The Chætognatha, or Primitive Mollusca

It is not necessary to attempt here a complete review of the stages by which our knowledge of the Chætognatha has been accumulated. Several of our predecessors have already published such historical résumés, and so it will suffice to refer the reader to the works of Hertwig, Langerhans, and Grassi for fuller details than we give here.

In his original description Martin Slabber (1775) expressed the opinion that the “arrow-shaped worms” should constitute a sub-division of the Linnean group of Vermes. Quoy and Gaimard (1827), after examining the large Sagitta bipunctata from Gibraltar, were uncertain as to whether it was Zoophyte or Mollusc. D’Orbigny (1834), however, decided in favour of the Mollusca, and associated the arrow-worms with the Heteropoda, an association in which he was followed by Milne-Edwards (1845), Troschel (1845), Siebold (1848), and Burmeister (1856). Milne-Edwards, indeed, went so far as to point out that the “prépuce” together with the head of Sagitta were the equivalents of the Molluscan head.

The anatomical work of Krohn (1844) marks a new epoch. Having been fortunate in obtaining specimens of large size at Messina, Krohn was able to investigate many hitherto undiscovered details of the internal organisation, and amongst others the nervous system, testes, and buccal apparatus. The result was au alliance of the Arrow-worms with the Annelida. Five years later Oersted (1849) suggested that Sagitta should rather be considered as a Nematode, an association favoured by Leuckart (1854) and by several writers of zoological text-books (e.g. Claus). The comparison rests principally upon a certain similarity of arrangement of the musculature.

Passing over a suggestion of relationship with the Tardigrada and lower Arthropods by Huxley (1852 and 1878), which has been reconsidered by Grassi (1883) on account of fancied resemblances between the cerebral ganglia, we come to the more serious proposal that a new subdivision of the worms should be constituted for the reception of the Sagittæ ; and of the names proposed, that of Leuckart (1854)—Chætognatha—has taken precedence of Oesthelminthes of Gegenbaur (1856) and of Pterhelminthes of Harting (circ. 1865).

In 1857 a fantastic theory of Vertebrate affinity was put forward by Meissner. It has, however, failed to find farther support than that accorded to it by Haeckel. Metschnikoff (1867) included the Chætognatha in his new group of Ambulacrata, together with the Echinodermata, Brachiopoda, and Enteropneusta. This view has been supported by Bütschli, who, while believing the Chætognatha to have some affinity to the Annelida, considered that the method of the development of the hypoblast is a sufficient justification for regarding them as being near akin to the Echinodermata and to the Tunicata; still more recently (1876) he has proclaimed the Brachiopoda to be their nearest relations.

A revival of the Molluscan theory is due to Langerhans (1878), who re-investigated the nervous system, and discovered the circumœsophageal nerve ring.

In more recent times, while the Annelidan theory has found some support, notably by Giard (1875) and Gourret (1884), it has also found a vehement opponent in Grassi, who maintains that among the Chætognatha there is not to be found any trace either of the metamerisation of the body, or of setæ, or of parapodia. On turning to recent zoological text-books we find that the majority of writers content themselves with expressing their uncertainty concerning the exact position of the group in the Zoological System, but, on the whole, there is a distinct feeling in favour of an association with the “Vermes.”

The theory that the Chætognatha are primitive Mollusca is on first thoughts open to the criticism that they do not possess features which have hitherto been generally accepted as essential to members of the Molluscan phylum, and that the obvious points of resemblance are the results of convergence rather than indications of phylogenetic affinity.

We shall endeavour to meet these objections, and to show that the resemblances are shared by all the chief organs in the Chætognath anatomy, namely, by the nervous, alimentary, genital, and other systems ; that also in their development the Chætognath approaches very close to the primitive Molluscan type, and that in the case of those features in which the Chætognath organisation appears not to be of the Molluscan type, interesting parallels are to be found within the limits of the Molluscan phylum itself.

At the same time it is not proposed to merge the Chætognatha in any one of the existing Molluscan classes; they are rather to be regarded as modern and modified representatives of a free-swimming ancestor from which the Molluscan phylum has sprung. They represent, in fact, Archimollusca modified for an active swimming life in the open waters of the ocean, with affinities to all existing classes of the phylum. They have, to our mind, more primitive features than the “Archimollusc” of Lankester, which we shall term the creeping Archimollusc to distinguish it from our swimming type of archimollusc; and to emphasize this relationship we propose to consider the Chætognatha as belonging to the lowest class in the Molluscan phylum from which the others have sprung.

The external configuration of the Chætognath body has justly earned for them their common name of “Arrow-worms.” Human ingenuity has not got further in striving after the form best adapted to fast swimming beneath the surface of the water. The lines of the Sagitta and the Whitehead torpedo are the same. Comparatively few of the Mollusca, which have taken to a swimming mode of life, have reached such perfection of outline; in all the difficulty of disencumbering themselves of unwieldy shells and feet and mantle flaps has had to be overcome, and even when these have all gone, as in the languid Phyllirhoe, the result is not satisfactory. The Pteropods are the butterflies of the sea rather than its dragon-flies ; and the elaborate hydraulics of the Cephalopods effect locomotion by a totally different process, which we are not considering here.

