Origin of Vertebrates/Chapter II

by Walter Holbrook Gaskell

CHAPTER II

THE EVIDENCE OF THE ORGANS OF VISION

Different kinds of eye.—Simple and compound retinas.—Upright and inverted retinas.—Median eyes.—Median or pineal eyes of Ammocœtes and their optic ganglia.—Comparison with other median eyes.—Lateral eyes of vertebrates compared with lateral eyes of crustaceans.—Peculiarities of the lateral eye of the lamprey.—Meaning of the optic diverticula.—Evolution of vertebrate eyes.—Summary.

The Different Kinds of Eye.

In all animals the eyes are composed of two parts. 1. A set of special sensory cells called the retina. 2. A dioptric apparatus for the purpose of forming an image on the sensory cells. The simplest eye is formed from a modified patch of the surface-epithelium; certain of the hypodermal cells, as they are called, elongate, and their cuticular surface becomes bulged to form a simple lens. These elongated cells form the retinal cells, and are connected with the central nervous system by nerve-fibres which constitute an optic nerve; the cells themselves may contain pigment.

The more complicated eyes are modifications of this type for the purpose of making both the retina and the dioptric apparatus more perfect. According to a very prevalent view, these modifications have been brought about by invaginations of the surface-epithelium. Thus if ABCD (Fig. 28) represents a portion of the surface-epithelium, the chitinous cuticle being represented by the dark line, with the hypodermal cells beneath, and if the part C is modified to form an optic sense-plate, then an invagination occurring between A and B will throw the retinal sense-cells with the optic nerve further from the surface, and the layers B and A between the retina and the source of light will be available for the formation of the dioptric apparatus. The lens is now formed from the cuticular surface of A, and the hypodermal cells of A elongate to form the layer known by the name of corneagen, or vitreogen, the cells of B remaining small and forming the pre-retinal layer of cells. The large optic nerve end-cells of the retinal layer, C, take up the position shown in the figure, and their cuticular surface becomes modified to form rods of varying shape called rhabdites, which are attached to the retinal cells. Frequently the rhabdites of neighbouring cells form definite groups, each group being called a rhabdome. Whatever shape they take it is invariably found that these little rods (bacilli), or rhabdites, are modifications of the cuticular surface of the cells which form the retinal layer. Also, as must necessarily be the case from the method of formation, the optic nerve arises from the nuclear end of the retinal cells, never from the bacillary end. As in the case first mentioned, so in this case, the light strikes direct upon the bacillary end of the retinal cells; such eyes, therefore, are eyes with an upright retina.

Fig. 28.—Diagram of Formation of an Upright Simple Retina.

It may happen that the part invaginated is the optic sense-plate itself, as would be the case if in the former figure, instead of C, the part B was modified to form a sense-plate. This will give rise to an eye of a character different from the former (Fig. 29). The optic nerve-fibres now lie between the source of light and the retinal end-cells, the layer A as before forms the cuticular lens, and its hypodermal cells elongate to form the corneagen; there is no pre-retinal layer, but, on the contrary, a post-retinal layer, C, called the tapetum, and, as is seen, the light passes through the retinal layer to the tapetum. The cuticular surface of the retinal cells forming the rods or bacilli is directed towards the tapetal layer away from the source of light, and the nuclei of the retinal cells are pre-bacillary in position, in contradistinction to the upright eye, where they are post-bacillary. The retinal end-cells are devoid of pigment, the pigment being in the tapetal layer.

Such an eye, in contradistinction to the former type, is an eye with an inverted retina; but still the same law holds as in the former case—the optic nerve-fibres enter at the nuclear ends of the cells, and the rods are formed from the cuticular surface.

In these eyes the pigmented tapetal layer is believed to act as a looking-glass; the dioptric apparatus throws the image on to its shiny surface, from whence it is reflected directly on to the rods, which are in close contact with the tapetum. A similar process has been suggested in the case of the mammalian lateral eye, with its inverted retina. Johnson describes the post-retinal pigmented layer as being frequently coloured and shiny, and imagines that it reflects the image directly back on to the rods.

Fig. 29.—Diagram of Formation of an Inverted Simple Retina.

The arrow shows the direction of the source of light in this as in the preceding figure. In both figures the cuticular rhabdites are represented by thick black lines.

Thus we see that eyes can be placed in different categories, e.g. those with an upright retina and those with an inverted retina; also, according to the presence or absence of a tapetum, eyes have been grouped as tapetal or non-tapetal. All the eyes considered so far are called simple eyes, or ocelli; and a number of ocelli may be contiguous though separate, as in the lateral eyes of the scorpion. They may, however, come into close contact and form one single, large, compound eye. Such ocelli, in a very large number of cases, retain each its own dioptric apparatus, and therefore the external appearance of the compound eye represents not a single lens, but a large number of facets, as is seen in the eyes of insects. Owing to these differences, eyes have been divided into simple and compound, and into facetted and non-facetted.

Yet another complication occurs in the formation of eyes, which is, perhaps, the most important of all: the retinal portion of the eye, instead of consisting of simple retinal cells, with their accompanying rhabdites, may include within itself a portion of the central nervous system.

The rationale of such a formation is as follows: The external covering of the body is formed by a layer of external epithelial cells—the ectodermal cell-layer—and an underlying neural layer, of which the latter gives origin to the central nervous system. As development proceeds, this central nervous system sinks inwards, leaving as its connection with the ectoderm the sensory nerves of the skin. That part of the neural layer which underlies the optic plate forms the optic ganglion, and when the central nervous system leaves the surface to take up its deeper position, the strand of nerve-fibres known as the optic nerve, is left connecting it with the retinal cells as seen in Figs. 28, 29. It may, however, happen that part of the optic ganglion remains at the surface, in close connection with the retinal end-cells, when the rest of the central nervous system sinks inwards. The retina of such an eye is composed of the combined optic ganglion and retinal end-cells; the strand of nerve-fibres which is left as the connection between it and the rest of the brain, which is also called the optic nerve, is not a true peripheral nerve, as in the first case, but rather a tract of fibres connecting two parts of the brain, of which one has remained at the periphery. Such a retina, in contradistinction to the first kind, may be called a compound retina.

The optic ganglion, as seen in eyes with a simple retina, consists of a cortical layer of small, round nerve-cells, and an internal medulla of fine nerve-fibres, which form a thick network known as 'Punctsubstanz,' or in modern terminology, 'Neuropil.' Fibres which pass into this 'neuropil' from other parts of the brain connect them with the optic ganglion.

At the present time, owing to the researches of Golgi, Ramón y Cajal, and others, the nervous system is considered to be composed of a number of separate nerve-units, called neurones, each neurone consisting of a nerve-cell with its various processes; one of these—the neuraxon—constitutes the nerve-fibre belonging to that nerve-cell, the other processes—the dendrites—establish communication with other neurones. The place where these processes come together is called a synapse, and the tangle of fine fibres formed at a number of synapses forms the 'neuropil.'

Fig. 30.—Diagram of Formation of an Upright Compound Retina.

ABCD, as in Fig. 28. Op. g. I. and Op. g. II., two optic ganglia which combine to form the retinal ganglion, Rt. g.

When, therefore, a compound retina is formed by the amalgamation of the ectodermal part—the retinal cells proper—with the neurodermic part—to which the name 'retinal ganglion' may be given,—such a retina consists of neuropil substance and nerve-cells, as well as the retinal end-cells. In all such compound retinas, the retinal ganglion is not single, but two optic ganglia at least are included in it, so that there are two sets of nerve-cells and two synapses are always formed; one between the retinal end-cells and the neurones of the first optic ganglion, which may be called the ganglion of the retina, the other between the first and second ganglia, which, seeing that the neuraxons of its cells form the optic nerve, may be called the ganglion of the optic nerve. The 'neuropil' formed by these synapses forms the molecular layers of the compound retina, and the cells themselves form the nuclear layers. Thus an upright compound retina, formed in the same way as the upright simple retina, would be illustrated by Fig. 30.

Further, in precisely the same way as in the case of the simple retina, such a compound retina may be upright or inverted. Thus, in the lateral eyes of crustaceans and insects, a compound retina of this kind is formed, which is upright; while in the vertebrates the compound retina of the lateral eyes is inverted.

The compound retina of vertebrates is usually described as composed of a series of layers, which may be analyzed into their several components as follows:—

Layer of rods and cones External nuclear layer External molecular layer Internal nuclear layer Internal molecular layer Optic nerve-cell layer Layer of optic nerve fibres	$\scriptstyle {\left.{\begin{matrix}\ \\\ \end{matrix}}\right\}\,}$	retina proper	$\scriptstyle {\left.{\begin{matrix}\ \end{matrix}}\right\}\,}$	Ectodermic part
	$\scriptstyle {\left.{\begin{matrix}\ \\\ \end{matrix}}\right\}\,}$	ganglion of retina	$\scriptstyle {\left.{\begin{matrix}\ \\\\\ \ \end{matrix}}\right\}\,}$	retinal ganglion	$\scriptstyle {\left.{\begin{matrix}\ \\\ \end{matrix}}\right\}\,}$	neurodermic part
	$\scriptstyle {\left.{\begin{matrix}\ \\\ \end{matrix}}\right\}\,}$	ganglion of optic nerve

The difference between the development of these two types of eye—those with a simple retina and those with a compound retina—has led, in the most natural manner, to the conception that the retina is developed, in the higher animals, sometimes from the cells of the peripheral epidermis, sometimes from the tissue of the brain—two modes of development termed by Balfour 'peripheral' and 'cerebral.' An historical survey of the question shows most conclusively that all investigators are agreed in ascribing the origin of the simple retina to the peripheral method of development, the retina being formed from the hypodermal cells by a process of invagination, while the cerebral type of development has been described only in the development of the compound retina. The natural conclusion from this fact is that, in watching the development of the compound retina, it is more difficult to differentiate the layers formed from the epidermal retinal cells and those formed from the epidermal optic ganglion-cells, than in the case of the simple retina, where the latter cells withdraw entirely from the surface. This is the conclusion to which Patten has come, and, indeed, judging from the text-book of Korschelt and Heider, it is the generally received opinion of the day that, as far as the Appendiculata are concerned, the retina, in the true sense—the retinal end-cells, with their cuticular rods,—is formed, in all cases, from the peripheral cells of the hypodermal layer, the cuticular rods being modifications of the general cuticular surface of the body. The apparent cerebral development of the crustacean retina, as quoted from Bobretsky by Balfour, is therefore in reality the development of the retinal ganglion, and not of the retina proper.

There is, I imagine, a universal belief that the natural mode of origin of a sense-organ, such as the eye, must always have been from the cells forming the external surface of the animal, and that direct origin from the central nervous system is a priori most improbable. It is, therefore, a matter of satisfaction to find that the evidence for the latter origin has universally broken down, with the single exception of the eyes of vertebrates and their degenerated allies; a fact which points strongly to the probability that a reconsideration of the evidence upon which the present teaching of the origin of the vertebrate eye is based will show that here, too, a confusion has arisen between that part formed from the epidermal surface and that from the optic ganglion.

The Median or Pineal Eyes.

