The account given above chiefly differs from that of Remak by not supposing that the mesoblast-cells which form the heart are in any way split off from the wall of the alimentary canal.

There can be no doubt that His is wrong in supposing that the heart originates from the mesoblast of the splanchnopleure and somatopleure uniting to form its walls, thus leaving a cavity between them in the centre. The heart is undoubtedly formed out of the mesoblast of the splanchnopleure only.

Afanassiev's observations are nearer to the truth, but there are some points in which he has misinterpreted his sections.

Sections Pl. 2, figs. 10 and 11, explain what I have just said about the origin of the heart. Immediately around the notochord the mesoblast is not split, but a very little way outside it is seen to be split into two parts so and sp; the former of these follows the epiblast, and together with it forms the somatopleure, which has hardly begun to be folded at the line where the sections are taken. The latter (sp) forms with the hypoblast (hy) the splanchnopleure, and thus has become folded in to form the walls of the alimentary canal (d). In fig. 11 the folds have not united in the central line, but in fig. 10 they have so united. In fig. 11, where the mesoblast, still following the hypoblast, turns back to assume its normal direction, it is seen to be thickened and to have become split, so that a cavity (of) (of the omphalomeseraic vein) is formed in it on each side, lined by endothelium.

In the section immediately behind section fig. 11 the mesoblast was thickened, but had not become split.

In fig. 10 the hypoblast folds are seen to have united in the centre, so as to form a completely closed digestive canal (d); the folds of the mesoblast have also united, so that there is only a single cavity in the heart (hz), lined, as was the case with the omphalomeseraic veins, by endothelium.

EXPLANATION OF PLATE 2.

Fig. 1 is taken from the anterior part of the pellucid area of a thirty hours' chick, with four protovertebræ. At n is a nucleus with two nucleoli.

Figs. 2 and 3 are taken from the posterior end of the pellucid area of a chick with eight protovertebræ. In fig. 3 the nuclei are seen to have considerably increased in number at the points of starting of the protoplasmic processes. At n is seen a nucleus with two nucleoli.

Fig. 4 is taken from the anterior part of the pellucid area of an embryo of thirty-six hours. It shews the narrow processes characteristic of the anterior part of the pellucid area, and the fewer nuclei. Small spaces, which have the appearance of vacuoles, are shewn at v.

Fig. 5 is taken from the posterior part of the pellucid area of a thirty-six hours' embryo. It shews the nuclei, with somewhat irregular nucleoli, which have begun to acquire the red colour of blood-corpuscles; the protoplasmic processes containing the nuclei; the nuclei in the protoplasm surrounding the corpuscles, as shewn at a, a´.

Fig. 6 shews fully formed blood-vessels, in part filled with blood-corpuscles and in part empty. The walls of the capillaries, formed of cells, spindle-shaped in section, are shewn, and also the secondary investment of Klein at k, and at b is seen a narrow protoplasmic process filled with blood-corpuscles.

Fig. 7 is taken from the anterior part of the pellucid area of a thirty-six hours' embryo. It shews a collection of nuclei which are beginning to become blood-corpuscles.

Figs. 1-5 are drawn with an 1/8 object-glass. Fig. 6 is on a much smaller scale. Fig. 7 is intermediate.

Fig. 8.—A transverse section through the dorsal region of a forty-five hours' embryo; ao, aorta with a few blood-corpuscles. v, Blood-vessels, all of them being formed in the splanchnopleure, and all of them provided with the secondary investment of Klein; pe, pellucid area; op, opaque area.

Fig. 9.—Small portion of a section through the opaque area of a thirty-five hours' embryo, showing protoplasmic processes, with nuclei passing from the somatopleure to the splanchnopleure.

Fig. 10.—Section through the heart of a thirty-four hours' embryo. a. Alimentary canal; hb, hind brain; nc, notochord; e, epiblast; so, mesoblast of the somatopleure; sp, mesoblast of the splanchnopleure; hy, hypoblast; hz, cavity of the heart.