The separate regions of body merge gently into one another, and all projections unnecessary for functions of swimming or feeding have been smoothed down. For this reason the visceral sac is all contained within the cigar-shaped contour of the body-wall. In Mollusca the integument of the visceral sac is typically folded near its junction with the head and foot, but no such mantle-flap or pallium is distinguishable in Sagitta. The palliai ectoderm is, however, drawn out in the horizontal plane to form the lateral and caudal fins of many of those Dibranchiate Cephalopoda (text-fig. 2) in which the shell is most degenerate, and the resemblance is strengthened by the deposition of a chitinous or conchyolinous substance within the fold either in the form of fin-rays or of a continuous lamina.

If we may be permitted to imagine the alterations of structure which a typical Mollusc would undergo if it were to become adapted for an active, free-swimming, predacious life; I think that, judging by living examples, we should find that they would be in the direction of the structure of the Chætognatha.

That foot, visceral hump, shell, and mantle cavity might disappear, is clearly shown by the structure of Phyllirhoe. But the Chætognatha differ from Phyllirhoe in their smaller size, and in the far greater activity of their swimming movements. To these two factors we attribute the relatively large development of certain organs, such as the longitudinal muscular system and the non-existence of others, e. g. separate vascular and respiratory systems respectively.

It may be premised at once that there is nothing in the Chætognath anatomy or development which we have felt able to homologise with the shell or shell-gland of the Mollusca. We cannot, however, consider that the absence of this palliai organ is very material to the main issue, for even among Mollusca of very moderate swimming powers the shell is usually the first structure to dwindle and disappear (Heteropods, Phyllirhoe, etc.), and might therefore be expected to be altogether absent in a Mollusc built upon such perfect lines for swimming as Sagitta. But it may be argued, if the ancestor of Chætognatha had a typical Molluscan shell, surely some trace of a shell-gland would be noticeable in the younger stages of their growth, it being so conspicuous a feature in Molluscan developments. For its absence two explanations may be given : firstly, by our hypothesis, the Chætognatha are presumed to have become modified for their pelagic life at a very early period, before the Molluscan phylum and the Molluscan shell had been in existence for a long time, in other words, the Chætognatha are a far older group than any of the shell-less Mollusca.

If the loss of all trace of the shell-gland may have been the effect of time, its loss has no doubt also been accelerated by the marvellous rapidity with which Chætognath development takes place.

Although the body cavity appears to be divided into three paired cavities by septa, there is no justification for regarding the Chætognatha as exhibiting any real metamerie segmentation at all. In the first place the two transverse septa cannot be regarded as homologous with one another, or with the septa which separate the metameres of Annelida, because in time, mode, and purpose of origin they are absolutely different. Embryological examination shows that the anterior septum is produced by the meeting and fusion of the somatic and splanchnic mesoblast very early in embryonic life, at a time when differentiation of tissues has not begun and the mesoblast is still continuous with the hypoblast ; and that the posterior septum appears in close connection with the genital cells, and is not formed by the whole thickness of the mesoderm, but probably only by the cellular envelopes of the genital cells, and certainly by the splanchnic layer exclusively.

In the light of our new knowledge concerning these developmental phenomena, the theory of a close relationship of Chætognatha to an unsegmented group of animals of a high grade of organisation is in no way prejudiced by the division of the body into three sections which have the guise but not the reality of metameres.

The Chætognatha differ from the majority of Mollusca in the absence of a more or less extensive ciliated epithelial covering of the body both in the larval and adult stages. But though devoid of a general investment of cilia which might impede their active movements in the water, most of them possess a circular or oval ring of ciliated cells on the dorsal side of the head and behind the eyes. This ciliated organ is innervated directly from the cerebral ganglion, and is believed to have an olfactory or gustatory function. It will be referred to again as the homologue of the velum or preoral circlet of cilia of the Trochophor larva of the Mollusca (text-fig. 1, v).

The external epithelium of the Chætognatha, however, agrees with that of the Mollusca in the presence of gland-cells, which are especially abundant upon the ventral surface of those littoral species, which live among algæ, such as Spadella cephaloptera; and similar cells, though less well developed, have been noticed in Sagitta darwinii, Grassi, and to less extent in Spadella marioni by Gourret.

The two large glands (text-fig. 1, gl.) on the upper side of the head have been described by Gourret in Spadella marioni. They are covered by the “hood,” and may be variously regarded as mucous glands for the lubrication of the hood, as poison glands for the spines, or as extra-buccal glands. It is unlikely that their function is excretory, as Gourret suggests.

In some species of Chætognatha, e. g. Spadella draco, the sub-epithelial tissue presents a peculiar vesicular condition which is particularly noticeable in the neck region anterior to the lateral fin. Similar cells occur in the sub-epithelial (mesodermal) connective tissue of many Mollusca.