Undoubtedly, in recent times, the most important clue to the ancestry of vertebrates has been given by the discovery that the so-called pineal gland in the vertebrate brain is all that remains of a pair of median or pineal eyes, the existence of which is manifest in the earliest vertebrates; so that the vertebrate, when it first arose, possessed a pair of median eyes as well as a pair of lateral eyes. The ancestor of the vertebrate, therefore, must also have possessed a pair of median eyes as well as a pair of lateral eyes.

Very instructive, indeed, is the evidence with regard to these median eyes, for one of the great characteristics of the ancient palæostracan forms is the invariable presence of a pair of median eyes as well as a pair of lateral eyes. In the living representative of such forms—Limulus—the pair of median eyes (Fig. 5) is well shown, and it is significant that here, according to Lankester and Bourne, these eyes are already in a condition of degeneration; so also in many of the Palæostraca (Fig. 7) the lateral eyes are the large, well-developed eyes, while the median eyes resemble those of Limulus in their insignificance.

We see, then, that in the dominant arthropod race at the time when the fishes first appeared, the type of eyes consisted of a pair of well-developed lateral eyes and a pair of insignificant, partially degenerated, median eyes. Further, according to all palæontologists, in the best-preserved head-shields of the most ancient fishes, especially well seen in the Osteostraci, in Cephalaspis, Tremataspis, Auchenaspis, Keraspis, a pair of large, prominent lateral eyes existed, between which, in the mid-line, are seen a pair of small, insignificant median eyes.

The evidence of the rocks, therefore, proves that the pair of median eyes which were originally the principal eyes (Hauptaugen), had already, in the dominant arthropod group been supplanted by a pair of lateral eyes, and had, in consequence, become small and insignificant, at the time when vertebrates first appeared. This dwindling process thus initiated in the arthropod itself has steadily continued ever since through the whole development of the vertebrates, with the result that, in the highest vertebrates, these median or pineal eyes have become converted into the pineal gland with its 'brain-sand.'

In the earliest vertebrate these median eyes may have been functional; they certainly were more conspicuous than in later forms. Alone among living vertebrates the right median eye of Ammocœtes is so perfect and the skin covering it so transparent that I have always felt doubtful whether it may not be of use to the animal, especially when one takes into consideration the undeveloped state of the lateral eyes in this animal, hidden as they are under the skin. Thus the one living vertebrate which is comparable with these extinct fishes is the one in which one of the pineal eyes is most well defined, most nearly functional.

Before passing to the consideration of the structure of the median eyes of Ammocœtes, it is advisable to see whether these median eyes in other animals, such as arachnids and crustaceans, belong to any particular type of eyes, for then assuredly the median eyes of Ammocœtes ought to belong to the same type if they are derived from them.

In the specialized crustacean, as in the specialized vertebrate, the median eyes have disappeared, at all events in the adult, but still exist in the primitive forms, such as Branchipus, which resemble the trilobites in some respects. On the other hand, the median eyes have persisted, and are well developed in the arachnids, both scorpions and spiders possessing a well-developed pair. The characteristics of the median eyes must then be especially sought for in the arachnid group.

Both scorpions and spiders possess many eyes, of which two are always separate and median in position, while the others form lateral groups; all these eyes possess a simple retina and a simple corneal lens. Grenacher was the first to point out that in the spiders two very distinct types of eye are found. In the one the retina is upright; in the other the retina is inverted, and the eye possesses a tapetal layer. The distribution of these two types is most suggestive, for the inverted retina is always found in the lateral eyes, never in the two median eyes; these always possess a simple upright retina.

In the crustaceans, the lateral eyes differ also from the median eyes, but not in the same way as in the arachnids; for here both types of eye possess an upright retina, but the retina of the lateral eyes is compound, while that of the median eyes is simple. In other words, the median eyes are in all cases eyes with a simple upright retina and a simple cuticular lens, while the retina of the lateral eyes is compound or may be inverted, according as the animal in question possesses crustacean or arachnid affinities. The lateral eye of the vertebrate, possessing, as it does, an inverted compound retina, indicates that the vertebrate arose from a stock which was neither arachnid nor crustacean, but gave rise to both groups—in fact, was a member of the great palæostracan group. What, then, is the nature of the median eyes in the vertebrate?

The Median Eyes of Ammocœtes.

The evidence of Ammocœtes is so conclusive that I, for one, cannot conceive how it is possible for any zoologist to doubt whether the parietal organ, as they insist on calling it, had ever been an eye, or rather a pair of eyes.

Anyone who examines the head of the larval lamprey will see on the dorsal side, in the median line, first, a somewhat circular orifice—the unpaired nasal opening; and then, tailwards to this, a well-marked circular spot, where the skin is distinctly more transparent than elsewhere. This spot coincides in position with the underlying dorsal pineal eye, which shines out conspicuously owing to the glistening whiteness of its pigment. Upon opening the brain-case the appearance as in Fig. 20 is seen, and the mass of the right ganglion habenulæ (G.H.R.), as it has been called, stands out conspicuously as well as the right or dorsal pineal eye (Pn.). Both eye and ganglion appear at first sight to be one-sided, but further examination shows that a left ganglion habenulæ is present, though much smaller than on the right side. In connection with this is another eye-like organ—the left or ventral pineal eye,—much more aborted, much less like an eye than the dorsal one; so also there are two bundles of peculiar fibres called Meynert's bundles, which connect this region with the infra-infundibular region of the brain; of these, the right Meynert's bundle is much larger than the left.

Fig. 31.—One of a Series of Horizontal Sections through the Head of Ammocœtes.

l.m., upper lip muscles; m.c., muco-cartilage; n., nose; na.c., nasal cartilage; pn., right pineal eye and nerve; g.h.r., right ganglion habenulæ; s.m., somatic muscles; cr., membranous wall of cranium; ch., choroid plexus; gl., glandular substance and pigment filling up brain-case.

Fig. 32.—Eye of Acilius Larva, with its Optic Ganglion. On the right side the nerve end-cells have been drawn free from pigment.	Fig. 33.—Pineal Eye of Ammocœtes, with its Ganglion Habenulæ. On the left side the eye is drawn as it appeared in the section. On the right side I have removed the pigment from the nerve end-cells, and drawn the eye as, in my opinion, it would appear if it were functional.

This difference between right and left indicates a greater degeneration on the left side, and points distinctly to a close relationship between the nerve-masses known as ganglia habenulæ and the median eyes. In my opinion this ganglion is, in part, at all events, the optic ganglion of the median eye on each side. It is built up on the same type as the optic ganglia of invertebrate simple eyes, with a cortex of small round cells and a medulla of fine nerve-fibres. Into this ganglion, on the right side, there passes a very well-defined nerve—the nerve of the dorsal eye. The eye itself with its nerve, pn., and its optic ganglion, g.h.r., is beautifully shown by means of a horizontal section through the head of Ammocœtes (Fig. 31). Originally, as described by Scott, the eye stood vertically above its optic ganglion, and presented an appearance remarkably like Fig. 32, which represents one of the simple eyes and optic ganglia of a larva of Acilius as described by Patten; then, with the forward growth of the upper lip, the right pineal eye was dragged forward and its nerve pulled horizontally over the ganglion habenulæ. For this reason the eye, nerve, and ganglion are better shown in a nearly horizontal than in a transverse section.

The optic nerve belonging to this eye is most evident and clearly shown in Fig. 31, and in the series of consecutive sections which follow upon this section; no doubt can arise as to the structure in question having been the nerve of the eye, even though, as is possible, it does not contain any functional nerve-fibres.

Fig. 34.—Horizontal Section through Brain of Ammocœtes, to show the Left, or Ventral Pineal Eye. pn.₂, left or ventral pineal eye; pn.₁, last remnant of right, or dorsal pineal eye; g.h.r., right ganglion habenulæ; g.h.l.₁, g.h.l.₃, parts of left ganglion habenulæ; pi., fold of pia mater which separates the left ganglion habenulæ from the left pineal eye; f., strands of nerve-fibres connecting the left eye with its ganglion, g.h.l.₃; V₃, third ventricle; V.aq., ventricle of aquæduct.

The second, ventral or left, eye, belonging to the left ganglion habenulæ is very different in appearance, being much less evidently an eye. Fig. 34 is one of the same series of horizontal sections as Fig. 31, pn.₁ being the last remnant of the right, or dorsal, eye, while pn.₂ shows the left, or ventral, eye with its connection with the left ganglion habenulæ.

In a series of sections I have followed the nerve of the right pineal eye to its destination, as described in my paper in the Quarterly Journal of Microscopical Science, and have found that it enters into the ganglion habenulæ just as the nerve to any simple eye enters into its optic ganglion. This nerve, as I have shown, is a very distinct, well-defined nerve, with no admixture of ganglion-cells or of connective tissue, very different indeed to the connection between the left pineal eye and its optic ganglion. Here there is no defined nerve at all; but the cells of the ganglion habenulæ stretch right up to the remains of the eye itself. Seeing, then, that both the eye and ganglion on this side have reached a much further grade of degeneration than on the right side, it may be fairly concluded that the original condition of these two median eyes is more nearly represented by the right eye, with its well-defined nerve and optic ganglion, than by the left eye, or by the eyes in lizards and other animals which do not show so well-defined a nerve as is possessed by Ammocœtes. Quite recently Dendy has examined the two median eyes in the New Zealand lamprey Geotria australis. In this species the second eye is much better defined than in the European lamprey, and its connection with the ganglion habenulæ is more nerve-like. In neither eye, however, is the nerve so clean cut and isolated as is the nerve of the dorsal, or right, eye in the Ammocœtes stage of Petromyzon Planeri; in both, cells resembling those of the cortex of the ganglion habenulæ and connective tissues are mixed up with the nerve-fibres which pass from each eye to its respective optic ganglion.

The Right Pineal Eye of Ammocœtes.

The optic fibres of the right median eye of Ammocœtes are connected with a well-defined retina, the limits of which are defined by the white pigment so characteristic of this eye. This pigment is apparently calcium phosphate, which still remains as the 'brain-sand' of the human pineal gland. The cells, which are hidden by this pigment, were described by me in 1890 as the retinal end-cells with large nuclei. In 1893, Studniçka examined them more closely, and concluded that the retinal cells are of two kinds: the one, nerve end-cells, the sensory cells proper; the other, pigmented epithelial cells, which surround the sense-cells. The sense-cells may contain some of the white pigment, but not so much as the other cells. Similarly, in the median eyes of Limulus, Lankester and Bourne find it difficult to determine how far the retinal end-cells contain pigment and how far that pigment really is in the cells surrounding these nerve end-cells.

The interior of the eye presents the appearance of a cavity in shape like a cornucopia, the stalk of which terminates at the place where the nerve enters. This cavity is not empty, but the posterior part of it is filled with the termination of the nerve end-cells of the retina, as pointed out by me and confirmed by Studniçka. These terminations are free from pigment, and contain strikingly translucent bodies, which I have described in my paper in the Quarterly Journal, and called rhabdites, for they present the same appearance and are situated in the same position as are many of the rhabdites on the terminations of the retinal end-cells of arthropod eyes. Studniçka has also seen these appearances, and figures them in his second paper on the nerve end-cells of the pineal eye of Ammocœtes.

Up to this point the following conclusions may be drawn:—

1. Ammocœtes possesses a pair of median eyes, just as was the case with the most ancient fishes, and with the members of the contemporary palæostracan group.

