CHAPTER VIII.

DEVELOPMENT OF THE SPINAL NERVES AND OF THE SYMPATHETIC NERVOUS SYSTEM.

_The spinal nerves._

The development of the spinal nerves has been already treated by me at considerable length in a paper read before the Royal Society in December, 1875[247], and I have but little fresh matter to add to the facts narrated in that paper. The succeeding account, though fairly complete, is much less full than the previous one in the _Philosophical Transactions_, but a number of morphological considerations bearing on this subject are discussed.

Footnote 247: _Phil. Trans._ Vol. 166, p. 175. [This Edition, No. VIII.]

The rudiments of the posterior roots make their appearance considerably before those of the anterior roots. They arise during stage I, as outgrowths from the spinal cord, at a time when the muscle-plates do not extend beyond a third of the way up the sides of the spinal cord, and in a part where no scattered mesoblast-cells are present. They are formed first in the anterior part of the body and successively in the posterior parts, in the following way. At a point where a spinal nerve is about to arise, the cells of the dorsal part of the cord begin to proliferate, and the uniform outline of the cord becomes broken (Pl. 14, fig. 3). There is formed in this way a small prominence of cells springing from the summit of the spinal cord, and constituting a rudiment of a pair of posterior roots. In sections anterior to the point where a nerve is about to appear, the nerve-rudiments are always very distinctly formed. Such a section is shewn in Pl. 14, fig. 2, and the rudiments may there be seen as two club-shaped masses of cells, which have grown outwards and downwards from the extreme dorsal summit of the neural canal and in contact with its walls. The rudiments of the two sides meet at their point of origin at the dorsal median line, and are dorsally perfectly continuous with the walls of the canal.

It is a remarkable fact that rudiments of posterior roots are to be seen in every section. This may be interpreted as meaning that the rudiments are in very close contact with each other, but more probably means, as I hope to shew in the sequel, that there arises from the spinal cord a continuous outgrowth from which discontinuous processes (the rudiments of posterior roots) grow out.

After their first formation these rudiments grow rapidly ventralwards in close contact with the spinal cord (vide Pl. 14, fig. 1, and Pl. 11, figs. 6 and 7), but soon meet with and become partially enclosed in the mesoblastic tissue (Pl. 11, fig. 7). The similarity of the mesoblast and nerve-tissue in Scyllium and Pristiurus embryos hardened in picric or chromic acid, render the nerves in these genera, at the stage when they first become enveloped in mesoblast, difficult objects to observe; but no similar difficulty is encountered in the case of Torpedo embryos.

While the rudiments of the posterior roots are still quite short, those of the anterior roots make their first appearance. Each of these (Pl. 14, fig. 4, _a.r._) arises as a very small but distinct conical outgrowth from a ventral corner of the spinal cord. From the very first the rudiments of the anterior roots have an indistinct form of peripheral termination and somewhat fibrous appearance, while the protoplasm of which they are composed becomes attenuated towards its end. The points of origin of the anterior roots from the spinal cord are separated by considerable intervals. In this fact, and also in the fact of the nerves of the two sides never being united with each other in the median line, the anterior roots exhibit a marked contrast to the posterior. There are thus constituted, before the close of stage I, the rudiments of both the anterior and posterior roots of the spinal nerves. The rudiments of both of these take their origin from the involuted epiblast of the neural canal, and the two roots of each spinal nerve are at first quite unconnected with each other. It is scarcely necessary to state that the pairs of roots correspond in number with the muscle-plates.

It is not my intention to enter with any detail into the subsequent changes of the rudiments whose origin has been described, but a few points especially connected with their early development are sufficiently important to call for attention.

One feature of the posterior roots at their first formation is the fact that they appear as processes of a continuous outgrowth of the spinal cord. This state of affairs is not of long continuance, and before the close of stage I each posterior root has a separate junction with the spinal cord. What then becomes of the originally continuous outgrowth? It has not been possible for me to trace the fate of this step by step; but the discovery that at a slightly later period (stage K) there is present a continuous commissure independent of the spinal cord connecting the dorsal and central extremities of all the spinal nerves, renders it very probable that the original continuous outgrowth becomes converted into this commissure. Like all the other nervous structures, this commissure is far more easily seen in embryos hardened in a mixture of osmic and chromic acids or osmic acid, than in those hardened in picric acid. Its existence must be regarded as one of the most remarkable results of my researches upon the Elasmobranch nervous system. At stage K it is fairly thick, though it becomes much thinner at a slightly later period. Its condition during stage K is shewn in Pl. 12, fig. 18, _com_. What it has been possible for me to make out of its eventual fate is mentioned subsequently[248].

