On the Ancestral Form of the Chordata

Since then the aspect of things has changed and later investigations, more especially those of Dohrn and of Ray Lankester, have rendered it nearly certain that the Tunicata must, on the contrary, be looked upon as degenerate Vertebrates which can be hardly of much use in helping us to the failing clue.

Dohrn, Semper, Hatschek, Leydig, Kleinenberg, and Eisig are amongst the foremost who have suggested, and most brilliantly expounded and argued, that the Annelids offer the greatest number of points of resemblance with the Vertebrates; that they and the Arthropods have descended together with the Vertebrata from a primitive type, distantly agreeing in shape with Polygordius, and that the only postulate which this assumption necessarily implies, is the old idea of Geoffroy St. Hilaire that the ventral side of Annelids and Arthropods is homologous with the dorsal side of the Vertebrata.

These naturalists explain the difference in situation of the mouth and oesophagus, with respect to the cerebral ganglion, by divers subtle hypotheses, which, however, generally disagree with each other. Their views are nevertheless rapidly gaining ground, although the school of Gegenbaur and Haeckel has never been reconciled to them. Gegenbaur looks upon two lateral cords such as are present in Nemertines as a very primitive arrangement, from whence might at any rate be derived the ventral nerve-cord of Annelids and Arthropods. Harting (‘Leerboek der Dierkunde,’ 1874) was inclined to accept the possibility of a similar dorsal coalescence out of which a spinal cord might take origin. Balfour and myself felt strongly inclined to choose this side in the contest—he in once more tracing the outlines of a similar explanation (‘Development of Elasmobranch Fishes’), I in recapitulating the facts, such as they had made themselves known to me in the organisation of certain Nemertines, in which, indeed, a tendency towards approximation of the lateral cords on the dorsal side was unmistakable (‘Verh. der Kon. Akad. van Wetenschappen,’ Amsterdam, 1880). When Balfour, in the second volume of his ‘Comparative Embryology,’ made himself a definite advocate of this view in opposition to the combatants for the Annelidan affinities, it may be safely inferred that many of the younger naturalists paused to reconsider the claims of both hypotheses.

The great difficulty which is encountered in any attempt to point out a definite group amongst invertebrates most closely related to the primitive Vertebrata is the total absence of anything resembling so important and so early-formed an organ as the Vertebrate Chorda dorsalis. Attempts to find anything like it amongst the Annelids, even amongst Polygordius and its archaic allies have proved either futile or barren.

I will at present attempt to point out in what group of invertebrate animals we do indeed find an organ which, in my estimation, ranks equal to the vertebrate notochord, and thus supplies the much-desired transitional form by which the Chordata are allied to the lower Metazoa, and in fact to such forms as have neither the much specialised organisation of the segmented animals (Arthropods and Annelids), nor require to be turned upside down before their homology with the lowest Vertebrata is admissible.

That I venture to state this hypothesis before I am able to bring into the field a well-arranged host of facts in its support, must be ascribed to my desire to induce such fellowworkers in biology as have more leisure and better occasion for investigating the numerous problems it suggests than I have, into taking up a question which cannot be but looked upon as of the highest significance for modern morphology.

According to my opinion the proboscis of the Nemerteans, which arises as an invaginable structure (entirely derived, both phylo- and ontogenetically, from the epiblast), and which passes through a part of the cerebral ganglion, is homologous with the rudimentary organ which is found in the whole series of Vertebrates without exception—the hypophysis cerebri. The proboscidian sheath of the Nemerteans is comparable in situation (and development?) with the chorda dorsalis of Vertebrates.

After this brief statement of my hypothesis I will enter into a short discussion of its different details.

It is not my intention to consider the numerous modifications of structure which the hypophysis cerebri presents in different adult Vertebrates, nor its glandular appearance, the connection into which it enters with blood-vessels, &c.; but I wish to restrict myself to the comparison of its very first ontogenetic stages, in which it may be presumed most purely to reproduce its ancestral character.

