Origin of Vertebrates/Chapter VIII

From the evidence given in the last chapter, combined with that given in Chapter IV., the probability of the theory that the trigeminal group of nerves of the vertebrate have been derived from the prosomatic group of nerves of the invertebrate can be put to the test by the answers to the following morphological and anatomical questions:—

1. Do we find in the vertebrate two segmentations in this region corresponding to the two segmentations in the branchial region, i.e. a somatic or dorsal series of segments, and a splanchnic or ventral series of segments? The latter would not be branchial, but rather of the nature of free tactile appendages; so that it is useless to look for or talk about gill-slits, although such appendages, being serially homologous with the branchial mesosomatic appendages, would readily give rise to the conception of branchial segments.

2. Is there morphological evidence that the trigeminal nerve is not the nerve belonging to a single segment, or even to two segments, but is really a concentration of at least six, probably seven, segmental nerves?

3. Is there morphological evidence that the oculomotor and trochlear nerves, which on all sides are regarded as belonging to the trigeminal segments, are not single nerves corresponding each to a single segment, but are the somatic motor roots belonging to the same segments as those to which the trigeminal supplies the splanchnic roots?

4. Do the mesoderm segments, which give origin to the eye-muscles, and therefore do the head-cavities of this region, correspond with the trigeminal segments? Considering the concentration of parts in this region and the difficulty already presented by the want of numerical agreement between the prosomatic appendages and the prosomatic cœlomic cavities in Limulus, it may very probably be difficult to determine the actual number of the mesoderm segments.

5. Is there anatomical evidence that the ganglion of origin of the motor part of the trigeminal nerve is not a single ganglion, but a representative of many, probably seven?

6. Is there anatomical evidence that the ganglia of origin of the oculomotor and trochlear nerves represent many ganglia?

7. Is there any evidence that the organs originally supplied by the motor part of the trigeminal nerve are directly comparable with prosomatic appendages?

It is agreed on all sides that in this region of the head there is distinct evidence of double segmentation, the dorsal mesoderm segments giving origin to the eye-muscles, and the ventral segments to the musculature innervated by the trigeminal nerve. Originally, according to the scheme of van Wijhe, two segments only were recognized, the dorsal parts of which were innervated by the IIIrd and IVth nerves respectively. Since his paper, the tendency has been to increase the number of segments in this region, as is seen in the following sketch, taken from Rabl, of the history of cranial segmentation.

The first attempt to deal with this question was made by Goethe and Oken. They considered that the cranial skeleton was composed of a series of vertebræ, but as early as 1842 Vogt pointed out that only the occipital segments could be reduced to vertebræ. In 1869, Huxley showed that vertebræ were insufficient to explain the cranial segmentation, and that the nerves must be specially considered. The olfactory and optic nerves he regarded as parts of the brain, not true segmental nerves; the rest of the cranial nerves were segmental, with special reference to branchial arches and clefts, the facial, glossopharyngeal, and separate vagus branches supplying the walls of the various branchial pouches. In a similar manner, the supra- and infra-maxillary branches of the trigeminal were arranged on each side of the mouth, and the inner and outer twigs of the first (ophthalmic) branch of the trigeminal on each side of the orbito-nasal cleft, the trabecular and the supra-maxillary arches being those on each side of this cleft. Thus Huxley considered that there was evidence of a series of pairs of ventral arches belonging to the skull, viz. the trabecular and maxillary in front of the mouth, the mandibular, hyoid, and branchial arches behind, and that the Vth, VIIth, IXth, and Xth nerves were segmental in relation to these arches and clefts. Gegenbaur, in 1871 and 1872, considered that the branchial arches represented the lower arches of cranial vertebræ, and therefore corresponded to lower arches in the spinal region, i.e. the skull was composed of as many vertebræ as there are branchial arches. These vertebræ were confined to the notochordal part of the skull, the prechordal part having arisen secondarily from the vertebral part, while the number of vertebræ are at least nine, possibly more. The nerves which could be homologized with spinal nerves were, he thought, divisible into two great groups—(1) the trigeminal group, which included the eye-muscle nerves, the facial, and its dorsal branch, the auditory; (2) the vagus group, which included the glossopharyngeal and vagus.

Such was the outcome of the purely comparative anatomical work of Huxley and Gegenbaur—work that has profoundly influenced all the views of segmentation up to the present day.

Now came the investigations of the embryologists, of whom I will take, in the first instance, Balfour, whose observations on the embryology of the Selachians led him to the conclusion that besides the evidence of segmentation to be found in the cranial nerves and in the branchial clefts, further evidence was afforded by the existence of head-cavities, the walls of which formed muscles just as they do in the spinal region. He came to the conclusion that the first head-cavity belonged to one or more pre-oral segments, of which the nerves were the oculomotor, trochlearis, and possibly abducens; while there were seven post-oral segments, each with its head-cavity and its visceral arch, of which the trigeminal, facial, glossopharyngeal, and the four parts of the vagus were the respective nerves.

Marshall, in 1882, considered that the cranial segments were all originally respiratory, and that all the segmental nerves are arranged uniformly with respect to a series of gill-clefts which have become modified anteriorly and have been lost, to a certain extent, posteriorly. He included the olfactory nerves among the segmental nerves, and looked upon the olfactory pit, the orbito-nasal lacrymal duct, the mouth, and the spiracle as all modified gill-slits, so that he reckoned three pre-oral and oral segments belonging to the Ist, IIIrd, IVth, and Vth nerves, and eight post-oral segments belonging respectively to the VIIth and VIth nerves, and to the IXth nerve, and six segments belonging to the Xth nerve. He pointed out that muscles supplied by the oculomotor nerve develop from the outer wall of the first head-cavity; not, however, the obliquus superior and rectus externus, the latter originating probably from the walls of the third cavity.