Footnote 248: It is not by any means always possible to detect this commissure in transverse sections. As I have suggested, in connection with a similar commissure connecting the vagus branches, it perhaps easily falls out of the section, and is always so small that the hole left would certainly be invisible.

A second feature of the earliest condition of the posterior roots is their attachment to the extreme dorsal summit of the spinal cord--a point of attachment very different from that which they eventually acquire. Before the commencement of stage K this state of things has become altered; and the posterior roots spring from the spinal cord in the position normal for Vertebrates.

This apparent migration caused me at first great perplexity, and I do not feel quite satisfied that I have yet got completely to the bottom of its meaning. The explanation which appears to me most probable has suggested itself in the course of some observations on the development of the thin roof of the fourth ventricle. A growth of cells appears to take place in the median dorsal line of the roof of the spinal cord. This growth tends to divaricate the two lateral parts of the cord, which are originally contiguous in the dorsal line, and causes therefore the posterior roots, which at first spring from the dorsal summit, to assume an apparent attachment to the side of the cord at some little distance from the summit. If this is the true explanation of the change of position which takes place, it must be regarded as due rather to peculiar growths in the spinal cord, than to any alteration in the absolute attachment of the nerves.

By stage K the rudiment of the posterior root has become greatly elongated, and exhibits a division into three distinct portions (Pl. 14, fig. 6):

(1) A proximal portion, in which is situated the pedicle of attachment to the wall of the neural canal.

(2) An enlarged portion, which may conveniently from its future fate be called the spinal ganglion.

(3) A distal portion beyond this.

The proximal portion presents a fairly uniform diameter, and ends dorsally in a rounded expansion; it is attached, remarkably enough, _not by its extremity, but by its side, to the spinal cord. The dorsal extremities of the posterior roots are therefore free._ It seems almost certain that the free dorsal extremities of these roots serve as the starting points for the dorsal commissure before mentioned, which connects the roots together. The attachment of the posterior nerve-root to the spinal cord is, on account of its small size, very difficult to observe. In favourable specimens there may however be seen a distinct cellular prominence from the spinal cord, which becomes continuous with a small prominence on the lateral border of the nerve-root near its distal extremity. The proximal extremity of the rudiment is composed of cells, which, by their small size and circular form, are easily distinguished from those which form the succeeding or ganglionic portion of the nerve. This succeeding part has a swollen configuration, and is composed of large elongated cells with oval nuclei. The remainder of the rudiment forms the commencement of the true nerve.

The anterior root, which, at the close of stage I, formed a small and inconspicuous prominence from the spinal cord, grows rapidly during the succeeding stages, and soon forms an elongated cellular structure with a wide attachment to the spinal cord (Pl. 14, fig. 5). At first it passes obliquely and nearly horizontally outwards, but, before reaching the muscle-plate of its side, takes a bend downwards (Pl. 14, fig. 7).

I have not definitely made out when the anterior and posterior roots unite, but this may easily be seen to take place before the close of stage K (Pl. 12, fig. 18).

One feature of some interest with reference to the anterior roots, is the fact that they arise not vertically below, but alternately with the dorsal roots, a condition which persists in the adult.

Although I have made some efforts to determine the eventual fate of the commissure uniting the dorsal roots, these have not hitherto been crowned with success. It grows thinner and thinner, becoming at the same time composed of fibrous protoplasm with imbedded nuclei (Pl. 14, figs. 8 and 9). By stage M it is so small as to be quite indistinguishable in transverse sections; and I have failed in stage P to recognize it at all. I can only conclude that it gradually atrophies, and finally vanishes without leaving a trace. Both its appearance and history are very remarkable, and deserve the careful attention of future investigators.

There can be little doubt that it is some sort of remnant of an ancestral structure in the nervous system; and it would appear to indicate that the central nervous system must originally have been formed of a median and two lateral strands. At the same time I very much doubt whether it can be brought into relation with the three rows of ganglion-cells (a median and two lateral) which are so frequently present on the ventral side of annelidan nerve-cords.

_My results may be summarised as follows_:--Along the extreme dorsal summit of the spinal cord there arises on each side a continuous outgrowth. From each outgrowth processes corresponding in number to the muscle-plates grow downwards. These are the rudiments of the posterior nerve-roots. The outgrowths, though at first attached to the spinal cord throughout their whole length, soon cease to be so, and remain in connection with it at certain points only, which form the primitive junctions of the posterior roots with the spinal cord. The original outgrowth on each side remains as a bridge, uniting together the dorsal extremities of all the posterior roots. The posterior roots, though primitively attached to the dorsal summit of the spinal cord, eventually come to arise from its sides. The original homogeneous rudiments before the close of stage K become differentiated into a root, a ganglion, and a nerve.