We find the hypophysis originating as an invagination from the epiblast, arising either independently on the outer surface (this, according to Dohrn’s interesting researches, is the case in one of the lowest of the vertebrate scale—Petromyzon), this invagination at the same time being directed towards the anterior termination of the notochord, and lying in its direct continuation (figs. 1 and 2), or (as is the case in the higher Vertebrates) not directly on the outer surface, but on that portion of the epiblast which has become the stomodæum (fig. 6). In the latter case it arises as a median dorsal outgrowth from the mouth-cavity, directed towards that portion of the under surface of the brain where, between Prosencephalon and Metencephalon, the infundibulum travels downwards, this being at the same time the limit up to which the notochord extends forwards under the brain. The fact that an outgrowth from the brain thus grows downwards to meet this epiblastic invagination sufficiently indicates that in ancestral generations, where the hypophysis was a less rudimentary organ, some sort of connection existed between it and the cerebral thickening of the central nervous system.

The constant presence in all Vertebrates of an organ so rudimentary as the hypophysis, and about the significance of which no plausible explanation has as yet been offered, has already been insisted upon above.

Both facts are in favour of regarding it as a very ancient structure, which was once of great importance, and had a different and at the same time a more definite physiological value.

In tracing this ancestral significance, the relation to the brain and the somewhat less direct but, nevertheless, unmistakable relation to the notochord must not be lost sight of.

We will now consider the ontogenetic and phylogenetic history of the Nemertine proboscis. In the lower Platyelminths the researches of v. Graff, lately crowned by his brilliant monograph, have brought to light the different stages through which retractility of a portion of the tactile anterior extremity of the body, in which urticating elements are present, leads to the appearance of a definite proboscidian structure, which obtains a special musculature, and which finally (in Graff’s Rhabdocoel family of the Proboscida) has definitely become a proboscis, which is directly comparable to that of Nemertines, situated like it above the intestine, internally (externally when everted) clothed by the direct continuation of the outer layer of epiblast, and serving tactile and at the same time—through its nematocysts—aggressive purposes.

The proboscis of Nemertines is thus directly related to this important structure of the lower flat-worms, as was already noticed in Gegenbaur’s ‘Grundzüge’ (1870). We find urticating elements largely developed in the proboscidian coating of Palaeo- and Schizonemertini, whereas in the Hoplonemertini the tactile significance may perhaps have come to predominate, if we judge from the extremely complicate arrangement and massive development of nervous tissue in the proboscis of these forms, which, moreover, is here provided with a central stylet-shaped armature.

As to the ontogenetic development of the Nemertean proboscis, the great majority of authorities are in accordance that it develops as an invagination from the epiblast, which commences at the anterior extremity and gradually pushes backwards. Extensive details as to its successive developmental stages are, however, not yet to hand, only the principal fact above mentioned being generally accepted.

It is highly important to notice that in this backward course the proboscis takes its way between the two anterior thickenings of the lateral nerve-cords, which in Carinella constitute the simplest Nemertean brain, and in other genera become more or less subdivided, the right and left halves being united by a thick commissure (fig. 3), ventral in relation to the proboscis, and by a thinner one dorsal to it. In all cases the proboscis passes through the ring of nerve tissue thus formed; in all cases the proboscidian sheath reaches forwards to the level of this nervous commissure, through which the proboscis passes (fig. 4).

If we may look upon the spinal cord and brain of vertebrates as a dorsal coalescence of lateral trunks similar to those of the Nemertines (as was already advocated in my paper “zur Anatomie und Physiologie des Nervensystems der Nemertinen,”Amsterdam, 1880), then the double proposition above enunciated necessarily leads to the conclusion that the spot just mentioned corresponds to that part of the vertebrate brain where the hypophysis (proboscis) bends upwards towards the central nervous apparatus and where the notochord (proboscidian sheath) terminates, i.e. the region of the primitive fore-brain. This proposition at the same time implies the homology between the vertebrate fore-brain and part of the nervous lobes of Platyelminth ancestors.