There is no distinct organ in the Chætognatha which can be compared with the Gastropod or Lamellibranch foot, but it must be remembered that the Chætognatha are not creeping animals, nor is there any reason to suppose that their recent ancestors (like those of the Heteropoda) were so, or, indeed that they have ever had a creeping ancestor at all. On the other hand the circumoral hood, “Kappe” or “prépuce,” situated, as it is, between the head and visceral regions, occupies the same relative position as the “foot” of a Dibranchiate Cephalopod, or the lateral lobes of Nautilus, and to these structures we desire to compare it. The method of innervation would, no doubt, settle the question of the homology of these organs easily were the hood a more powerfully developed muscular organ than it is. According to our theory the Chætognath “hood” is the Cephalopod circumoral foot in its most primitive condition, or else it represents that organ in a reduced condition as a mere reduplicature of the ectoderm into which the body cavity extends but a short distance. It is therefore to be compared with the interbraehial membrane of many of the Cephalopoda (Histioteuthis, Tremoctopus [text-fig. 8], Leioglossa) rather than with the arms themselves. In the deep-sea form Sagitta macrocephala it is in a much reduced condition (text-fig. 1, c.f.). In development, its bilateral origin already indicated by the notch in the mid-ventral line is very clear, for it is formed from two separate ectodermal thickenings, one on either side of the mouth (Doncaster).

The two glands near its dorsal attachments may secrete mucus of the inner surface of the hood (text-fig. 1, A, B, gl.). According to the theory that hood and Cephalopod inter-brachial membrane are homologous organs, these hood-glands and the aquiferous pores of Tremoctopus and other Cephalopoda, cannot be regarded as homologous.

The musculature of the Chætognatha, which has so often been cited as their point of resemblance to the Nematoda, is as distinct as possible from our ordinary conception of Molluscan musculature, but among the varied members of the Molluscan phylum there are some which present features closely resembling the Chætognath condition.

Further, the Chætognath muscle fibres are striped, whereas Molluscan fibres are as a rule unstriped.¹ But were striped muscle entirely unknown in living Mollusca we should not feel inclined to attribute much importance to this difference, believing that muscle striation is merely an indication of efficiency from the point of view of rapidity of contraction. Molluscs are slow in their movements, the Chætognatha are quick, that is all. However, Spillmann has shown that well-striated muscle may occasionally appear among the Mollusca, for instance, in the hearts of Turbo rugo sus and Acmæa virginea.

The arrangement of the longitudinal musculature in a sheath surrounding the viscera is an adaptation for swimming similar to that in Phyllirhoe, whose ancestors, no doubt, possessed both foot and shell. The subdivision of the muscle bands into four quadrants finds an exact counterpart in Chætoderma (von Graff, text-fig. 4). The homology of these muscles with the four longitudinal retractor muscles of the head of the Cephalopoda will be noticed later.

The occurrence of three pairs of supporting plates in connection with the spines on the surface of the buccal mass is an important feature in Chætognath anatomy, but of far wider significance is the complicated chitinous (?) skeletal piece with its large lamella and two lateral wings serving for the attachment of the retractor, extensor, and other muscles in the head. This internal cephalic skeleton we believe to be the homologue of the cartilaginous cephalic skeleton of the Cephalopoda, and bears a superficial resemblance to that of Nautilus (text-fig. 5). Its presence in addition to the many other points of resemblance between the two groups should be something more than a mere case of analogy.

The remarkable resemblance of the Chætognath and Molluscan nervous systems was recognised as long ago as 1844 by Krohn, the discoverer of the nervous system in Sagitta hexaptera, and again by Langerhans, who discovered the buccal ganglia. The two most recent accounts by Hertwig and Gourret are not entirely in agreement, but both may be readily referred to the Molluscan plan. Indeed, the resemblance is so close that, in a description of the Chætognath nervous system, the Molluscan terminology may be appropriately employed.

Certain characteristics which are primitive among the Mollusca are retained in the nervous system of the Chætognatha, but these co-exist with others which appear to be directly associated with the more active mode of life of the latter. In two features, namely, in the widely distributed nerve plexus which enwraps the body, and in the close adhesion of the central nervous system to the ectoderm, the Chætognatha are more primitive than the Mollusca; the ladder-like arrangement of the nerves that occurs in Amphineura, for instance, is not a Chætognath feature, although the nerves issuing from the ventral ganglion are highly suggestive of it. The uniform sheathing of ganglionic cells, indicating lack of concentration, which gives peculiar interest to primitive nerve-cords, as in the Chitonidæ, is not to be found in Chætognatha, whose active movements demand a more perfect apparatus for their co-ordination, and in which a more centralised nervous system has accordingly been evolved.

The cerebral ganglion, consisting of a median group of nerve-cells and of two lateral ones joined by commissural fibres, gives off posteriorly a pair of rhinophoral and a pair of optic nerves to the olfactory organs and eyes respectively, and the great cerebro-visceral connectives which lead back to the “ventral” or visceral ganglion. Lateral extensions of the cerebral ganglion give off two nerves, which are, at any rate, partly motor, to the cephalic or buccal muscles, and a pair of lateral nerves to the integument of the head.

