looking at a question, and it seems to me possible and more probable to turn the matter round and to argue that the original condition of the surface-covering was that of large plates, as in Pteraspis. By the subsequent splitting up of such plates, Drepanaspis was formed, and later on, by further splitting, the elasmobranch, Thelodus being a stage on the way to the formation of an elasmobranch, and not a backward stage from the elasmobranch towards Pteraspis.

This method of looking at the problem seems to me to be more in consonance with the facts than the reverse; for, as pointed out by Jaekel, the fishes with large plates are the oldest, and in Cyathaspis, the very oldest of all, the size of the plates is most conspicuous; he considers, therefore, this preconceived view that large plates are formed by the fusion of small ones must give way to the opposite belief.



So also Rohon, as quoted by Traquair, who, in his first paper accepted Lankester's view that the ridges of the pteraspidian shield were formed by the fusion of a linear arrangement of numbers of placoid scales, suggests in his second paper that these ridges may have been the most primitive condition of the dermal skeleton of the vertebrate, out of which, by differentiation, the dermal denticles (placoid scales) of the selachian, as well as their modifications in the ganoids, teleosteans, and amphibians, have arisen.

One thing is agreed upon on all sides; no sign of bone-corpuscles is to be found in this dermal covering of Pteraspis. In the deeper layers are large spaces, the so-called pulp-cavities leading into narrow canaliculi, the so-called dentine canals. The structure is looked upon as similar to that of the pulp and dentine canals of many fish-scales.

On the other hand, this dermal covering of Pteraspis has been compared by Patten with the arrangement of the chitinous structure of certain parts of the external covering of Limulus, a comparison which to my mind presents a great difficulty. The chitin-layers in Limulus are external to the epidermal cells, being formed by them; the layers in Pteraspis which look like chitin must have been internal to the epidermal layer; for each vascular canal which passes from a pulp-cavity on its way to be distributed into the dentine canals of the ridge gives off short side branches, which open directly into the groove between the ridges. If these canals were filled with blood they could not possibly open directly into the open grooves between the ridges; these openings must, therefore, have been covered over with an epithelial layer which covered over the surface of the animal, and consequently the chitin-like structure must have been internal to the epidermis, and not external, as on Patten's view. The comparison of this structure with the dentine of fish-scales signifies the same thing, for in the latter the epidermis is external to the dentine-plates, the hard skeletal structure is in the position of the cutis, not of the cuticle.

The position appears to me to be this: the dermal cranial skeleton of vertebrates, whether it takes the form of a bony skull or of the dorsal plates of a cephalaspid or a pteraspid is, in all cases, not cuticular, i.e. is not an external formation of the epidermal cells, but is formed in tissue of the nature of connective tissue underlying the epidermis. On the contrary, the hard part of the head-carapace of the palæostracan is an external formation of the epidermal cells.

If, then, this tissue of Pteraspis is not to be looked upon as chitin, how can we imagine its formation? It is certainly not bone, for there are no bone-corpuscles; it is a very regular laminated structure resembling in appearance chitin rather than anything else.

As in all cases of difficulty, turn to Ammocœtes and let us see what clue there is to be found there. The skin of Ammocœtes is peculiar among vertebrates in many respects. It consists of a number of epidermal cells, as in Fig. 140, the varying function of which need not be considered here, covered over with a cuticular layer which is extraordinarily thick for the cuticle of a vertebrate skin; this cuticular layer is perforated with fine canaliculi, through which the secretion of the underlying cells passes, as is seen in Fig. 140, A and B. This cuticle corresponds to the chitinous covering of the arthropod, and like it is perforated with canaliculi, and, according to Lwoff, possibly contains chitin. The epidermal cells rest on a thick layer of most striking appearance (Fig. 141), for it resembles, in an extraordinary degree, when examined superficially, a layer of chitin; it is called the laminated layer, and is characterized by the extreme regularity of the laminæ. This appearance is due, as the observations of Miss Alcock show, to alternate layers of connective tissue fibres arranged at right angles to each other, each fibre running a straight course and possessing its own nucleus. Although the fibres in each layer are packed close together, they are sufficiently apart to form with the fibres of the alternate layers a meshwork rather than a homogeneous structure, and thus the surface view of this layer shows a regular network of very fine spaces through which nerve-fibres and fluid pass. This layer is easily dissolved in a solution of hypochlorite of soda, a fluid which dissolves chitin. Any one looking at Ammocœtes would say that the only part of its skin which resembles chitin is this laminated layer, and therefore the only part of its skin which would afford an indication of the nature of the skeleton of Pteraspis is this laminated layer, which belongs to the cutis, and not to the cuticle. Yet another significant peculiarity of this layer is its entire disappearance at transformation. Miss Alcock, in a research not yet published, has shown that this layer is completely broken up and absorbed at transformation; the cutis of Petromyzon is formed entirely anew, and no longer presents any regular laminated character, but resembles rather the sub-epidermal connective tissue layer of the skin of higher vertebrates. This laminated layer, then, just like the muco-cartilage, shows, by its complete disappearance at transformation, its ancestral character.