It remains to be further inquired into—and the facts as they lie before us are very suggestive in this direction—whether perhaps the distinction between the two pair of lobes as they are present in most Nemertines may not have been perpetuated in the vertebrates, these superior lobes (after dorsal coalescence of the two halves of the nervous system) becoming the forebrain, the inferior ones the equivalents of mid- and hind-brain. The following two points are in favour of such an interpretation: (1) the nerves for the higher sense organs, eyes,¹ and olfactory (?) pits start from the superior brain lobes; (2) the strong nerve which on both sides supplies the anterior (respiratory, M’Intosh !) region of the oesophagus, and for which in a former paper I have proposed the name of N. vagus, takes its origin in the inferior lobes (figs. 3 and 5).

Upon the dorso-median coalescence of these inferior lobes and of the lateral stems, above the intestine and the proboscidian sheath, the latter must have become severed anteriorly from its connection with both nerve-system and proboscis. Might not the fact of the anterior end of the notochord being bent upwards in several of the lower Elasmobranchs (cf. Gegenbaur, ‘Das Kopfskelet der Selachier,’ pl. ix, figs. 1 and 2) be interpreted as a reminiscence of this connection ?

A further character which is common to the two epiblastic invaginations, known as hypophysis and as proboscis respectively, is the shifting of their external opening. Amongst Nemertines examples are found which form a parallel to the larger bulk of Vertebrates (fig. 6) where the hypophysis does not arise (as in Petromyzon) independently on the outer surface, but where it is an invagination directed upwards from the roof of the mouth cavity. Both in Malacobdella and in Akrostomum (a genus of Hoplonemertini, instituted by Grube, in which I place, for example, M‘Intosh’s Amphiporus bioculatus and Amph. hastatus, and of which I have myself examined several specimens) the opening for the proboscis is not independently situated at the anterior extremity, but is found on the dorsal wall of the intestinal tract, just inside the mouth (figs. 7—10). I have the strongest reason to believe, upon which I will not here further enter, that this is a secondary modification, and that the separate opening is the original state of things, phylogenetically related to the separate proboscis of certain Rhabdocoels.

The facts here advanced may justify us in looking upon the Platyelminth (s. str. Nemertean) proboscis as the homologue of the vertebrate hypophysis, as was implied in the first part of our proposition.

The proboscidian sheath in Nemertines is a cavity closed on all sides and lined by an epithelium. It is situated in the median dorsal line, above the intestine, just inside the muscular body wall, to which it is more or less firmly attached. Muscular fibres serve to a large extent towards the thickening of the tube we are considering.

It terminates in the immediate vicinity of the anus, and reaches forwards to just in front of the cerebral ganglia, which in Schizo- and Palæonemertini are situated at a short distance in front of the ventrally situated mouth. In the Hoplonemertini the mouth has travelled forwards till close to the tip of the head, the intestinal tract thus passing beyond the proboscidian sheath anteriorly. In certain other Nemertines the proboscidian cavity does not extend through the whole length of the body posteriorly. This is, for example, the case in that genus which must be regarded for several reasons as the least differentiated, primitive type—the genus Carinella. It is only in the anterior region of the body that the proboscis and the cavity surrounding it are found, the latter situated as usual above the intestine. Here, too, the mouth is found ventrally, the opening for the proboscis terminally. One other genus—Drepanophorus—deserves special mention, in so far as the proboscidian sheath has the bulk of its cavity increased by lateral thin-walled sacs, metamerically placed, one above each lateral cæcum of the intestine, and communicating with the cavity of the sheath by narrow perforations of the muscular tissue of its wall. Nemertes carcinophila is said both by M’Intosh and Barrois to be without a special proboscidian sheath. Barrois found the proboscis much reduced (according to him as an effect of parasitism) and floating in the general body-cavity. Not having examined this species myself, and not having either ever met with a general body-cavity in other Nemertines, I would venture to suggest the necessity of a careful reexamination of this species, which might prove to be not without importance for the problem we are considering.