These two sections shew that the heart is entirely formed from the mesoblast of the splanchnopleure, and that it is formed by the splitting of that part of the mesoblast which has turned to assume its normal direction after being folded in to form the muscular wall of the alimentary canal. In fig. 11 the cavities so formed on each side have not yet united, but in fig. 10 they have united. When the folding becomes more complete the cavities (of, of) in fig. 11 will unite, and in this way the origin of the omphalomeseraic veins will be carried further backwards. In the section immediately behind section 11 the mesoblast had become thickened, but had not split.

Without further preface I will pass on to my investigations.

The Egg-shell.

In the eggs of all the species of Dog-fishes which I have examined the yolk lies nearest that end of the quadrilateral shell which has the shortest pair of strings for attachment. This is probably due to the shape of the cavity of the shell, and is certainly not due to the presence of any structures similar to chalazæ.

The Yolk.

The yolk is not enclosed in any membrane comparable to the vitelline membrane of Birds, but lies freely in a viscid albumen which fills up the egg-capsule. It possesses considerable consistency, so that it can be removed into a basin, in spite of the absence of a vitelline membrane, without falling to pieces. This consistency is not merely a property of the yolk-sphere as a whole, but is shared by every individual part of it.

With the exception of some finely granular matter around the blastoderm, the yolk consists of rather small, elliptical, highly refracting bodies, whose shape is very characteristic and renders them easily recognizable. A number of striæ like those of muscle are generally visible on most of the spherules, which give them the appearance of being in the act of breaking up into a series of discs; but whether these striæ are normal, or produced by the action of water I have not determined.

Position of the Blastoderm.

The blastoderm is always situated, immediately after impregnation, near the pole of the yolk which lies close to the end of the egg-capsule. Its position varies a little in the different species and is not quite constant in different eggs of the same species. But this general situation is quite invariable. It is of about the same proportional size as the blastoderm of a bird.

Segmentation.

These appearances are proved by transverse sections to be deceptive. There is no sharp line either at the sides or below separating the blastoderm from the yolk. In the passage between the fine granular matter of the germ to the coarser yolk-spheres every intermediate size of granule is present; and, though the space between the two is rather narrow, in no sense of the word can there be said to be any break or line between them.

This gradual passage stands in marked contrast with what we shall find to be the case at the close of the segmentation. In the youngest egg which I had, the germinal disc was already divided into four segments by two furrows at right angles. These furrows, however, did not reach its edge; and from my sections I have found that they were not cut off below by any horizontal furrow. So that the four segments were continuous below with the remainder of the germ without a break.

In the next youngest specimen which I had, there were already present eighteen segments, somewhat irregular in size, but which might roughly be divided into an outer ring of larger spheres, separated, as it were, by a circular furrow from an inner series of smaller segments. The furrows in this case reached quite to the edge of the germinal disc.

In each segment a nucleus was generally to be seen in sections. I will, however, reserve my remarks upon the nature of the nuclei till I discuss the nuclei of the blastoderm as a whole.

For some little time the peripheral segments continue larger than the more central ones, but this difference of size becomes less and less marked, and before the segments have become too small to be seen with the simple microscope, their size appears to be uniform over the whole surface of the blastoderm.

In the blastoderms somewhat older than the one last described the segments have already become completely separate masses, and each of them already possesses a distinct nucleus. They form a layer one or two segments deep. The limits of the blastoderm are not, however, defined by the already completed segments, but outside these new segments continue to be formed around nuclei which appear in the yolk. At this stage there is, therefore, no line of demarcation between the germ and the yolk, but the yolk is being bored into, so to speak, by a continuous process of fresh segmentation.

The further segmentation of the already existing spheres, and the formation of new ones from the yolk below and to the sides, continues till the central cells acquire their final size, the peripheral ones being still large, and undefined towards the yolk. These also soon reach the final size, and the blastoderm then becomes rounded off towards the yolk and sharply separated from it.

