Origin of Vertebrates/Chapter IV

The respiratory apparatus in all the terrestrial vertebrates is of the same kind—one single pair of lungs. These lungs originate as a diverticulum of the alimentary canal. On the other hand, the aquatic vertebrates breathe by means of a series of branchiæ, or gills, which are arranged segmentally, being supported by the segmental branchial cartilaginous bars, as already mentioned in the last chapter.

The transition from the gill-bearing to the lung-bearing vertebrates is most interesting, for it has been proved that the lungs are formed by the modification of the swim-bladder of fishes; and in a group of fishes, the Dipnoi, or lung-fishes, of which three representatives still exist on the earth, the mode of transition from the fish to the amphibian is plainly visible, for they possess both lungs and gills, and yet are not amphibians, but true fishes. But for the fortunate existence of Ceratodus in Australia, Lepidosiren in South America, and Protopterus in Africa, it would have been impossible from the fossil remains to have asserted that any fish had ever existed which possessed at the same moment of time the two kinds of respiratory organs, although from our knowledge of the development of the amphibian we might have felt sure that such a transitional stage must have existed. Unfortunately, there is at present no likelihood of any corresponding transitional stage being discovered living on the earth in which both the dorsal arthropod alimentary canal and the ventral vertebrate one should simultaneously exist in a functional condition; still it seems to me that even if Ceratodus, Lepidosiren, and Protopterus had ceased to exist on the earth, yet the facts of comparative anatomy, together with our conception of evolution as portrayed in the theory of natural selection, would have forced us to conclude rightly that the amphibian stage in the evolution of the vertebrate phylum was preceded by fishes which possessed simultaneously lungs and gills.

In the preceding chapter the primitive cartilaginous vertebrate skeleton, as found in Ammocœtes, was shown to correspond in a marvellous manner to the cartilaginous skeleton of Limulus. In a later chapter I will deal with the formation of the cranium from the prosomatic skeleton; in this chapter it is the mesosomatic skeleton which is of interest, and the consideration of the necessary consequences which logically follow upon the supposition that the branchial cartilaginous bars of Limulus are homologous with the branchial basket-work of Ammocœtes.

Seeing that in both cases the cartilaginous bars of Limulus and Ammocœtes are confined to the branchial region, their homology of necessity implies an homology of the two branchial regions, and leads directly to the conclusion that the branchiæ of the vertebrate were derived from the branchiæ of the arthropod, a conclusion which, according to the generally accepted view of the origin of the respiratory region in the vertebrate, is extremely difficult to accept; for the branchiæ of Limulus and of the Arthropoda in general are part of the mesosomatic appendages, while the branchiæ of vertebrates are derived from the anterior part of the alimentary canal. This conclusion, therefore, implies that the vertebrate has utilized in the formation of the anterior portion of its new alimentary canal the branchial appendages of the palæostracan ancestor.



The segments and appendages on the right are numbered in correspondence with the cranial system of lateral nerve-roots as found in vertebrates. M., metastoma. The surface ornamentation is represented on the first segment posterior to the branchial segments. The opercular appendage is marked out by dots.

Let us consider dispassionately whether such a suggestion is a priori so impossible as it at first appears. One of the principles of evolution is that any change which is supposed to have taken place in the process of formation of one animal or group of animals from a lower group must be in harmony with changes which are known to have occurred in that lower group. On the assumption, therefore, that the vertebrate branchiæ represent the branchial portion of the arthropod mesosomatic appendages which have sunk in and so become internal, we ought to find that in members of this very group such inclusion of branchial appendages has taken place. This, indeed, is exactly what we do find, for in all the scorpion tribe, which is acknowledged to be closely related to Limulus, there are no external mesosomatic appendages, but in all cases these appendages have sunk into the body, have disappeared as such, and retained only the vital part of them—the branchiæ. In this way the so-called lung-books of the scorpion are formed, which are in all respects homologous with the branchiæ or gill-books of Limulus. Now, as already mentioned, the lords of creation in the palæostracan times were the sea-scorpions, which, as is seen in Fig. 62, resembled the land-scorpions of the present day in the entire absence of any external appendages on the segments of the mesosomatic region. As they lived in the sea, they must have breathed with gills, and those branchial appendages must have been internal, just as in the land-scorpions of the present time. Indeed, markings have been found on the internal side of the segments 1-5, Fig. 62, which are supposed to indicate branchiæ, and these segments are therefore supposed to have borne the branchiæ. Up to the present time no indication of gill-slits has been found, and we cannot say with certainty how these animals breathed. Further, in the Upper Silurian of Lesmahago, Lanarkshire, a scorpion (Palæophonus Hunteri), closely resembling the modern scorpion, has been found, which, as Lankester states, was in all probability aquatic, and not terrestrial in its habits. How it breathed is unknown; it shows no signs of stigmata, such as exist in the scorpion of to-day.

Although we possess as yet no certain knowledge of the position of the gill-openings in these ancient scorpion-like forms, what we can say with certainty—and that is the important fact—is, that at the time when the vertebrates appeared, a very large number of the dominant arthropod race possessed internally-situated branchiæ, which had been directly derived from the branchiæ-bearing appendages of their Limulus-like kinsfolk.

This abolition of the branchiæ-bearing appendages as external organs of locomotion, with the retention of the important branchial portion of the appendage as internal branchiæ, is a very important suggestion in any discussion of the way vertebrates have arisen from arthropods; for, if the same principle is of universal application, it leads directly to the conclusion that whenever an appendage possesses an organ of vital importance to the animal, that organ will remain, even though the appendage as such completely vanishes. Thus, as will be shown later, special sense-organs such as the olfactory remain, though the animal no longer possesses antennæ; the important excretory organs, the coxal glands, and important respiratory organs, the branchiæ, are still present in the vertebrate, although the appendages to which they originally belonged have dwindled away, or, at all events, are no longer recognizable as arthropod appendages.

Passing from a priori considerations to actual facts, it is advisable to commence with the innervation of the branchial segments; for, seeing that the foundation of the whole of this comparative study of the vertebrate and the arthropod is based upon the similarity of the two central nervous systems, it follows that we must look in the first instance to the innervation of any organ or group of organs in order to find out their relationship in the two groups of animals.

The great characteristic of the vertebrate branchial organs is their segmental arrangement and their innervation by the vagus group of nerves, i.e. by the hindermost group of the cranial segmental nerves. These cranial nerves are divided by Gegenbaur into two great groups—an anterior group, the trigeminal, which supplies the muscles of mastication, and a posterior group, the vagus, which is essentially respiratory in function. Of these two groups, I will consider the latter group first.

In Limulus the great characteristic of the branchial region is its pronounced segmental arrangement, each pair of branchial appendages belonging to a separate segment. This group of segments forms the mesosoma, and these branchial appendages are the mesosomatic appendages. Anterior to them are the segments of the prosoma, which bear the prosomatic or locomotor appendages. The latter are provided at their base with gnathites or masticating apparatus, so that the prosomatic group of nerves, like the trigeminal group in the vertebrate, comprises essentially the nerves subserving the important function of mastication. As already pointed out, the brain-region of the vertebrate is comparable to the supra-œsophageal and infra-œsophageal ganglia of the invertebrate, and it has been shown (p. 54) how, by a process of concentration and cephalization, the foremost region of the infra-œsophageal ganglia becomes the prosomatic region, and is directly comparable to the trigeminal region in the vertebrate; while the hindermost region is formed from the concentration of the mesosomatic ganglia, and is directly comparable to the medulla oblongata, i.e. to the vagus region of the vertebrate brain.