The anterior roots, like the posterior, are outgrowths from the spinal cord, but are united independently with it, and the points from which they spring originally, remain as those by which they are permanently attached. The anterior roots arise, not vertically below, but in the intervals between the posterior roots. They are at first quite separate from the posterior roots; but before the close of stage K a junction is effected between each posterior root and the corresponding anterior root. The anterior root joins the posterior at some little distance below its ganglion.

* * * * *

The results here arrived at are nearly in direct opposition to those of the majority of investigators, though in accordance, at least so far as the posterior roots are concerned, with the beautiful observations of Hensen 'on the Development of Mammalia[249].'

Footnote 249: _Zeit. f. Anat. u. Entwicklungsgeschichte_, Vol. I.

Mr Marshall[250] has more recently published a paper on the development of the nerves in Birds, in which he shews in a most striking manner that the observations recorded here for Elasmobranchii hold good for the posterior roots of Birds. The similarity between his figures and my own is very noticeable. A further discussion of the literature would be quite unprofitable, and I proceed at once to certain considerations suggested by the above observations.

Footnote 250: _Journal of Anatomy and Physiology_, Vol. XI. April, 1877.

_General considerations._ One point of general anatomy upon which my observations throw considerable light, is the _primitive origin of nerves_. So long as it was admitted that the spinal and cerebral nerves developed in the embryo independently of the central nervous system, their mode of origin always presented to my mind considerable difficulties. It never appeared clear how it was possible for a state of things to have arisen in which the central nervous system as well as the peripheral terminations of nerves, whether motor or sensory, were formed independently of each other; while between them a third structure was developed, which, growing out either towards the centre or towards the periphery, ultimately brought the two into connection. That such a condition could be a primitive one seemed scarcely possible.

Still more remarkable did it appear, on the supposition that the primitive mode of formation of these parts was represented in the developmental history of Vertebrates, that we should find similar structural elements in the central and in the peripheral nervous systems. The central nervous system arises from the epiblast, and yet contains precisely similar nerve-cells and nerve-fibres to the peripheral nervous system, which, when derived from the mesoblast, was necessarily supposed to have an origin completely different from that of the central nervous system. Both of these difficulties are to a great extent removed by the facts of the development of these parts in Elasmobranchii.

It is possible to suppose that in their primitive differentiation contractile and sensory systems may, as in Hydra[251], have been developed from the protoplasm of even the same cell. As the sensory and motor systems became more complicated, the sensory portion of a cell would become separated by an increasing interval from the muscular part of a cell, and the two parts of a cell would only be connected by a long protoplasmic process. When such a condition as that was reached, the sensory portion of the cell would be called a ganglion-cell or terminal sensory organ, the connecting process a nerve, and the contractile portion of the cell a muscle-cell. When these organs were in this condition, it might not impossibly happen for the general developmental growth which tended to separate the ganglion-cell and the muscle-cell to be so rapid as to render it impossible for the growth of the connecting nerve to keep pace with it, and that thus the process connecting the ganglion-cell and the muscle-cell might become ruptured. Nevertheless the tendency of the process to grow from the ganglion-cell to the muscle-cell, would remain, and when the rapid developmental growth had ceased, the two would become united again by the growth of the process which had previously been ruptured. It will be seen that this hypothesis, which I have considered only with reference to a single nerve and muscle-cell, might be extended so as to apply to a complicated central nervous system and peripheral nerves and muscles, and also could apply equally as well to the sensory as to the motor terminations of a nerve. In the case of the sensory termination, we should only have to suppose that the centre nervous cell became more and more separated by the general growth from the recipient terminal sensory cell, and that during the general growth the connection between the two was mechanically ruptured but restored again on the termination of the more rapid growth.

Footnote 251: Kleinenberg Hydra.

As the descendants of the animal in which the rupture occurred became progressively more complicated, the two terminal cells must have become widely separated at a continually earlier period, till finally they may have been separated at a period of development when they were indistinguishable from the surrounding embryonic cells; and since the rupture would also occur at this period, the primitive junction between the nerve-centre and termination would escape detection. The object of this hypothesis is to explain the facts, so far as they are known, of the development of the nervous system in Vertebrates.