In the same year, 1882, came van Wijhe's well-known paper, in which he showed that the mesoderm of the head in the selachian divided into two sets of segments, dorsal and ventral; that the dorsal segments were continuous with the body-somites, and that the ventral segments formed the lateral plates of mesoblast between each of the visceral and branchial pouches. He concluded that the dorsal somites were originally nine in number, that each was supplied with a ventral nerve-root, in the same way as the somites in the trunk, and that to each one a visceral pouch corresponded, whose walls were supplied by the corresponding dorsal nerve-root; of these nine segments, the ventral nerve-roots of the first three segments were respectively the oculomotor, trochlearis, and abducens nerves. The next three segments possessed no definable ventral root or muscles, and the seventh, eighth, and ninth segments possessed as ventral roots the hypoglossal nerve, with its muscular supply. The corresponding dorsal nerve-roots were the trigeminal, facial, auditory, glossopharyngeal and vagus nerves, the difference between cranial and spinal dorsal roots being that the former contain motor fibres.

Ahlborn, in 1884, drew a sharp distinction between the segments of the mesoderm and those of the endoderm. The former segmentation he called mesomeric, the latter branchiomeric. He considered the two segmentations to be independent, and concluded that the branchiomeric was secondary to the mesomeric, and therefore not of segmental value. As to the segments of the mesoderm in the head, the three hindmost or occipital in Petromyzontidæ remain permanently, and correspond to the three last segments in the selachian head. Of the anterior mesoderm segments, he considered that there were originally six, and that there are six typical eye-muscles in all Craniota, which have been compressed into three segments, as in Selachia.

Froriep (1885) showed in sheep-embryos and in chicks that the hypoglossal nerve belongs to three proto-vertebræ posterior to the vagus region, which were true spinal segments. He therefore modified Gegenbaur's conceptions to this extent: that portion of the skull designated by Gegenbaur as vertebral must be divided into two parts—a hind or occipital region, which is clearly composed of modified vertebræ and is the region of the hypoglossal nerves, and a front region, extending from the oculomotor to the accessorius nerves, which is characterized segmentally by the formation of branchial arches, but in which there is no evidence that proto-vertebræ were ever formed. He therefore divides the head-skeleton into three parts—

1. Gegenbaur's evertebral part—the region of the olfactory and optic nerves—which cannot be referred to any metameric segmentation.

2. The pseudo-vertebral, pre-spinal, or branchial part, clearly shown to be segmented from the consideration of the nerves and branchial arches, but not referable to proto-vertebræ—the region of the trigeminal and vagus nerves.

3. The vertebral spinal part—the region of the hypoglossal nerves.

He further showed that the ganglia of the specially branchial nerves, the facial, glossopharyngeal, and vagus, are at one stage in connection with the epidermis, so that these parts of the epidermis represent sense-organs which do not develop; these organs probably belonged to the lateral line system. As the connection takes place at the dorsal edge of the gill-slits, they may also be called rudimentary branchial sense-organs.

Since this paper of Froriep's, it has been generally recognized, and Gegenbaur has accepted Froriep's view, that the three hindmost metameres, which distinctly show the characteristics of vertebræ, belong to the spinal and not to the cranial region, so that the metameric segmentation of the cranial region proper has become more and more associated with the branchial segmentation. Froriep's discovery of the rudimentary branchial sense-organs as a factor in the segmentation question has led Beard to the conclusion that the olfactory and auditory organs represent in a permanent form two of these rudimentary branchial sense-organs. He therefore includes both the olfactory and auditory nerves in his list of cranial segmental nerves, and makes eleven cranial branchial segments in front of the spinal segments represented by the hypoglossal.

A still larger number of cranial segments is supposed to exist, according to the researches of Dohrn and Killian, in the embryos of Torpedo ocellata. The former, holding to the view that vertebrates arose from annelids, considered that the head was formed of a series of metameres, to each one of which a mesoderm-segment, a gill-arch, a gill-cleft, a segmental nerve and vessel belonged. He found in the front head-region of a Torpedo embryo, corresponding to van Wijhe's first four somites, no less than twelve to fifteen mesoderm segments, and concluded, therefore, that the eye-muscle nerves, especially the oculomotor, represented many segmental nerves, and were not the nerves of single segments; so, also, that the inferior maxillary part of the trigeminal and the hyoid nerve of the facial are probably not single nerves, but a fusion of several. Killian comes to much the same conclusion as Dohrn, for he finds seventeen to eighteen separate mesoderm segments in the head, of which twelve belong to the trigeminal and facial region.

Since Rabl's paper, a number of papers have appeared, especially from America, dealing with yet another criterion of the original segmentation of the head, viz. a series of divisions of the central nervous system itself, which are seen at a very early stage of development, and are called neuromeres; the divisions in the cranial region being known as encephalomeres, and those of the spinal region as myomeres. Locy's paper has especially brought these divisions into prominence as a factor in the question of segmentation. They are essentially segments of the epiblast and not of the mesoblast; they are conspicuous in very early stages, and appear to be in relation with the cranial nerves, according to Locy. He recognizes in Squalus acanthias, in front of the spino-occipital region, fourteen pairs of such encephalomeres and a median unsegmented termination, which may represent one more pair fused in the middle line, making at least fifteen. He distributes these fifteen segments as follows: fore-brain three and unsegmented termination, mid-brain two, and hind-brain nine.

Again, Kupffer, in his recent papers on the embryology of Ammocœtes, asserts that especial information as to the number of primitive segments is afforded by the appearance in the early stages of a series of epibranchial ganglia in connection with the cranial nerves, which remain permanently in the case of the vagus nerves, but disappear in the case of pro-otic nerves. He considers that the evidence points to the number of segments in the mid- and hind-brain region as being primitively fifteen, viz. six segments belonging to the trigeminal and abducens group, three segments belonging respectively to the facial, auditory, and glossopharyngeal, and six to the vagus.

From this sketch we see that the modern tendency is to make six segments at least out of the region of the trigeminal nerves rather than two. In this region, as already mentioned, the evidence of segmentation is based more clearly on the somatic than on the splanchnic segments. We ought, therefore, in the first place, to consider the teaching of the eye-muscles and their nerves and the cœlomic cavities in connection with them, and see whether the hypothesis that such muscles represent the original dorso-ventral somatic muscles of the palæostracan ancestor is in harmony with and explains the facts of modern research.