As far, then, as concerns the centres of origin of these two groups of nerves and their exits from the central nervous system, they are markedly homologous in the two groups of animals.

It has often been held that the arrangements of the vertebrate nervous system differ from those of other segmented animals in one important particular. The characteristic of the vertebrate is the origin of every segmental nerve from two roots, of which one contains the efferent fibres, while the other possesses a sensory ganglion, and contains only afferent fibres. This arrangement, which is found along the whole spinal cord of all vertebrates, is not found in the segmental nerves of the invertebrates; and as it is supposed that the simpler arrangement of the spinal cord was the primitive arrangement from which the vertebrate central nervous system was built up, it is often concluded that the animal from which the vertebrate arose must have possessed a series of nerve-segments, from each of which there arose bilaterally ventral (efferent) and dorsal (afferent) roots.

Now, the striking fact of the vertebrate segmental nerves consists in this, that, as far as their structure and the tissues which they innervate are concerned, the cranial segmental nerves are built up on the same plan as the spinal; but as far as concerns their exit from the central nervous system they are markedly different. A large amount of ingenuity, it is true, has been spent in the endeavour to force the cranial nerves into a series of segmental nerves, which arise in the same way as the spinal by two roots, of which the ventral series ought to be efferent and the dorsal series afferent, but without success. We must, therefore, consider the arrangement of the cranial segmental nerves by itself, separately from that of the spinal nerves, and the problem of the origin of the vertebrate segmental nerves admits of two solutions—either the cranial arrangement has arisen from a modification of the spinal, or the spinal from a simplification of the cranial. The first solution implies that the spinal cord arrangement is older than the cranial, the second that the cranial is the oldest.

In my opinion, the evidence of the greater antiquity of the cranial region is overwhelming.

The evidence of embryology points directly to the greater phylogenetic antiquity of the cranial region, for we see how, quite early in the development, the head is folded off, and the organs in that region thereby completed at a time when the spinal region is only at an early stage of development. We see how the first of the trunk somites is formed just posteriorly to the head region, and then more and more somites are formed by the addition of fresh segments posteriorly to the one first formed. We see how, in Ammocœtes, the first formed parts of the skeleton are the branchial bars and the basi-cranial system, while the rudiments of the vertebræ do not appear until the Petromyzon stage. We see how, with the elongation of the animal by the later addition of more and more spinal segments, organs, such as the heart, which were originally in the head, travel down, and the vagus and lateral-line nerves reach their ultimate destination. Again, we see that, whereas the cranial nerves, viz. the ocular motor, the trigeminal, facial, auditory, glossopharyngeal, and vagus nerves, are wonderfully fixed and constant in all vertebrates, the only shifting being in the spino-occipital region, in fact, at the junction of the cranial and spinal region, the spinal nerves, on the other hand, are not only remarkably variable in number in different groups of animals, but that even in the same animal great variations are found, especially in the manner of formation of the limb-plexuses. Such marked meristic variation in the spinal nerves, in contrast to the fixed character of the cranial nerves, certainly points to a more recent formation of the former nerves.

Also the observations of Assheton on the primitive streak of the rabbit, and on the growth in length of the frog embryo, have led him to the conclusion that, as in the rabbit so in the frog, there is evidence to show that the embryo is derived from two definite centres of growth: the first, phylogenetically the oldest, being a protoplasmic activity, which gives rise to the anterior end of the embryo; the second, one which gives rise to the growth in length of the embryo. This secondary area of proliferation coincides with the area of the primitive streak, and he has shown, in a subsequent paper, by means of the insertion of sable hairs into the unincubated blastoderm of the chick, that a hair inserted into the centre of the blastoderm appears at the anterior end of the primitive streak, and subsequently is found at the level of the most anterior pair of somites.

He then goes on to say—

"From these specimens it seems clear that all those parts in front of the first pair of mesoblastic somites—that is to say, the heart, the brain and medulla oblongata, the olfactory, optic, auditory organs and foregut—are developed from that portion of the unincubated blastoderm which lies anterior to the centre of the blastoderm, and that all the rest of the embryo is formed by the activity of the primitive streak area."

In other words, the secondary area of growth, i.e. the primitive streak area, includes the whole of the spinal cord region, while the older primary centre of growth is coincident with the cranial region.

In searching, then, for the origin of the segmental nerves, we must consider the type on which the cranial nerves are arranged rather than that of the spinal nerves.

The first striking fact occurs at the spino-occipital region, where the spinal cord merges into the medulla oblongata, for here in the cervical region we find each spinal segment gives origin to three distinct roots, not two—a dorsal root, a ventral root, and a lateral root. This third root gives origin to the spinal accessory nerve, and in the region of the medulla oblongata these lateral roots merge directly into the roots of the vagus nerve; more anteriorly the same system continues as the roots of the glossopharyngeal nerve, as the roots of the facial nerve, and as a portion, especially the motor portion, of the trigeminal nerve. Now, all these nerves belong to a well-defined system of nerves, as Charles Bell pointed out in 1830, a system of nerves concerned with respiration and allied mechanisms, such as laughing, sneezing, mastication, deglutition, etc., nerves innervating a set of muscles of very different kind from the ordinary body-muscles concerned with locomotion and equilibration. Also the centres from which these motor nerves arise are well defined, and form cell-masses in the central nervous system, quite separate from those which give origin to somatic muscles.

This original idea of Charles Bell, after having been ignored for so long a time, is now seen to be a very right one, and it is an extraordinary thing that his enunciation of the dual nature of the spinal roots, which was, to his mind, of subordinate importance, should so entirely have overshadowed his suggestion, that in addition to the dorsal and ventral roots, a lateral system of nerves existed, which were not exclusively sensory or exclusively motor, but formed a separate system of respiratory nerves.

Further, anatomists divide the striated muscles of the body into two great natural groups, characterized by a difference of origin and largely by a difference of appearance. The one set is concerned with the movements of internal organs, and is called visceral, the other is derived from the longitudinal sheet of musculature which forms the myotomes of the fish, and has been called parietal or somatic. The motor nerves of these two sets of muscles correspond with the lateral or respiratory and ventral roots respectively.

Finally, it has been shown that the segments of which a vertebrate is composed are recognizable in the embryo by the segmented manner in which the musculature is laid down, and van Wijhe has shown that in the cranial region two sets of muscles are laid down segmentally, thus forming a dorsal and ventral series of commencing muscular segments. Of these the anterior segments of the dorsal series give origin to the striated muscles of the eye which are innervated by the IIIrd (oculomotor), IVth (trochlearis), and VIth (abducens) nerves, while the posterior segments give origin to the muscles from the cranium to the shoulder-girdle, innervated by the XIIth (hypoglossal) nerve. The ventral series of segments give origin to the musculature supplied by the trigeminal, facial, glossopharyngeal, and vagus nerves.