In Vertebrates we certainly appear to have an outgrowth from the nervous system, which eventually becomes united with the muscle or sensory terminal organs. The ingenious hypothetical scheme of development of the nerves given by Hensen[252] would be far preferable to the one suggested if it could be brought into conformity with the facts. There is, however, at present no evidence for Hensen's view, as he himself admits, but considering how little we know of the finer details of the development of nerves, it seems not impossible that such evidence may be eventually forthcoming. The evidence from my own observation is, so far as it goes, against it. At a time anterior to the outgrowth of the spinal nerves, I have shewn[253] that the spinal cord is completely invested by a delicate hyaline membrane. It is difficult to believe that this is pierced by a number of fine processes, which completely escape detection, but which must, nevertheless, be present on the hypothesis of Hensen.

Footnote 252: Virchow's _Archiv_, Vol. XXXI. 1864.

Footnote 253: _Phil. Trans._, 1876. [This Edition, No. VIII.]

The facts of the development of nerves in Vertebrates are unquestionably still involved in considerable doubt. It may, I think, be considered as certain, that in Elasmobranchii the roots of the spinal and cranial nerves are outgrowths of the central nervous system. How the final terminations of the nerves are formed is, however, far from being settled. Götte[254], whose account of the development of the spinal ganglia is completely in accordance with the ordinary views, yet states[255] that the growth of the nerve fibres themselves is a centrifugal one from the ganglia. My own investigations prove that the ganglia have a centrifugal development, and also appear to demonstrate that the nerves themselves near the ganglion have a similar manner of growth. Moreover, the account given in the preceding chapter of the manner in which the nerves become connected with the mucous canals of the head, goes far to prove that the whole growth of the nerves is a centrifugal one. The combination of all these converging observations tells strongly in favour of this view.

Footnote 254: _Entwicklungsgeschichte der Unke._

Footnote 255: _Loc. cit._ p. 516.

On the other hand, Calberla[256] believes that in the tails of larval Amphibians he has seen connective-tissue cells unite with nerve-processes, and become converted into nerves, but he admits that he cannot definitely prove that the axis-cylinder has not a centrifugal growth, while the connective-tissue cells merely become converted into the sheath of the nerve. If Calberla's view be adopted, that the nerves are developed directly out of a chain of originally indifferent cells, each cell of the chain being converted in turn into a section of the nerve, an altogether different origin of nerves from that I have just suggested would seem to be indicated.

Footnote 256: _Archiv für Micros. Anat._, Vol. XI. 1875.

The obvious difficulty, already alluded to, of understanding how it is, according to the generally accepted mode of development of the spinal nerves, that precisely similar nerve-cells and nerves should arise in structures which have such different origins as the central nervous system and the spinal nerves, is completely removed if my statements on the development of the nerves in Elasmobranch represent the truth.

One point brought out in my investigations appears to me to have bearings upon the origin of the central canal of the vertebrate nervous system, and in consequence upon the origin of the vertebrate nervous system itself. This point is, that the posterior nerve-rudiments make their first appearance at the extreme dorsal summit of the spinal cord. The transverse section of the ventral nervous cord of an ordinary segmented Annelid consists of two symmetrical halves placed side by side. If by a mechanical folding the two lateral halves of the nervous cord became bent towards each other, while into the groove between the two the external skin became pushed, we should have an approximation to the vertebrate nervous system. Such a folding as this might take place to give extra rigidity to the body in the absence of a vertebral column.

If this folding were then completed in such a way that the groove, lined by external skin and situated between the two lateral columns of the nervous system, became converted into a canal, above and below which the two columns of the nervous system united, we should have in the transformed nervous cord an organ strongly resembling the spinal cord of Vertebrates.

It is well known that the nerve-cells are always situated on the ventral side of the abdominal nerve-cord of Annelids, either as a continuous layer, or in the form of two, or more usually, three bands. The dorsal side of the cord is composed of nerve-fibres or white matter. If the folding I have supposed were to take place in the Annelid nervous cord, the grey and white matters would have very nearly the same relative situations as they have in the Vertebrate spinal cord. The grey matter would be situated in the interior and line the central canal, and the white matter would nearly surround the grey. The nerves would then arise, not from the sides of the nervous cord as in existing Annelids, but from its extreme ventral summit. One of the most striking features which I have brought to light with reference to the development of the posterior roots, is the fact of their growing out from the extreme dorsal summit of the neural canal, a position analogous to the ventral summit of the Annelidan nervous cord. Thus the posterior roots of the nerves in Elasmobranchii[257] arise, in the exact manner which might have been anticipated, were the spinal canal due to such a folding as I have suggested.

Footnote 257: There are strong reasons for regarding the posterior roots as the primitive ones. These are spoken of later, but I may state that they depend:

(1) On the fact that only _posterior_ roots exist in the brain.

(2) That only posterior roots exist in Amphioxus.

(3) That the posterior roots develop at an earlier period than the anterior.