The only universally recognized somatic nerves belonging to these segments which exist in the adult are the nerves to the eye-muscles, of which, according to van Wijhe, the oculomotor is the nerve of the 1st segment, the trochlearis of the 2nd, and the abducens of the 3rd; while the nerves and muscles belonging to the 4th and 5th segments, i.e. the 2nd facial and glossopharyngeal segments respectively, show only the merest rudiments, and do not exist in the adult. One significant fact appears in this statement of van Wijhe, and is accepted by all those who follow him, viz. that the oculomotor nerve has equal segmental value with the trochlearis and the abducens, although it supplies a number of muscles, each of which, on the face of it, has the same anatomical value as the superior oblique or external rectus. Dohrn alone, as far as I know, as already pointed out, insists upon the multiple character of the oculomotor nerve.

As far as the anatomist is concerned, the evidence is becoming clearer and clearer that the nucleus of the IIIrd nerve is a composite ganglion composed of a number of nuclei, each similar to that of the trochlearis, so that if the trochlearis nucleus is a segmental motor nucleus, then the oculomotor nucleus is a combined nucleus belonging to at least four segmental nerves, each of which has the same value as that of the trochlearis.

The investigations of a number of anatomists, among whom may be mentioned Gudden, Obersteiner, Edinger, Kölliker, Gehuchten, all lead directly to the conclusion that this oculomotor nucleus is composed of a number of separate nuclei, of which the most anterior as also the Edinger-Westphal nucleus contains small cells, while the others contain large cells. Thus Edinger divides the origin of the oculomotor nerve into a small-celled anterior part and a larger posterior part, of which the cells are larger and distinctly arranged in three groups—(1) dorsal, (2) ventral, and (3) median. Between the anterior and posterior groups lies the Edinger-Westphal nucleus, which is small-celled; naturally, the large-celled group is that which gives origin to the motor nerves of the eye-muscles, the small-celled being possibly concerned with the motor nerves of the pupillary and ciliary muscles. I may mention that Kölliker considers that the anterior lateral nucleus has nothing to do with the oculomotor nerve, but is a group of cells in which the fibres of the posterior longitudinal bundle and of the deep part of the posterior commissure terminate.

These conclusions of Edinger are the outcome of work done in his laboratory by Perlia, who says that in new-born animals the nucleus of origin of the oculomotor nerve is made up of a number of groups quite distinct from each other, each group being of the same character as that of the trochlearis. He finds the same arrangement in various mammals and birds. Further, he finds that some of the fibres arise from the nucleus of the opposite side, thus crossing, as in the trochlearis; these crossing fibres belong to the most posterior of the dorsal group of nuclei, i.e. to the nerve to the inferior oblique muscle.

The evidence, therefore, points to the conclusion that the oculomotor nucleus is a multiple nucleus, each part of which gives origin to one of the nerves of one of the eye-muscles.

Edinger says that such an array of clinical observations exists, and of facts derived from post-mortem dissections, that one may venture to designate the portion of the nucleus from which the innervation of each individual ocular muscle comes. He gives Starr's table, the latest of these numerous attempts, begun by Pick. According to Starr, the nuclei of the nerves to the individual muscles are arranged from before backward, thus—

Further, the evidence of the well-known physiological experiments of Hensen and Völckers that the terminal branches of the oculomotor nerve arise from a series of segments of the nucleus, arranged more or less one behind the other in a longitudinal row, leads them to the conclusion that the nuclei of origin are arranged as follows, proceeding from head to tail:—

It is instructive to compare this arrangement of Hensen and Völckers with the arrangement of the origin of these muscles from the premandibular cavity as given by Miss Platt.

Thus she states that the most posterior part of the premandibular cavity is cut off so as to form a separate cavity, resembling, except in position, the anterior cavity; this separate, most posterior part gives origin to the inferior oblique muscle. She then goes on to describe how the dorsal wall of the remainder of the premandibular cavity becomes thickened, to form posteriorly the rudiment of the inferior rectus and anteriorly the rudiments of the superior and internal recti, a slight depression in the wall of the cavity separating these rudiments. The internal rectus is the more median of the two anterior muscles. In other words, her evidence points not only to a fusion of somites to form the premandibular cavity, but also to the arrangement of these somites as follows, from head to tail: (1) internal rectus, (2) superior rectus, (3) inferior rectus, (4) inferior oblique—an order precisely the same as that of Hensen and Völckers, and of Starr.

I conclude, from the agreement between the anatomical, physiological, and morphological evidence, that the IIIrd and IVth nerves contain the motor somatic nerves belonging to the same segments as the motor trigeminal, in other words, to the prosomatic segments, so that the eye-muscles, innervated by III. and IV., represent segmental muscles belonging to the prosoma. Further, I conclude that originally there were seven prosomatic segments, the first of which is represented by the anterior cavity described by Miss Platt, and does not form any permanent muscles; that the next four belong to the premandibular cavity, and the muscles formed are the superior rectus, internal rectus, inferior rectus, and inferior oblique; and that the last two belong to the mandibular cavity, the muscles formed being Miss Platt's mandibular muscle and the superior oblique. It is, to say the least of it, a striking coincidence that such an arrangement of the cœlomic cavities as here given should be so closely mimicked by the arrangement in the prosomatic region of Limulus as already mentioned; it suggests inevitably that the head-cavities of the vertebrate are nothing more than the prosomatic and mesosomatic segmental cœlomic cavities, as found in animals such as Limulus. In the table on p. 253, I have inserted the segments in the vertebrate for comparison with those of Limulus.

Before we can come to any conclusion as to the original position of these eye-muscles, it is necessary to consider the VIth nerve and the external rectus muscle. This nerve and this muscle belong to van Wijhe's 4th segment. The muscle is, therefore, the somatic segmental muscle belonging to the same segment as the facial and is, in fact, a segmental muscle belonging not to the prosoma, but to the mesosoma. Neal comes to the conclusion that the existing abducens is the only root which remains permanent among a whole series of corresponding ventral roots belonging to the opisthotic segments, and further points out that the external rectus was originally an opisthotic muscle which has taken up a pro-otic position, or, translating this statement into the language of Limulus, etc., it is a mesosomatic muscle which has taken up a prosomatic position.

There is, however, another muscle—the Retractor oculi—belonging to the same group which is innervated by the VIth nerve. Quite recently Edgeworth has shown that in birds and reptiles this muscle belongs to the hyoid segment; so that in this respect also the hyoid segment proclaims its double nature.

With respect to the external rectus muscle, Miss Platt has shown that the mandibular muscle is formed close alongside the external rectus, so that the two are in close relationship as long as the former exists.