To demonstrate the existence of ventral commissures below the œsophagus is not always an easy matter. Langerhans definitely affirmed the presence of a circumcesophageal ring completed by a subcutaneous commissural nerve just behind the mouth, a statement which Hertwig was not able to confirm. My own observations on Neapolitan material have led me to believe that there is a commissural plexus of ganglion cells beneath the skin, which, if more concentrated, might give rise to a commissural nerve strand such as Langerhans figures. Such a nerve-loop is to be compared with the stomatogastric loop of the Mollusca, the commissural nerves of which are often extremely attenuated, and, owing to their position on the buccal muscles, difficult to demonstrate; in Nautilus, for example, the completion of the buccal nerve-loop has only been recently proved by Graham Kerr. The co-existence of two pairs of ganglia, which may be termed buccal and pharyngeal in Chætognatha, in many Gastropoda (e. g. Patella), and in Nautilus is a significant feature.

The descriptions of Hertwig and of Gourret differ in regard to the mode of origin of certain nerves from the cerebral mass, but they may be reconciled if we suppose the cerebro-buccal connectives and motor nerves to the muscles of the mandibles, to run side by side for a short distance after leaving the cerebral ganglion in Sagitta hexaptera (Hertwig), but to be separate from the start in Spadella marioni (Gourret).

If it could be established that the buccal ganglia of the Chætognatha are derived from the stomodæal ectoderm the resemblance to the Mollusca would be complete.

Concerning the existence of the posterior or visceral loop there can be no doubt. The visceral ganglia appear early as ectodermal thickenings of relatively enormous size. Though at first lateral in position, they soon become approximated, and give rise to the “ventral ganglion,” which, however, never loses its primitive character of lying immediately beneath the ectoderm.

The pedal ganglia and their commissural loop do not exist in any conspicuous form, nor should we expect to find them in the absence of a well-developed and muscular foot. It might be possible to trace their homologues by making a more minute study of the innervation of the hood; but at present we merely suggest that cells corresponding to those of Molluscan pedal ganglia may be merged in the lateral expansions of the cerebral ganglion already mentioned. Von Jhering, on the other hand, believed that the “ventral ganglion” consisted of visceral and pedal ganglia united.

Pedal ganglia and nerves excepted, the Chætognath nervous system admits of a close comparison in point of detail with that of Mollusca, but the particular type to which it exhibits the closest affinity is that of Nautilus, as will be seen by reference to the diagrams (text-fig. 6).

A peculiar feature in Nautilus is the undivided nature of the cerebral ganglion, which is also found in Chætognatha, though unusual among Mollusca. The chief differences are the absence of the pedal loop, the more centralised arrangement of ganglionic cells on the visceral loop, the close adhesion of the nervous system to the ectoderm and the superficial nerve plexus.

The tactile organs of Chætognatha consist of isolated neuro-epithelial cells terminating in tactile hairs, and of groups of tactile bristles which belong to a type of sense organ which is widely distributed among Mollusca.

The “olfactory” (?) ring of ciliated cells and the area within it (text-fig. 1, v.) is undoubtedly an organ of great importance. It seems to be the homologue of the velum and the rhinophoral organs enclosed within it of the Mollusc, for the innervation from the cerebral ganglion by two nerves lying just within the optic nerves, is precisely similar. It is possible that the minute organs which Weiss has described in certain Oigopsid Cephalopoda are to be compared with them. And, in this connection, both the tentacles of Spadella cephaloptera and the olfactory spoon-shaped organs of Chiroteuthis should be made objects of further study.

The eyes are undoubtedly of the Molluscan vesicular type. The retinal cells contain rods; the crystalline lens is secreted within an invagination of the ectoderm, which has become overgrown by a flattened epithelium. The fact that in the Molluscan eye the pigment forms a complete cup to the retina, but is mainly restricted to one side of the ocellus in Sagitta, is a minor point resulting from the peculiar method of the grouping of Chætognath eyes in threes.

As in the Amphineura, otocysts are unknown.

If our hypothesis of the origin of the Mollusca be correct, it follows that the straight and symmetrical alimentary canal of the Chætognatha is iii a more primitive condition than that of any other Mollusc. Excepting for the absence of the radula, an organ better adapted to quiet browsing than to predatory Chætognath habits, the buccal armature is singularly like the armature which predatory Molluscs develop. The resemblance both in position and structure between the lateral groups of mandibles of a Gymnosomatous Pteropod, such as Clio (text-fig. 7, h.) and the “hooks” of Sagitta, is very close. No chitinous teeth are developed in the mid-ventral line, but the repeated series of spines in marginal and lateral groups above the entrance to the buccal cavity suggests a possibility of the successional development and replacement of chitinous elements near the mouth. And if this were to happen in the mid-ventral line a radula would result. At the same time it must be remembered that the radula is absent in several Amphineura and in Lamellibranchia, in the Leioglossal Cephalopoda, and that it even tends to disappear in certain groups of carnivorous Gastropoda.

A comparison between the forms of teeth characteristic of Chætognatha and of Amphineura is instructive (text-fig. 8).

Compare the saw-like hooks of Krohnia hamata and S. serrato-dentata (a, b) with a tooth of Lepidomenia (d), and the teeth of S. bipunctata (c) with those of Proneomenia (e).

No special salivary glands have been noticed, but the alimentary canal of Sagitta cephaloptera is furnished with two (liver?) diverticula (text-fig. 1, A, h.).