2. The retina of one of these eyes is well-defined and upright, not inverted, and therefore in this respect agrees with that of all median eyes.

3. The presence of nerve end-cells, with pigment either in them or in cells around them, to the unpigmented ends of which translucent bodies resembling rhabdites are attached, is another proof that this retina agrees with that of the median eyes of arthropods.

4. The simple nature of the nerve with its termination in an optic ganglion closely resembling in structure an arthropod optic ganglion, together with Studniçka's statement that the nerve end-cells pass directly into the nerve, points directly to the conclusion that this retina is a simple, not a compound, retina, and that it therefore in this respect also agrees with the retina of all median eyes.

With respect to this last conclusion, neither I myself nor Studniçka have been able to see any definite groups of cells between the nerve end-cells and the optic nerve such as a compound retina necessitates.

On the other hand, Dendy describes in the New Zealand lamprey, Geotria australis, a cavity where the nerve enters into the eye, which he calls the atrium. This cavity is distinct from the main cavity of the eye, and is separated from it by a mass of cells similar in appearance to those of the cortex of the ganglion habenulæ. In these two eyes then, groups of cells, resembling in appearance those belonging to an optic ganglion, exist in the eyes themselves. This atrium is evidently that part of the central cavity which I have called the handle of the cornucopia in the European lamprey, and the very fact that it is separated from the rest of the central cavity is evidence that we are dealing here with a later stage in the history of the pineal eyes than in the case of the Ammocœtes of Petromyzon Planeri. Taking also into consideration the continuity of the mass of small ganglion-cells which surround this atrium with the cells of the ganglion habenulæ by means of the similar cells scattered along the course of the nerve, and also bearing in mind the fact already stated that in the more degenerate left eye of Ammocœtes the cells of the ganglion habenulæ extend right up to the eye itself, it seems more likely than not that these cells do not represent the original optic ganglion of a compound retina, but rather the subsequent invasion, by way of the pineal nerve, of ganglion-cells belonging to a portion of the brain. In the last chapter it has been suggested that the presence of the trochlear or fourth cranial nerve has given rise to the formation of the cerebellum by a similar spreading.

There is certainly no appearance in the least resembling a compound retina such as is seen in the vertebrate or crustacean lateral eye. In the median eyes of scorpions and of Limulus, cells are found within the capsule of the eye among the nerve-fibres and the nerve end-cells. These are especially numerous in the median eyes of Limulus, as described by Lankester and Bourne, and are called by them intrusive connective tissue cells. The meaning of these cells is not, to my mind, yet settled. It is sufficient for my purpose to point out that the presence of cells interneural in position among the nerve end-cells of the retina of the median eyes of Ammocœtes is more probable than not, on the assumption that the retina of these eyes is built up on the same plan as that of the median eyes in Limulus and the scorpions.

It is further to be borne in mind that these specimens of Geotria worked at by Dendy were in the 'Velasia' stage of the New Zealand lamprey, and correspond, therefore, more nearly to the Petromyzon than to the Ammocœtes stage of the European lamprey.

The Dioptric Apparatus.

Besides the retina, all eyes possess a dioptric apparatus. What is the evidence as to its nature in these vertebrate median eyes? Lankester and Bourne have divided the eyes of scorpions and Limulus into two kinds, monostichous and diplostichous. In the first the retinal cells are supposed to give rise to not only rhabdites but also the cuticular chitinous lens, so that the eye is one-layered; in the second the lens is formed by a well-marked hypodermal layer, in front of the retina, composed of elongated cells, so that these eyes are two-layered or diplostichous. The lateral eyes, according to them, are all monostichous, but the median eyes are diplostichous.

Fig. 35.—Eye of Acilius Larvæ. (After Patten.)

l., chitinous lens; c., corneagen; pr., pre-retinal layer; rh., rhabdites; ret., retinal end-cells.

This distinction is not considered valid by other observers. Thus, as already indicated, Patten looks on all these eyes as three-layered, and states that in all cases a corneagen or vitreogen layer exists, which gives origin to the lens. For my own part I agree with Patten, but we are not concerned here with the lateral eyes. It is sufficient to note that all observers are agreed that the median eyes are characterized by this well-marked cell-layer, the so-called vitreous or corneagen layer of cells.

Fig. 36.—Eye of Hydrophilus Larva, with the Pigment over the Retinal End-cells. l., chitinous lens; c., corneagen; pr., pre-retinal layer; rh., rhabdites; ret., retinal end-cells.

This layer (c., Fig. 35) is composed of much-elongated cells of the hypodermal layer, in each of which the large nucleus is always situated towards the base of the cell. The space between it and the retina contains, according to Patten the cells of the pre-retinal layer (pr.). These may be so few and insignificant as to give the impression that the vitreous layer is immediately adjacent to the retina (ret.).

Let us turn now to the right pineal eye of Ammocœtes (Fig. 37) and see what its further structure is. The anterior part of the eye is free from pigment, and is composed, as is seen in hæmatoxylin or carmine specimens, of an inner layer of nuclei which are frequently arranged in a wavy line. From this nucleated layer, strands of tissue, free from nuclei, pass to the anterior edge of the eye.

In the horizontal longitudinal sections it is seen that these strands are confined to the middle of the eye. On each side of them the nuclear layer reaches the periphery, so that if we consider these strands to represent long cylindrical cells, as described by Beard, then the anterior wall may be described as consisting of long cylindrical cells, which are flanked on either side by shorter cells of a similar kind. The nuclei at the base of these cylindrical cells are not all alike. We see, in the first place, large nuclei resembling the large nuclei belonging to the nerve end-cells; these are the nuclei of the long cylindrical cells. We see also smaller nuclei in among these larger ones, which look like nuclei of intrusive connective tissue, or may perhaps form a distinct layer of cells, situated between the cells of the anterior wall and the terminations of the nerve end-cells already referred to.

These elongated cells are in exactly the same position and present the same appearance as the cells of the corneagen layer of any median eye. Like the latter they are free from pigment and never show with osmic staining any sign of the presence of translucent rhabdite-like bodies, such as are seen in the termination of the retinal cells, and like the latter their nuclei are at the base. The resemblance between this layer and the corneagen cells of any median eye is absolute. Between it and the terminations of the retinal cells there exists some ill-defined material certainly containing cells which may well correspond to Patten's pre-retinal layer of cells.

Retina, corneagen, nerve, optic ganglion, all are there, all in their right position, all of the right structure, what more is needed to complete the picture?

Fig. 37.—Pineal Eye of Ammocœtes, with its Ganglion Habenulæ.

In order to complete the dioptric apparatus a lens is necessary. Where, then, is the lens in these pineal eyes? In all the arachnid eyes, whether median or lateral, the lens is a single corneal lens composed of the external cuticle, which is thickened over the corneagen cells. This thickened cuticle is composed of chitin, and is not cellular, but is dead material formed out of the living underlying corneagen cells. Such a lens is in marked contrast to the lens of the lateral vertebrate eye, which is formed by living cells themselves. This thickening of the cuticular layer to form a lens could only exist as long as that layer is absolutely external, so that the light strikes immediately upon it; for, if from any cause the eye became situated internally, the place of such a lens must be filled by the structures situated between it and the surface, and the thickened cuticle would no longer be formed.

In all vertebrates these pineal eyes are separated from the external surface by a greater or less thickness of tissues; in the case of Ammocœtes, as is seen in Fig. 31, the eye lies within the membranous cranial wall, and is attached closely to it. The position, then, of the cuticular, or corneal lens, as it is often called, on the supposition that this is a median eye of the arachnid type, is taken by the membranous cranium, and, as I have described in my paper in the Quarterly Journal, on carefully lifting the eye in the fresh condition from the cranial wall, it can be seen under a dissecting microscope that the cranial wall often forms at this spot a lens-like bulging, which fits the shallow concavity of the surface of the eye, and requires some little force to separate it from the eye.

As will appear in a subsequent chapter, this cranial wall has been formed by the growth, laterally and dorsally, of a skeletal structure known by the name of the plastron. The last part of it to be completed would be that part in the mid-dorsal line, where apparently, in consequence of the insinking of the degenerating eyes, a dermal and subdermal layer already intervened between the source of light and the eyes themselves.

When the membranous cranium was completed in the mid-dorsal region, it was situated here as elsewhere just internally to the subdermal layer, and therefore enclosed the pineal eyes. This, to my mind, is the reason why the pineal eyes, which, in all other respects, conform to the type of the median eyes of an arachnid-like animal, do not possess a cuticular lens. Other observers have attempted to make a lens out of the elongated cells of the anterior wall of the eye (my corneagen layer), but without success.

Studniçka, who calls this layer the pellucida, does not look upon it as the lens, but considers, strangely enough, that the translucent appearances at the ends of each nerve end-cell represent a lens for that cell, so that every nerve end-cell has its own lens. Still more strange is it that, holding this view, he should yet consider these knobs to be joined by filaments to the cells in the anterior wall of the eye, a conception fatal to the action of such knobs as lenses.

The discovery that the vertebrate possesses, in addition to the lateral eyes, a pair of median eyes, which are most conspicuous in the lowest living vertebrate, together with the fact that such eyes are built up on the same plan as the median eyes of living crustaceans or arachnids, not only with respect to the eye itself but also to its nerve and optic ganglion, constitutes a fact of the very greatest importance for any theory of the origin of vertebrates; especially in view of the further fact, that similar eyes in the same position are found not only in all the members of the Palæostraca, but also in all those ancient forms (classed as fishes) which lived at that time. At one and the same moment it proves the utter impossibility of reversing dorsal and ventral surfaces, points in the very strongest manner to the origin of the vertebrate from some member or other of the palæostracan group, and insists that the advocates of the origin of vertebrates from the Hemichordata, etc., should give an explanation of the presence of these two median eyes of a more convincing character than that given here.

The Lateral Eyes.

Turning now to the consideration of the lateral eyes, we see that these eyes in the arachnids often possess an inverted retina, in the crustaceans always an upright retina. In the arachnids they possess a simple retina, while in the crustaceans their retina is compound; so that in the latter case the so-called optic nerve is in reality a tract of fibres connecting together the brain-region with a variable number of optic ganglia, which have been left at the periphery in close contact with the retinal cells, when the brain sunk away from the superficial epithelial covering.

There is, then, this difference between the lateral eyes of crustaceans and arachnids, that the retina of the former is compound, but never inverted, while that of the latter may be inverted, but is always simple.

The retina of the lateral eyes of the vertebrate resembles both of these, for it is compound, as in the crustacean, and inverted as in the arachnid.