Further, as already mentioned, the eye-muscles in Ammocœtes must be considered by themselves; they do not belong in structure or position to the longitudinal somatic muscles innervated by the spinal nerves; their structure is not the same as that of the tubular constrictor or branchial muscles, but resembles that structure somewhat; their position is dorso-ventral rather than longitudinal; they may be looked upon as a primitive type of somatic muscles segmentally arranged, the direction of which was dorso-ventral.

Anderson also has shown that the time of medullation of the nerves supplying these muscles is much earlier than that of the nerves belonging to the somatic trunk-muscles, their medullation taking place at the same time as that of the motor nerves supplying the striated visceral muscles; and Sherrington has observed that these muscles do not possess muscle-spindles, while all somatic trunk-muscles do. Both these observations are strong confirmation of the view that the eye-muscles must be classified in a different category to the ordinary somatic trunk muscle group.

What, then, is the interpretation of these various embryological and anatomical facts?

Remembering the tripartite division of each segmental nerve-group in Limulus into (1) dorsal or sensory somatic nerve, (2) appendage-nerve, and (3) ventral somatic nerve, I venture to suggest that the three nerves—the oculomotorius, the trochlearis, and the abducens—represent the ventral somatic nerves of the prosoma, and partly also of the mesosoma; that they are nerves, therefore, which may have originally contained sensory fibres, and which still contain the sensory fibres of the eye-muscles themselves, as stated by Sherrington. According to this suggestion, the eye-muscles are the sole survivors of the segmental dorso-ventral somatic muscles, so characteristic of the group from which I imagine the vertebrates to have sprung. In the mesosomatic region the dorso-ventral muscles which were retained were those of the appendages and not of the mesosoma itself, because the presumed ancestor breathed after the fashion of the water-breathing Limulus, by means of the dorso-ventral muscles of its branchial appendages, and not after the fashion of the air-breathing scorpion, by means of the dorso-ventral muscles of the mesosoma. The only mesosomatic dorso-ventral muscles which were retained were those of the foremost mesosomatic segments, i.e. those supplied by the VIth nerve, which were preserved owing to their having taken on a prosomatic position and become utilized to assist in the movements of the lateral eyes.

Let us turn now to the consideration of the corresponding musculature in Limulus and in the scorpion group. These muscles constitute the markedly segmental muscles to which I have given the name 'dorso-ventral somatic muscles.' They are most markedly segmental in the mesosomatic region, both in Limulus and in Scorpio, each mesosomatic segment possessing a single pair of these vertical mesosomatic muscles, as Benham calls them (cf. Fig. 58 (Dv.)). In the prosomatic region the corresponding muscles are not so clearly defined in Limulus; they are apparently attached to the plastron forming the group of plastro-tergal muscles. From Benham's description it is sufficiently evident that they formed originally a single pair to each prosomatic segment.

In Scorpio, according to Miss Beck, the dorso-ventral prosomatic muscles are situated near the middle line on each side and form the following well-marked series of pairs of muscles, shown in Fig. 110, A, taken from her paper, and thus described by her:—

1. The dorso-cheliceral-sternal muscle (61) is the most anterior of the dorso-ventral muscles. It is very small, and is attached to the carapace near the median line anteriorly to the central eyes.

2. The median dorso-preoral-entosclerite muscle (62) is a large muscle, between which and its fellow of the opposite side the eyes are situated. It is attached dorsally to the carapace and ventrally to the pre-oral entosclerite.

3. The anterior dorso-plastron muscle (63) is attached dorsally to the carapace in the middle line, being joined to its fellow of the opposite side. They separate, and are attached ventrally to the plastron. Through the arch thus formed the alimentary canal and the dorsal vessel pass.

4. The median dorso-plastron muscle (64) is attached dorsally to the posterior part of the carapace. It runs forward on the anterior surface of the posterior flap of the plastron to the body of the plastron, to which it is attached.

To these may be added, owing to its attachment to the plastron,

5. The posterior dorso-plastron muscle (65). This is the first of the dorso-ventral muscles attached to the mesosomatic tergites, being attached to the tergite of the first segment of the mesosoma.

This muscle is of interest, in connection with the prosomatic dorso-ventral muscles, because it is attached to the plastron, and runs a course in close contact with the muscle (64), the two muscles being attached dorsally close together, on each side of the middle line, the one at the very posterior edge of the prosomatic carapace, and the other at the very anterior edge of the mesosomatic carapace.

Taking these muscles separately into consideration, it may be remarked with respect to (61) that the cheliceral segment in its paired dorso-ventral muscles, as in its tergo-coxal muscles, takes up a separate position isolated from the rest of the prosomatic segments.

Next comes (62) the median dorso-preoral-entosclerite muscle, which is strikingly different from all the other dorso-ventral muscles in its large size and the extent of its attachment to the dorsal carapace, according to Miss Beck's figures. The reason of its large size is clearly seen upon dissection of the muscles in Buthus, for I find that, strictly speaking, it is not a single muscle, but is composed of a series of muscle-bundles, separated from each other by connective tissue. There are certainly three separate muscles included in this large muscle, which are attached in a distinct series along the pre-oral entosclerite, and present the appearance given in Fig. 110, A, at their attachment to the prosomatic carapace. Of this muscle-group the most anterior and the most posterior bundle are distinctly separate muscles; I am not, however, clear whether the middle bundle represents one or two muscles.

This division of Miss Beck's muscle (62) into three or four muscles brings the prosomatic region of the scorpion into line with the mesosomatic, and enables us to feel sure that a single pair of dorso-ventral somatic muscles belongs to each prosomatic segment just as to each mesosomatic, and, conversely, that each such single pair of muscles possesses segmental value in this region as much as in the mesosomatic.

It is very striking to see how in all the Scorpionidæ, in which the two median eyes are the principal eyes, this muscle group (62) on the two sides closely surrounds these two eyes, so that with a fixed pre-oral entosclerite, a slight movement of the eyes, laterally or anteriorly, owing to the flexibility of the carapace, might result as the consequence of their contraction. But this cannot be the main object of these muscles. The pre-oral entosclerite is firmly fixed to the camerostome, as is seen in Fig. 94, ''pr. ent.'', so that the main object of these muscles is, as Huxley has pointed out, the movement of this organ.