The Nuclei of the Yolk.

Intimately connected with the segmentation is the appearance and history of a number of nuclei which arise in the yolk surrounding the blastoderm.

When the horizontal furrows appear which first separate the blastoderm from the yolk, the separation does not occur along the line of passage from the fine to the coarse yolk, but in the former at some distance from this line.

They vary immensely in size—from that of an ordinary nucleus to a size greater than the largest blastoderm-cell.

In Pl. 3, fig. 1, n, is shewn their distribution in this finely granular matter and their variation in size. But whatever may be their size, they always possess the same characteristic structure. This is shewn in Pl. 3, figs. 1 and 2, n.

They are rather irregular in shape, with a tendency when small to be roundish, and are divided by a number of lines into distinct areas, in each of which a nucleolus is to be seen. The lines dividing them into these areas have a tendency (in the smaller specimens) to radiate from the centre, as shewn in Pl. 3, fig. 1.

These nuclei colour red with hematoxylin and carmine and brown with osmic acid, while the nucleoli or granules contained in the areas also colour very intensely with all the three above-named reagents.

With such a peculiar structure, in favourable specimens these nuclei are very easily recognised, and their distribution can be determined without difficulty. They are not present alone in the finely granular yolk, but also in the coarsely granular yolk adjoining it. They form very often a special row, sometimes still more markedly than in Pl. 3, fig. 1, along the floor of the segmentation cavity. They are not, however, found alone in the yolk. All the blastoderm-cells in the earlier stages possess precisely similar nuclei! From the appearance of the first nucleus in a segmentation-sphere till a comparatively late period in development, every nucleus which can be distinctly seen is found to be of this character. In Pl. 3, fig. 2, this is very distinctly shewn.

They appear in small numbers around the blastoderm at the close of segmentation, and round each one of them there may at this time be seen in osmic acid specimens, and with high powers, a fine network similar to but finer than that represented in Pl. 3, fig. 2. This network cannot, as a general rule, be traced far into the yolk, but in some exceptionally thin specimens it may be seen in any part of the fine granular yolk around the blastoderm, the meshes of the network being, however, considerably coarser between than around the nuclei. This network may be seen in the fine granular material around the germ till the latest period of which I have yet cut sections of the blastoderm. In the later specimens, indeed, it is very much more distinctly seen than in the earlier, owing to the fact that in parts of the blastoderm, especially under the embryo, the yolk-granules have disappeared partly or entirely, leaving only this fine network with the nuclei in it.

A specimen of this kind is represented in Pl. 3, fig. 2, where the meshes of the network are seen to be finer immediately around the nuclei, and coarser in the intervals. The specimen further shows in the clearest manner that this network is not divided into areas, each representing a cell and each containing a nucleus. I do not know to what extent this network extends into the yolk. I have never yet seen the limits of it, though it is very common to see the coarsest yolk-granules lying in its meshes. Some of these are shewn in Pl. 3, fig. 2, yk.

Admitting, as I think it is necessary to do, the organized condition of the whole yolk-sphere, there are two possible views as to its nature. We may either take the view that it is one gigantic cell, the ovum, which has grown at the expense of the other cells of the egg-follicle, and that these cells in becoming absorbed have completely lost their individuality; or we may look upon the true formative yolk (as far as we can separate it from the remainder of the food-yolk) as the remains of one cell (the primitive ovum), and the remainder of the yolk as a body formed from the coalescence of the other cells of the egg-follicle, which is adherent to, but has not coalesced with, the primitive ovum, the cells in this case not having completely lost their individuality; and to these cells, the nuclei, I have found, must be supposed to belong.

The former view I think, for many reasons, the most probable. The share of these nuclei in the segmentation, and the presence of similar nuclei in the cells of the germ, both support it, and are at the same time difficulties in the way of the other view. Leaving this question which cannot be discussed fully in a preliminary paper like the present one, I will pass on to another important question, viz.:

I will conclude my account of these nuclei by briefly summarizing the points I have arrived at in reference to them.