Also, the afferent or sensory nerves of the skin over the whole of this head-region are supplied by the trigeminal nerve, while the afferent nerves to the visceral surfaces are supplied by the vagus, glossopharyngeal and facial nerves.

In van Wijhe's original paper he arranged the segments belonging to the cranial nerves in the following table:—

As is seen in the table, van Wijhe attempts to arrange the cranial segmental nerves into dorsal and ventral roots, in accordance with the arrangement in the spinal region. In order to do this he calls the Vth, VIIth, IXth, and Xth nerves dorsal roots, although they are not purely sensory nerves, but contain motor fibres as well.

It is not accidental that he should have picked out for his dorsal roots the very nerves which form Charles Bell's lateral series of roots, inasmuch as this system of lateral roots, apart from dorsal and ventral roots, really is, as Charles Bell thought, an important separate system, dependent upon a separate segmentation in the embryo of the musculature supplied by these roots. This segmentation may receive the name of visceral or splanchnic in contradistinction to somatic, since all the muscles without exception belong to the visceral group of striated muscles.

These observations of van Wijhe lead directly to the following conclusion. In the cranial region there is evidence of a double set of segments, which may be called somatic and splanchnic. The somatic segments, consisting of the outer skin and the body musculature, are doubly innervated as are those of the spinal cord by a series of ventral motor roots, the oculomotor or IIIrd nerve, the trochlear or IVth nerve, the abducens or VIth nerve, and the hypoglossal or XIIth nerve, and by a series of dorsal sensory roots, the sensory part of the trigeminal or Vth nerve. But the splanchnic segments are innervated by single roots, the vagus or Xth nerve, glossopharyngeal or IXth nerve, facial or VIIth nerve, and trigeminal or Vth nerve, which are mixed, containing both sensory and motor fibres, thus differing markedly from the arrangement of the spinal nerves.

From this sketch it follows that the arrangement seen in the spinal cord, would result from the cranial arrangement if this third system of lateral roots were left out. Further, since the cranial system is the oldest, we must search in the invertebrate ancestor for a tripartite rather than a dual system of nerve-roots for each segment; a system composed of a dorsal root supplying only the sensory nerves of the skin-surfaces, a lateral mixed root supplying the system connected with respiration with both sensory and motor fibres, and a ventral root supplying the motor nerves to the body-musculature.

If the argument used so far is correct, and this tripartite system of nerve-roots, as seen in the cranial nerves of the vertebrate, really represents the original scheme of innervation in the palæostracan ancestor, then it follows that each segment of Limulus ought to be supplied by three nerves—(1), a sensory nerve supplying its own portion of the skin-surface of the prosomatic and mesosomatic carapaces; (2), a lateral mixed nerve supplying exclusively the appendage of the segment, for the appendages carry the respiratory organs; and (3), a motor nerve supplying the body-muscles of the segment.

It is a striking fact that Milne-Edwards describes the nerve-roots in exactly this manner. The great characteristic of the nerve-roots in Limulus as in other arthropods is the large appendage-nerve, which is always a mixed nerve; in addition, there is a system of sensory nerves to the prosomatic and mesosomatic carapaces, called by him the epimeral nerves, which are purely sensory, and a third set of roots which are motor to the body-muscles, and possibly also sensory to the ventral surface between the appendages.

Moreover, just as in the vertebrate central nervous system the centres of origin of the motor nerves of the branchial segmentation are distinct from those of the somatic segmentation, so we find, from the researches of Hardy, that a similar well-marked separation exists between the centres of origin of the motor nerves of the appendages and those of the somatic muscles in the central nervous system of Branchipus and Astacus.

In the first place, he points out that the nervous system of Branchipus is of a very primitive arthropod type; that it is, in fact, as good an example of an ancient type as we are likely to find in the present day; a matter of some importance in connection with my argument, since the arthropod ancestor of the vertebrate, such as I am deducing from the study of Ammocœtes, must undoubtedly have been of an ancient type, more nearly connected with the strange forms of the trilobite era than with the crabs and spiders of the present day.

His conclusions with respect to Branchipus may be tabulated as follows:—

1. Each ganglion of the ventral chain is formed mainly for the innervation of the appendages.

2. Each ganglion is divided into an anterior and posterior division, which are connected respectively with the motor and sensory nerves of the appendages.

3. The motor nerves of the appendages arise as well-defined axis-cylinder processes of nerve-cells, which are arranged in well-defined groups in the anterior division of the ganglion.

4. A separate innervation exists for the muscles and sensory surfaces of the trunk. The trunk-muscles consist of long bundles, from which slips pass off to the skin in each segment; they are thus imperfectly segmented. In accordance with this, a diffuse system of nerve-fibres passes to them from certain cells on the dorsal surface of each lateral half of the ganglion. These cell-groups are therefore very distinct from those which give origin to the motor appendage-nerves, and, moreover, are not confined to the ganglion, but extend for some distance into the interganglionic region of the nerve-cords which connect together the ganglia of the ventral chain.

Hardy's observations, therefore, combined with those of Milne-Edwards, lead to the conclusion that in such a primitive arthropod type as my theory postulates, each segment was supplied with separate sensory and motor somatic nerves, and with a pair of nerves of mixed function, devoted entirely to the innervation of the pair of appendages; that also, in the central nervous system, the motor nerve-centres were arranged in accordance with a double set of segmented muscles in two separate groups of nerve-cells. These nerve-cells in the one case were aggregated into well-defined groups, which formed the centres for the motor nerves of the markedly segmented muscles of the appendages, and in the other case formed a system of more diffused cells, less markedly aggregated into distinct groups, which formed the centres for the imperfectly segmented somatic muscles.

Such an arrangement suggests that in the ancient arthropod type a double segmentation existed, viz. a segmentation of the body, and a segmentation due to the appendages. Undoubtedly, the segments originally corresponded absolutely as in Branchipus, and every appendage was attached to a well-defined separate body-segment. In, however, such an ancient type as Limulus, though the segmentation may be spoken of as twofold, yet the number of segments in the prosomatic and mesosomatic regions are much more clearly marked out by the appendages than by the divisions of the soma; for, in the prosomatic region such a fusion of somatic segments to form the tergal prosomatic carapace has taken place that the segments of which it is composed are visible only in the young condition, while in the mesosomatic region the separate somatic segments, though fused to form the mesosomatic carapace, are still indicated by the entapophysial indentations.

Clearly, then, if the mesosomatic branchial appendages of forms related to Limulus were reduced to the branchial portion of the appendage, and that branchial portion became internal, just as is known to be the case in the scorpion group, we should obtain an animal in which the mesosomatic region would be characterized by a segmentation predominantly branchial, which might be termed, as in vertebrates, the branchiomeric segmentation, but yet would show indications of a corresponding somatic or mesomeric segmentation. The nerve supply to these segments would consist of—

1. The epimeral purely sensory nerves to the somatic surface, equivalent in the vertebrate to the ascending root of the trigeminal.

2. The mixed nerves to the internal branchial segments, equivalent in the vertebrate to the vagus, glossopharyngeal, and facial.