In order to avoid repetition of the long name given to this muscle group (62) by Miss Beck, because of their position, and for other reasons which will appear in the sequel, I will call this group of muscles the group of recti muscles. These recti muscles belong clearly to the segments posterior to the first prosomatic or cheliceral segment, and represent certainly three, probably four, of these segments, i.e. belong to the segments corresponding to the second, third, fourth, and fifth prosomatic locomotor appendages—the endognaths of the old Eurypterids.

The next pair of muscles is the pair of anterior dorso-plastron muscles (63). This muscle-pair evidently belongs to a segment posterior to the segments represented by the group already discussed, and belongs, therefore, in all probability to the same segment as the sixth pair of prosomatic appendages—the ectognaths of the old Eurypterids. This can be settled by considering either the nerve-supply or the embryological development. In the Eurypteridæ it seems most highly probable that the dorso-ventral muscles of each half of the segments belonging to the endognaths should be compressed together and separate from the dorso-ventral muscle belonging to the ectognathal segment, on account of the evident concentration and small size of the endognathal segments in contradistinction to the separateness and large size of the ectognathal segment.

The striking peculiarity of this muscle-pair, which distinguishes it from all other muscles in the scorpion, is the common attachment of the muscles of the two sides in the mid-dorsal line, so that the pair of muscles forms an arch through which the alimentary canal and dorsal blood-vessel pass.

The same dorso-ventral muscles are present in Phrynus, and in this animal the fibres of this pair of muscles (63) actually interlace before the attachment to the prosomatic carapace, so that the attachment of the muscle on each side overpasses the mid-dorsal line, and a true crossing occurs. In Fig. 108 the position of this pair of muscles is shown just posteriorly to the brain-mass. This muscle I will call the oblique muscle.

Finally we come to the muscles (64) and (65), the median and posterior dorso-plastron muscles, which run close together. Both muscles are attached to the plastron, and, therefore, to that extent belong to the prosomatic region; they are attached dorsally close to the junction of the prosoma and mesosoma. This position of the first mesosomatic dorso-ventral muscle belonging to the opercular segment may be compared with the position of the first mesosomatic dorso-ventral muscle in Limulus which has become attached to the prosomatic carapace; in both cases we see an indication that the foremost pair of mesosomatic dorso-ventral somatic muscles tend to take up a prosomatic position.

As to the pair of small muscles (64), I believe that they represent the dorso-ventral muscles of the seventh prosomatic segment (if the pair of muscles (63) belongs to the segment of the sixth locomotor prosomatic appendages), i.e. they belong to the chilarial segment or metastoma.

I desire to draw especial attention to the fact that the dorso-ventral muscle (64), which represents the seventh segment, always runs close alongside the dorso-ventral muscle (65), which represents the first mesosomatic or opercular segment.

The comparison, then, of these two sets of facts leads to the following conclusions:—

The foremost prosomatic or trigeminal segment stood separate and apart, being situated most anteriorly; the musculature of this segment does not develop, so that the only evidence of its presence is given by the anterior cœlomic cavity. This corresponds, according to my scheme, with the first or anterior cœlomic cavity of Limulus, and therefore represents, as far as the prosomatic appendages are concerned, the first prosomatic appendage-pair, or the cheliceræ; the appendage-muscles being the muscles of the cheliceræ, and the dorso-ventral somatic muscles the pair of dorso-cheliceral sternal muscles (61) in the scorpion. Both these sets of muscles, therefore, dwindle and disappear in the vertebrate.

Then came four segments fused together to form the premandibular segment, the characteristic of which is the apparent non-formation of any permanent musculature from the ventral mesoderm-segments, and the formation of the eye-muscles innervated by the oculomotor nerve from the dorsal mesoderm segments. These four segments have been so fused together that van Wijhe looked upon them as a single segment, and the premandibular cavity as the cavity of a single segment. They represent, according to my scheme, the segments belonging to the endognaths, i.e. the second, third, fourth, fifth pairs of prosomatic appendages; the premandibular cavity, therefore, represents the second cœlomic cavity in Limulus, which, according to Kishinouye, is the sole representative of the cœlomic cavities of the second, third, fourth, fifth prosomatic segments. The muscles derived from the ventral mesoderm-segments represent the muscles of these appendages, which therefore dwindle and disappear in the vertebrate, with the possible exception of the muscles innervated by the descending root of the trigeminal. The muscles derived from the dorsal mesoderm-segments, i.e. the eye-muscles supplied by the oculomotor nerve, represent the dorso-ventral somatic muscles of these four segments, muscles which are represented in the scorpion by the recti group of muscles, i.e. the median dorso-preoral-entosclerite muscles (62).

Then came two segments, the mandibular, in which muscles are formed both from the ventral and from the dorsal mesoderm-segments. From the former arose the main mass of muscles innervated by the motor root of the trigeminal, from the latter the superior oblique muscle and the mandibular muscle of Miss Platt, of which the former alone survives in the adult condition. These two segments are looked upon as a single segment by van Wijhe, of which the mandibular cavity is the cœlomic cavity. They represent, according to my scheme, the segments belonging to the sixth pair of prosomatic appendages or ectognaths, and the seventh pair, i.e. the chilaria or metastoma.

The first part, then, of the mandibular cavity represents the third cœlomic cavity in Limulus and the muscles derived from the ventral mesoderm, in all probability the muscles of the tongue in the lamprey (cf. Chap. IX.), which represents the ectognaths or sixth pair of prosomatic appendages, while the muscles derived from the dorsal mesoderm, i.e. the superior oblique muscles, represent the dorso-ventral somatic muscles of this segment, muscles which are represented in the scorpion group by the pair of anterior dorso-plastron or oblique muscles (63).

The second part of the mandibular cavity represents the 4th cœlomic cavity in Limulus and the muscles derived from the ventral mesoderm, in all probability the muscles of the lower lip in the lamprey (cf. Chap. IX.), which represents the metastoma; while the muscles derived from the dorsal mesoderm, i.e. Miss Platt's pair of mandibular muscles, represent the dorso-ventral somatic muscles of this segment, muscles which are represented in the scorpion group by the pair of median dorso-plastron muscles (64).