The type according to which the proboscidian sheath is built up is very similar throughout the whole group, although the muscular elements in its wall may increase in number (fig, 16) and become more complicately arranged, or its size may be considerably reduced. It is capable of a very considerable increase in width corresponding to the movements, the rapid retraction, or the mode of coiling up of the proboscis it encloses. It is, moreover, filled with a fluid, containing corpuscles characteristic in shape, and in one case—Cerebratulus urticans—characteristic in chemical properties, viz. by the presence of haemoglobin, This fluid is in no way connected with the fluid circulating in the longitudinal and transverse blood-vessels. The dorsal blood-vessel takes its course beneath the proboscidian sheath, between it and the intestine; in many instances it is enclosed in the muscular wall of the sheath in the foremost part of the body, above the oesophagus. The possibility of a comparison with the subnotochordal rod of Vertebrates ought to be considered.

The inner epithelium lining the cavity of the sheath is very marked and everywhere present; it is least conspicuous in Carinella, owing perhaps to the considerable extension which the sheath had undergone in all specimens that have hitherto been examined in view to this point.

This being the general arrangement of the proboscidian sheath, it now concerns us to examine what is known about its development in the embryo. The data available are very scanty, and in some respects contradictory. Barrois describes it in certain species of Lineus as developing from the reticulum, the mesoblastic tissue between the epi- and hypoblast, and gradually extending backwards at the same rate as the developing proboscis pushes it in that direction. In Amphiporus the development of the proboscidian sheath was studied by the same observer, and according to his description there is a remarkable divergence from the development in Lineus.

In Amphiporus the proboscidian sheath is not formed gradually, travelling slowly backwards along the median dorsal line, but the sheath suddenly appears all round the whole length of the proboscis. It is here formed out of the fatty mass, which also gives rise to the digestive tube.

Tetraslemma, another Hoplonemertine, corresponds closely, according to the same observer, with Amphiporus just described.

Salensky, who has lately given a very short account (‘Biologisches Centralblatt,’ 1883) of the development of Nemertes (Borlasia) vivipara, ascribes a mesoblastic origin to the proboscidian sheath. Nevertheless, he noticed what appeared to him to be a connection between the first origin (Anlage) of the oesophagus and that of the proboscis. As he has postponed giving the details of this connection to a later publication we cannot at present judge of its significance.

Hoffman is the only other author who gives any details about the formation of the proboscidian sheath. According to his account of sections made of Tetrastemma, a portion of the proboscis is split off from the dorsal surface of the alimentary canal. The muscular proboscidian sheath is mesoblastic in origin. This observation, which can hardly be brought to agree with the epiblastic origin of the proboscis, noticed above, might perhaps allow of a different interpretation. As a simple suggestion I would advance, that perhaps Hoffmann may have mistaken the formation of the inner portion of the proboscidian sheath (so often confused with the proboscis !) for that of a part of the proboscis itself.

Hypoblastic formative elements internally would then coalesce with mesoblastic derivates, more especially muscular elements, exteriorly applied to the former and constituting together the proboscidian sheath, i.e. the wall of the proboscidian chamber.

Such an interpretation would appear to be more acceptable than the coalescence of a tubiform derivate of the hypoblast, with an invagination of the epiblast travelling backwards, the fusion of these two giving rise to the definite, cylindrical, eversible proboscis. Balfour, in his ‘Comparative Embryology,’ is not inclined to accept Hoffmann’s statements without further confirmation.

Still this observation, if it may be interpreted as proposed, would be of importance and its repetition much to be desired. This and Barrois’-description above cited appear to open the prospect that embryology may eventually succeed in demonstrating for the proboscidian sheath, or for one of its constituent parts, a hypoblastic origin.