The genital organs of the Chætognatha are situated in the hinder part of the visceral sac. The gonads are paired and hermaphrodite, but the male and female gonads develop in separate cavities, and pass to the exterior by separate paired ducts. The Mollusca too are typically hermaphrodite, and although there are some reasons for regarding the dioecious condition as primitive in the group, both ova and spermatozoa are derived from the epithelial lining of the same cavity ; but in Mollusca belonging to very different groups there is a tendency for the male and female gonads to develop apart, either in separate male and female acini (Pleurobranchidæ, Nudibranchs, Cardium oblongum, etc.), of in separate regions of the same gland (Cycladidæ), or in different glands (Anatinacea and Septibranchia). The longitudinal partition between the paired gouadial cavities of the Chætognatha reminds us of the division between the paired gonads of the Aplacophora, and is typical, there is good reason to believe, of the primitive condition in Mollusca (text-fig. 10).

Thanks to the minute observations of Doncaster we are able to draw attention to an astonishing similarity between the behaviour of the gonads both of the Mollusca and of the Chætognatha, which we believe may throw a new light upon the morphology of the transverse septum of the latter group. The following are the details of the process as they have been described among Mollusca :

In Chiton, Haller observed the young ova to sink below the ovarial epithelium, and then growing larger bulge it out, so as to form a follicle. By a process of growth each egg-cell gets carried out into the perigonadial coelom on the end of a stalk made by the follicular epithelium.

In the Cephalopoda the same process obtains. In Nautilus the result is exactly as in Chiton. In Octopus, Argonauta, and others a higher grade of complexity is reached by the formation of branches from the simple egg-stalks, resulting in the production of “egg-trees.” In the Oigopsidæ the region that bears the egg-stalks projects right across the genital cavity as a “spindle-shaped body beset all over with stalked eggs.”

In Sagitta the process is as follows : The genital cells in the embryo lie close to the longitudinal partition, bedded in mesoblast, one behind the other (Doncaster, figs. 14, 17). At a particular stage they become enclosed in a sort of cellular envelope of mesoderm, which does not yet form a definite epithelial sheath; a condition to be compared with the stage in Chiton and other Molluscs before stalk-formation has commenced. Fig. 15 of Doncaster’s paper shows some of these follicular nuclei closely adpressed to a genital cell. About the fourth day genital cells which have lain quiet for some time “move slowly across the cœlomic cavity until they reach the body-wall on each side.” While traversing the cœlom the male and female cells on each side move together, and during their progress the transverse septum is formed between them by the cells, which we homologise with the follicular cells of Mollusca. The process of the formation of the transverse septum is therefore absolutely distinct and unlike the process of the growth of septa in Annelids, and although it is impossible to say whether the genital cells themselves or the follicle cells which lie about them are the veræ causæ of the movement, yet the fact of its occurrence in the two groups we are considering is significant.

The origin of the genital cells side by side from the walls of the same cavity, and the subsequent division of this cavity into anterior and posterior parts by a septum of quite peculiar growth, may be given as reasons against the view that the so-called trunk and tail cavities of Chætognatha are indications of metamerism.

Again, if the trunk and tail cavities were homomeric, we should expect to find their ducts also homomeric. This is certainly not the case, for the efferent genital ducts of the Chætognatha are essentially different both in structure and in development. The sperm-ducts being ciliated, and the oviducts being devoid of ciliation, was clearly established by Hertwig in 1880, and should have made later writers hesitate before adopting a theory of metamerie segmentation; but quite recently the embryological studies of Doncaster have indicated that the two ducts are derived from different germ-layers—the sperm duct from epiblast, and the oviduct from mesoblast.

We suggest that the oviduct of Chætognatha is merely a gonocœl, partly lined by the follicular epithelium covering the egg-cells, which has acquired an opening of its own to the exterior (text-fig. 9).

In the absence of complicated copulatory organs the Chætognatha resemble the Amphineura.

The spermatozoa are pin shaped, as in many Mollusca.

When we come to compare the arrangement of body cavities, genital organs, and their ducts in the Chætognatha with the various arrangements of these organs among the Mollusca, we find that the same structural plan is common to both groups, and that in the Chætognatha certain primitive features are retained (text-fig. 10).

The ancestor common to both Chætognatha and Mollusca had, we imagine, paired but unsegmented cœlomic spaces. Gonad mother-cells developed in each half, and were passed to the exterior through paired coelomoducts, which also served for the discharge of renal products. This ancestor we believe to have been hermaphrodite (text-fig. 10, A).

In the Aplacophoran Proneomenia, Chætognatha, and Cephalopoda certain of these primitive conditions persist, but not all in the same form.

In Proneomenia the coelom has become restricted to a relatively small portion of the whole body. The anterior ends of the paired cœlomic spaces remain separate as gonoccels; the posterior portions have become confluent, forming the pericardium, which communicates with the exterior by right and left coelomoducts, which serve the double purpose of genital and renal ducts (text-fig. 10, c).

In Cephalopoda the posterior portion of the cœlom is the gonocœl; it is undivided, and communicates with the exterior by a pair of gonad ducts (D). The anterior portion of the cœlom is paired in Octopoda, and forms the two pericardial spaces, each of which has its own duct to the exterior (E).