3. The motor nerves to the somatic muscles, equivalent in the vertebrate to the original nerve-supply to the somatic muscles belonging to these segments, i.e. to the muscles derived from van Wijhe's 4th, 5th, and 6th somites.

Further, the centres of origin of these appendage-nerves would form centres in the central nervous system separate from the centres of the motor nerves to the somatic muscles, just as the centres of origin of the motor parts of the facial, vagus, and glossopharyngeal nerves form groups of cells quite distinct from the centres for the hypoglossal, abducens, trochlear, and oculomotor nerves.

In fact, if the vertebrate branchial nerves are looked upon as the descendants of nerves which originally supplied branchial appendages, then every question connected with the branchial segmentation, with the origin and distribution of these nerves, receives a simple and adequate solution—a solution in exact agreement with the conclusion that the vertebrate arose from a palæostracan ancestor.

It would, therefore, be natural to expect that the earliest fishes breathed by means of branchial appendages situated internally, and that the evidence for such appendages would be much stronger in them than in more recent fishes.

Although we know nothing of the nature of the respiratory apparatus in the extinct fishes of Silurian times, we have still living, in the shape of Ammocœtes, a possible representative of such types. If, then, we find, as is the case, that the respiratory apparatus of Ammocœtes differs markedly from that of the rest of the fishes, and, indeed, from that of the adult form or Petromyzon, and that that very difference consists in a greater resemblance to internal branchial appendages in the case of Ammocœtes, then we may feel that the proof of the origin of the branchial apparatus of the vertebrate from the internal branchial appendages of the invertebrate has gained enormously.

In order to make clear the nature of the branchial segments in Ammocœtes, I have divided the head-part of the animal by means of a longitudinal horizontal section into halves—ventral and dorsal—as shown in Figs. 63 and 64. These figures are each a combination of a section and a solid drawing. The animal was slit open by a longitudinal section in the neighbourhood of the gill-slits, and each half was slightly flattened out, so as to expose the ventral and dorsal internal surfaces respectively. The structures in the cut surface were drawn from one of a series of horizontal longitudinal sections taken through the head of the animal. These figures show that the head-region of Ammocœtes consists of two chambers, the contents of which are different. In front, an oral or stomodæal chamber, which contains the velum and tentacles, is enclosed by the upper and lower lips, and was originally separated by a septum from the larger respiratory chamber, which contains the separate pairs of branchiæ. A glance at the two drawings shows clearly that Rathke's original description of this chamber is the natural one, for he at that time, looking upon Ammocœtes branchialis as a separate species, described the branchial chamber as containing a series of paired gills, with the gill-openings between consecutive gills. His branchial unit or gill, therefore, was represented by each of the so-called diaphragms, which, as seen in Figs. 63, 64, are all exactly alike, except the first and the last. Any one of these is represented in section in Fig. 65, and represents a branchial unit in Rathke's view and in mine. Clearly, it may be described as a branchial appendage which projects into an open pharyngeal chamber, so that the series of such appendages divides the chamber into a series of compartments, each of which communicates with the exterior by means of a gill-slit, and with each other by means of the open space between opposing appendages.

Each of these appendages possesses its own cartilaginous bar (Br. cart.), as explained in Chapter III.; each possesses its own branchial or visceral muscles (coloured blue in Figs. 63 and 64), separated absolutely from the longitudinal somatic muscles (coloured dark red in Figs. 63 and 64) by a space (Sp.) containing blood and peculiar fat-cells, etc. Each possesses its own afferent branchial blood-vessel from the ventral aorta, and its own efferent vessel to the dorsal aorta (Fig. 65, a. br. and v. br.). Each possesses its own segmental nerve, which supplies its own branchial muscles and no others with motor fibres, and sends sensory fibres to the general surface of each appendage, as also to the special sense-organs in the shape of the epithelial pits (S., Fig. 65) arranged along the free edges of the diaphragms; each of these nerves possesses its own ganglion—the epibranchial ganglion.



Somatic muscles coloured red. Branchial and visceral muscles coloured blue. Tubular constrictor muscles distinguished from striated constrictor muscles by simple hatching. Tent., tentacles; ''Tent. m.c., muco-cartilage of tentacles; Vel. m.c., muco-cartilage of the velum; Hy. m.c., muco-cartilage of the hyoid segment; Ps. br., pseudo-branchial groove; Br. cart., branchial cartilages; Sp., space between somatic and splanchnic muscles; Th. op., orifice of thyroid; H.'', heart.



The work of Miss Alcock has shown that the segmental branchial nerve supplies solely and absolutely such an appendage or branchial segment, and does not supply any portion of the neighbouring branchial segments. The nerve-supply in Ammocœtes gives no countenance to the view that the original unit was a branchial pouch, the two sides of which each nerve supplied, but is strong evidence that the original unit was a branchial appendage, which was supplied by a single nerve with both motor and sensory fibres.

Any observer having before him only this picture of the respiratory chamber of Ammocœtes, upon which to base his view of a vertebrate respiratory chamber, would naturally look upon the branchial unit of a vertebrate as a gilled appendage projecting into the open cavity of the anterior part of the alimentary canal or pharynx. This is not, however, the usual conception. The branchial unit is ordinarily described as a gill-pouch, which possesses two openings or slits, an internal one into the lumen of the alimentary canal, and an external one into the surrounding medium. This view is based upon embryological evidence of the following character:—

The alimentary canal of all vertebrates forms a tube stretching the whole length of the animal; the anterior part of this tube becomes pouched on each side at regular intervals, and the walls of each pouch becoming folded form the respiratory surfaces or gills. The openings of these separate pouches into the central lumen of the gut form the internal gill-pouch openings; the other extremity of the pouch approaches the external surface of the animal, and finally breaks through to form a series of external gill-pouch openings.

From the mesoblastic tissue, between each gill-pouch, there is formed a supporting cartilaginous bar, to which are attached a system of branchial muscles, with their nerves and blood-vessels. These cartilaginous bars, in all fishes above the Cyclostomata, form a supporting framework for the internal gill-slit, so that the gills are situated externally to them; the more primitive arrangement is, as already mentioned, a system of cartilaginous bars, extra-branchial in position, so that the gills are situated internally to them.

From this description of the mode of formation of the respiratory apparatus in water-breathing vertebrates the conception has arisen of the gill-pouch as the branchial unit, a conception which is absolutely removed from all idea of a branchial unit such as is found in an arthropod, viz. an appendage.

This conception of spaces as units pervades the whole of embryology, and is the outcome of the gastrula theory—a theory which teaches that all animals above the Protozoa are derived from a form which by invagination of its external surface formed an internal cavity or primitive gut. From pouches of this gut other cavities were said to be formed, called cœlomic cavities, and thus arose the group of cœlomatous animals. To speak of the developmental history of animals in terms of spaces; to speak of the atrophy of a cavity as though such a thing were possible, is, to my mind, the wrong way of looking at the facts of anatomy. It resembles the description of a net as a number of holes tied together with string, which is not usually considered the best method of description.