If such be proved to be the case, not only its situations but also its development would correspond to that of the notochord of the lower Vertebrates. Still, considering in how many cases the origin of the notochord in Vertebrates is apparently mesoblastic (this phenomenon being considered as secondary, the hypoblastic origin as the primary, or ancestral arrangement), it cannot be considered as absolutely necessary, that in the other offshoot, the Nemertines, the hypoblastic origin of the proboscidian sheath should first be demonstrated before any homology between notochord and proboscidian sheath may be accepted. In Nemertines as well as in Vertebrata the mesoblastic origin of the proboscidian sheath might be a secondary condition. In that case much value could not be attributed to special cases of coincidence in the embryological data, and it would be the most primitive representatives of both groups which would more especially be likely to furnish evidence of a conclusive character. However, we must here wait for more circumstantial evidence before being justified in further advancing in the domain of speculation. Yet, with respect to this question, it must not be overlooked that the fact of the cavity of the proboscidian sheath carrying peculiar corpuscles, and of its being a closed sac lined by an epithelium, tends far to give it the general character of a body-cavity, a coelomic diverticulum, which is indeed situated dorsally and longitudinally, but which by those general characters would lead us to expect a derivation from the archenteron, rather than a schizoccelic origin in mesoblastic tissues.

If we were inclined to accept the mesoblastic derivation as primary, and wished to picture to ourselves a possible common origin of notochord and proboscidian sheath in the common ancestor of Vertebrates and Nemertines, we should have to postulate as a still more primitive arrangement an axial thickening of the mesoblastic tissues, which became more solid in the one and hollowed out by the proboscis in the other. This would clash both with the hypoblastic origin of the vertebrate notochord and with the phylogenetic significance of the hypophysis.

We have now to consider certain aspects of the suggested homology between notochord and proboscidian sheath.

There is no doubt that the fully developed notochord of a Vertebrate is a structure of an entirely different character from the proboscidian sheath of a Nemertean. The one is a solid, rod-like organ, the other a hollow tube. However, at the very earliest stages of its formation the notochord of the primitive vertebrates (cf. Hatchek’s ‘Development of Amphioxus ‘) possesses a central groove, which is a derivate of the archenteron, and which only secondarily, in accordance with the further differentiation of the tissues of the notochord, obliterates.¹

The ulterior difference in histological structure of the one, a cellular tissue eminently vacuolar, and of the other: the cellular lining and fluid contents of a tube, the cavity of which does, as a rule, not obliterate, is, however, no serious objection to their eventual homology. In more than one instance modern morphology recognises solid cellular strings to be homologous with others containing a cavity inside them.

The different degree in which muscular tissue takes part in the constitution of the proboscidian sheath (figs. 16—18) must of course not be overlooked, the more so as it is entirely absent in the notochord and its envelopes. However, in Nemertines this muscular tissue can be shown to be most closely related to the function of the proboscis, and in fact to be sometimes exceedingly reduced. For this reason its importance as a point of comparison must not be over-rated.

All these differences are in the last instance due to the different significance in the animal economy which has in the two groups been attained by this organ. In the Vertebrates this central rod-like structure, sustaining the mesoblastic somites in their progressive development, has a significance as a temporary axis, round which these processes take place. Its important character as a primitive, i. e. as an ancestral organ, is recognised notwithstanding, or rather just because of, its gradual disappearance in the adult forms of the higher groups, where its significance as a sustaining axis has been replaced by that of the vertebral column.

That in a far distant ancestor of the vertebrates it may similarly have been subservient to the lodging of a retractile proboscis, tactile in function, appears to me to follow from a careful consideration of the relations between the hypophysis and the notochord, and between the first-named rudimentary organ and the brain.

An important fact, which in conclusion I must make mention of, is a phenomenon which I have repeatedly observed in the posterior portion of the proboscidian sheath of different species of Cerebratulus, very long Nemertines, where the sheath reaches down to the posterior extremity of the body. Whereas in young specimens of this species the sheath was a hollow tube down to the very end, in older and larger specimens the aspect of things had changed. In the posterior extremity of the body the cavity was here nearly filled up by a continuous cellular tissue with distinct nuclei (fig. 18), sometimes even entirely obliterated. This cellular tissue is sometimes apparently glandular, the arrangement in some cases even such that it must be interpreted as a set of radial acini, by which the surface is considerably enlarged. Future investigations will have to decide whether the cases of evident obliteration may be interpreted as a step towards real solidification of that part of the tube which is comparatively of the smallest value for the general function in connection with the expulsion of the proboscis. This change of function and histological appearance is only present in the more primitively organised groups, which rarely make use of their proboscis; in the more highly specialised Hoplonemertini, where the proboscis is in constant play, and the development of the muscular elements in the proboscidian sheath vary considerably, it was nowhere found.