Very suggestive is the arrangement of the different skeletal tissues in the head-region of Ammocœtes. Fig. 141 represents a section through the head near the pineal eye. Most internally is a, a section of the membranous cranium, then comes b, the muco-cartilaginous skeleton, then c, the laminated layer, and finally d, the external cuticle. If in Ammocœtes we possess an epitome of the history of the vertebrate, how would these layers be represented in the past ages, supposing they could be fossilized?



a, cranial wall; b, muco-cartilage; c, laminated layer; d, external cuticular layer.

The most internal layer a, by the formation of cartilage and then bone, represents the great mass of vertebrate fossils; the next layer b, by a process of calcification, as previously argued, represents the head-shield of the Osteostracan fishes; while the cuticular layer d, no longer thin, is the remnant of the Palæostracan head-carapace. Between these two layers, b and d, lies the laminated layer c. Intermediate to the Palæostracan and the Osteostracan comes the Heterostracan, with its peculiar head-shield—a head-shield whose origin is more easily conceivable as arising from something of the nature of the laminated layer than from any other structure represented in Ammocœtes.

My present suggestion, then, is this: the transition from the skeletal covering of the Palæostracan to that of the highest vertebrates was brought about by the calcification of successive layers from without inwards, all of which still remain in Ammocœtes and show how the external chitinous covering of the arthropod was gradually replaced by the deep-lying internal bony cranium of the higher vertebrates.

In Ammocœtes the layer which represents the covering of the Palæostracan has already almost disappeared. At transformation the layers representing the stage arrived at by the Heterostracan and the Osteostracan disappear; but the stage representing the higher vertebrates, far from disappearing, by the formation of cartilage reaches a higher stage and prepares the way for the ultimate stage of all—the formation of the bony cranium.

So much for the evidence as to the nature of the structure of the head-shield of the Pteraspidæ.

It suggests that these fishes were covered anteriorly with armoured plates derived from the cutis layer of the skin, a layer which was specially thickened and very vascular, apparently, to enable respiration to be very largely, if not entirely, effected by the surface of the body. It is difficult to understand how the sea-scorpions breathed, and it is easy to see how the formation of ventral and dorsal plates enclosing the mesosomatic appendages may at the outset have hindered the action of the branchiæ. The respiratory chamber, according to my view, had at first the double function of respiration and digestion. A new digestive apparatus was the pressing need at the time; it would, therefore, be of distinct advantage to remove, as much as possible, the burden of respiration from this incipient alimentary canal.

What can be said as to the shape of these ancient forms of fishes? Certain parts of them are absolutely known, other parts are guesswork. They are known to have possessed a dorsal shield, a ventral shield formerly looked upon as belonging to a separate species, called Scaphaspis, and a spine attached to the dorsal shield. The rest of their configuration, as given in Smith Woodward's restoration (Fig. 142) is guesswork; the fish-like body with its scales, the heterocercal tail, is based on the most insufficient evidence of something of the nature of scales having being found near the head-plates.

The dorsal shield is characterized by a pair of lateral eyes situated on the edge of the shield, not as in Cephalaspis near the middle line. In the middle line, where the rostrum meets the large dorsal plate, median eyes were situated. But the slightest sign of any median single nasal opening, such as is so characteristic of the head-shield of the Osteostraci and of Ammocœtes has never been discovered. The olfactory organ must have been situated on the ventral side as in the larval stage of Ammocœtes, or in the Palæostraca. Many of these head-shields are remarkably well preserved, and it is difficult to believe that an olfactory opening would not be seen if any such had existed, as it does in Thyestes.