There are two ways in which a series of pouches can be formed from a simple tube without folding, either by a thinning at regular intervals of the original tissue surrounding the tube, or by the ingrowth into the tube of the surrounding tissue at regular intervals, thus—



In the first case (A) the formation of a pouch is the significant act, and therefore the branchial segments might be expressed in terms of pouches. In the second case (B) the formation of a pouch is brought about in consequence of the ingrowth of the mesoblastic tissues at intervals; here, although the end-result is the same as in the first case, the pouch-formation is only secondary, the true branchial unit is the mesoblastic ingrowth.

The evidence all points directly to the second method of formation. Thus Shipley, in his description of the development of the lamprey, says—

"The gill-slits appear to me to be the result of the ventral downgrowth of mesoblast taking place only at certain places, these forming the gill-bars. Between each downgrowth the hypoblastic lining of the alimentary canal remains in contact with the epiblast; here the gill-opening subsequently appears about the twenty-second day."

Dohrn describes and gives excellent pictures of the growth of the diaphragms, as the Ammocœtes grows in size, pictures which are distinctly reminiscent of the corresponding illustrations given by Brauer of the growth of the internal gills in the scorpion embryo.

Another piece of evidence confirmatory of the view that the branchial segments are really of the nature of internal appendages, as the result of which gill-pouches are formed, is given by the presence in each of these branchial bars or diaphragms of a separate cœlomic cavity. From the walls of this cavity the branchial muscles and cartilaginous bar are formed.

Now, from an embryological point of view, the vertebrate shows that it is a segmented animal by the formation of somites, which consist of a series of divisions of the cœlom, of which the walls form a series of muscular and skeletal segments. In the head-region, as already mentioned, such cœlomic divisions form two rows—a dorsal and a ventral set. From the walls of the dorsal set the somatic musculature is formed. From those of the ventral set the branchial musculature. From the latter also the branchial cartilaginous bars are formed. Thus Shipley, in his description of the development of the lamprey, says: "The mesoblast between the gills arranges itself into head-cavities, and the walls of these cavities ultimately form the skeleton of the gill-arches."

Similarly, in the arthropod, the segments in the embryo are marked out by a series of cœlomic cavities and Kishinouye has described in Limulus a separate cœlomic cavity for every one of the mesosomatic or branchial segments, and he states that in Arachnida the segmental cœlomic cavities extend into the limbs. These cavities both in the vertebrate and in the arthropod disappear before the adult condition is reached.

The whole evidence thus points strongly to the conclusion that the true branchial segmental units are the branchial bars or diaphragms, not the pouches between them.

It is possible to understand why such prominence has been given to the conception of the branchial unit as a gill-pouch rather than as a gill-appendage, when the extraordinary change of appearance in the respiratory chamber of the lamprey which occurs at transformation, is taken into consideration. This change is of a very far-reaching character, and consists essentially of the formation of a new alimentary canal in this region, whereby the pharyngeal chamber of Ammocœtes is cut off posteriorly from the alimentary canal, and is confined entirely to respiratory purposes, its original lumen now forming a tube called the bronchus, which opens into the mouth and into a series of branchial pouches.

In Fig. 68 I give diagrammatic illustrations taken from Nestler's paper to show the striking change which takes place at transformation, (A) representing three branchial segments of Ammocœtes, and (B) the corresponding three segments of Petromyzon. The corresponding parts in the two diagrams are shown by the cartilages (br. cart.), the sense-organs (S), and the branchial veins (V. br.); the corresponding diaphragms are marked by the figures 1, 2, 3 respectively. As is clearly seen, it is perfectly possible in the latter case to describe the respiratory chamber, as Nestler has done, as divided into a series of separate smaller chambers—the gill-pouches—by means of a series of diaphragms or branchial bars. The surface of these gill-pouches is in part thrown into folds for respiratory purposes, and each gill-pouch opens, on the one hand, into the bronchus (Bro.), and, on the other, to the exterior by means of the gill-slit. The branchial unit in Petromyzon is, therefore, according to Nestler and other morphologists, the folded opposed surfaces of two contiguous diaphragms, and each one of the diaphragms is intersegmental between two gill-pouches.

Nestler then goes on to describe the arrangement in Ammocœtes in the same terms, although there is no bronchus or gill-pouch, but only an open chamber into which these gill-bearing diaphragms project, which open chamber serves both for the passage of food and of the water for respiration. This is manifestly the wrong way to look at the matter: the adult form is derived from the larval, not vice versâ, and the transformation process shows exactly how the gills, in Rathke's sense, come together to form the bronchus and so make the gill-pouches of Petromyzon.

When we bear in mind that almost all observers consider that the internal branchiæ of the scorpion group are directly derived from branchial appendages of a kind similar to those of Limulus, it is evident that a branchial appendage such as that of Ammocœtes might also have arisen from such an appendage, because in various respects it is easier to compare the branchial appendage of Ammocœtes, than that of the scorpion group, with that of Limulus.

(After .)

In both figures the branchial cartilages (br. cart.), the branchial view (V. br.), and the sense-organs (S), are marked out in order to show corresponding points. The muscles, blood-spaces, branchial arteries, etc., of each branchial segment are not distinguished, being represented a uniform black colour. Bro., the bronchus into which each gill-pouch opens.

In the case of the scorpions, various suggestions have been made as to the manner in which such a conversion may have taken place. The most probable explanation is that given by Macleod, in which each of the branchiæ of the scorpion group is directly compared with the branchial part of the Limulus appendage which has sunk into and amalgamated with the ventral surface.

According to this view, the modification which has taken place in transforming the branchial Limulus-appendage into the branchial scorpion-appendage is a further stage of the process by which the Limulus branchial appendage itself has been formed, viz. the getting rid of the free locomotor segments of the original appendage, thus confining the appendage more and more to the basal branchial portion. So far has this process been carried in the scorpion that all the free part of the appendage has disappeared; apparently, also, the intrinsic muscles of the appendage have vanished, with the possible exception of the post-stigmatic muscle, so that any direct comparison between the branchial appendages of Limulus and the scorpions is limited to the comparison of their branchiæ, their nerves, and their afferent and efferent blood-vessels.

In the case of Ammocœtes the comparison must be made not with air-breathing but with water-breathing scorpions, such as existed in past ages in the forms of Eurypterus, Pterygotus, Slimonia, and with the crowd of trilobite and Limulus-like forms which were in past ages so predominant in the sea; forms in some of which the branchial appendages had already become internal, but which, from the very fact of these forms being water-breathers, probably resembled, in respect of their respiratory apparatus, Limulus rather than the present-day scorpion.

On the assumption that the branchial appendages of Ammocœtes, like the branchial appendages of the scorpion group, are to a certain extent comparable with those of Limulus, it becomes a matter of great interest to inquire whether the mode in which respiration is effected in Ammocœtes resembles most that of Limulus or of the scorpion.

The difference between the movements of respiration in Limulus and those of the scorpions consists in the fact that, although in both cases respiration is effected mainly by dorso-ventral muscles, these muscles are not homologous in the two cases: in the former, the dorso-ventral appendage-muscles are mainly concerned, in the latter, the dorso-ventral somatic muscles.