A portion, or more probably the whole, of the yolk of the Dog-fish consists of organized material, in which nuclei appear and increase either by division or by a process of independent formation, and a great number of these subsequently become the nuclei of cells formed around them, frequently at a distance from the germ, which then travel up and enter it.

Leaving these nuclei, I will now pass on to the formation of the layers.

At the close of segmentation the surface of the blastoderm is composed of cells of a uniform size, which, however, are too small to be seen by the aid of the simple microscope.

The cells of this uppermost layer are somewhat columnar, and can be distinguished from the remainder of the cells of the blastoderm as a separate layer. This layer forms the epiblast; and the Dog-fish agree with Birds, Batrachians, and Osseous fish in the very early differentiation of it.

The remainder of the cells of the blastoderm form a mass, many cells deep, in which it is impossible as yet or till a very considerably later period to distinguish two layers. They may be called the lower layer cells. Some of them near the edge of this mass are still considerably larger than the rest, but they are, as a whole, of a fairly uniform size. Their nuclei are of the same character as the nuclei in the yolk.

There is one point to be noticed in the shape of the blastoderm as a whole. It is unsymmetrical, and a much larger number of its cells are found collected at one end than at the other. This absence of symmetry is found in all sections which are cut parallel to the long axis of the egg-capsule. The thicker end is the region where the embryo will subsequently appear.

The embryonic end of the germ is always the one which points towards the pole of the yolk farthest removed from the egg-capsule.

The germ grows, but not very rapidly, and without otherwise undergoing any very appreciable change, for some time.

The growth at these early periods appears to be particularly slow, especially when compared with the rapid manner in which some of the later stages of the development are passed through.

The next important change which occurs is the formation of the so-called “segmentation cavity.”

This forms a very marked feature throughout the early stages. It appears, however, to have somewhat different relations to the blastoderm than the homologous structure in other vertebrates. In its earliest stage which I have observed, it appears as a small cavity in the centre of the lower layer cells. This grows rapidly, and its roof becomes composed of epiblast and only a thin lining of “lower layer” cells, while its floor is formed by the yolk (Pl. 3, fig. 3, sg). In the next and third stage (Pl. 3, fig. 4, sg) its floor is formed by a thin layer of cells, its roof remaining as before. It has, however, become a less conspicuous formation than it was; and in the last (fourth) stage in which it can be distinguished it is very inconspicuous, and almost filled up by cells.

What I have called the second stage corresponds to a period in which no trace of the embryo is to be seen. In the third stage the embryonic end of the blastoderm projects outwards to form a structure which I shall speak of as the “embryonic rim,” and in the fourth and last stage a distinct medullary groove is formed. For a considerable period during the second stage the segmentation cavity remains of about the same size; during the third stage it begins to be encroached upon, and becomes smaller both absolutely, and relatively to the increased size of the germ.

Dog-fish resemble Osseous fish in the fact that their embryos are entirely formed from a portion of the germ which does not form part of the roof of the segmentation cavity, so that the cells forming the roof of the segmentation cavity take no share at any time in the formation of their embryos. They further agree with Osseous fish (always supposing that the descriptions of Oellacher, loc. cit., and Götte, Archiv. für Micr. Anat. Bd. IX. are correct) in the floor of the segmentation cavity being formed at one period by yolk. Together with these points of similarity there are some important differences.

(1) The segmentation cavity in the Osseous fish from the first arises as a cavity between the yolk and the blastoderm, and its floor is never at any period covered with cells. In the Dog-fish, as we have said above, both in the earlier and later periods the floor is covered with cells.

(2) The roof in the Dog-fish is invariably formed by the epiblast and a row of flattened lower layer cells.