The difficulty of interpreting these types is the difficulty of understanding their method of locomotion; that is largely the reason why the spine has been placed as if projecting from the back, and a fish-like body with a heterocercal tail-fin added. If, on the contrary, the spine is a terminal tail-spine, then, as far as the fossilized remains indicate, the animal consisted of a dorsal shield, a ventral shield, and a tail-spine, to which must be added two apparently lateral pieces and a few scales. If the animal did not possess a flexible body with a tail-fin, but terminated in a rigid spike after the fashion of a Limulus-like animal, then it must have moved by means of appendages. At present we have not sufficient evidence to decide this question.

That the animal crawled about in the mud by means of free appendages is by no means an impossible view, seeing how difficult it is to find the remains of appendages in the fossils of this far-back time, even when we are sure that they existed. Thus, for many generations, the appendages of trilobites, which occur in such countless numbers, and in such great variety of form, were absolutely unknown, until at last, in consequence of a fortunate infiltration by pyrites, they were found by Beecher preserved down to the minutest detail. Even to this day no trace of appendages has been found in such forms as Hemiaspis, Bunodes, Belinurus, Prestwichia.

The whole question of the evidence of any prosomatic appendages in these ancient fishes is one of very great interest, and of late years has been investigated by Patten. It has long been known that forms such as Pterichthys and Bothriolepis possessed two large, jointed locomotor appendages, and Patten has lately obtained better specimens of Bothriolepis than have ever been found before, which show not only the general configuration of the fish, but also the presence of mandibles or gnathites in the mouth-region resembling those of an arthropod. These mandibles had been seen before (Smith Woodward), but Patten's specimens are more perfect than any previously described, and cause him to conclude that these ancient fish were of the nature of arthropods rather than of vertebrates.

Patten has also been able to obtain some excellent specimens of the under surface of the head of Tremataspis, which, as evident in Fig. 143, show the presence of a series of holes, ranging on each side from the mouth-opening, in a semicircular fashion towards the middle line. He considers that these openings indicate the attachments of appendages, in opposition to other observers, such as Jaekel, who look upon them as gill-slits. To my mind, they are not in the right position for gill-slits; they are certainly in a prosomatic rather than in a mesosomatic position, and I should not be at all surprised if further research justified Patten's position. So convinced is he of the presence of appendages in all these old forms, that he considers them to be arthropods rather than vertebrates, although, at the same time, he looks upon them as indicating the origin of vertebrates from arthropods. Here, perhaps, it is advisable to say a few words on Patten's attitude towards this question.

Two years after I had put forward my theory of the derivation of vertebrates from arthropods, Patten published, in the Quarterly Journal of Microscopical Science, simultaneously with my paper in that journal, a paper entitled "The Origin of Vertebrates from Arachnids." In this paper he made no reference to my former publications, but he made it clear that there was an absolutely fundamental difference between our treatment of the problem; for he took the old view that of necessity there must be a reversal of surfaces in order that the internal organs should be in the same relative positions in the vertebrate and in the invertebrate. He simply, therefore, substituted Arachnid for Annelid in the old theory. Because of this necessity for the reversal of surfaces he discarded the terms dorsal and ventral as indicative of the surfaces of an animal, and substituted hæmal and neural, thereby hopelessly confusing the issue and making it often very difficult to understand his meaning.



He still holds to his original opinion, and I am still waiting to find out when the reversal of surfaces took place, for his investigations lead him, as must naturally be the case, to compare the dorsal (or, as he would call it, the hæmal) surface of Bothriolepis, of the Cephalaspidæ, and of the Pteraspidæ with the dorsal surface of the Palæostraca.

All these ancient fishes are, according to him, still in the arthropod stage, have not yet turned over, though in a peculiarly unscientific manner he argues elaborately that they must have swum on their back rather than on their front, and so indicated the coming reversal. Because they were arthropods they cannot have had a frontal nose-organ; therefore, Patten looks upon the nose and the two lateral eyes of the Osteostraci as a complex median eye, regardless of the fact that the median eyes already existed.

Every atom of evidence Patten has brought forward, every new fact he has discovered, confirms my position and makes his still more hopelessly confused. Keep the animal the right side uppermost, and the evidence of the rocks confirms the transition from the Palæostracan to the Cyclostome; reverse the surfaces, and the attempt to derive the vertebrate from the palæostracan becomes so confused and hopelessly muddled as to throw discredit on any theory of the origin of vertebrates from arthropods. For my own part, I fully expect that appendages will be found not only in the Cephalaspidæ but also in the Pteraspidæ, and I hope Patten will continue his researches with increasing success. I feel sure, however, his task will be much simplified if he abandons his present position and views the question from my standpoint.