The paper by Benham gives a full description of the musculature of Limulus, and according to his arrangement the muscles are divided into two sets, longitudinal and dorso-ventral. Of the latter, each mesosomatic segment possesses a pair of dorso-ventral muscles, attached to the mid-ventral mesosomatic entochondrite, and to the tergal surface (Fig. 58, Dv.). These muscles are called by Benham the vertical mesosomatic muscles. I shall call them the somatic dorso-ventral muscles, in contradistinction to the dorso-ventral muscles of the branchial appendages. Of the latter, the two chief are the external branchial (Fig. 66, m$1$) and the posterior entapophysio-branchial (Fig. 66, m$1$); a third muscle is the anterior entapophysio-branchial (Fig. 66, m$2$). Of these muscles, the posterior entapophysio-branchial (m$2$) is closely attached along the branchial cartilaginous bar up to its round-headed termination on the anterior surface of the appendage. The anterior entapophysio-branchial muscle (m$1$) is attached to the branchial cartilage near the entapophysis.

In the case of the scorpion, as described by Miss Beck, the branchial appendage has become reduced to the branchiæ, and the intrinsic appendage-muscles have entirely disappeared, with the possible exception of the small post-stigmatic muscle; on the other hand, the dorso-ventral somatic muscles, which are clearly homologous with the corresponding muscles of Limulus, have remained, and become the essential respiratory muscles.

Of these two possible types of respiratory movement it is quite conceivable that in the water-breathing scorpions of olden times and in their allies, the dorso-ventral muscles of their branchial appendages may have continued their rôle of respiratory muscles, and so have given origin to the respiratory muscles of the ancestors of Ammocœtes.

The respiratory muscles of Ammocœtes are three in number, and have been described by Nestler and Miss Alcock as the adductor muscle, the striated constrictor muscle, and the tubular constrictor muscle (Fig. 65, m. add., m.c.s., and m.c.t.). Of these, the constrictor muscle (Fig. 71, m. con. str.) is in close contact with its cartilaginous bar, while the adductor (Fig. 71, m. add.) is attached to the cartilage only at its origin and insertion, and the tubular muscles (Fig. 71, m. con. tub.) have nothing whatever to do with the cartilage at all, being attached ventrally to the connective tissue in the neighbourhood of the ventral aorta (V.A.), and dorsally to the mid-line between the dorsal aorta (D.A.) and the notochord.

The close relationship of the constrictor muscle to the cartilaginous branchial bar does not favour the surmise that this muscle is homologous with the dorso-ventral somatic muscle of the scorpion. It is, however, directly in accordance with the view that this muscle is homologous with one of the dorso-ventral appendage-muscles, such as the posterior entapophysio-branchial muscle (m$2$, Fig. 66) of the Limulus appendage, especially when the homology of the Ammocœtes branchial cartilage with the Limulus branchial cartilage is borne in mind. I am, therefore, inclined to look upon the constrictor and adductor muscles of the Ammocœtes branchial segment as more likely to have been derived from dorso-ventral muscles which belonged originally to a branchial appendage, such as we see in Limulus, than from dorso-ventral somatic muscles, such as the vertical mesosomatic muscles which are found both in Limulus and scorpion. In other words, I am inclined to hold the view that the somatic dorso-ventral muscles have disappeared in this region in Ammocœtes, while dorso-ventral appendage-muscles have been retained, i.e. the exact reverse to what has taken place in the air-breathing scorpion.

I am especially inclined to this view because of the manner in which it fits in with and explains van Wijhe's results. Ever since Schneider divided the striated muscles of vertebrates into parietal and visceral, such a division has received general acceptance and, as far as the head-region is concerned, has received an explanation in van Wijhe's work; for Schneider's grouping corresponds exactly to the two segmentations of the head-mesoblast, discovered by van Wijhe, i.e. to the somatic and splanchnic striated muscles according to my nomenclature. Of these two groups the splanchnic or visceral striated musculature, innervated by the Vth, VIIth, IXth, and Xth nerves, which ought on this theory to be derived from the musculature of the corresponding appendages, is, speaking generally, dorso-ventral in direction in Ammocœtes and of the same character throughout; the somatic musculature, on the other hand, is clearly divisible, in the head region, into two sets—a spinal and a cranial set. The somatic muscles innervated by the spinal set of nerves, including in this term the spino-occipital or so-called hypoglossal nerves, are in Ammocœtes most sharply defined from all the other muscles of the body. They form the great dorsal and ventral longitudinal body-muscles, which extend dorsally as far forward as the nose and are developed embryologically quite distinctly from the others, being formed as muscle-plates (Kästchen). On the other hand, the cranial somatic muscles are the eye-muscles, the formation of which resembles that of the visceral muscles, and not of the spinal somatic. Their direction is not longitudinal, but dorso-ventral; they cannot, in my opinion, be referred to the somatic trunk-muscles, and must, therefore, form a separate group to themselves. Thus the striated musculature of the Ammocœtes must be divided into (1) the visceral muscles; (2) the longitudinal somatic muscles; and (3) the dorso-ventral somatic muscles. Of these the 1st, on the view just stated, represent the original appendage-muscles; the 2nd belong to the spinal region, and will be considered with that region; the 3rd represent the original segmental dorso-ventral somatic muscles, which are so conspicuous in the musculature of the Limulus and the scorpion group.

The discussion of this last statement will be given when I come to deal with the prosomatic segments of Ammocœtes. I wish, here, simply to point out that van Wijhe has shown that the eye-muscles develop from his 1st, 2nd, and 3rd dorsal mesoblastic segments, and therefore represent the somatic muscles belonging to those segments, while no development of any corresponding muscles takes place in the 4th, 5th, and 6th segments; so that if the eye-muscles represent a group of dorso-ventral somatic muscles, such muscles have been lost in the 4th, 5th, and 6th segments. The latter segments are, however, the glossopharyngeal and vagus segments, the branchial musculature of which is derived from the ventral segments of the mesoderm. In other words, van Wijhe's observations mean that the dorso-ventral somatic musculature has been lost in the branchial or mesosomatic region, while the dorso-ventral appendage musculature has been retained, and that, therefore, the mode of respiration in Ammocœtes more closely resembles that of Limulus than of Scorpio.

In addition to these branchial muscles, another and very striking set of muscles is found in the respiratory region of Ammocœtes—the so-called tubular muscles. These muscles are of great interest, but as they are especially connected with the VIIth nerve, their consideration is best postponed to the chapter dealing with that nerve.

Also, in connection with the vagus group of nerves, special sense-organs are found in the skin covering this mesosomatic region, the so-called epithelial pit-organs (Ep. pit., Fig. 71). They, too, are of great interest, but their consideration may also better be deferred to the chapter dealing with those special sense-systems known as the lateral line and auditory systems.

Closely bound up with the respiratory system is the nature of the circulation of blood through the gills. Before, therefore, proceeding to the consideration of the segments in front of those which carry branchiæ, it is worth while to compare the circulation of the blood in the gills of Limulus and of Ammocœtes respectively.