It must always be borne in mind that in the palæostracan epoch the dominant race was neither crustacean nor arachnid, but partook of the characters of both; also, as is characteristic of dominance, there was very great variety of form, so that it seems more probable than not that some of these forms may have combined the arachnid and crustacean characteristics to the extent of possessing lateral eyes with an inverted yet compound retina. A certain amount of evidence points in this direction. As already stated, the compound retina which characterizes the vertebrate lateral eyes is characteristic of all facetted eyes, and in the trilobites facetted lateral eyes are commonly found. From this it may be concluded that many of the trilobites possessed eyes with a compound retina. There have, however, been found in certain species, e.g. Harpes vittatus and Harpes ungula, lateral eyes which were not facetted, and are believed by Korschelt and Heider to be of an arachnid nature. They say, "Palæontologists have appropriately described them as ocelli, although, from a zoological point of view, they do not deserve this name, having most probably arisen in a way similar to that conjectured in connection with the lateral eyes of scorpions." If this conjecture is right, then in these forms the retina may have been inverted, but because they belonged to the trilobite group, the retina was most probably compound, so that here we may have had the combination of the arachnid and crustacean characteristics. On the other hand, in some forms of Branchipus, and many of the Gammaridæ, a single corneal lens is found in conjunction with an eye of the crustacean type, so that a non-facetted lateral eye, found in a fossil form, would not necessarily imply the arachnid type of eye with the possibility of an inverted retina. Whatever may be the ultimate decision upon these particular forms, the striking fact remains, that both in the vertebrate and in the arachnid the median eyes possess a simple upright retina, while the lateral eyes possess an inverted retina, and that both in the vertebrate and the crustacean the median eyes possess a simple upright retina, while the lateral eyes possess a compound retina.

The resemblance of the retina of the lateral eyes of vertebrates to that of the lateral eyes of many arthropods, especially crustaceans, has been pointed out by nearly every one who has worked at these invertebrate lateral eyes. The foundation of our knowledge of the compound retina is Berger's well-known paper, the results of which are summed up by him in the following two main conclusions.

1. The optic ganglion of the Arthropoda consists of two parts, of which the one stands in direct inseparable connection with the facetted eye, and together with the layer of retinal rods forms the retina of the facetted eye, while the other part is connected rather with the brain, and is to be considered as an integral part of the brain in the narrower sense of the word.

Fig. 38.—The Retina of Musca. (After Berger.)

Br., brain; O.n., optic nerve; n.l.o.g., nuclear layer of ganglion of optic nerve; m.l., molecular layer (Punktsubstanz); n.l.r.g.i. and n.l.r.g.o., inner and outer nuclear layers of the ganglion of the retina; f.br.r., terminal fibre-layer of retina; r., layer of retinal end-cells (indicated only).

2. In all arthropods examined by him, the retina consists of five layers, as follows:—

(1) The layer of rods and their nuclei.
(2) The layer of nerve-bundles.
(3) The nuclear layer.
(4) The molecular layer.
(5) The ganglion cell layer.

Berger passes under review the structure and arrangement of the optic ganglion in a large number of different groups of arthropods, and concludes that in all cases one part of the optic ganglion is always closely attached to the visual end-cells, and this combination he calls the retina. On the other hand, the nerve-fibres which connect the peripheral part of the optic ganglion with the brain, the so-called optic nerve, are by no means homologous in the different groups; for in some cases, as in many of the stalk-eyed crustaceans, the whole optic ganglion is at the periphery, while in others, as in the Diptera, only the retinal ganglion is at the periphery, and the nerve-stalk connects this with the rest of the optic ganglion, the latter being fused with the main brain-mass. In the Diptera, in fact, according to Berger, the optic nerve and retina are most nearly comparable to those of the vertebrate. For this reason I give Berger's picture of the retina of Musca (Fig. 38), in order to show the arrangement there of the retinal layers.

Fig. 39.—The Brain of Sphæroma serratum. (After Bellonci.)

Ant. I. and Ant. II., nerves to 1st and 2nd antennæ. f.br.r., terminal fibre-layer of retina; Op. g. I., first optic ganglion; Op. g. II., second optic ganglion; O.n., optic nerve-fibres forming an optic chiasma.

In Branchipus and other primitive Crustacea, Berger also finds the same retinal layers, but is unable to distinguish in the brain the rest of the optic ganglion. Judging from Berger's description of Branchipus, and Bellonci's of Sphæroma, it would almost appear as though the cerebral part of the retina in the higher forms originated from two ganglionic enlargements, an external and internal enlargement, as Bellonci calls them. The external ganglion (Op. g. I., Fig. 39) may be called the ganglion of the retina, the cells of which form the nuclear layer of the higher forms, and the internal ganglion (Op. g. II., Fig. 39), from which the optic nerve-fibres to the brain arise, may therefore be called the ganglion of the optic nerve. Bellonci describes how in this latter ganglion cells are found very different to the small ones of the external ganglion or ganglion of the retina. So also in Branchipus, judging from the pictures of Berger, Claus, and from my own observations (cf. Fig. 46, in which the double nature of the retinal ganglion is indicated), the peripheral part of the optic ganglion—i.e. the retinal ganglion—may be spoken of as composed of two ganglia. The external of these is clearly the ganglion of the retina; its cells form the nuclear layer, the striking character of which, and close resemblance to the corresponding layer in vertebrates, is shown by Claus' picture, which I reproduce (Fig. 40). The internal ganglion with which the optic nerve is in connection contains large ganglion cells, which, together with smaller ones, form the ganglionic layer of Berger.

The most recent observations of the structure of the compound retina of the crustacean eye are those of Parker, who, by the use of the methylene blue method, and Golgi's method of staining, has been able to follow out the structure of the compound retina in the arthropod on the same lines as had already been done for the vertebrate. These two methods have led to the conclusion that the arthropod central nervous system and the vertebrate central nervous system are built up in the same manner—viz. by means of a series of ganglia connected together, each ganglion being composed of nerve-cells, nerve-fibres, and a fine reticulated substance called by Leydig in arthropods 'Punktsubstanz,' and known in vertebrates and in invertebrates at the present time as 'neuropil.' A further analysis resolves the whole system into a combination of groups of neurones, the cells and fibres of which form the cells and fibres of the ganglia, while their dendritic connections with the terminations of other neurones, together with the neuroglia-cells form the 'neuropil.' As is natural to expect, that part of the central nervous system which helps to form the compound retina is built up in the same manner as the rest of the central nervous system.

Fig. 40.—Bipolar Cells of Nuclear Layer in Retina of Branchipus. (After Claus.) f.br.r., terminal fibre-layer of retina; n.l.r.g., bipolar cells of the ganglion of the retina = inner nuclear layer; m.l., Punktsubstanz = inner molecular layer; b.m., basement membrane formed by neurilemma round central nervous system.

Thus, according to Parker, the mass of nervous tissue which occupies the central part of the optic stalk in Astacus is composed of four distinct ganglia; the retina is connected with the first of these by means of the retinal fibres, and the optic nerve extends proximally from the fourth ganglion to the brain. Each ganglion consists of ganglion-cells, nerve-fibres, and 'neuropil,' and, in addition, supporting cells of a neuroglial type. By means of the methylene blue method and the Golgi method, it is seen that the retinal end-cells, with their visual rods, are connected with the fibres of the optic nerve by means of a system of neurones, the synapses of which take place in and help to form the 'neuropil' of the various ganglia. Thus, an impulse in passing from the retina to the brain would ordinarily travel over five neurones, beginning with one of the first order and ending with one of the fifth. He makes five neurones although there are only four ganglia, because he reckons the retinal cell with its elongated fibre as a neurone of the first order, such fibre terminating in dendritic processes which form synapses in the 'neuropil' of the first ganglion with the neurones of the second order.

Similarly the neurones of the second order terminate in the 'neuropil' of the second ganglion, and so on, until we reach the neurones of the fifth order, which terminate on the one hand in the 'neuropil' of the fourth ganglion, and on the other pass to the optic lobes of the brain by their long neuraxons—the fibres of the optic nerve.

He compares this arrangement with that of Branchipus, Apus, Estheria, Daphnia, etc., and shows that in the more primitive crustaceans the peripheral optic apparatus was composed, not of four but of two optic ganglia, not, therefore, of five but of three neurones, viz.—

1. The neurone of the first order—i.e. the retinal cell with its fibre terminating in the 'neuropil' of the first optic ganglion (ganglion of the retina).

2. The neurone of the second order, which terminates in the 'neuropil' of the second ganglion (ganglion of the optic nerve).

3. The neurone of the third order, which terminates in the optic lobes of the brain by means of its neuraxons (the optic nerve).

We see, then, that the most recent researches agree with the older ones of Berger, Claus, and Bellonci, in picturing the retina of the primitive crustacean forms as formed of two ganglia only, and not of four, as in the specialized crustacean group the Malacostraca.

The comparison of the arthropod compound retina with that of the vertebrate shows, as one would expect upon the theory of the origin of vertebrates put forward in this book, that the latter retina is built up of two ganglia, as in the more primitive less specialized crustacean forms. The modern description of the vertebrate retina, based upon the Golgi method of staining, is exactly Parker's description of the simpler form of crustacean retina in which the 'neuropil' of the first ganglion is represented by the external molecular layer, and that of the second ganglion by the internal molecular layer; the three sets of neurones being, according to Parker's terminology:—

1. The neurones of the first order—viz. the visual cells—the nuclei of which form the external nuclear layer, and their long attenuated processes form synapses in the external molecular layer with

2. The neurones of the second order, the cells of which form the internal nuclear layer, and their processes form synapses in the internal molecular layer with

3. The neurones of the third order, the cells of which form the ganglionic layer and their neuraxons constitute the fibres of the optic nerve which end in the optic lobes of the brain.

Strictly speaking, of course, the visual cells with their elongated processes have no right to be called neurones: I only use Parker's phraseology in order to show how closely the two retinas agree even to the formation of synapses between the fine drawn-out processes of the visual cells and the neurones of the ganglion of the retina.

The Retina of the Lateral Eye of Ammocœtes.

As in the case of all other organs, it follows that if we are dealing here with a true genetic relationship, then the lower we go in the vertebrate kingdom the more nearly ought the structure of the retina to approach the arthropod type. It is therefore a matter of intense interest to determine the nature of the retina in Ammocœtes in order to see whether it differs from that of the higher vertebrates, and if so, whether such differences are explicable by reference to the structure of the arthropod eye.

Before describing the structure of this retina it is necessary to clear away a remarkable misconception, shared among others by Balfour, that this eye is an aborted eye, and that it cannot be considered as a primitive type. Thus Balfour says: "Considering the degraded character of the Ammocœte eye, evidence derived from its structure must be received with caution," and later on, "the most interesting cases of partial degeneration are those of Myxine and the Ammocœte. The development of such aborted eyes has as yet been studied only in the Ammocœte, in which it resembles in most important features that of other Vertebrata."

Again and again the aborted character of the eye is stated to be evidence of degeneration in the case of the lamprey. What such a statement means, why the eye is in any way to be considered as aborted, is to me a matter of absolute wonderment: it is true that in the larval form it lies under the skin, but it is equally true that at transformation it comes to the surface, and is most evidently as perfect an eye as could be desired. There is not the slightest sign of any degeneration or abortion, but simply of normal development, which takes a longer time than usual, lasting as it does throughout the life-time of the larval form.

Kohl, who has especially studied degenerated vertebrate eyes, discusses with considerable fulness the question of the Ammocœtes eye, and concludes that in aborted eyes a retarded development occurs, and this applies on the whole to Ammocœtes, "but with the important difference that in this case the period of retarded development is not followed by a stoppage, but on the contrary by a period of very highly intensified progressive development during the metamorphosis," with the result that "the adult eye of Petromyzon Planeri does not diverge from the ordinary type."