The argument from the position of the outgrowth of nerves becomes the more striking from its great peculiarity, and forms a feature which would be most perplexing without some such explanation as I have proposed. The central epithelium of the neural canal, according to this view, represents the external skin, and its ciliation in certain cases may, perhaps, be explained as a remnant of the ciliation of the external skin still found amongst many of the lower Annelids.

I have employed the comparison of the Vertebrate and Annelidan nervous cords, not so much to prove a genetic relation between the two, as to shew the _à priori_ possibility of the formation of a spinal cord, and the _à posteriori_ evidence we have of the vertebrate canal having been formed in the way indicated. I have not made use of what is really my strongest argument, viz. that the embryological mode of formation of the spinal canal by a folding in of the external epiblast is the very method by which I supposed the spinal canal to have been formed in the ancestors of Vertebrates. My object has been to suggest a meaning for the peculiar primitive position of the posterior roots, rather than to attempt to explain in full the origin of the spinal canal.

Although the homologies between the Vertebrate and the Annelidan nervous systems are not necessarily involved in the questions which arise with reference to the formation of the spinal canal, they have nevertheless considerable bearings on it.

Two views have recently been put forward on this subject. Professor Gegenbaur[258] looks upon the central nervous system of Vertebrates as equivalent to the superior oesophageal ganglia of Annelids and Arthropods only, while Professors Leydig[259] and Semper[260] and Dr Dohrn[261] compare it with the whole Annelidan nervous system.

Footnote 258: _Grundriss d. vergleichenden Anat._ p. 264.

Footnote 259: _Bau des thierischen Körpers._

Footnote 260: _Stammesverwandtschaft d. Wirbelthiere u. Wirbellosen_ and _Die Verwandtschaftsbeziehungen d. gegliederten Thiere_. This latter work, for a copy of which I return my best thanks to the author, came into my hands after what follows was written, and I much regret only to have been able to make one or two passing allusions to it. The work is a most important contribution to the questions about to be discussed, and contains a great deal that is very suggestive; some of the conclusions with reference to the Nervous System appear to me however to be directly opposed to the observations on Spinal Nerves above recorded.

Footnote 261: _Ursprung d. Wirbelthiere u. Princip des Functionswechsels._

The first of these two views is only possible on the supposition that Vertebrates are descended from unsegmented ancestors, and even then presents considerable difficulties. If the ancestors of Vertebrates were segmented animals, and several of the recent researches tend to shew that they were, they must almost certainly have possessed a nervous cord like that of existing Annelids. If such were the case, it is almost inconceivable that the greater portion of the nervous system which forms the ventral cord can have become lost, and the system reduced to the superior oesophageal ganglia. Dr Dohrn[262], who has speculated very profoundly on this matter, has attempted to explain and remove some of the difficulties which arise in comparing the nervous systems of Vertebrates and Annelids. He supposes that the segmented Annelids, from which Vertebrates are descended, were swimming animals. He further supposes that their alimentary canal was pierced by a number of gill-slits, and that the anterior amongst these served for the introduction of nutriment into the alimentary canal, in fact as supplementary mouths as well as for respiration. Eventually the old mouth and throat atrophied, and one pair of coalesced gill-slits came to serve as the sole mouth. Thus it came about that on the disappearance of that portion of the alimentary canal, which penetrated the oesophageal nervous ring, the latter structure ceased to be visible as such, and no part of the alimentary canal was any longer enclosed by a commissure of the central nervous system. With the change of mouth Dr Dohrn also supposes that there took place a change, which would for a swimming animal be one of no great difficulty, of the ventral for the dorsal surface. This general explanation of Dr Dohrn's, apart from the considerable difficulty of the fresh mouth, appears to me to be fairly satisfactory. Dr Dohrn has not however in my opinion satisfactorily dealt with the questions of detail which arise in connection with this comparison. One of the most important points for his theory is to settle the position where the nervous system was formerly pierced by the oesophagus. This position he fixes in the fourth ventricle, and supports his hypothesis by the thinness of the roof of the spinal canal in this place, and the absence (?) of nervous structures in it.

Footnote 262: _Loc. cit._

It appears to me that this thinness cannot be used as an argument. In the first place, if the hypothesis I have suggested as to the formation of the spinal canal be accepted, the formation of the canal must be supposed to have occurred in point of time either after or before the loss of the primitive mouth. If, on the one hand, the spinal canal made its appearance before the atrophy of the primitive mouth, the folding to form it must necessarily have ceased behind the mouth; and, on the supposition of the oesophageal ring having been situated in the region of the fourth ventricle, a continuation of the spinal canal could not be present in front of this part. If, on the other hand, the cerebro-spinal canal appeared after the disappearance of the primitive mouth, its roof must necessarily also be a formation subsequent to the atrophy of the mouth, and varieties of structure in it can have no bearing upon the previous position of the mouth.