In connection with this last pair of muscles we find that the external rectus in the vertebrate represents the first dorso-ventral mesosomatic muscle in the scorpion, i.e. the posterior dorso-plastron muscle (65), and, as already mentioned (p. 267), that it always lies closely alongside the mandibular muscle, just as in the scorpion group muscle (65) always lies alongside muscle (64).

In the invertebrate as well as in the vertebrate this muscle is a mesosomatic muscle which has taken up a prosomatic position.

The question naturally arises, what explanation can be given of the fact that these dorso-ventral muscles attached on each side of the mid-dorsal line to the prosomatic carapace became converted into the muscles moving the eyeballs of the two lateral eyes? An explanation which must take into account not only the isolated position of the abducens nerve, but also the extraordinary course of the trochlearis. The natural and straightforward answer to this question appears to me quite satisfactory, and I therefore venture to commend it to my readers.

I have argued the case out to myself as follows: The lateral eyes must have been originally situated externally to the group of muscles innervated by the oculomotor nerve, for a sheet of muscle representing the superior internal and inferior rectus muscles could only wrap round the internal surface of each lateral eye; i.e. the arrangement of the muscle-sheet, as in the scorpion, about two median eyes, is in the wrong position, for if those two eyes, which are the main eyes in the scorpion, were to move outwards to become two lateral eyes, then such a muscle-group would form a superior external and inferior rectus group. The evidence, however, of Eurypterus and similar forms is to the effect that the lateral eyes became big and the median eyes insignificant and degenerate. If, then, with the degeneration of the one and the increasing importance of the other, these lateral eyes came near the middle line, then the muscular group (62), which I have called the recti group, would naturally be pressed into their service, and would form an internal and not an external group of eye-muscles.

In Fig. 110, A, taken from Miss Beck's paper, I have shown the relative position of the eyes and the segmental dorso-ventral prosomatic muscles on the carapace of the scorpion. In Fig. 110, B, I have drawn the prosomatic carapace of Eurypterus Scouleri, taken from Woodward's paper, with the eyes as represented there; in this I have inserted the segmental dorso-ventral muscles as met with in the scorpion, thereby demonstrating how, with the degeneration of the median eyes and the large size of the lateral eyes, the recti muscles of the scorpion would approach the position of an internal recti group to the lateral eyes, and so give origin to the group of muscles innervated by the oculomotor nerve. In the Eurypterus these large eyes are large single eyes, not separate ocelli, as in the scorpion.

All, then, that is required is that in the first formed fishes, which still possessed the dorso-ventral muscles of their Eurypterid ancestors, the lateral eyes should be the important organs of sight, large and near the mid-dorsal line. Such, indeed, is found to be the case. In amongst the masses of Eurypterids found in the upper Silurian deposits at Oesel, as described by Rohon, numbers of the most ancient forms of fish are found belonging to the genera Thyestes and Tremataspis. The nature of the dorsal head-shields of these fishes is shown in Fig. 14, which represents the dorsal head-shield of Thyestes verrucosus, and Fig. 111 that of Tremataspis Mickwitzi. They show how the two lateral eyes were situated close on each side of the mid-dorsal line in these Eurypterus-like fishes, in the very position where they must have been if the eye-muscles were derived from the dorso-ventral somatic muscles of a Eurypterid ancestor.



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

In Lankester's words, one of the characteristics of the Osteostraci (Cephalaspis, Auchenaspis, etc.), as distinguished from the Heterostraci (Pteraspis), are the large orbits placed near the centre of the shield. The apparent exception of Thyestes mentioned by him is no exception, for orbits of the same character have since been discovered, as is seen in Rohon's figure (Fig. 14). In Fig. 110, C, I give an outline of the frontal part of the head-shield of a Cephalaspid, in which I have drawn the eye-muscles as in the other two figures.

Although all the members of the Osteostraci possess large lateral eyes towards the centre of the head-shield, the other group of ancient fishes, the Heterostraci, are characterized by the presence of lateral eyes far apart, situated on the margin of the head-shield on each side (cf. Fig. 142, o, p. 350).

So, also, on the invertebrate side, the lateral eyes of Pterygotus and Slimonia are situated on the margin of the prosomatic carapace, while those of Eurypterus and Stylonurus are situated much nearer the middle line of the prosomatic carapace.

Next comes the question of the superior oblique muscle and the trochlearis nerve. Why does this nerve (n.IV. in Fig. 106, C and D) alone of all the nerves in the body take the peculiar position it always does take? The only suggestion that I know of which sounds reasonable and worth consideration is that put forward by Fürbringer, which is an elaboration of the original suggestion of Hoffmann. Hoffmann suggested in 1889 that the trochlearis nerve represented originally a nerve for a protecting organ of the pineal eye, which became secondarily a motor nerve for the lateral eye as the pineal eye degenerated. Fürbringer differs from Hoffmann in that he considers that the nerve was originally a motor nerve, and was not transformed from sensory to motor, yet thinks Hoffmann's suggestion is in the right direction.

He points out that the crossing of the trochlearis is not a crossing of fibres between two centres in the central nervous system, but may be explained by the shifting of the peripheral organ, i.e. the muscle, from one side to the other, and the nerve following this shift. Consequently, says Fürbringer, the course of the nerve indicates the original position of the muscle, and therefore he imagines that the ancestor of the superior oblique muscle was a muscle the fibres of which were attached in the mid-dorsal line, and interlaced with those of the other side, the two muscles thus forming an arch through which the nervous system with its central canal passed. Then, for the sake of getting a more efficient pull, the crossing muscle-fibres became more definitely attached to the opposite side of the middle line, and finally obtained a new attachment on the opposite side, with the obliteration of the muscular arch; the nerve on each side, following the shifts of the muscle, naturally took up the position of the original muscular arch, and so formed the trochlear nerve, with its dorsal crossing. This explanation of Fürbringer's was associated by him with movements of the median pineal eyes, the length of their nerve, according to him, even yet indicating their previous mobility. This assumption is not, it seems to me, necessary. The length of the nerve is certainly no indication of mobility, for in Limulus and the scorpion group the nerve to each median eye is remarkably long, yet these eyes are immovably fixed in the carapace. All that is required is a pair of dorso-ventral muscles belonging to the segment immediately following the group of segments represented by the oculomotor nerves, the fibres of which should cross the mid-dorsal line at their attachment; for, seeing that the lateral eyes were originally so near this position, it follows that such muscles might form part of the muscular group belonging to the lateral eye without having previously moved the pineal eyes. In fact, Fürbringer's explanation requires as starting-point that the pair of muscles which ultimately become the superior oblique should have the exact position of the pair of dorso-ventral muscles in the scorpion, called by Miss Beck the anterior dorso-plastron muscles (63), which I have named the oblique muscles. Here, and here only, do we find an interlacement, across the mid-dorsal line, of the fibres of attachment of the muscles on the two sides, in consequence of which this pair of muscles is described by her as forming an arch encircling the alimentary canal and dorsal vessel. If, then, as I have previously argued, the primitive plastron formed a pair of trabeculæ, and the nervous system grew round the alimentary canal, such an arch would encircle the tubular central nervous system of the vertebrate.