In all the higher vertebrates the blood circulates in a closed system of capillaries, which unite the arterial with the venous systems. In all the higher invertebrates this capillary system can hardly be said to exist; the blood is pumped from the arterial system into blood spaces or lacunæ, and thus comes into immediate contact with the tissues. From these it is collected into veins, and so returned to the heart. There is, in fact, no separate lymph-system in the higher invertebrates; the blood-system and lymph-system are not yet differentiated from each other. This also is the case in Ammocœtes; here, too, in many places the blood is poured into a lacunar space, and collected thence by the venous system; a capillary system is only in its commencement and a lymph-system does not yet exist. In this part of its vascular system Ammocœtes again resembles the higher invertebrates more than the higher vertebrates.

This resemblance is still more striking when the circulation in the respiratory organs of the two animals is compared. A branchial appendage is essentially an appendage whose vascular system is arranged for the special purpose of aerating blood. In the higher vertebrates such a purpose is attained by the pulmonary capillaries, in Limulus by the division of the posterior surface of the basal part of the appendage into thin lamellar plates, the interior of each of which is filled with blood. The two surfaces of each lamella are kept parallel to each other by means of fibrous or cellular strands forming little pillars at intervals, called by Macleod "colonettes." A precisely similar arrangement is found in the scorpion gill-lamella, as seen in Fig. 69, A, taken from Macleod. In Ammocœtes there are no well-defined branchial capillaries, but the blood circulates, as in the invertebrate gill, in a lamellar space; here, also, as Nestler has shown, the opposing walls of the gill-lamella are held in position by little pillar-like cells, as seen in Fig. 69, B, taken from his paper.

In this representative of the earliest vertebrates the method of manufacturing an efficient gill out of a lacunar blood-space is precisely the same as that which existed in Limulus and the scorpion, and, therefore, as that which existed in the dominant invertebrate group at the time when vertebrates first appeared. This similarity indicates a close resemblance between the circulatory systems of the two groups of animals, and therefore, to the superficial inquirer, would indicate an homology between the heart of the vertebrate and the heart of the higher invertebrate; but the former is situated ventrally to the gut and the nervous system, while the latter is composed of a long vessel which lies in the mid-dorsal line immediately under the external dorsal covering. Indeed, this ventral position of the heart in the one group of animals and its dorsal position in the other, combined with the corresponding positions of the central nervous system, is one of the principal reasons why all the advocates of the origin of vertebrates from the Appendiculata, with the single exception of myself, feel compelled to reverse the dorsal and ventral surfaces in deriving the vertebrate from the invertebrate. But there is one most important fact which ought to make us hesitate before accepting the homology of the dorsal heart of the arthropod with the ventral heart of the vertebrate—The heart in all invertebrates is a systemic heart, i.e. drives the arterial blood to the different organs of the body, and then the veins carry it back to the respiratory organ, from whence it passes to the heart.



A, Branchial lamellæ of Scorpio (after Macleod); B, Branchial lamellæ of Ammocœtes (after Nestler).

The only exception to this scheme is found in the vertebrate where the heart is essentially a branchial heart, the blood being driven from the heart to the ventral aorta, from which by the branchial arteries it is carried to the gills, and then, after aeration, is collected into the dorsal aorta, whence it is distributed over the body. The distributing systemic vessel is the dorsal aorta, not the heart which belongs essentially to the ventral venous system. This constitutes a very strong reason for believing that the systemic heart of the invertebrate is not homologous with the heart of the vertebrate. How, then, did the vertebrate heart arise?

Let us first see how the blood is supplied to the gills in Limulus.

(After .)

In Limulus the blood flows into the lamellæ from sinuses or blood-spaces (b.s., Fig. 66) at the base of each of the lamellæ, which sinuses are filled by a vessel which may be called the branchial artery, since it is the afferent branchial vessel. On each side of the middle line of the ventral surface of the body a large longitudinal venous sinus exists, called by Milne-Edwards the venous collecting sinus, L.V.S., (Fig. 70 and Fig. 58), which gives off to each of the branchial appendages on that side a well-defined afferent branchial vessel—the branchial artery (a. br.). The blood of the branchial artery flows into the blood-spaces between the anterior and posterior laminæ of the appendage and thence into the gill-lamellæ, from which it is collected into an efferent vessel or branchial vein, termed by Milne-Edwards the branchio-cardiac canal, which carries it back to the dorsal heart. The position of the branchial artery and vein is shown in Fig. 66, which represents a section through the branchial appendage of Limulus at right angles to the cartilaginous branchial bar (br. cart.), just as Fig. 65 represents a section through the branchial appendage of Ammocœtes at right angles to the cartilaginous branchial bar.

Further, the observations of Blanchard, Milne-Edwards, Ray Lankester, and Benham concur in showing that in both Limulus and the scorpion group a striking and most useful connection exists between the heart and these two collecting venous sinuses, in the shape of a segmentally arranged series of muscular bands (V.p., Fig. 70 and Fig. 58), attached, on the one hand, to the pericardium, and on the other to the venous collecting sinus on each side. These muscular bands, to which Lankester and Benham have given the name of 'veno-pericardial muscles,' are so different in appearance from the rest of the muscular substance, that Milne-Edwards did not recognize them as muscular, but called them 'brides transparentes.' Blanchard speaks of them in the scorpion as 'ligaments contractiles,' and considers that they play an important part in assisting the pulmonary circulation; for, he says, "en mettant a nu une portion du cœur, on remarque que ces battements se font sentir sur les ligaments contractiles, et determinent sur les poches pulmonaires une pression qui fait aussitot refluer et remonter le sang dans les vaisseaux pneumocardiaques." Lankester, in discussing the veno-pericardial muscles of Limulus and of the scorpions, says that these muscles probably contract simultaneously with the heart and are of great importance in assisting the flow through the pulmonary system. More recently Carlson has investigated the action of these muscles in the living Limulus and found that they act simultaneously with the muscles of respiration.

Precisely the same arrangement of veno-pericardial muscles and of longitudinal venous collecting sinuses occurs in the scorpions. It is one of the fundamental characters of the group, and we may fairly assume that a similar arrangement existed in the extinct forms from which I imagine the vertebrate to have arisen. The further consideration of this group of muscles will be given in Chapter IX.

Passing now to the condition of the branchial blood-vessels of Ammocœtes, we see that the blood passes into the gill-lamellæ from a blood-space in the appendage, which can hardly be dignified by the name of a blood-vessel. This blood-space is supplied by the branchial artery which arises segmentally from the ventral aorta (V.A.), as seen in Fig. 71 (taken from Miss Alcock's paper). From the gill-lamellæ the blood is collected into an efferent or branchial vein (v. br.), which runs, as seen in Fig. 65, along the free edge of the diaphragm, and terminates in the dorsal aorta.

The ventral aorta is a single vessel near the heart, but at the commencement of the thyroid it divides into two, and so forms two ventral longitudinal vessels, from which the branchial arteries arise segmentally.



Nerves coloured red are the motor nerves to the branchial muscles. Nerves coloured blue are the internal sensory nerves to the diaphragms and the external sensory nerves to the sense-organs of the lateral line system. ''Br. cart., branchial cartilage; M. con. str., striated constrictor muscles; M. con. tub., tubular constrictor muscles; M. add., adductor muscle; D.A., dorsal aorta; V.A., ventral aorta; S., sense-organs on diaphragm; n. Lat., lateral line nerve; X., epibranchial ganglia of vagus; R. br. prof. VII., ramus branchialis profundus of facial; J.v., jugular vein; Ep. pit.'', epithelial pit.