According to both Götte and Oellacher the roof of the segmentation cavity in Osseous fishes is in the earlier stages formed alone of the two layers which correspond with the single layer forming the epiblast in the Dog-fish. In Osseous fishes it is very difficult to distinguish the various layers, owing to the similarity of their component cells. In Dog-fish this is very easy, owing to the great distinctness of the epiblast, and it appears to me, on this account, very probable that the two above-named observers may be in error as to the constitution of its roof in the Osseous fish. With both the Bird and the Frog the segmentation cavity of the Dog-fish has some points of agreement, and some points of difference, but it would take me too far from my present subject to discuss them.

This thickened edge of the blastoderm is still more conspicuous in the next and second stage (Pl. 3, fig. 3).

In the second stage the chief points of progress, in addition to the increased thickness of the edge of the blastoderm, are—

(1) The increased thickness and distinctness of the epiblast, caused by its cells becoming more columnar, though it remains as a one-cell-thick layer.

(2) The disappearance of the cells from the floor of the segmentation cavity.

The lower layer cells have undergone no important changes, and the blastoderm has increased very little if at all in size.

From Pl. 3, fig. 3, it is seen that there is a far larger collection of cells at the embryonic than at the opposite end.

Passing over some rather unimportant stages, I will come to the next important one.

The general features of this (the third) stage in a surface view are—

(1) The increase in size of the blastoderm.

(2) The diminution in size of the segmentation cavity, both relatively and absolutely.

(3) The appearance of a portion of the blastoderm projecting beyond the rest over the yolk. This projecting rim extends for nearly half the circumference of the yolk, but is most marked at the point where the embryo will shortly appear. I will call it the “embryonic rim.”

These points are still better seen from sections than from surface views, and will be gathered at once from an inspection of Pl. 3, fig. 4.

The only other important feature of this stage is the appearance of a layer of cells on the floor of the segmentation cavity.

Does this layer come from an ingrowth from the thickened edge of the blastoderm, or does it arise from the formation of new cells in the yolk?

It is almost impossible to answer this question with certainty. The following facts, however, make me believe that the newly formed cells do play an important part in the formation of this layer.

(1) The presence at an earlier date of almost a row of nuclei under the floor of the segmentation cavity (Pl. 3, fig. 1).

(2) The presence on the floor of the cavity of such large cells as those represented in fig. 1, bd, cells which are very different, as far as the size and granules are concerned, from the remainder of the cells of the blastoderm.

On the other hand, from this as well as other sections, I have satisfied myself that there is a distinct ingrowth of cells from the embryonic swelling. It is therefore most probable that both these processes, viz. a fresh formation and an ingrowth, have a share in the formation of the layer of cells on the floor of the segmentation cavity.

In the next stage we find the embryo rising up as a distinct body from the blastoderm, and I shall in future speak of the body, which now becomes distinct as the embryo. It corresponds with what Kupffer (loc. cit.) in his paper on the “Osseous Fishes” has called the “embryonic keel.”This starting-point for speaking of the embryo as a distinct body is purely arbitrary and one merely of convenience. If I wished to fix more correctly upon a period which could be spoken of as marking the commencing formation of the embryo, I should select the time when structures first appear to mark out the portion of the germ from which the embryo becomes formed; this period would be in the Elasmobranchii, as in the Osseous fish, at the termination of segmentation, when the want of symmetry between the embryonic end of the germ and the opposite end first appears.

Almost from its first appearance this ridge acquires a shallow groove—the medullary groove (Pl. 3, fig. 5, mg)—along its middle line, where the epiblast and hypoblast are in absolute contact (vide fig. 6a, 7a, 7b, &c.) and where the mesoblast (which is already formed by this stage) is totally absent. This groove ends abruptly a little before the front end of the embryo, and is deepest in the middle and wide and shallow behind.