Still more striking is this pair of muscles (63) in Phrynus (Fig. 108), where we see how the arch formed by them almost touches the posterior extremity of the supra-œsophageal brain-mass, crossing, therefore, over the beginning of the stomach region of the animal. The angle formed by the arch is much more obtuse than that formed in Scorpio, so that an actual crossing of the muscle-fibres has taken place at the point of attachment to the carapace. Also, only the part nearest the carapace is muscular, the rest forming a long tendinous prolongation of the plastron wall (the primordial cranium), as seen in the figure.

This muscle-pair is, as it should be, the pair of dorso-ventral muscles belonging to the segment immediately following on the group of segments represented by the recti muscles, i.e. according to previous argument, the segment belonging to the sixth pair of locomotor appendages or ectognaths; a muscle, therefore, which would arise in the vertebrate from the mandibular, and not from the premandibular cavity. A similar muscle probably existed in Eurypterus (M.obl. in Fig. 106, B), and, as in the case of the formation of the oculomotor group, derived from the recti group of the scorpion, would form the commencement of the superior oblique muscle in Thyestes and Tremataspis.



It is instructive to notice that the original position of attachment of this muscle is naturally posterior to that of the oculomotor group of muscles, and that Fürbringer, in his description of the eye-muscles of Petromyzon, asserts that this muscle in this primitive vertebrate form is not attached as in other vertebrates, but is posterior to the other muscles, so that he calls it the posterior rather than the superior oblique. The nature of the change by which the muscle known in the scorpion as the anterior dorso-plastron muscle (63) was probably converted into the superior oblique muscle of the vertebrate, is represented in the drawings Fig. 112, in which also are indicated the dwindling of the median eyes, and the progressive superiority of the lateral eyes, as well as the transformation of the recti muscle-group of the scorpion into the muscles supplied by the oculomotor nerve of the vertebrate.

With respect to the external rectus muscle, it follows naturally that if the muscles (64) and (65) are to follow suit with the rest of the group and become attached to the lateral eyes, they must take up an external position. These two muscles, which always run together, as seen in Fig. 110, A, the one belonging to the prosoma and the other to the mesosoma, are represented by the mandibular muscle of Miss Platt and the external rectus, the former derived from the walls of the last pro-otic head-cavity, the latter from the foremost of the opisthotic head-cavities.

Such, then, is the simple explanation of the origin of the eye-muscles which follows from my theory, and we see that the successive alterations of the position of the orbit, and, therefore, of the globe of the eye with its muscles, as we pass from Thyestes to man, is the natural consequence of the growth of the frontal bone, i.e. of the brain.

Turning now to the evidence as to the number of ventral segments, i.e. the motor and sensory supply to the prosomatic appendages afforded by the trigeminal nerve, we must, I think, come to the same conclusion as Dohrn, viz. that if there were originally seven dorsal or somatic segments in this region represented by: 1, Anterior cavity, muscle lost; 2, 3, 4, 5, muscles of the premandibular cavity, ''sup. rectus, inf. rectus, int. rectus, inf. oblique, supplied by IIIrd nerve; 6, 7, muscles of the mandibular cavity, sup. oblique'', supplied by IVth nerve and muscle lost, there must have been also seven corresponding ventral or splanchnic segments supplied by the trigeminal. At present the evidence for such segments is nothing like so strong as for the corresponding somatic ones; there are, however, certain suggestive facts which point distinctly in this direction in connection with both the motor and sensory parts of the trigeminal. The origin of the trigeminal motor fibres in the central nervous system is most striking. We may take it for granted that a nucleus of cells giving origin to one or more segmental motor nerves will possess a greater or less longitudinal extension in the central nervous system, according to the number of fused separate segmental centres it represents. Thus a nucleus such as that of the IVth nerve or of the facial is small and compact in comparison to the extensive conjoint nucleus of the vagus and cranial accessory.

Upon examination of the motor nucleus of the trigeminal, we find a compact or well-defined nucleus, the ''nucl. masticatorius'', the nerves of which supply the masseter, temporal, and other muscles, so that the anatomical evidence at first sight appears to bear out van Wijhe's conclusion that the motor trigeminal supplies at most two segments. Further examination, however, shows that this is not all, for the extraordinary so-called descending root of the Vth must be taken into consideration in any question of the origin of the motor elements, just as the equally striking ascending root enters into the consideration of the meaning of the sensory elements of the Vth.

It is not necessary here to discuss the controversy as to whether this descending root is motor or sensory. It is universally considered at present to be motor, and is believed to supply, as Kölliker suggested, among other muscles, the m. tensor tympani and the m. tensor veli palati. It is thus described by Obersteiner—

"From the region of the mid-brain the motor root receives an important addition of thick fibres, which form the cerebral or descending root. The large, round vesicular cells from which the fibres of the descending root arise form no single compact group, but are partly single, partly arranged like little bunches of grapes, as far as the region of the anterior corpora quadrigemina. The further we go brainwards, the smaller is the number of fibres. In the region of the anterior corpora quadrigemina, the few cells of origin are found more and more median; so that the uppermost trigeminal fibres descend in curves almost from the mid-line, as is shown by the exceptional occurrence of one or more of the characteristic cells above the aqueduct. At the height of the posterior commissure one finds the last of these trigeminal cells."