Apart from the argument which can be derived from the nature of this cellular coating the significance of this phenomenon will have to be carefully inquired into. Along such a line of development we might picture to ourselves the eventual conversion of a hollow proboscidian sheath into a solid notochord, the more so as functionally the proboscidian sheath in Nemertines may already be looked upon as an axis, around which the other organs symmetrically arrange themselves as they do around the notochord in Vertebrates. It must at the same time be borne in mind that the muscular coating in this posterior portion is found to be considerably reduced and replaced by a more or less homogeneous and comparatively thin sheath.

Having thus far considered the arguments which are at present available for insisting upon the homology between proboscis and hypophysis, on the one hand, and between proboscidian sheath and notochord on the other, it now remains to inquire whether there are points in the anatomy of Nemertines, beyond those just now exposed, which either corroborate or weaken the evidence hitherto advanced in favour of the suggestion that the Nemertines, more than any other known group of Invertebrates, resemble the ancestors of the Protochordata.

I need hardly insist upon the fact that I do not advocate any direct relation between existing Nemertines and existing Vertebrates; my argument goes no further than the attempt to show that the general plan of structure of a Nemertine is more in accordance with that of a vertebrate animal than is, for example, that of the Archiannelida, and that the link connecting Cœlenterate ancestors with Vertebrate descendants has most probably comprised forms in which two lateral nerve-cords were present, ultimately coalescing dors ally, and in which an epiblastic proboscis served for purposes which have either been given up or have been replaced by others when the animals gradually exchanged the Platyelminth for the Chordate type.

Simultaneously with this passage from the Cœlenterate type to the Chordate the highly important processes must have been gone through which lead to the formation of a body-cavity, separate from the archenteron with which, as embryology teaches us, certain diverticula are originally in open communication, ultimately becoming constricted off and developing into the splanchnic and somatic layers, which have the body-cavity between them.

The brilliant researches of Lang on Gunda segmentata, and of Hatschek on the ‘Development of the Amphioxus,’ must here in the first instance guide us; and anybody having carefully perused those important contributions, and having compared them with each other, must have been struck by the great probability of the view advocated by Lang, that the alimentary diverticula of these Platyelminths are the fore-runners of the arrangement of the cœlom in the higher enteroccelous worms, and that through this link a glimpse is gained at the road along which Annelids may have developed out of an ancestral Platyelminth stock.¹

On the other hand, the stages in the development of Amphioxus, where a double set of lateral diverticula of the archenteron is present (fig. 12), which ultimately become converted into the mesoblastic somites, appear to be of very great importance, in so far as they render it highly probable that in the ancestry of vertebrates certain forms with metamerically placed alimentary cæca must have obtained, of which the larval stage of Amphioxus is the reminiscence. In the remaining vertebrates the primitive alimentary diverticula giving rise to the cœlom are reduced to two. This appears to be an ulterior simplification. An attempt to explain this simplification, and to bring the process of the formation of the coelom in Amphioxus under the same head with that in the other vertebrates, has at the present moment not yet been made. It is, however, sure to be made some day by the leading authorities on the subject. For the present it may suffice to point out that the ulterior development of the mesoblastic somites in the bulk of the vertebrates re-establishes the homology with the more primitive processes in Amphioxus.

For us this larval stage of Amphioxus is all the more interesting, because it must lead up to Platyelminths, corresponding with Gunda in the presence of alimentary cæca, metarnerically placed and of a general internal métamérisation, but differing from Gunda in such important respects as the presence of the forerunners, both of the hypophysis and of the notochord, two structures no trace of which is found in the salt-water Triclades. Such Platyelminths must needs have resembled the present Nemertines more than anything else.