On each side of it is a plate of mesoblast equivalent to the combined vertebral and lateral plates of the Chick. These, though they cannot be considered as entirely the cause of the medullary groove, may perhaps help to make it deeper. In the parts of the germ outside the embryo the mesoblast is again totally absent, or, more correctly, we might say that outside the embryo the lower layer cells do not become differentiated into hypoblast and mesoblast, and remain continuous only with the lower of the two layers into which the lower layer cells become differentiated in the body of embryo. This state of things is not really very different from what we find in the Chick. Here outside the embryo (i.e. in the opaque area) there is a layer of cells in which no differentiation into hypoblast and mesoblast takes place, but the layer remains continuous rather with the hypoblast than the mesoblast.

From longitudinal sections I have found that the segmentation cavity has ceased by this stage to have any distinct existence, but that the whole space between the epiblast and the yolk is filled up with a mass of elongated cells, which probably are solely concerned in the formation of the vascular system. The thickened posterior edge of the blastoderm is still visible.

At the embryonic end of the blastoderm, as I pointed out in an earlier stage, the epiblast and the lower layer cells are perfectly continuous.

Where they join the epiblast, the lower layer cells become distinctly divided, and this division commenced even in the earlier stage, into two layers; a lower one, more directly continuous with the epiblast, consisting of cells somewhat resembling the epiblast-cells, and an upper one of more flattened cells (Pl. 3, fig. 4, m). The first of these forms the hypoblast, and the latter the mesoblast. They are indicated by hy and m in the figures. The hypoblast, as I said before, remains continuous with the whole of the rest of lower layer cells of the blastoderm (vide fig. 7b). This division into hypoblast and mesoblast commences at the earlier stage, but becomes much more marked during this one.

In this case we would be compelled to suppose that the mass of lower layer cells which forms the embryonic swelling is used as food for the growth of the involuted epiblast, or else employed solely in the growth over the yolk of the non-embryonic portion of the blastoderm; but the latter possibility does not seem compatible with my sections.

I do not believe that it is possible, from the examination of sections alone, to decide which of these two views (viz. whether the epiblast is involuted, or whether it becomes merely continuous with the lower layer cells) is the true one. The question must be decided from other considerations.

The following ones have induced me to take the view that there is no involution, but that the mesoblast and hypoblast are formed from the lower layer cells.

(1) That it would be rather surprising to find the mass of lower layer cells which forms the “embryo swelling” playing no part in the formation of embryo.

(2) That the view that it is the lower layer cells from which the hypoblast and mesoblast are derived agrees with the mode of formation of these two layers in the Bird, and also in the Frog; since although, in the latter animal, there is an involution, this is not of the epiblast, but of the larger cells of the lower pole of the yolk, which in part correspond with what I have called the lower layer cells in the Dog-fish.

If the view be accepted that it is from the lower layer cells that the hypoblast and mesoblast are formed, it becomes necessary to explain what the continuity of the hypoblast with the epiblast means.

The explanation of this is, I believe, the keystone to the whole position. The vertebrates may be divided as to their early development into two classes, viz. those with holoblastic ova, in which the digestive canal is formed by an involution with the presence of an “anus of Rusconi.”

This class includes “Amphioxus,” the “Lamprey,” the “Sturgeon,” and “Batrachians.”

This class includes the “Elasmobranchii,” “Osseous fish,” “Reptiles,” and “Aves.”

The mode of formation of the alimentary canal in the first class is clearly the more primitive; and it is equally clear that its mode of formation in the second class is an adaptation due to the presence of the large quantity of food-yolk.

In the Dog-fish I believe that we can see, to a certain extent, how the change from the one to the other of these modes of development of the alimentary canal took place.

I believe that in the Dog-fish we have before our eyes one of the steps by which a direct mode of formation comes to be substituted for an indirect one by involution. We find, in fact, in the Dog-fish, that the cells from which are derived the mesoblast and hypoblast come to occupy their final position in the primitive arrangement of the cells during segmentation, and not by a subsequent and secondary involution.

This change in the mode of formation of the alimentary canal is clearly a result of change of mechanical conditions from the presence of the large food-yolk.

Excellent parallels to it will be found amongst the Mollusca. In this class the presence or absence of food-yolk produces not very dissimilar changes to those which are produced amongst vertebrates from the same cause.