But apart from speculations upon the origin of the spinal cord, there are strong arguments against Dr Dohrn's view about the fourth ventricle. In the first place, were the fourth ventricle to be the part of the nervous system which previously formed the oesophageal commissures, we should expect to find the opening in the nervous system at this point to be visible at an early period of development, and at a later period to cease to be so. The reverse is however the case. In early embryonic life the roof of the fourth ventricle is indistinguishable from other parts of the nervous system, and only thins out at a later period. Further than this, any explanation of the thin roof of the fourth ventricle ought also to elucidate the nearly similar structure in the sinus rhomboidalis, and cannot be considered satisfactory unless it does so.

The peculiarities of the cerebro-spinal canal in the region of the brain appear to me to present considerable difficulties in the way of comparing the central nervous system of Vertebrates and segmented Annelids. The manner in which the cerebro-spinal canal is prolonged into the optic vesicles, the cerebral and the optic lobes is certainly opposed both to an intelligible explanation of the spinal canal itself, and also to a comparison of the two nervous systems under consideration.

Its continuation into the cerebral hemispheres and into the optic lobes (mid-brain) may perhaps be looked upon as due to peculiar secondary growths of those two ganglia, but it is very difficult to understand its continuation into the optic vesicles.

If it be granted that the spinal canal has arisen from a folding in of the external skin, then the present inner surface of the optic vesicle must also have been its original outer surface, and it follows as a necessary consequence that the present position of the rods and cones behind and not in front of the nervous structures of the retina was not the primitive one. The rods and cones arise, as is well known, from the inner surface of the outer portion of the optic vesicle, and must, according to the above view, be supposed originally to have been situated on the external surface, and have only come to occupy their present position during the folding in, which resulted in the spinal canal. On _à priori_ grounds we should certainly expect the rods and cones to have resulted from the differentiation of a layer of cells external to the conducting nervous structures. The position of the rods and cones posterior to these suggests therefore that some peculiar infolding has occurred, and may be used as an argument to prove that the medullary groove is no mere embryonic structure, but the embryonic repetition of an ancestral change. The supposition of such a change of position in the rods and cones necessarily implies that the folding in to form the spinal canal must have been a very slow one. It must have given time to the refracting media of the eye gradually to travel round, so as still to maintain their primitive position, while in successive generations a rudimentary spinal furrow carrying with it the retina became gradually converted into a canal[263].

Footnote 263: Professor Huxley informs me that he has for many years entertained somewhat similar views to those in the text about the position of the rods and cones, and has been accustomed to teach them in his lectures.

If Dr Dohrn's comparison of the vertebrate nervous system with that of segmented Annelids be accepted, the following two points must in my opinion be admitted:--

(1) That the formation of the cerebro-spinal canal was subsequent to the loss of the old mouth.

(2) That the position of the old mouth is still unknown.

The well-known view of looking at the pituitary and pineal growths as the remnants of the primitive oesophagus, has no doubt some features to recommend it. Nearly conclusive against it is the fact that the pituitary involution is not, as used to be supposed, a growth towards the infundibulum of the hypoblast of the oesophagus, but of the epiblast of the mouth. It is almost inconceivable that an involution from the present mouth can have assisted in forming part of the old oesophagus.

There is a view not involving the difficulty of the oesophageal ring, fresh mouth[264], and of the change of the ventral to the dorsal surface, which, though so far unsupported by any firm basis of observed facts, nevertheless appears to me worth suggesting. It assumes that Vertebrates are descended _not_ through the present line of segmented Vermes, but through some other line which has now, so far as is known, completely vanished. This line must be supposed to have originated from the same _unsegmented Vermes_ as the present segmented Annelids. They therefore acquired fundamentally similar segmental and other Annelidan organs.