Referring in his summing up to this retarded development, he says: "Such reminiscences of embryonic conditions are after all present here and there in normally developed organs, and by no means entitle us to speak of abnormal development."

The evidence, then, is quite clear that the eye of Petromyzon, or, indeed, of the full-grown Ammocœtes, is in no sense an abnormal eye, but simply that its development is slow during the ammocœte stage. The retina of Petromyzon was figured and described by Langerhans in 1873. He describes it as composed of the following layers:—

⁠(1) Membrana limitans interna.
⁠(2) Thick inner molecular layer.
⁠(3) Optic fibre layer.
⁠(4) Thick inner nuclear layer.
⁠(5) Peculiar double-layered ganglionic layer.
⁠(6) External molecular layer.
⁠(7) External nuclear layer.
⁠(8) Membrana limitans externa.
⁠(9) Layer of rods.
(10) Pigment-epithelium.

Fig. 41.—Retina and Optic Nerve of Petromyzon. (After Müller and Langerhans.)

On the left side the Müllerian fibres and pigment-epithelium are represented alone. The retina is divided into an epithelial part, C (the layer of visual rod-cells), and a neurodermal or cerebral part which is formed of, A, the ganglion of the optic nerve and, B, the ganglion of the retina. 1, int. limiting membrane; 2, int. molecular layer with its two layers of cells; 3, layer of optic nerve fibres; 4, int. nuclear layer; 5, double row of tangential fulcrum cells; 6, layer of terminal retinal fibres; 7, ext. nuclear layer; 8, ext. limiting membrane; 9, layer of rods; 10, layer of pigment-epithelium. D, axial cell layer (Axenstrang) in optic nerve. The layer 6 is drawn rather too thick.

He points out especially the peculiarity of layer (2) (2, Fig. 41), the inner molecular, in which two rows of nuclei are arranged with great regularity, the one row closely touching the membrana limitans interna, the other at the inner boundary of the middle third of the molecular layer. Of these two rows of nuclei, he describes the innermost as composed almost entirely of large nuclei belonging to ganglion cells, while the outermost is composed mainly of distinctly smaller nuclei, which in staining and appearance appear to belong not to nerve-cells but to the true reticular tissue of the molecular layer.

He also draws special attention to the remarkable layer (5) (5, Fig. 41), which is not found in the retina of the higher vertebrates, the cells of which, in his opinion, are of the nature of ganglion-cells.

W. Müller, in 1874, gave a most careful description of the eye of Ammocœtes and Petromyzon, and traced the development of the retina; the subsequent paper of Kohl does not add anything new, and his drawings are manifestly diagrams, and do not represent the appearances so accurately as Müller's illustrations. In the accompanying figure (Fig. 41) I reproduce on the right-hand side Müller's picture of the retina of Petromyzon, but have drawn it, as in Langerhans' picture, at the place of entry of the optic nerve.

From his comparison of this retina with a large number of other vertebrate retinas, he comes to the conclusion that the retina of all vertebrates is divisible into

A. An ectodermal (epithelial) part consisting of the layer of the visual cells, and

B. A neurodermal (cerebral) part which forms the rest of the retina.

Further, Müller points out that the neuroderm gives origin throughout the central nervous system to two totally different structures, on the one hand to the true nervous elements, on the other to a system of supporting cells and fibres which cannot be classed as connective tissue, for they do not arise from mesoblast, and are therefore called by him 'fulcrum-cells.' In the retina he recognizes two distinct groups of such supporting structures—(1) a system of radial fibres with well-marked elongated nuclei, which extend between the two limiting layers, and form at their outer ends a membrane-like expansion which was originally the outer limit of the retina, but becomes afterwards co-terminous with the membrana limitans externa, owing to the piercing through it of the external limbs of the rods. This system, which is known by the name of the radial Müllerian fibres (shown on the left-hand side of Fig. 41), has no connection with (2) the spongioblasts and neurospongium, which form a framework of neuroglia, in which the terminations of the optic ganglion and of the retinal ganglion ramify to form the molecular layers.

It is evident from Fig. 41 that the retina of Ammocœtes and Petromyzon differs in a striking manner from the typical vertebrate retina. The epithelial part (C) remains the same—viz. the visual rods, the external limiting membrane, and the external nuclear layer; but the cerebral part, the retinal ganglion (A and B), is remarkably different. It is true, it consists in the main of the small-celled mass known as the inner nuclear layer, and of the reticulated tissue or 'neuropil' known as the inner molecular layer, just as in all other compound retinal eyes; but neither the ganglion cell-layer nor the optic fibre-layer is clearly defined as separate from this molecular layer; on the contrary, it is matter of dispute as to what cells represent the ganglionic layer of higher vertebrates, and the optic fibres do not form a distinct innermost layer, but pass into the inner molecular layer at its junction with the inner nuclear layer. A comparison of this innermost part of the retina (A, Fig. 41), with the corresponding part in Berger's picture of Musca (n.l.o.g., Fig. 38), shows a most striking similarity between the two. In both cases the fibres of the optic nerve (O.n., Fig. 38) which cross at their entrance pass into the 'neuropil' of this part of the retinal ganglion, and are connected probably (though that is not proved in either case) with the cells of the ganglionic layer. In both cases we find two well-marked parallel rows of cells in this part of the retina, of which one, the innermost, is composed in Ammocœtes of large ganglion-cells, and the other mainly of smaller, deeper staining cells apparently supporting in function. Similarly, also, in Branchipus, as I conclude from my own observations as well as from those of Berger and Claus, the ganglionic layer is composed partly of true ganglion-cells and partly of supporting cells arranged in a distinct layer. This part, then, of the retina of Ammocœtes is remarkably like that of a typical arthropod retina, and forms that part of the retinal ganglion which may be called the ganglion of the optic nerve.

Next comes the ganglion of the retina (B, Fig. 41) (Parker's first optic ganglion), the cells of which form the small bipolar granule-cells of the inner nuclear layer; granule-cells arranged in rows just as they are shown in Claus' picture of the same layer in the retina of Branchipus (Fig. 40), just as they are found in the cortical layers of the optic ganglion of the pineal eye (ganglion habenulæ), in the optic lobes and other parts of the Ammocœtes brain, or in the cortical layers of the optic ganglia of all arthropods.

Between this small-celled nuclear layer (4, Fig. 41) and the layer of nuclei of the visual rod cells (7, Fig. 41) (the external nuclear layer), we find in the eye of Ammocœtes and Petromyzon two well-marked rows of cells of a most striking character—viz. the two remarkably regular rows of large epithelial-like cells with large conspicuous nuclei, which give the appearance of two opposing rows of limiting epithelium (5, Fig. 41), already mentioned in connection with the researches of Langerhans and W. Müller. Here, then, is a striking peculiarity of the retina of the lamprey, and according to Müller the obliteration of these two layers can be traced as we pass upwards in the vertebrate kingdom. Among fishes, they are especially well seen in the perch; in the higher vertebrates the whole layer is only a rudiment represented, he thinks, by the simple layer of round cells which lies close against the inner surface of the layer of terminal fibres (Nervenansätze), and is especially evident in birds and reptiles. In man and the higher mammals they are probably represented by the horizontal cells of the outer part of the inner nuclear layer.

Seeing, then, that they are most evident in Ammocœtes, and become less and less marked in the higher vertebrates, it is clear that their origin cannot be sought among the animals higher in the scale than Ammocœtes, but must, therefore, be searched for in the opposite direction.

Müller describes them as forming a very conspicuous landmark in the embryology of the retina, dividing it distinctly into two parts, an outer thinner, and an inner somewhat thicker part, the zone formed by them standing out conspicuously on account of the size and regularity of the cells and their lighter appearance when stained. Thus in his description of the retina of an Ammocœtes 95 mm. in length, he says, "The layer of pale tangentially elongated cells formed a double layer and produced the appearance of a pale, very characteristic zone between the outer and inner parts of the retina."

Let us now turn to the retina of the crustacean and see whether there is any evidence there that the retina is divisible into an outer and inner part, separated by a zone of characteristically pale staining cells with conspicuous nuclei. The most elaborate description of the development of the retina of Astacus is given by Reichenbach, according to whom the earliest sign of the formation of the retina is an ectodermic involution (Augen-einstülpung), which soon closes, so that the retinal area appears as a thickening. In close contiguity to this thickening, the thickening of the optic ganglion arises, so that that part of the optic ganglion which will form the retinal ganglion fuses with the thickened optic plate and forms a single mass of tissue. Later on a fold (Augen-falte) appears in this mass of tissue, in consequence of which it becomes divided into two parts. The lining walls of this fold form a double row of cells, the nuclei of which are most conspicuous because they are larger and lighter in colour than the surrounding nuclei, so that by this fold the retina is divided into an outer and an inner wall, the line of demarcation being conspicuous by reason of these two rows of large, lightly-staining nuclei.

Reichenbach is unable to say that this secondary fold is coincident with the primary involution, and that therefore the junction between the two rows of large pale nuclei is the line of junction between the retinal ganglion and the retina proper, because all sign of the primary involution is lost before the secondary fold appears.

Parker compares the appearances in the lobster with Reichenbach's description in the crayfish, and says that he finds only a thickening, no primary involution; at the same time he expressly states that in the very early stages his material was deficient, and that he had not grounds sufficient to warrant the statement that no involution occurs. He also finds that in the lobster the ganglionic tissue which arises by proliferation is divided into an outer and inner part; the separation is effected by a band of large, lightly-staining nuclei, which, in position and structure, resemble the band figured by Reichenbach. According to Parker, then, the line of separation indicated in the development by Reichenbach's outer and inner walls is not the line of junction between the retina and the retinal ganglion, as Reichenbach was inclined to think, but rather a separation of two rows of large ganglion-cells belonging to the retinal ganglion.

The similarity between these conspicuous layers of lightly-staining cells in Ammocœtes and in crustaceans is remarkably close, and in both cases observers have found the same difficulty in interpreting their meaning. In each case one group of observers looks upon them as ganglion-cells, the other as supporting structures. Thus in the lamprey, Müller considers them to belong to the supporting elements, while Langerhans and Kohl describe them as a double layer of ganglion-cells. In the crustacean, Berger in Squilla, Grenacher in Mysis, and Parker in Astacus, look upon them as supporting elements, while Viallanes in Palinurus considers them to be true ganglionic cells.

Whatever the final interpretation of these cells may prove to be, we may, it seems to me, represent an ideal compound retina of the crustacean type by combining the investigations of Berger, Claus, Reichenbach, and Parker in the following figure.

Fig. 42.—Ideal Diagram of the Layers in a Crustacean Eye.

The retina is divided into an epithelial part, C (the layer of retinular cells and rhabdomes), and a neurodermal or cerebral part, which is formed of, A, the ganglion of the optic nerve, and, B, the ganglion of the retina. 1, optic nerve fibres which cross at their entrance into the retina; 2, int. molecular layer with its two rows of cells; 3, int. nuclear layer; 4, Reichenbach's double row of large lightly-staining cells; 5, layer of terminal retinal fibres; 6, ext. nuclear layer; 7, ext. limiting membrane; 8, layer of crystalline cones; 9, cornea.