In Chiton the gonadial cœlom has become separated from the posterior pericardial cœlom, and each has its own pair of ducts to the exterior (F).

The universal occurrence of paired ducts in these primitive types is, we think, explained by the theory that in the common ancestor the cœlomic spaces were divided by a median longitudinal partition.

In Sagitta this primitive independence of right and left cœlomic spaces is retained, but the male and female portions of the hermaphrodite gland have become separated by the transverse genital septum, and a second pair of gonad ducts has come into existence in consequence (text-fig. 10, B).

See Addendum, page 394.

It now remains to be demonstrated that the developmental histories of the Chætognatha and Mollusca are not so diverse as has been usually supposed, for many close comparisons are possible between them.

In the larval stage the Chætognatha have no velum or other trace of external ciliation, and, in this respect, they resemble the Cephalopoda. The late appearance of cilia may be accounted for by the early development taking place inside an egg-shell, and by the larva, on hatching, being already so advanced as to have its adult swimming organs sufficiently developed. The typical Mollusc, on the other hand, hatches at an earlier stage, and goes through a period of swimming by cilia before settling down to the method of locomotion peculiar to its adult condition. So, although we have no evidence that the Chætognath ancestor was provided with any ciliary swimming organ, such a thing is very probable; and Grassi’s suggestion that the ciliated olfactory ring is the survival of a larval organ (although he did not indicate the right one) may not be far from the truth.

We will now pass to the chief features of resemblance in the embryology. The invaginate gastrula is common to both. The mouth has been proved to develop by a stomodæal invagination at the pole opposite the blastopore both in those Gastropoda which have little food-yolk and in the Chætognath, and the same statement is true of the Scaphopoda and Lamellibranchia and of Dondersia, though not of Chiton.

In Mollusca the mesoblast cells usually arrange themselves in two lateral masses, in which paired cœlomic cavities subsequently appear as schizoccelic spaces. Erlanger was certainly in error in attributing an enteroccelic origin to them in Paladina vivipara, and his statement has been refuted by Tönniges—and this is practically identical with what happens in Sagitta, for even here the lateral mesodermal pouches lose their enteroccelic lumen, according to Doncaster, and the permanent cœlomic spaces appeal’ as splits at a later stage. The only important point of difference is that the paired lateral cœlomic spaces of Sagitta remain separate throughout life, the splanchnic mesoblast giving rise to the dorsal and ventral mesenteries, whereas in the Mollusca they generally become confluent, although in every case indications of their paired origin maybe perceived in the fact that all organs derived from them are typically paired.

The four large cells which lie at the posterior end of meso-blastic pouches in a young Sagitta larva remind us of the first large mesoderm cells which are so conspicuous in the larvæ of Chiton, Teredo, and other Mollusca. Both ultimately give rise to the genital cells, but in the Chætognath these cells do not form anything but ova and certain ovarian structures (perhaps the oviduct). In the Mollusca the largo cells give rise to other mesodermic structures as well.

The resemblances between Chætognath and Molluscan nervous systems have already been described ; it remains but to observe that in their development they are absolutely identical. The dorsal and ventral ganglia appear as independent local thickenings of the ectoderm in which a considerable proliferation of nuclei takes place. A fact of importance is that the “ventral ganglion” of Sagitta, like the visceral ganglia of Mollusca, is paired in its origin, being derived from two cell-bands on the ventro-lateral regions of the body (Doncaster).

The buccal nervous system was said to be mesodermal by Hertwig, but this statement needs further confirmation. The development of the buccal ganglia from the stomodæal ectoderm has been demonstrated in Paladina, and a similar origin is possible in Sagitta.

The origin of the hood from two lateral ectodermal thickenings on either side of the mouth has already been described.

The fate of the mesoderm and of its paired cavities is of the highest importance. In Sagitta, after a transient stage, in which the body cavities are entirely obliterated, two bilaterally symmetrical pairs of cavities appear as schizocœls, the head cavities and the gonadial cavities respectively. The chief function of the mesoblast of the anterior or head pair of cavities is apparently to give rise to the musculature of the pharyngeal apparatus. The cavities themselves during the process of growth become extended into the base of the hood thus proving this organ not to be purely ectodermal in origin, a fact which is of considerable importance from the point of view of homology with the interbrachial membrane of the Cephalopoda.

The fate of the perigonadial cœlomic cavities is very similar in Chætognatha and Mollusca, and, in our opinion, the Chætognath type of cœlom, gonads and their ducts represents the ancestral Molluscan condition with greater exactitude than any scheme hitherto proposed.

In spite of the great variations in the arrangement of the Molluscan cœlom it is generally agreed that the most primitive condition is that in which the gonadial and reno-pericardial cavities are united as in the Cephalopoda and in the Aplaco-phoran Amphineura, as in a modified degree in the more archaic Gastropoda (Trochus) and Lamellibranchia (Solenomya), and as is indicated by the origin of the gonads from the wall of the pericardial space in Paladina and Dreissensia.