Footnote 264: Professor Semper ("Die Verwandtschaftsbeziehungen d. gegliederten Thiere," _Arbeiten aus d. Zool.-zoot. Institut_, Würzburg, 1876) has some interesting speculations on the difficult question of the vertebrate mouth, which have unfortunately come to my knowledge too late to be either fully discussed or incorporated in the text. These speculations are founded on a comparison of the condition of the mouth in Turbellarians and Nemertines. He comes to the conclusion that there was a primitive mouth on the cardiac side of the supra-oesophageal ganglion, which is the existing mouth of Turbellarians and Vertebrates and the opening of the proboscis of Nemertines, but which has been replaced by a fresh mouth on the neural side in Annelids and Nemertines. In Nemertines however the two mouths co-exist--the vertebrate mouth as the opening of the proboscis, and the Annelid mouth as the opening for the alimentary tract. This ingenious hypothesis is supported by certain anatomical facts, which do not appear to me of great weight, but for which the reader must refer to the original paper. It no doubt avoids the difficulty of the present position of the vertebrate mouth, but unfortunately at the same time substitutes an equal difficulty in the origin of the Annelidan mouth. This Professor Semper attempts to get over by an hypothesis which to my mind is not very satisfactory (p. 378), which, however, and this Professor Semper does not appear to have noticed, _could equally well be employed to explain the origin of a Vertebrate mouth as a secondary formation subsequent to the Annelidan mouth_. Under these circumstances this fresh hypothesis does not bring us very much nearer to a solution of the vertebrate-annelid mouth question, but merely substitutes one difficulty for another; and does not appear to me so satisfactory as the hypothesis suggested in the text.

At the same time Professor Semper's hypothesis suggests an explanation of that curious organ the Nemertine proboscis. If the order of changes suggested by him were altered it might be possible to suppose that there never was more than one mouth for all Vermes, but that the proboscis in Nemertines gradually split itself off from the oesophagus to which it originally belonged, and became quite free and provided with a separate opening and perhaps carried with it the so-called vagus of Professors Semper and Leydig.

The difference between the two branches of the Vermes lay in the nervous system. The unsegmented ancestors of the _present_ Annelids seem to have had a pair of super-oesophageal ganglia, from which two main nervous stems extended backwards, one on each side of the body. Such a nervous system in fact as is possessed by existing Nemertines or Turbellarians[265]. As the Vermes became segmented and formed the Annelids, these side nerves seem to have developed ganglia, corresponding in number with the segments, and finally, approximating on the ventral surface, to have formed the ventral cord[266].

Footnote 265: It is not of course to be supposed that the primitive nervous system was pierced by a proboscis like that of the Nemertines.

Footnote 266: This is Gegenbaur's view of the development of the ventral cord, and I regard it in the meantime as the most probable view which has been suggested.

The other branch of Vermes which I suppose to have been the ancestors of Vertebrates started from the same stock as existing Annelids, but I conceive the lateral nerve-cords, instead of approximating ventrally, to have done so dorsally, and thus a dorsal cord to have become formed analogous to the ventral cord of living Annelids, only without an oesophageal nerve-ring[267].

Footnote 267: A dorsal instead of a ventral approximation of the lateral nerve-cords would be possible in the descendants of such living segmented Vermes as Saccocirrus and Polygordius.

It appears to me, (if the difficulties of comparing the Annelidan ventral cord with the spinal cord of Vertebrates are found to be insurmountable), that this hypothesis would involve far fewer improbabilities than one which supposes the whole central nervous system of Vertebrates to be homologous with the super-oesophageal ganglia. The mode of formation of a nervous system presupposed in my hypothesis, well accords with what we know of the formation of the ventral cord in existing Annelids.

The supposition of the existence of another branch of segmented Vermes is not a very great difficulty. Even at the present day we have possibly more than one branch of Vermes which have independently acquired segmentation. viz.: the Chætopodous Annelids and the Hirudinea. If the latter is an isolated branch, it is especially interesting from having independently developed a series of segmental organs like those of Chætopodous Annelids, which we must suppose the ancestors of Vertebrates also to have done if they too form an independent branch.

In addition to the difficulty of imagining a fresh line of segmented Vermes, there is another difficulty to my view, viz.: the fact that in almost all Vermes, the blood flows forwards in the dorsal vessel, and backwards in the ventral vessel. This condition of the circulation very well suits the view of a change of the dorsal for the ventral surfaces, but is opposed to these surfaces being the same for Vertebrates and Vermes. I cannot however regard this point as a very serious difficulty to my view, considering how undefined is the circulation in the unsegmented groups of the Vermes.