The comparison of this figure (Fig. 42) with that of the Petromyzon retina (Fig. 41) shows how great is the similarity of the latter with the arthropod type, and how the very points in which it deviates from the recognized vertebrate type are explainable by comparison with that of the arthropod. The most striking difference between the retinas in the two figures is that the layer of terminal nerve fibres (5, Fig. 42), which, after all, are only the elongated terminations of the retinal cells belonging to Parker's neurones of the first order, is very much longer than in Petromyzon or in any vertebrate, for the external molecular layer (6, Fig. 41) (Müller's layer of Nervenansätze) is very short and inconspicuous (in Fig. 41 it is drawn too thick).

Turning from the retina to the fibres of the optic nerve we again find a remarkable resemblance, for in Ammocœtes, as pointed out by Langerhans and carefully figured by Kohl, a crossing of the fibres of the optic nerve occurs as the nerve leaves the retina, just as is so universally the case in all compound retinas. To this crossing Kohl has given the name chiasma nervi optici, in distinction to the cerebral chiasma, which he calls chiasma nervorum opticorum. Further, we find that even this latter chiasma is well represented in the arthropod brain; thus Bellonci in Sphæroma, Berger, Dietl, and Krieger in Astacus, all describe a true optic chiasma, the only difference in opinion being, whether the crossing of the optic nerves is complete or not. Especially instructive are Bellonci's figures and description. He describes the brain of Sphæroma as composed of three segments—a superior segment, the cerebrum proper, a middle segment, and an inferior segment; the optic fibres, as is seen in Fig. 39, after crossing, pass direct into the middle segment, in the ganglia of which they terminate. From this segment also arises the nerve to the first antenna of that side—i.e. the olfactory nerve. The optic part, then, of this middle segment is clearly the brain portion of the optic ganglionic apparatus, and may be called the optic lobes, in contradistinction to the peripheral part, which is usually called the optic ganglion, and is composed of two ganglia, Op. g. I. and Op. g. II., as already mentioned. These optic lobes are therefore homologous with the optic lobes of the vertebrate brain.

The resemblance throughout is so striking as to force one to the conclusion that the retina of the vertebrate eye is a compound retina, composed of a retina and retinal ganglion of the type found in arthropods. From this it follows that the development of the vertebrate retina ought to show the formation of (1) an optic plate formed from the peripheral epidermis and not from the brain; (2) a part of the brain closely attached to this optic plate forming the retinal ganglion, which remains at the surface when the rest of the optic ganglion withdraws; (3) an optic nerve formed in consequence of this withdrawal, as the connection between the retinal and cerebral parts of the optic ganglion.

This appears to me exactly what the developmental process does show according to Götte's investigations. He asserts that the retina arises from an optic plate, being the optical portion of his 'Sinnes-platte.' At an early stage this is separated by a furrow (Furche) from the general mass of epidermal cells which ultimately form the brain. This separation then vanishes, and the retina and brain-mass become inextricably united into a mass of cells, which are still situated at the surface. By the closure of the cephalic plate and the withdrawal of the brain away from the surface, a retinal mass of cells is left at the surface connected with the tubular central nervous system by the hollow optic diverticulum or primary optic vesicle. If we regard only the retinal and nervous elements, and for the moment pay no attention to the existence of the tube, Götte's observation that the true retina has been formed from the optic plate (Sinnes-platte) to which the retinal portion of the brain (retinal ganglion) has become firmly fixed, and that then the optic nerve has been formed by the withdrawal of the rest of the brain (optic lobes), is word for word applicable to the description of the development of the compound retina of the arthropod eye, as has been already stated.

The Significance of the Optic Diverticula.

The origin of the retina from an optic epidermal plate in vertebrates, as in all other animals, brings the cephalic eyes of all animals into the same category, and leaves the vertebrate eye no longer in an isolated and unnatural position. In one point the retina of the vertebrate eye differs from that of a compound retina of an invertebrate; in the former, a striking supporting tissue exists, known as Müller's fibres, which is absent in the latter. This difference of structure is closely associated with another of the same character as in the central nervous system, viz. the apparent development of the nervous part from a tube. We see, in fact, that the retinal and nervous arrangements of the vertebrate eye are comparable with those of the arthropod eye, in precisely the same way and to the same extent as the nervous matter of the brain of the vertebrate is comparable with the brain of the arthropod. In both cases the nervous matter is, in structure, position, and function, absolutely homologous; in both cases there is found in the vertebrate something extra which is not found in the invertebrate—viz. a hollow tube, the walls of which, in the case of the brain, are utilized as supporting tissues for the nerve structures. The explanation of this difference in the case of the brain is the fundamental idea of my whole theory, namely, that the hollow tube is in reality the cephalic stomach of the invertebrate, around which the nervous brain-matter was originally grouped in precisely the same manner as in the invertebrate. What, then, are the optic diverticula?

"The formation of the eye," as taught by Balfour, "commences with the appearance of a pair of hollow outgrowths from the anterior cerebral vesicle. These outgrowths, known as the optic vesicles, at first open freely into the cavity of the anterior cerebral vesicle. From this they soon, however, become partially constricted, and form vesicles united to the base of the brain by comparatively narrow, hollow stalks, the rudiments of the optic nerves."

"After the establishment of the optic nerves, there takes place (1) the formation of the lens, and (2) the formation of the optic cup from the walls of the primary optic vesicle."

He then goes on to explain how the formation of the lens forms the optic cup with its double walls from the primary optic vesicle, and says—

"Of its double walls, the inner, or anterior, is formed from the front portion, the outer, or posterior, from the hind portion of the wall of the primary optic vesicle. The inner, or anterior, which very speedily becomes thicker than the other, is converted into the retina; in the outer, or posterior, which remains thin, pigment is eventually deposited, and it ultimately becomes the tesselated pigment-layer of the choroid."

The difficulties in connection with this view of the origin of the eye are exceedingly great, so great as to have caused Balfour to discuss seriously Lankester's suggestion that the eye must have been at one time within the brain, and that the ancestor of the vertebrate was therefore a transparent animal, so that light might get to the eye through the outer covering and the brain-mass; a suggestion, the unsatisfactory nature of which Balfour himself confessed. Is there really evidence of any part of either retina or optic nerve being formed from the epithelial lining of the tube?

This tube is formed as a direct continuation of the tube of the central nervous system, and we can therefore apply to it the same arguments as have been used in the discussion of the meaning of the latter tube. Now, the striking point in the latter case is the fact that the lining membrane of the central canal is in so many parts absolutely free from nervous matter, and so shows, as in the so-called choroid plexuses, its simple, non-nervous epithelial structure. This also we find in the optic diverticulum. Where there is no evidence of any invasion of the tube by nervous elements, there it retains its simple non-nervous character of a tube composed of a single layer of epithelial cells—viz. in that part of the tube which, as Balfour says, remains thin, in which pigment is eventually deposited, and which ultimately becomes the tesselated pigment-layer of the choroid. Nobody has ever suggested that this pigment-layer is nervous matter, or ever was, or ever will be, nervous matter; it is in precisely the same category as the membranous roof of the brain in Ammocœtes, which never was, and never will be, nervous matter. Yet, according to the old embryology both in the case of the eye and the brain, the pigment-layer and the so-called choroid plexuses are a part of the tubular nervous system.

Turning now to the optic nerve, Balfour describes it as derived from the hollow stalk of the optic vesicle. He says—

"At first the optic nerve is equally continuous with both walls of the optic cup, as must of necessity be the case, since the interval which primarily exists between the two walls is continuous with the cavity of the stalk. When the cavity within the optic nerve vanishes, and the fibres of the optic nerve appear, all connection is ruptured between the outer wall of the optic cup and the optic nerve, and the optic nerve simply perforates the outer wall, and becomes continuous with the inner one."

In this description Balfour, because he derived the optic nerve fibres from the epithelial wall of the optic stalk, of necessity supposed that such fibres originally supplied both the outer and inner walls of the optic cup and, therefore, seeing that when the fibres of the optic nerve appear they do not supply the outer wall, he supposes that their original connection with the outer wall is ruptured, because a discontinuity of the epithelial lining takes place coincidently with the appearance of the optic nerve-fibres, and, according to him, the optic nerve simply perforates the outer wall and becomes continuous with the inner one. This last statement is very difficult to understand. I presume he meant that some of the fibres of the optic nerve supplied from the beginning the inner wall of the optic cup, but that others which originally supplied the outer wall were first ruptured, then perforated the outer wall, and finally completed the supply to the inner wall or retina.

This statement of Balfour's is the necessary consequence of his belief, that the epithelial cells of the optic stalk gave rise to the fibres of the optic nerve. If, instead of this, we follow Kölliker and His, who state that the optic nerve-fibres are formed outside the epithelial walls of the optic stalk, and that the cells of the latter form supporting structures for the nerve-fibres, then the position of the optic nerve becomes perfectly simple and satisfactory without any rupturing of its connection with the outer wall and subsequent perforation, for the optic nerve-fibres from their very first appearance pass directly to supply the retina—i.e. the inner wall of the optic cup and nothing else.

They pass, as is well known, without any perforation by way of the choroidal slit to the inner surface of the inner wall (retina) of the optic cup; then, when the choroidal slit becomes closed by the expansion of the optic cup, the optic nerve naturally becomes situated in the centre of the base of the cup and spreads over its inner surface as that surface expands.

A section across the optic cup at an early stage at the junction of the optic stalk and optic cup would be represented by the upper diagram in Fig. 43; at a later stage, when the choroidal slit is closed, by the lower diagram.

Fig. 43.—Diagram of the Relation of the Optic Nerve to the Optic Cup.

The upper diagram represents a stage before the formation of the choroidal slit, the lower one the stage of closure of the choroidal slit. R., retina; O.n., optic nerve; p., pigment epithelium.

The evident truth of this manner of looking at the origin of the optic nerve is demonstrated by the appearance of the optic nerve in Ammocœtes and Petromyzon. In the latter, although the development is complete, and the eye, and consequently also the optic nerve-fibres, are fully functional, there is still present in the axial core of the nerve a row of epithelial cells (Axenstrang) which are altered so as to form supporting structures, in the same way as a row of epithelial cells in the retina is altered to form the system of supporting cells known by the name of the Müllerian fibres.

The origin of this axial core of cells is perfectly clear, as has been pointed out by W. Müller. He says—

"The development of the optic nerve shows peculiarities in Petromyzon of such a character as to make this animal one of the most valuable objects for deciding the various controversial questions connected with the genesis of its elements. The lumen of the stalk of the primary optic vesicle is obliterated quite early by a proliferation of its lining epithelium. Also the original continuity of this epithelium with that of the pigment-layer is at an early period interrupted at the point of attachment of the optic stalk. This interruption occurs at the time when the fibres of the optic nerve first become visible."

Further on he says—

"The epithelium of the optic stalk develops entirely into supporting cells, which in Petromyzon fill up the original lumen and so form an axial core (Axenstrang) to the nerve-fibres which are formed entirely outside them; the projections of these supporting cells are directed towards the periphery, and so separate the bundles of the optic nerve-fibres. The mesodermal coat of the optic stalk takes no part in this separation; it simply forms the connective tissue sheath of the optic nerve. The development of the optic nerve in the higher vertebrates also obeys the same law, as I am bound to conclude from my own observations."