This view has been expressed by Dr. Pelseneer in a series of diagrams indicating the probable transformations of the genital duct (Mollusca, fig. 5 bis, p. 14), but in delineating the “ancestral hypothetical form” Dr. Pelseneer has not carried the history as far back as is now possible, for he has omitted to emphasise the real cause of the bilateral symmetry of the parts, namely, the primitive independence of right and left cœlomic cavities, necessitating the development of independent right and left ducts whether genital or renal.

When, as is typically the case in Mollusca, the right and left cœlomic cavities become confluent, one of the original pair of ducts may be dispensed with. Confirmatory evidence for the primitive separation of the cœlomic cavities into right and left among Mollusca is to be found in the paired structure of the gonads of Aplacophora and Lamellibranchia, in the paired origin of the pericardial coelom in Paludina, and in the very widely-distributed, paired, renal cavities. In the Chætognatha the primitive division of cœlomic spaces into right and left halves is retained throughout life.

Thus far our comparison lies on certain ground. The next point to be considered is the significance of the division of the perigonadial cœlom by the transverse genital septum in the Chætognatha, and this is capable of more than one interpretation. If we regard this genital septum as a division between two homomeric segments we shall adopt the simplest explanation. The condition of transversely-divided cœlomic spaces in Chiton would be evolved from the Chætognath condition on the assumption that the posterior cavity has lost the power of developing gonads, and that its ducts have become purely renal in function.

The genital coelom of Chiton would, on this theory, be the homologue of the anterior pair of body cavities of Sagitta.

Unfortunately the observed facts of development do not support the view of a metamerie repetition of these cavities. The genital rudiments in Mollusca first appear as ridges on the pericardial walls, and therefore in a cavity which is not metamerically repeated, and grave doubts have been recently thrown upon the view that the genital septum of Sagitta is an indication of metamerie segmentation by the above-mentioned observations of Doncaster.

We are thus led to the conclusion that Sagitta and primitive Mollusca, like the Aplacophora, had but a single pair of gonadial-renal-cœlomic cavities, into which the hermaphrodite gonads were dehisced and from which they passed to the exterior by a single pair of ducts, together with the nitrogenous waste. By the differentiation of the walls of this cavity and its ducts into excretory and gonad-bearing portions we arrive at the condition typical of the higher types of Mollusca in which a second pair of ducts is developed in connection with the discharge of the genital products. Such is the condition in the Cephalopoda in which the gonadial pericardial cavity remains undivided, but in the Polyplacophora the gonadial cœlom has become completely separated from the renal cœlom. In the Chætognatha the gonadial cœlomic spaces have become separated for the purpose of dividing the male and female sexual cells (text-fig. 10).

Both Chætognatha and Mollusca are hermaphrodite. The parent germ-cells originate at the same time, and are indistinguishable. It might be suggested that the determining factors of sex only come into operation after the germ-cells have migrated into separate cavities in which different nutritive conditions prevail.

The very numerous and close resemblances which exist between the Chætognatha and every class of Mollusca show that there is no single structure, or absence of structure, of importance in the Chætognath anatomy which is not capable of being developed somewhere within the limits of the Molluscan phylum, and that many apparently insignificant features have their exact counterparts therein. In short, granted similar conditions of life, a Chætognath type might be expected to arise from the Molluscan stock. On the other hand, the claims which have been put forward in favour of the alliance of the Chætognatha with the Annelida or Nematoda are not capable of being supported by so extensive or consistent an argument.

Can the Chætognatha be definitely associated with any one class of Mollusca, or are their relationships more general ?

The following morphological characters are among those obviously of importance :

1. The original bilateral symmetry of the Mollusca is presented by the Chætognatha in its most perfect form, especially in respect of the body cavities.

2. The Chætognatha resemble many Mollusca of undoubtedly primitive type, in the absence of apparent segmentation.

3. The vermiform shape of the body, recalling that of the Amphineura aplacophora.

4. No extensive hæmocœl has been hitherto identified.

5. The alimentary canal is straight ; the anus opens in front of part of the visceral sac.

6. There is no evidence of a radula, either in the Chætognatha or in their ancestors. The buccal armature is other-wise very like that of many Mollusca.

7. The nervous system is of the Molluscan type.

8. The “hood,” the suggested homologue of the circumoral Cephalopod “foot.”

9. The growth of the genital cells within a follicular epithelium and upon stalks. Hermaphroditism.

10. The two pairs of openings from the perigonadial coelom to the exterior. These are believed to be the homologues of the two pairs of ducts leading from the pericardial-gonadial cœlom of the more primitive Mollusca.

11. The cephalic endo-skeleton and lateral fins.

12. The preoral ciliated ring ; the suggested homologue of the velum.

The characters of the Chætognatha are, on the whole, just those of the more archiac types of Mollusca rather than of the Gastropoda or Lamellibranchia. Their affinity to the Aplacophoran Amphineura is indicated by the vermiform shape and bilateral symmetry of the body, by the straight alimentary canal, by many negative characters, such as absence of shell, foot, and radula. The Amphineura may have a more primitive nervous system, but in the Chætognatha the paired arrangement of body cavities is the more primitive. Molluscan nephridia and Chætognath sperm-ducts would appear to be homologous structures, but the genital ducts of Chiton may only be the analogues of the oviducts of Sagitta.