Here the important question at once thrusts itself upon us: Has the formation of a coelom already been arrived at in the Nemertines or not? i.e. have these animals a body-cavity developed out of and separated from the primitive digestive tract or not ? Although I have formerly, when attempts were made to bring the Nemertines under the so-called Parenchymatous Flat-worms, combated those attempts, and endeavoured to show that the regular arrangement of digestive and generative cæca, the development of muscular septa between them, &c., went contrary to it, yet now that our ideas about the significance of a true bodycavity as an ultimate derivate of the archenteron have of late years gained so considerably in clearness and definition, I should hesitate to affirm that any such body-cavity is present in Nemertines, and would be inclined to answer the question proposed above negatively.

Both in the more highly, differentiated Hoplonemertini and in the more primitive Schizo- and Palæonemertini, I have met with numerous instances in which all the space which remained free between the muscular body-wall on. the one hand, and the intestinal, generative, proboscidian, and circulatory cavities on the other, was one unbroken mass of connective tissue.

Sometimes, more especially around the oesophagus, occurred what were apparently fissures and cavities in this tissue. They were not lined by an epithelium (are perhaps in communication with the vascular system ?), and could best be compared to a true Schizocœlom (Huxley), i.e. fissures in a mesoblastic tissue.

All this makes me very much inclined to look upon the alimentary diverticula of the Nemertines in the same light as Lang does upon those of Gunda; incipient coelomic sacs, comparable to those of the larval stage of Amphioxus.

A question very difficult to answer is this: How do these alimentary diverticula eventually come to exchange their function and significance to such an extent? If they were originally acquired with a view to an enlargement of the digestive surface, they must in the course of time, as they became constricted off, have lost this function, and in its stead have developed powerful layers of epithelial muscular tissue in their walls, which then represent the successive myomeres, and which finally supplant the original muscular body-wall (Hautmuskelschlauch), itself never divided into myomeres, and originally derived from the epiblast.

Traces of this epiblastic muscular sheath, primitively enveloping the myomeres, which secondarily spring from the alimentary diverticula, appear to be found in certain Vertebrates, externally to their general musculature.

It remains for the present unsolved what were the leading factors in this important transformation, the general outlines of which we have here only touched upon.

We have now to compare Nemertines and primitive Vertebrates under another important head: foremost oesophageal diverticula and their relation to respiratory functions and sensorial (?) apparatus. Here, too, I do not wish to enter into a thorough discussion of the subject; an enumeration of the chief points may suffice for the present.

A special respiratory apparatus in the form of external branchiæ has never been met with in Nemertines. In a very early stage of embryological development, however, two lateral diverticula, situated in the very foremost portion of the oesophagus in front of the mouth, bud out from its wall (Bütschli, Barrois, and others), and are in this stage directly comparable to similar diverticula which arise in the same region in the Balanoglossus larva, and there give rise to the first pair of branchial slits (figs. 14 and 15),

In Nemertines these diverticula become constricted off from their point of origin—the oesophagus—and entering into connection with invaginations from the epiblast, which bring about a free access of the external sea-water, they become converted—at least in the large section of the Schizonemertines—into an apparatus which I have proved to be subservient (‘Zur Anatomie u. Physiologie der Nemertinen,’ p. 28) to a process of cerebral respiration, in which oxygen is carried to the nervous system itself, the cellular elements of which are in this subdivision profusely provided with haemoglobin.