_Sympathetic nervous system._

Between stages K and L there may be seen short branches from the spinal nerves, which take a course towards the median line of the body, and terminate in small irregular cellular masses immediately dorsal to the cardinal veins (Pl. 18, fig. 1, _sy.g._). These form the first traces that have come under my notice of the sympathetic nervous system. In the youngest of my embryos in which I have detected these it has not been possible for me either definitely to determine the antero-posterior limits of the system, or to make certain whether the terminal masses of cells which form the ganglia are connected by a longitudinal commissure. In a stage slightly younger than L the ganglia are much more definite, the anterior one is situated in the cardiac region close to the end of the intestinal branch of the vagus, and the last of them quite at the posterior end of the abdominal cavity. The anterior ganglia are the largest; the commissural cord, if developed, is still very indistinct. In stage L the commissural cord becomes definite, though not very easy to see even in longitudinal sections, and the ganglia become so considerable as not to be easily overlooked. They are represented in Pl. 13, fig. 1, _sy.g._ and in Pl. 18, fig. 2, in the normal position immediately above the cardinal veins. The branches connecting them with the trunks of the spinal nerves may still be seen without difficulty. In later stages these branches cannot so easily be made out in sections, but the ganglia themselves continue as fairly conspicuous objects. The segmental arrangement of the ganglia is shewn in Pl. 18, fig. 3, a longitudinal and vertical section of an embryo between stages L and M with the junctions of the sympathetic ganglia and spinal nerves. The ganglia occupy the intervals between the successive segments of the kidneys.

The sympathetic system only came under my notice at a comparatively late period in my investigations, and the above facts do not in all points clear up its development[268]. My observations seem to point to the sympathetic system arising as an off-shoot from the cerebrospinal system. Intestinal branches would seem to be developed on the main nerve stems of this in the thoracic and abdominal regions, each of these then develops a ganglion, and the ganglia become connected by a longitudinal commissure. On this view a typical spinal nerve has the following parts: two roots, a dorsal and ventral, the dorsal one ganglionated, and three main branches, (1) a ramus dorsalis, (2) a ramus ventralis, and (3) a ramus intestinalis. This scheme may be advantageously compared with that of a typical cranial nerve according to Gegenbaur. It may be noted that it brings the sympathetic nervous system into accord with the other parts of the nervous system as a product of the epiblast, and derived from outgrowths from the neural axis. It is clear, however, that my investigations, though they may naturally be interpreted in this way, do not definitely exclude a completely different method of development for the sympathetic system.

Footnote 268: The formation out of the sympathetic ganglia of the so-called paired suprarenal bodies is dealt with in connection with the vascular system. The original views of Leydig on these bodies are fully borne out by the facts of their development.

EXPLANATION OF PLATE 14.

_This Plate illustrates the Formation of the Spinal Nerves._

COMPLETE LIST OF REFERENCE LETTERS.

_ar._ Anterior root of a spinal nerve. _ch._ Notochord. _com._ Commissure connecting the posterior roots of the spinal nerves. _i._ Mesoblastic investment of spinal cord. _mp._ Muscle-plate. _n._ Spinal nerve. _nc._ Neural canal. _pr._ Posterior root of a spinal nerve. _spg._ Ganglion on posterior root of spinal nerve. _v.r._ Vertebral rudiment. _w._ White matter of spinal cord. _y._ Point where the spinal cord became segmented off from the superjacent epiblast.

Figs. 1, 2, and 3. Three sections of a Pristiurus embryo belonging to stage I. Fig. 1 passes through the heart, fig. 2 through the anterior part of the dorsal region, fig. 3 through a point slightly behind this. (Zeiss CC, ocul. 2.) In fig. 3 there is visible a slight proliferation of cells from the dorsal summit of the neural canal. In fig. 2 this proliferation definitely constitutes two club-shaped masses of cells (_pr_)--the rudiments of the posterior nerve-roots,--both attached to the dorsal summit of the spinal cord. In fig. 1 the rudiments of the posterior roots are of considerable length.

Fig. 4. Section through the dorsal region of a Torpedo embryo slightly older than stage I, with three visceral clefts. (Zeiss CC, ocul. 2.) The section shews the formation of a pair of dorsal nerve-rudiments (_pr_) and a ventral nerve-rudiment (_ar_). The latter is shewn in its youngest condition, and is not distinctly cellular.

Fig. 5. Section through the dorsal region of a Torpedo embryo slightly younger than stage K. (Zeiss CC, ocul. 2.) The connective-tissue cells are omitted. The rudiment of the ganglion (_spg_) on the posterior root has appeared, and the junction of posterior root with the cord is difficult to detect. The anterior root forms an elongated cellular structure.

Fig. 6. Section through the dorsal region of a Pristiurus embryo of stage K. (Zeiss CC, ocul. 2.) The section especially illustrates the attachment of the posterior root to the spinal cord.

Fig. 7. Section through the same embryo as fig. 6. (Zeiss CC, ocul. 1.) The section contains an anterior root, which takes its origin at a point opposite the interval between two posterior roots.

Fig. 8. A series of posterior roots with their central ends united by a dorsal commissure, from a longitudinal and vertical section of a Scyllium embryo belonging to a stage intermediate between L and M. The embryo was hardened in a mixture of osmic and chromic acids.

Fig. 9. The central end of a posterior nerve-root from the same embryo, with the commissure springing out from it on either side.