The resemblance to certain Cephalopoda is also surprisingly close. The body-cavity of the Chætognatha is in the more primitive condition, in that it remains divided longitudinally throughout life, but owing to the fact that the Cephalopoda are dioecious, there is not the same need for a separation of the male and female gonads by a genital septum that there apparently is among Chætognatha. The nervous system of Nautilus and Chætognatha are both modifications of the same plan.. Endocephalic skeletons, germ-cells mounted on stalks, and last, but not least, the bilateral “foot,” hood-like and surrounding the mouth, are common to both Nautilus and Chætognatha.

Another very remarkable fact is that the peculiar characteristics of the Leioglossal Octopods—no radula, arms united by a membrane, fins developed on the sides of the body— are all characteristics of Sagitta.

Although a few naturalists have already suggested that the nervous system of the Mollusca and the Chætognatha may indicate relationship, no one has hitherto pointed out how perfect is the resemblance between the other organs of the two groups.

We have in the Chætognatha the key to many Molluscan mysteries, not the least important of which is the development of the Cephalopoda, which, owing to the presence of a large mass of inert food-yolk, has been profoundly modified. In fact, in the matter of developmental history, Sagitta would seem to bear the same relationship to the Cephalopoda that Amphioxus bears to the higher Vertebrata.

If our theory be correct the Cbætognath type of development may be that of the Pre-Silurian ancestors of Nautilus, and probably of the earlier straight-bodied Cephalopoda, such as the Orthoceratidæ and the Bactritidse.

A further inference of far-reaching importance is that the crawling Archimollusc of Lankester was not the common ancestor of the entire Molluscan phylum,, but only of the Polyplacophor-Gastropod-Lamellibranch section, and that the Cephalopoda have never had a creeping ancestor at all ; and, in our opinion, the Creeping Archimollusc itself has been derived from a Swimming Archimollusc from which the Amphineura, Aplacophora, and Cephalopoda have been independently evolved, and of which the Chætognatha are the nearest living representatives.

Several groups of Mollusca descended from the creeping Archimollusc have again taken to a predatory life among the Necton and Plankton, and have apparently thrown back to the earlier swimming Archimolluscan ancestor in several respects. It is thus that characteristics common to the Chætognatha and the Pteropoda are to be explained. The hereditary taint of asymmetry, however, is never lost ; even in Phyllirhoe the anus and the genital organs remain on one side of the body.

In the light of the undoubtedly primitive nature of the Chætognatha, the negative characteristics of the Aplacophora may well be reconsidered. Pelseneer and others consider that the Polyplacophora present the most archaic characters among the Amphineura, but until we have good evidence to the contrary there is at least equal justification for the view that certain structures in the Aplacophora are in an incipient rather than in a reduced condition. Foot, radula, and shell may never have attained to a higher grade of development among their ancestors than that which they have reached among the living Aplacophora.

The association of the Chætognatha with the Mollusca does not throw much light ou the position of Rhodope. The absence of heart, shell, radula, and foot are not, in our opinion, insuperable obstacles to including this curious minute form in the Molluscan phylum, but the fact of the presence of flame cells opens a vista of new difficulties.

By the inclusion of the Chætognatha within the Molluscan phylum no new suggestion is indicated as to the connection of the phylum with any other division of the animal kingdom, but the classes within the phylum will require some readjustment.

Our phylogenetic speculations are influenced by the following facts :

A. Mollusca typically pass through a free-swimming, bilaterally, symmetrical, “veliger” stage.

B. In creeping and sessile forms the foot and shell reach their highest development.

C. In free-swimming marine forms the shell tends to atrophy, and the creeping foot tends to become either a swimming organ or to disappear.

It may therefore be fairly argued that if a race of Mollusca had always been free swimming, either foot or shell might be absent.

The Chætognatha may therefore be fairly regarded as the living adult representatives of the phyletic stage indicated by the veliger larva, and it is from such an ancestor that we conceive the creeping Polypiacophora, the worm-like Aplacophora, and the swimming Cephalopoda to have been independently derived.

A scheme of classification which would represent this view would be—

Phylum.—MOLLUSCA.

Grade A.—Nectomalacia or Mollusca natantia.

Mollusca in which the primitive free-swimming habit has been retained : “foot” circumoral, a propodium developed from paired lateral Anlage.

Class 1. Chætognatha.

Without shell.

” 2. Cephalopoda.

With shell.

Grade B.—Herpetomalacia or Mollusca reptantia.

Mollusca in which a creeping habit has been developed : foot postoral, a metapodium, developed as an unpaired median structure.

Class 3. Amphineura Aplacophora.

Without shell.

” 4. Amphineura Polyplacophora.

Shells octuple.

” 5. Lamellibranchia.

Shell bivalve.

” 6. Gastropoda.

Shell single.

” 7. Scaphopoda.

Shell single, tubular.

ADDENDUM TO PAGE 380

Conant states that there is no longitudinal partition between the spermatic chambers in Spadella schizoptera, and that the right and left oviducts are in communication with one another by branches which meet and form a small blind tubule lying on the mid-ventral line beneath the intestine. This union is believed to be a means for helping a single copulation, affecting one side only, to fertilise the ova of both sides of the body. (MAT, 1907.)