From this description it is clear that the vascular supply of the branchial segment of Ammocœtes would resemble most closely the vascular supply of the Limulus branchial appendage, if the ventral aorta of the former was derived from two longitudinal veins, homologous with the paired longitudinal venous sinuses of the latter.

A priori, such a derivation seems highly improbable; and yet it is precisely the manner in which embryology teaches us that the heart and ventral aorta of the vertebrate have arisen.

Not only does the vertebrate heart differ from that of the invertebrate, in that it is branchial while the latter is systemic, but also it is unique in its mode of formation in the embryo. In the Appendiculata the heart is formed as a single organ in the mid-dorsal line by the growth of the two lateral plates of mesoblast dorsalwards, the heart being formed where they meet. In Mammalia and Aves, the heart and ventral aorta commence as a pair of longitudinal veins, one on each side of the commencing notochord.

If the embryo be removed from the yolk, the surface of the embryo covering these two venous trunks can be spoken of as the ventral surface of the embryo at that stage, and indeed we find that in the present day there is an increasing tendency to speak of this surface as the ventral surface of the embryo. Thus, Mitsukuri, in his studies of chelonian embryos, lays great stress on the importance of surface views and when the embryo has been removed from the yolk, figures and speaks of its ventral surface. So, also, Locy and Neal find that the best method of seeing the early segments of the embryo is to remove the embryo from the yolk, and examine what they speak of as a ventral view. At the period, then, before the formation of the throat, we may say that on the ventral surface of the embryo a pair of longitudinal venous sinuses are found, one on each side of the mid-ventral line, which are in the same position with respect to the mid-axis of the embryo as are the longitudinal venous sinuses in Limulus.

The next step is the formation of the throat by the extension of the layers of the embryo laterally to meet in the mid-line and so form the pharynx, with the consequence that a new ventral surface is formed; these two veins, as is well known, travel round also, and, meeting together in the new mid-ventral line, form the subintestinal vein, the heart, and the ventral aorta.

What is true of Mammalia and Aves, has been shown by P. Mayer to be true universally among vertebrates, so that in all cases the heart and ventral aorta have arisen by the coalescence in the new mid-ventral line of two longitudinal venous channels, which were originally situated one on each side of the notochord, in what was then the ventral surface of this part of the embryo. This history is especially instructive in showing how the pharyngeal region is formed by the growing round of the lateral mesoblast, i.e. the muscular and other mesoblastic tissues of the branchial segments, and how the two longitudinal veins take part in this process. The phylogenetic interpretation of this embryological fact seems to be, that the new ventral surface of the vertebrate in this region is formed, not only by the branchial appendages, but also by the growth ventrally of that part of the original ventral surface which covered each longitudinal venous sinus.

The following out of the consecutive clues, which one after the other arise in harmonious succession as the necessary sequence of the original working hypothesis, brings even now into view the manner in which the respiratory portion of the alimentary canal arose, and gives strong hints as to the position of that part of the arthropod which gave origin to the notochord. Here I will say no more at present, for the origin of the new alimentary canal of the vertebrate and of the notochord will be more fittingly discussed as a whole, after all the other organs of the vertebrate have been compared with the corresponding organs of the arthropod.



C.N.S., central nervous system; nc., notochord; m., myotome.

The lower half of figure shows comparative position of the longitudinal venous sinus (L.V.S.) in Limulus. C.N.S., central nervous system; Al., alimentary canal; H., heart; m., body-muscles.

The strong evidence that the vertebrate heart was formed from a pair of longitudinal venous sinuses on the ventral side of the central canal, carries with it the conclusion that the original single median dorsal heart of the arthropod is not represented in the vertebrate, for the dorsal aorta cannot by any possibility represent that heart.

Although it is not now functional the original existence of so important an organ as a dorsal heart may have left traces of its former presence; if so, such traces would be most likely to be visible in the lowest vertebrates, just as the median eyes are much more evident in them than in the higher forms. In Fig. 58 the position of the dorsal heart is shown in Limulus, and in Fig. 70 the shape and extent of this dorsal heart is shown. It extends slightly into the prosomatic region, and thins down to a point there, runs along the length of the animal and finally thins down to a point at the caudal end.

The heart is surrounded by a pericardium, from which at regular intervals a number of dorso-ventral muscles pass, to be inserted into the longitudinal venous sinus on each side. These veno-pericardial muscles are absolutely segmental with the mesosomatic segments, and are confined to that region, with the exception of two pairs in the prosomatic region. Their homologies will be discussed later.

Any trace of a heart such as we have just described must be sought for in Ammocœtes between the central nervous system and the mid-line dorsally. Now, in this very position a large striking mass of tissue is found, represented in section in Fig. 73, f. It forms a column of similar tissue along the whole mid-dorsal region, except at the two extremities; it tapers away in the caudal region, and headwards grows thinner and thinner, so that no trace of it is seen anterior to the commencement of the branchial region. It resembles in its dorsal position, in its shape, and in its size a dorsal heart-tube such as is seen in Limulus and elsewhere, but it differs from such a tube in its extension headwards. The heart-tube of Limulus ceases at the anterior end of the mesosomatic region, this fat-column of Ammocœtes at the posterior end. In its structure there is not the slightest sign of anything of the nature of a heart; it is a solid mass of closely compacted cells, and the cells are all very full of fat, staining intensely black with osmic acid. Nowhere else in the whole body of Ammocœtes is such a column of fat to be found. It is not skeletogenous tissue with cells of the nature of cartilage-cells, as Gegenbaur thought and as Balfour has depicted (Vol. II., Fig. 315) in his 'Comparative Embryology,' as though this tissue were a part of the vertebral column, but is simply fat-cells, such as might easily have taken the place of some other previously existing organ.

I do not know how to decide the question which thus arises. Supposing, for the sake of argument, that this column of fat-cells has really taken the place of the original dorsal heart, what criterion would there be as to this? The heart ex hypothesi having ceased to function, the muscular tissue would not remain, and the space would be filled up, presumably with some form of connective tissue. As likely as not, the connective tissue might take the form of fatty tissue, the storage of fat being a physiological necessity to an animal, while at the same time no special organ has been developed for such a purpose, but fat is being laid down in all manner of places in the body.

This dorsal fat-column, as it is seen in Ammocœtes, is not found in the higher vertebrates, so that it possesses, at all events, the significance of being a peculiarity of ancient times before the vertebrate skeletal column was formed.

I mention it here in connection with my view as to the origin of vertebrates, because there it is, in the very place where the dorsal heart ought to have been. For my own part, I should not have expected that a muscular organ such as the heart would leave any trace of itself if it disappeared, so that its absence in the dorsal region of the vertebrate does not seem to me in the slightest degree to invalidate my theory.



''sp. c., spinal cord; gl., glandular tissue filling the spinal canal; sk., Gegenbaur's skeletogenous cells; p.'', pigment.