The continuity of the hypoblast and epiblast at the embryonic rim is a remnant which, having no meaning or function, except in reference to the earlier mode of development, is likely to become lost, and in Birds no trace of it is any longer to be found.

In reference to the former point, however, I may mention that the Batrachians and Lampreys are to a certain extent intermediate in condition between the Amphioxus and the Dog-fishes, since in them the yolk becomes divided during segmentation into lower layer cells and epiblast, but a modified involution is still retained, while the Dog-fish may be looked upon as intermediate between Birds and Batrachians, the continuity at the hind end between the epiblast and hypoblast being retained by them, though not the involution.

It may be convenient here to call attention to some of the similarities and some of the differences which I have not yet spoken of between the development of Osseous fish and the Dog-fish in the early stages. The points of similarity are—(1) The swollen edge of the blastoderm. (2) The embryo-swelling. (3) The embryo-keel. (4) The spreading of the blastoderm over the yolk-sac from a point corresponding with the position of the embryo, and not with the centre of the germ. The growth is almost nothing at that point, and most rapid at the opposite pole of the blastoderm, being less and less rapid along points of the circumference in proportion to their proximity to the embryonic swelling. (5) The medullary groove.

Development of the Embryo.

After the embryo has become definitely established, for some time it grows rapidly in length, without externally undergoing other important changes, with the exception of the appearance of two swellings, one on each side of its tail.

These swellings, which I will call the Caudal lobes (figs. 8 and 9, ts), are also found in Osseous fishes, and have been called by Oellacher the Embryonal saum. They are caused by a thickening of mesoblast on each side of the hind end of the embryo, at the edge of the embryonic rim, and form a very conspicuous feature throughout the early stages of the development of the Dog-fish, and are still more marked in the Torpedo (Pl. 3, fig. 9). Although from the surface the other changes which are visible are very insignificant, sections shew that the notochord is commencing to be formed.

I mentioned that the mesoblast appeared to be primitively formed as two independent sheets, split off, so to speak, from the hypoblast, one on each side of the middle line of the embryo. If we looked upon the notochord as a third median sheet of mesoblast, split off from the hypoblast somewhat later than the other two, we should avoid having to admit its hypoblastic origin.

Professor Huxley, to whom I have shewn my specimens, strongly advocates this view.

(1) That this is the undoubtedly natural interpretation of the sections. (2) That the notochord becomes separated from the hypoblast after the latter has acquired its typical structure, and differs in that respect from the two lateral sheets of mesoblast, which are formed coincidently with the hypoblast by a homogeneous mass of cells becoming differentiated into two distinct layers. (3) That the first mode of looking at the matter really proves too much, since it is clear that by the same method of reasoning we could prove the mesoblastic origin of any organ derived from the hypoblast and budded off into the mesoblast. We would merely have to assert that it was really a mass of mesoblast budded off from the hypoblast rather later than the remainder of the mesoblast. Still, it must be admitted that the first view I have suggested is a possible, not to say a probable one, though the mode of arguing by which it can be upheld may be rather dangerous if generally applied. We ought not, however, for that reason necessarily to reject it in the present case. As Mr Ray Lankester pointed out to me, if we accept the hypoblastic origin of the notochord, we should find a partial parallel to it in the endostyle of Tunicates, and it is perhaps interesting to note in reference to it that the notochord is the only unsegmented portion of the axial skeleton.

Whether the strong à priori difficulties of the hypoblastic origin of the notochord are sufficient to counterbalance the natural interpretation of my sections, cannot, I think, be decided from the single case of the Dog-fish. It is to be hoped that more complete investigations of the Lamprey, &c., may throw further light upon the question.

Whichever view of the primitive origin of the notochord is the true one, its apparent origin is very instructive as illustrating the possible way in which an organ might come to change the layer to which it primarily belonged.