The anatomy of the Vth nerve reveals, then, three most striking facts:—

1. The motor nucleus of the Vth extends from the very commencement of the infra-infundibular region to nearly the commencement of the nucleus of the VIIth; in other words, the motor nucleus of the Vth extends through the whole prosomatic region, just as it must have done originally if its motor nerves supplied the muscles of the prosomatic appendages. Such an extended range of origin is indicative of the remains of an equally extended series of segmental centres or ganglia.

2. Of these centres the caudalmost have alone remained large and vigorous, constituting the nucleus masticatorius, which in the fish is divided into an anterior and posterior group, thus indicating a double rather than a single nucleus; while the foremost ones have dwindled away until they are represented only by the cells of the descending root, the muscles of these segments being still represented by possibly the tensor veli palati and the other muscles innervated from these cells.

3. The headmost of these cells takes up actually a position dorso-lateral to the central canal, so that the groups on each side nearly come together in the mid-dorsal line; a very unique and extraordinary position for a motor cell-group, but not improbable when we recall to mind Brauer's assertion as to the shifting of the foremost prosomatic ganglion-cells of the scorpion from the ventral to the dorsal side of the alimentary canal.

On the sensory side the evidence is also suggestive, the question here being not so much the distribution of the sensory nerves as the number of ganglia belonging to each of the cranial nerves.

With respect to this question, morphologists have come to the conclusion that there is a marked difference between spinal and cranial nerves, in that whereas the posterior root-ganglia of the spinal nerves arise from the central nervous system itself, i.e. from the neural crest, the ganglia of the cranial nerves arise partly from the neural crest, partly from the proliferation of cells on the surface of the animal; and because of the situation of these proliferating epidermal patches over the gill-clefts in the case of the vagus and glossopharyngeal nerves, they have been called by Froriep and Beard branchial sense-organs. Beard divides the cranial ganglia into two sets, one connected with the neural ridges, called the neural ganglia, and the other connected with the surface-cells, which he calls the lateral ganglia. This second set corresponds to Kupffer's epibranchial ganglia. Now it is clear that in the case of the vagus nerve, where, as is well shown in Ammocœtes, the nerve is not a single segmental nerve, but is in reality made up of a number of nerves going to separate branchial segments, the indication of such segments is not given by the main vagus ganglion or neural ganglion, but by the series of lateral ganglia. So also it is argued in the case of the trigeminal, that if in addition to the ganglion-cells arising from the neural crest separate ganglion-masses are found in the course of development, in connection with proliferating patches of the surface (plakodes, Kupffer calls them), then such isolated lateral ganglia are indications of separate segments, just as in the case of the vagus, even though the separate segments do not show themselves in the adult. So far the argument appears to me just, but the further conclusion that the presence of such plakodes shows the previous existence of branchial sense-organs, and, therefore, that such ganglia are epibranchial ganglia, indicating the position of a lost gill-slit, is not justified by the premises. If, as I suppose, the trigeminal nerve supplied a series of non-branchial appendages serially homologous with the branchial appendages supplied by the vagus, then it is highly probable that the trigeminal should behave with respect to its sensory ganglia similarly to the vagus nerve, without having anything to do with branchiæ.

Such plakodal ganglia, then, may give valuable indication of non-branchial segments as well as of branchial segments. The researches of Kupffer on the formation of the trigeminal ganglia in Ammocœtes are the chief attempt to find out from the side of the sensory ganglia the number of segments originally belonging to the trigeminal. The nature and result of these researches is described in my previous paper (Journal of Anatomy and Physiology, vol. xxxiv.), and it will suffice here to state that he himself concludes that the trigeminal originally supplied five at least, probably six, segments. As I have stated there, the evidence as given by him seems to me to indicate even as many as seven segments.

In the full-grown Ammocœtes, as is well known, there are two distinct ganglia belonging to the trigeminal, the one the ganglion of the ramus ophthalmicus, the other the main ganglion.

According to Kupffer the larval Ammocœtes possesses three sets of ganglia, not two, for between the foremost and hindmost ganglion he describes a nerve (x., Fig. 113), with four epibranchial ganglia, which do not persist as separate ganglia, but either disappear or are absorbed into the two main ganglia (Fig. 113). This discovery of Kupffer's is very suggestive, for, as already stated, a transformation takes place when the Ammocœtes is 5 mm. long, so that the arrangement of the parts before that period is distinctly more indicative of the ancestral arrangement than any later one.

If we use the name plakodal ganglia to represent that part of these ganglia which was originally connected with the skin, then Kupffer's researches assert that in the larval Ammocœtes there were seven such plakodal ganglia, one in front belonging to the foremost trigeminal ganglion, two behind, parts of the hindmost ganglion, and four in between, which do not exist later as separate ganglia.



A-B, the line of epibranchial ganglia; au., auditory capsule; nc., notochord; Hy., tube of hypophysis; Or., oral cavity; u.l., upper lip; l.l. lower lip; vel., septum between oral and respiratory cavities; V., VII., IX., X., cranial nerves; x., nerve with four epibranchial ganglia.

In accordance with the views put forward in this book, a possible interpretation of these plakodal ganglia would be given as follows:—

Beard, who, after Froriep, drew attention to this relation of the cranial ganglia to special skin-patches, has compared them with the parapodial ganglia of annelids, i.e. ganglia in connection with annelidan appendages; whether we are here obtaining a glimpse of the far-off annelidan ancestry of both arthropods and vertebrates it would be premature at present to say. It is natural enough to expect, on my view, to find evidence of annelidan ancestry in vertebrate embryology (as has been so often asserted to be the case), seeing that undoubtedly the Arthropoda are an advanced stage of Annelida; and, indeed, the way is not a long one when we consider Beecher's evidence that the Trilobita belong to the Phyllopoda, certainly a primitive crustacean group, which Bernard derives directly from the annelid group Chætopoda. If, then, these plakodal ganglia indicate the former presence of appendages, we obtain this result:—The foremost ganglion on each side possesses one plakodal ganglion, and therefore indicates an anterior pair of appendages, possibly the cheliceræ. Then comes the peculiar nerve with four plakodal ganglia indicating on each side four appendages close together, possibly the endognaths. Then, finally, on each side, the second large ganglion with two plakodal ganglia, indicating two pairs of appendages, possibly the ectognaths and the metastoma.