The evidence, then, of Ammocœtes is very conclusive. Originally a tube composed of a single layer of epithelial cells became expanded at the anterior end to form a bulb. On the outside of this tube or stalk the fibres of the optic nerve make their appearance, arising from the ganglion-cell layer of the retina, and, passing over the surface of the epithelial tube at the choroidal fissure, proceed to the brain by way of the optic chiasma. Owing to the large number of fibres, their crossing at the junction of the stalk with the bulb, and the narrowness at this neck, the obliteration of the lumen of the tube which takes place in the stalk is carried out to a still greater extent at this narrow part. The result of this is that all continuity of the cell-layers of the original tube of the optic stalk with those of both the inner and outer walls of the bulb is interrupted, and all that remains in this spot of the original continuous line of cells which connected the tube of the stalk with that of the bulb are possibly some of the groups of cells which are found scattered among the fibres of the optic nerve at their entrance into the retina. Such separation of the originally continuous elements of the epithelial wall of the optic stalk, which is apparent only at this neck of the nerve in Petromyzon, takes place along the whole of the optic nerve in the higher vertebrates, so that no continuous axial core of cells exist, but only scattered supporting cells.

If further proof in support of this view be wanted, it is given by the evidence of physiology, which shows that the fibres of the optic nerve are not different from other nerve-fibres of the central nervous system, but that they degenerate when separated from their nerve-cell, and that the nerve-cell of which the optic nerve-fibre is a process is the large ganglion-cell of the ganglionic layer of the retina. The origin of the ganglionic layer of the retina cannot therefore be separated from that of the optic nerve-fibres. If the one is outside the epithelial tube, so is the other, and what holds true of the ganglionic layer must hold good of the rest of the retinal ganglion and, from all that has been said, of the retina itself. We therefore come to the conclusion that the evidence is distinctly in favour of the view, that the retina and optic nerve in the true sense are structures which originally were outside a non-nervous tube, but, just like the central nervous system as a whole, have amalgamated so closely with the elements of this tube as to utilize them for supporting structures. One part of this non-nervous tube, its dorsal wall, like the corresponding part of the brain-tube, still retains its original character, and by the deposition of pigment has been pressed into the service of the eye to form the pigmented epithelial layer.

We can, however, go further than this, for we know definitely in the case of the retina what the fate of the epithelial cells lining this tube has been. They have become the system of supporting structures known as Müllerian fibres.

The epithelial layer of the primary optic vesicle can be traced into direct continuity with the lining epithelium of the brain cavity, as a single layer of epithelial cells in the core of the optic nerve, forming the optic stalk, which, in consequence of close contact, becomes the well-known axial layer of supporting cells. This epithelial layer of the optic stalk then expands to form the optic bulb, the outer or dorsal wall of which still remains as a single layer of epithelium and becomes the layer of pigment epithelium. This layer of epithelium becomes doubled on itself by the approximation of the inner or ventral wall of the optic cup to the outer or dorsal wall in consequence of the presence of the lens, and still remaining a single layer, forms the pars ciliaris retinæ; then suddenly, at the ora serrata, the single epithelial layer vanishes, and the layers of the retina take its place. It has long been known, however, that even throughout the retina this single epithelial layer still continues, being known as the fibres of Müller. This is how the fact is described in the last edition of Foster's "Text-book of Physiology," p. 1308—

"Stretching radially from the inner to the outer limiting membrane in all regions of the retina are certain peculiar-shaped bodies known as the radial fibres of Müller. Each fibre is the outcome of the changes undergone by what was at first a simple columnar epithelial cell. The changes are, in the main, that the columnar form is elongated into that of a more or less prismatic fibre, the edges of which become variously branched, and that while the nucleus is retained the cell substance becomes converted into neuro-keratin. And, indeed, at the ora serrata the fibres of Müller may be seen suddenly to lose their peculiar features and to pass into the ordinary columnar cells which form the pars ciliaris retinæ."

Fig. 44.—Diagram representing the Single-layered Epithelial Tube of the Vertebrate Eye after Removal of the Nervous and Retinal Elements. O.n., axial core of cells in optic nerve; p., pigment epithelium; p.c.r., pars ciliaris retinæ; m.f., Müllerian fibres; l., lens.

It is then absolutely clear that the essential parts of the eye may be considered as composed of two parts—

1. A tube or diverticulum from the tube of the central nervous system, composed throughout of a single layer of epithelium, which forms the supporting axial cells in the optic nerve, the pigment epithelium and the Müllerian fibres of the retina. Such a tube would be represented by the accompanying Fig. 44, and the left side of Fig. 41.

2. The retina proper with the retinal ganglion and the optic nerve-fibres as already described. In this part supporting elements are found, just as in any other compound retina, of the nature of neuroglia, which are independent of the Müllerian fibres.

Of these two parts we have already seen that the second is to all intents and purposes a compound retina of a crustacean eye, and seeing that the single-layered epithelial tube is continuous with the single-layered epithelial tube of the central nervous system—i.e. with the cephalic part of the gut of the arthropod ancestor—it follows with certainty that the ancestor of the vertebrates must have possessed two anterior diverticula of the gut, with the wall of which, near the anterior extremity, the compound retina has amalgamated on either side, just as the infra-œsophageal ganglia have amalgamated with the ventral wall of the main gut-tube. In this way, and in this way alone, does the interpretation of the structure of the vertebrate lateral eye harmonize in the most perfect manner with the rest of the conclusions already arrived at.

The question therefore arises:—Have we any grounds for believing that the ancient forms of primitive crustaceans and primitive arachnids, which were so abundant in the time when the Cephalaspids appeared, possessed two anterior diverticula of the stomach, such as the consideration of the vertebrate eye strongly indicates must have been the case?

The beautiful pictures of Blanchard, and his description, show how, on the arachnid side, paired diverticula of the stomach are nearly universal in the group. Thus, although they are not present in the scorpions, still, in the Thelyphonidæ, Phrynidæ, Solpugidæ, Mygalidæ, the most marked characteristic of the stomach-region is the presence of four pairs of cœcal diverticula, which spread laterally over the prosomatic region. In the spiders the number of such diverticula increases, and the whole prosomatic region becomes filled up with these tubes. Blanchard considers that they form nutrient tubes for the direct nutrition of the organs in the prosoma, especially the important brain-region of the central nervous system. He points out that these animals are blood-suckers, and that, therefore, their food is already in a suitable form for purposes of nutrition when it is taken in by them, so that, as it were, the anterior part of the gut is transformed into a series of vessels or diverticula conveying blood directly to the important organs in the prosoma, by means of which they obtain nourishment in addition to their own blood-supply.

The universality of such diverticula among the arachnids makes it highly probable that their progenitors did possess an alimentary canal with one or more pairs of anterior diverticula. In the vertebrate, however, the paired diverticula are associated with a compound retina, a combination which does not occur among living arachnids; we must, therefore, examine the crustacean group for the desired combination, and naturally the most likely group to examine is the Phyllopoda, especially such primitive forms as Branchipus and Artemia, for it is universally acknowledged that these forms are the nearest living representatives of the trilobites. If, therefore, it be found that the retina and optic nerve in Artemia is in specially close connection with an anterior diverticulum of the gut on each side, then it is almost certain that such a combination existed also in the trilobites.

Fig. 45.—Section through one of the two Anterior Diverticula of the Gut in Artemia and the Retinal Ganglion.

The section is through the extreme anterior end of the diverticulum, thus cutting through many of the columnar cells at right angles to their axis. Al., gut diverticulum; rt. gl., retinal ganglion.

Fig. 46.—The Brain, Eyes, and Anterior Termination of the Alimentary Canal of Artemia, viewed from the Dorsal Aspect. Br., brain; l.e., lateral eyes; c.e., median eyes; Al., alimentary canal.

Fig. 47.—A, The Formation of the Retina of the Eye of Ammocœtes (after Scott); B, The Formation of the Retina of the Eye of Ammocœtes, on my theory. R., retina; l., lens; O.n., optic nerve fibres; Al., cephalic end of invertebrate alimentary canal; V., cavity of ventricles of brain; Al.d., anterior diverticulum of alimentary canal; op.d., optic diverticulum.

My friend Mr. W. B. Hardy has especially investigated the nervous system of Artemia. In the course of his work he cut serial sections through the whole animal, and, as mentioned in my paper in the Journal of Anatomy and Physiology, he discovered that the retinal ganglion of each lateral eye is so closely attached to the end of the corresponding diverticulum of the gut that the lining cells of the ventral part of the diverticulum form a lining to the retinal ganglion (Fig. 45). In this animal there are only two gut-diverticula, which are situated most anteriorly. I have plotted out this series of sections by means of a camera lucida, with the result that the retina appears as a bulging attached ventro-laterally to the extremity of each gut-diverticulum, as is shown in Fig. 46. It is instructive to compare with this figure Scott's picture of the developing eye in Ammocœtes, where he figures the retina as a bulging attached ventrally to the extremity of the narrow tube of the optic diverticulum. In Fig. 47, A, I reproduce this figure of Scott, and by the side of it, Fig. 47, B, I have represented the origin of the vertebrate eye as I believe it to have occurred.

We see, then, this very striking fact, that in the most primitive of the Crustacea, not only are there two anterior diverticula of the gut, but also the retinal ganglion of the lateral eye is in specially close connection with the end of the diverticulum on each side. In fact, we find in the nearest living representative of the trilobites a retina and retinal ganglion and optic nerve, closely resembling that of the vertebrate, in close connection with an epithelial tube which has nothing to do with the organ of sight, but is one of a pair of anterior gut-diverticula. It is impossible to obtain more decisive evidence that the trilobites possessed a pair of gut-diverticula surrounded to a greater or less extent by the retina and optic nerve of each lateral eye.

Such anterior diverticula are commonly found in the lower Crustacea; they are usually known by the name of liver-diverticula, but as they take no part in digestion, and, on the contrary, represent that part of the gut which is most active in absorption, the term liver is not appropriate, and it is therefore better to call them simply the pair of anterior diverticula. Our knowledge of their function in Daphnia is given in a paper by Hardy and M‘Dougall, which does not appear to be widely known. Hardy succeeded in feeding Daphnia with yolk of egg in which carmine grains were mixed, and was able in the living animal to watch the whole process of deglutition, digestion, and absorption. The food, which is made into a bolus, is moved down to the middle region of the gut, and there digestion takes place. Then by an antiperistaltic movement the more fluid products of the digestion-process are sent right forward into the two anterior diverticula, where the single layer of columnar cells lining these diverticula absorbs these products, the cells becoming thickly studded with fat-drops after a feed of yolk of egg. The carmine particles, which were driven forward with the proteid- and fat-particles, are not absorbed, but are at intervals driven back by contractions of the anterior diverticula to the middle region of the gut.