The view seemingly held by many embryologists of the present day, that an organ, when it was primitively derived from one layer, can never be apparently formed in another layer, appears to me both unreasonable on à priori grounds, and also unsupported by facts.

I see no reason for doubting that the embryo in the earliest periods of development is as subject to the laws of natural selection as is the animal at any other period. Indeed, there appear to me grounds for the thinking that it is more so. The remarkable differences in allied species as to the amount of food-yolk, which always entail corresponding alterations in the development—the different modes of segmentation in allied species, such as are found in the Amphipoda and Isopoda—the suppression of many stages in freshwater species, which are retained in the allied marine species—are all instances of modifications due to natural selection affecting the earliest stages of development. If such points as these can be affected by natural selection I see no reason why the arrangement of individual cells (or rather primitive elements) should not also be modified; why, in fact, a mass of cells which was originally derived from one layer, but in the course of development became budded off from that layer and entered another layer, should not by a series of small steps cease ever to be attached to the original layer, but from the first moment it can be distinguished should be found as a separate mass in the second layer.

The change of layers will, of course, only take place where some economy is effected by it. The variations in the mode of development of the nervous system may probably be explained in this way.

If we admit that organs can undergo changes, as to the primitive layer from which they are derived, important consequences must follow.

Thus, the muscles, internal skeleton, and connective tissue are always placed in the adult between the skin (epidermis) and the epithelium of the alimentary canal.

We should therefore expect to find them, and, as a matter of fact, we always do find them, developed from a middle layer when this is present.

The upper layer must always and does always form the epidermis, and similarly the lower layer or hypoblast must form a part of the epithelium of the alimentary canal. A full discussion of this question would, however, lead me too far away from my present subject.

By the completion of its lower wall in the way described, the throat early becomes a closed tube, its closing taking place before any other important changes are visible in the embryo from surface views.

A considerable increase in length is attained before other changes than an increase in depth of the medullary groove and a more complete folding off of the embryo from the blastoderm take place. The first important change is the formation of the protovertebræ.

These are formed by the lateral plates of mesoblast, which I said were equivalent at once to the vertebral and lateral plates in the Bird, becoming split by transverse divisions into cubical masses.

At the time when this occurs, and, indeed, up till a considerably later period, the mesoblast is not split into somatopleure and splanchnopleure, and it is not divided into vertebral and lateral plates. The transverse lines of division of the protovertebræ do not, however, extend to the outer edge of the undivided lateral plates.

The differences between this mode of formation of the protovertebræ and that occurring in Birds are too obvious to require pointing out. I will speak of them more fully when I have given the whole history of the protovertebræ of the Dog-fish.

I will only now say that I have had in the early stages to investigate the formation of the protovertebræ entirely by means of sections, the objects being too opaque to be otherwise studied.

The next change of any importance is the commencement of the formation of the head. The region of the head first becomes distinguishable by the flattening out of the germ at its front end.

The flattened-out portion of the germ grows rapidly, and forms a spatula-like termination to the embryo (Pl. 3, fig. 8).

In the region of the head the medullary groove is at first totally absent (vide section, Pl. 3, fig. 8a).

I am at present quite unable even to form a guess what this peculiar feature of the brain means. It, no doubt, has some meaning in reference to the vertebrate ancestry if we could only discover it. The peculiar spatula-like flattened condition of the head is also (vide loc. ant. cit.) apparently found in the Sturgeons; it must therefore almost undoubtedly be looked upon as not merely an accidental peculiarity.

While these changes have been taking place in the head not less important changes have occurred in the remainder of the body. In the first place the two caudal lobes have increased in size, and have become, as it were, pushed in together, leaving a groove between them (fig. 8, ts). They are very conspicuous objects, and, together with the spatula-like head, give the whole embryo an almost comical appearance. The medullary canal has by this time become completely closed in the region of the tail (figs. 8 and 8b).

It is still widely open in the region of the back, and, though more nearly closed again in the neck, is, as I have said, flattened out to nothing in the head.