I am not prepared to say that in the great subdivision of the Hoplonemertini, where the central nervous apparatus is no longer provided with haemoglobin, but where, on the contrary, the circulating fluid is, these diverticula, which continue to develop in the same way out of the oesophagus, are also—and in the first place—subservient to a respiratory process. I am rather inclined to believe that in this group the cephalic grooves—as the epiblastic invaginations travelling inwards to meet the hypoblastic diverticula in question are called—remain more especially adapted for sensiferous purposes, probably of olfactory nature. The way in which the complicated organs in the adult, the so-called side organs, develop, remains quite the same: an outgrowth from the oesophagus coalesces with an ingrowth from the epiblast, the principal difference being that the connection with the brain-lobes is no more so intimate, and that the apparatus is connected with the brain by a special set of nerves. In some species it continues to be situated behind the brain, in others it becomes placed in front of the central nervous apparatus.

It appears to me that these facts are not without significance. However, I refrain for the present from a further discussion, and would merely wish to point to an interesting detail in the development of Amphioxus lately brought to light by Hatschek’s researches. It is the presence in the anterior region of the oesophagus, in front of the mouth, of two lateral hypoblastic diverticula, differing in their nature and in their further development, both from the archenteric diverticula (mesoblastic somites), and from the branchial outgrowths of the oesophagus. These two diverticula, originally symmetrical, become constricted off from the hypoblast, and in their further development they have a different fate, the left one communicating with the exterior by a ciliated opening, which appears in the epiblast, the right one forming an epithelial lining in the præoral body-region. The left one was looked upon by Kowalevsky as a special sense-organ of the larva.

Although I am not prepared for the present to furnish any evidence in this direction, I would call attention to the similarity in development between these structures and the cephalic diverticulum of Nemertines. Considering the amount of degeneration which in several respects Amphioxus appears to have undergone, it does not appear impossible that the left lobe is really a temporary olfactory organ, the right one having entered upon other functions, and having lost its original significance.

These oesophageal diverticula of Amphioxus would stand about in the same relation to the posterior paired outgrowths of the oesophagus which ulteriorly give rise to the branchial slits of this animal, as would the two primary larval diverticula of Balanoglossus, giving rise to the first pair of gill-slits, to the following ones successively appearing behind them. In Nemertines only one corresponding pair of respiratory diverticula is encountered, and they may remain in connection with those portions of epiblastic ingrowths which form the primary constituents of a sensorial (olfactory?) apparalus in certain of the higher differentiated genera in the way we have above traced.

The far reaching significance of our starting point has obliged us to throw a rapid glance at the principal points in which Nemertines allow of a certain degree of comparison with Vertebrates, and it would lead us too far off if we were to follow this up for the secondary, less important points, or for those which are at present not fully enough known to allow of any fruitful comparison. Among the latter I count the excretory and the generative apparatus. Do the closed generative sacs of Nemertines arise as part of the coelom (cf. Lang, ‘Gunda segmentata) ? What is the morphological significance of the generative ducts which establish a direct communication between these sacs and the exterior, and which are recognisable on the outer surface as a double set of symmetrical pores ? Is the paired nephridium provided with internal openings or is it not ? These and other questions will have to be diligently studied and solved before comparison can extend itself in the domain of these organs.

With respect to the vascular system, it is not unimportant that in Nemertines it is on the whole a closed system of vessels, sometimes carrying corpuscles charged with haemoglobin, sometimes colourless, and giving off a system of transverse connecting vessels, which link together the three longitudinal stems. These transverse vessels do not give off any capillaries, and are metamerically placed with unbroken regularity, one for each internal metamere (intestinal diverticulum). If indeed the suggested homology might prove to hold good between these diverticula and the mesoblastic somites of Amphioxus, the significance of this regular disposition, one for each of the transverse subdivisions of the body, corresponding in a general way to the arrangement of the aortic arches in vertebrate embryos could not be overlooked.

In conclusion I would point out that the speculations and suggestions contained in the last pages ought to be distinguished from the contents of the first part of this article. They have not in any way contributed to the formulating of the hypothesis there brought forward; they are merely the sequel in a train of thoughts which, starting from a comparison of such important and primitive organs in both Vertebrates and Nemertines as are the nervous system, the hypophysis and the notochord necessarily extended itself to other structures and organs occurring in both groups.

With respect to these, we must await more thorough investigations before pushing our speculations further.