These observations prove most clearly that the anterior diverticula have a special nutrient function, being the main channels by which new nutrient material is brought into the body, and, as pointed out by the authors, it is a remarkable exception in the animal kingdom that absorption should occur in that portion of the gut which is anterior to the part in which digestion occurs. In all these animals the two anterior diverticula extend forwards over the brain, and, as we have seen in Artemia, the anterior extremity of each one is so intimately related to a part of the brain—viz. the retinal ganglion—as to form a lining membrane to that mass of nerve-cells. It follows, therefore, that the nutrient fluid absorbed by the cells of this part of the gut-diverticulum must be primarily for the service of the retinal ganglion. In fact, the relations of this anterior portion of the gut to the brain as a whole suggest strongly that the marked absorptive function of this anterior portion of the gut exists in order to supply nutrient material in the first place to the most vital, most important organ in the animal—the brain and its sense-organs. This conclusion is borne out by the fact that in these lower crustaceans the circulation of blood is of a very inefficient character, so that the tissues are mainly dependent for their nutrition on the fluid immediately surrounding them. It stands to reason that the establishment of the anterior portion of the gut as a nutrient tube to the brain would necessitate a closer and closer application of the brain to that tube, so that the process of amalgamation of the brain with the single layer of columnar epithelial cells which constitutes the wall of the gut (which we see in its initial stage in the retinal ganglion of Artemia), would tend rapidly to increase as more and more demands were made upon the brain, until at last both the supra- and infra-œsophageal ganglia, as well as the retinal ganglia and optic nerves, were in such close intimate connection with the ventral wall of the anterior portion of the gut and its diverticula as to form a brain and retina closely resembling that of Ammocœtes.

Such an origin for the lateral eyes of the vertebrate explains in a simple and satisfactory manner why the vertebrate retina is a compound retina, and why both retina and optic nerve have an apparent tubular development.

At the same time one discrepancy still exists which requires consideration—viz. in no arthropod eye possessing a compound retina is the retina inverted. All the known cases of inversion among arthropods occur in eyes, the retina of which is simple, and are all natural consequences of the process of invagination by which the retina is formed. On the other hand, eyes with an inverted compound retina are not entirely unknown among invertebrates, for the eyes of Pecten and of Spondylus possess a retina which is inverted after the vertebrate fashion and still may be spoken of as compound rather than simple. It is clear that an invagination, the effect of which is an inversion of the retinal layer, would lead to the same result, whether the retinal optic nerves were short or long, whether, in fact, a retinal ganglion existed or not. Undoubtedly the presence of the retinal ganglion tends greatly to obscure any process of invagination, so that, as already mentioned, many observers, with Parker, consider the retina of the crustacean lateral eye to be formed by a thickening only, without any invagination, while Reichenbach says an obscure invagination does take place at a very early stage. So in the vertebrate eye most observers speak only of a thickening to form the retina, but Götte's observation points to an invagination of the optic plate at an early stage. So also in the eye of Pecten, Korschelt and Heider consider that the thickening, by which the retina is formed according to Patten, in reality hides an invagination process by means of which, as Bütschli suggests, an optic vesicle is formed in the usual manner. The retina is formed from the anterior wall of this vesicle, and is therefore inverted.

The origin of the inverted retina of the vertebrate eye does not seem to me to present any great difficulty, especially when one takes into consideration the fact that the retina is inverted in the arachnid group, only in the lateral eyes. The inversion is usually regarded as associated with the tubular formation of the vertebrate retina, and it is possible to suppose that the retina became inverted in consequence of the involvement of the eye with the gut-diverticulum. I do not myself think any such explanation is at all probable, because I cannot conceive such a process taking place without a temporary derangement—to say the least of it—of the power of vision, and as I do not believe that evolution was brought about by sudden, startling changes, but by gradual, orderly adaptations, and as I also believe in the paramount importance of the organs of vision for the evolution of all the higher types of the animal kingdom, I must believe that in the evolution from the Arthropod to the Cephalaspid, the lateral eyes remained throughout functional. I therefore, for my own part, would say that the inversion of the retina took place before the complete amalgamation with the gut-diverticulum, that, in fact, among the proto-crustacean, proto-arachnid forms there were some sufficiently arachnid to have an inverted retina, and at the same time sufficiently crustacean to possess a compound retina, and therefore a compound inverted retina after the vertebrate fashion existed in combination with the anterior gut-diverticula. Thus, when the eye and optic nerve sank into and amalgamated with the gut-diverticulum, neither the dioptric apparatus nor the nervous arrangements would suffer any alteration, and the animal throughout the whole process would possess organs of vision as good as before or after the period of transition.

Further, not only the retina but also the dioptric apparatus of the vertebrate eye point to its origin from a type that combined the peculiarities of the arachnids and the crustaceans. In the former it is difficult to speak of a true lens, the function of a lens being undertaken by the cuticular surface of the cells of the corneagen (Mark's 'lentigen'), while in the latter, in addition to the corneal covering, a true lens exists in the shape of the crystalline cones. Further, these crustacean lenses are true lenses in the vertebrate sense, in that they are formed by modified hypodermal cells, and not bulgings of the cuticle, as in the arachnid. We see, in fact, that in the compound crustacean eye an extra layer of hypodermal cells has become inserted between the cornea and the retina to form a lens. So also in the vertebrate eye the lens is formed by an extra layer of the epidermal cells between the cornea and the retina. The fact that the vertebrate eye possesses a single lens, though its retina is composed of a number of ommatidia, while the crustacean eye possesses a lens to each ommatidium, may well be a consequence of the inversion of the vertebrate retina. It is most probable, as Korschelt and Heider have pointed out, that the retina of the arachnid eyes is composed of a number of ommatidia, just as in the crustacean eyes and in the inverted eyes it is probable that the image is focussed on to the pigmented tapetal layer, and thence reflected on to the percipient visual rods. In such a method of vision a single lens is a necessity, and so it must also be if, as I suppose, eyes existed with an inverted compound retina. Owing to the crustacean affinities of such eyes, a lens would be formed and the retina would be compound: owing to the arachnid affinities, the retina would be inverted and the hypodermal cells which formed the lens would be massed together to form a single lens, instead of being collected in groups of four to form a series of crystalline cones.

To sum up: The study of the vertebrate eyes, both median and lateral, leads to most important conclusions as to the origin of the vertebrates, for it shows clearly that whereas, as pointed out in this and subsequent chapters, their ancestors possessed distinct arachnid characteristics, yet that they cannot have been specialized arachnids, such as our present-day forms, but rather they were of a primitive arachnid type, with distinct crustacean characteristics: animals that were both crustacean and arachnid, but not yet specialized in either direction: animals, in fact, of precisely the kind which swarmed in the seas at the time when the vertebrates first made their appearance. In the opinion of the present day, the ancestral forms of the Crustacea, which were directly derived from the Annelida, may be classed as an hypothetical group the Protostraca, the nearest approach to which is a primitive Phyllopod.

"Starting from the Protostraca," say Korschelt and Heider, "according to the present condition of our knowledge, we may, as has been already remarked, assume three great series of development of the Arthropodan stock, by the side of which a number of smaller independent branches have been retained. One of these series leads through the hypothetical primitive Phyllopod to the Crustacea; the second through the Palæostraca (Trilobita, Gigantostraca, Xiphosura) to the Arachnida; the third through forms resembling Peripatus to the Myriapoda and the Insecta. The Pantapoda and the Tardigrada must probably be regarded as smaller independent branches of the Arthropodan stock."

To these "three great series of development of the Arthropodan stock" the evidence of Ammocœtes shows that a fourth must be added, which, starting also from the Protostraca, and closely connected with the second, palæostracan branch, leads through the Cephalaspidæ to the great kingdom of the Vertebrata. Such a direct linking of the earliest vertebrates with the Annelida through the Protostraca is of the utmost importance, as will be shown later in the explanation of the origin of the vertebrate cœlom and urinary apparatus.

Summary.

The most important discovery of recent years which gives a direct clue to the ancestry of the vertebrates is undoubtedly the discovery that the pineal gland is all that remains of a pair of median eyes which must have been functional in the immediate ancestor of the vertebrate, seeing how perfect one of them still is in Ammocœtes. The vertebrate ancestor, then, possessed two pairs of eyes, one pair situated laterally, the other median. In striking confirmation of the origin of the vertebrate from Palæostracans it is universally admitted that all the Eurypterids and such-like forms resembled Limulus in the possession of a pair of median eyes, as well as of a pair of lateral eyes. Moreover, the ancient mailed fishes the Ostracodermata, which are the earliest fishes known, are all said to show the presence of a pair of median eyes as well as of a pair of lateral eyes. This evidence directly suggests that the structure of both the median and lateral vertebrate eyes ought to be very similar to that of the median and lateral arthropod eyes. Such is, indeed, found to be the case.

The retina of the simplest form of eye is formed from a group of the superficial epidermal cells, and the rods or rhabdites are formed from the cuticular covering of these cells; the optic nerve passes from these cells to the deeper-lying brain. This kind of retina may be called a simple retina, and characterizes the eyes, both median and lateral, of the scorpion group.

In other cases a portion of the optic ganglion remains at the surface, when the brain sinks inwards, in close contiguity to the epidermal sense-cells which form the retina; a tract of fibres connects this optic ganglion with the underlying brain, and is known as the optic nerve. Such a retina may be called a compound retina and characterizes the lateral eyes of both crustaceans and vertebrates. Also, owing to the method of formation of the retina by invagination, the cuticular surface of the retinal sense-cells, from which the rods are formed, may be directed towards the source of light or away from it. In the first case the retina may be called upright, in the second inverted.

Such inverted retinas are found in the vertebrate lateral eyes and in the lateral eyes of the arachnids, but not of the crustaceans.

The evidence shows that all the invertebrate median eyes possess a simple upright retina, and in structure are remarkably like the right median or pineal eye of Ammocœtes; while the lateral eyes possess, as in the crustaceans, an upright compound retina, or, as in many of the arachnids, a simple inverted retina. The lateral eyes of the vertebrates alone possess a compound inverted retina.

This retina, however, is extraordinarily similar in its structure to the compound crustacean retina, and these similarities are more accentuated in the retina of the lateral eye of Petromyzon than that of the higher vertebrates.

The evidence afforded by the lateral eye of the vertebrate points unmistakably to the conclusion that the ancestor of the vertebrate possessed both crustacean and arachnid characters—belonged, therefore, to a group of animals which gave rise to both the crustacean and arachnid groups. This is precisely the position of the Palæostracan group, which is regarded as the ancestor of both the crustaceans and arachnids. In two respects the retina of the lateral eyes of vertebrates differs from that of all arthropods, for it possesses a special supporting structure, the Müllerian fibres, which do not exist in the latter, and it is developed in connection with a tube, the optic diverticulum, which is connected on each side with the main tube of the central nervous system. These two differences are in reality one and the same, for the Müllerian fibres are the altered lining cells of the optic diverticulum, and this tube has the same significance as the rest of the tube of the nervous system; it is something which has nothing to do with the nervous portion of the retina but has become closely amalgamated with it. The explanation is, word for word, the same as for the tubular nervous system, and shows that the ancestor of the vertebrate possessed two anterior diverticula of its alimentary canal which were in close relationship to the optic ganglion and nerve of the lateral eye on each side. It is again a striking coincidence to find that Artemia, which with Branchipus represents a group of living crustaceans most nearly allied to the trilobites, does possess two anterior diverticula of the gut which are in extraordinarily close relationship with the optic ganglia of the retina of the lateral eyes on each side.

The evidence of the optic apparatus of the vertebrate points most remarkably to the derivation of the Vertebrata from the Palæostraca.