PART III: DESCRIPTION OF SPECIAL PARTS OF THE ANATOMY.

A: THE VASCULAR LAMELLÆ.

In Medusæ it is a common thing to find that in certain definite places of the gastro-vascular system two endodermal surfaces that were primarily separated by a space have come together and fused into a single lamella or plate. Such a structure is called indifferently a cathammal plate, an endodermal lamella, or a vascular lamella. In the adult animal the vascular lamellæ are by virtue of their very nature formations “with a past.” They are scaffolding left in the completed structure, giving us clues as to the way in which that structure was brought about; and in the Cubomedusæ, whose development is as yet unknown, they therefore afford an unusually interesting subject for special consideration.

The vascular lamellæ that are found in Charybdea and Tripedalia may for convenience be described as forming two systems, the internal and the marginal. The former comprises the endodermal fusions that separate the stomach from the stomach pockets (except for the spaces of communication left free, the gastric ostia) and those that separate the stomach pockets from one another. The marginal system consists of the lamella that connects _endoderm_ of the gastro-vascular system with _ectoderm_ of the surface in a ring all around the bell margin, and with it also the vascular lamella of the sensory niche, which has already been referred to in the general description of Charybdea. The lamellæ of the internal system have been described by previous writers, and especially by Claus in his paper on Charybdea, but they are still in need of comprehensive and clear treatment. The lamellæ of the margin and of the sensory niche have also been described by Claus, but not thoroughly or with entire accuracy, nor did he recognize the vascular lamellæ of the sensory niche as originally a part of the lamellæ of the margin. This last was first determined by H. V. Wilson upon specimens of Chiropsalmus quadrumanus obtained at Beaufort, North Carolina. Professor Wilson’s unpublished notes on Chiropsalmus were very kindly placed in my hands, and so far as the vascular lamellæ are concerned my own work is only a confirmation and amplification of his, since Charybdea and Tripedalia in this respect agree with Chiropsalmus.

The vascular lamellæ of the internal system are the most prominent and morphologically the most important. They comprise the four vertical strips of fusion that separate the four stomach pockets in the interradii (_ivl_ in the figures of the series of cross-sections of Charybdea and Tripedalia, Nos. 6-15 and 21-29), and four curved horizontal cross-pieces at the top of these which separate the stomach from the stomach pockets, and would make the separation complete did they not leave in each perradius a free space between their ends, which makes possible the gastric ostia.

The arrangement of this internal system of vascular lamellæ is simple. What they amount to is a certain definite number of linear adhesions between the two walls of an originally undivided gastro-vascular space, by which that space is divided up into a central stomach and a peripheral portion, and the peripheral portion thus further divided into the four stomach pockets. Perhaps the idea may be conveyed by likening the whole medusa to a couple of bowls fitting closely one within another and plastered together at the margins. The exumbrella then would correspond to the outer bowl, the subumbrella to the smaller inner bowl, and the original undivided gastro-vascular space to the space between the two. If now the walls of the space be cemented together in four horizontal curved lines just in the plane where the bottoms are bending round to become the sides of the bowls, leaving four interspaces between the ends of the lines, we should have the original space divided into a central horizontal somewhat lens-shaped region between the bottoms of the two bowls that would correspond to the central stomach, and a peripheral vertical portion between the sides of the bowls that would correspond to the peripheral gastro-vascular system; central and peripheral portions would communicate by the four interspaces between the lines of fusion, which would correspond to the four gastric ostia. If, further, the vertical peripheral portion be subdivided by four more lines of fusion running vertically at equal distances apart, each connecting above with the middle point of the corresponding horizontal line of fusion, we should have the simple peripheral portion divided into four parts, corresponding to the stomach pockets, by four vertical lines of fusion, corresponding to the four interradial vascular lamellæ, the _ivl_ of the figures.

These mutual relations of stomach, stomach pockets and lamellæ will perhaps be made clearer if a comparison is drawn between them and the similar structures of a Hydromedusa. Liriope, one of the Trachomedusæ, is a good form to take for such a comparison, since by reason of its direct development from the egg it is free from the complications of hydroid medusæ. The young medusa has at first a simple, undivided gastro-vascular cavity which later is divided up into the central stomach and the typical radial to circular canals of the Hydromedusæ by means of fusions between the two endodermal surfaces. Diagrams _a_, _b_ and _c_ of Fig. 35 represent very schematically this process of division into stomach and canals. In _a_ we have a projection upon a plane surface of the primary, undivided gastro-vascular cavity, as seen from above; _b_ shows the first four points of fusion in the interradii; _c_ represents those four points expanded by growth in all directions into broad cathammal plates in such a way as to leave the stomach in the centre, the radial canals in the perradii, and the circular canal in the periphery as all that remains open of the primary simple cavity. These broad plates of vascular lamella, separating the narrow radial canals, persist in the adult Liriope to tell the tale of the formation of the definitive gastro-vascular system. It seems to me that we are justified by analogy in drawing a similar conclusion for the Cubomedusæ. In _d_ of Fig. 35 is represented a projection of a Cubomedusa, in which the homology of the stomach pockets with the radial canals of the Hydromedusa, and of the narrow strips of fusion with the broad cathammal plates, is shown at a glance. To make the comparison more perfect we have only to remember that in the Cubomedusæ there exists below each interradial vascular lamella a connecting canal (Figs. 16, 29 and 35 _d_, _cc_) uniting the two separate adjacent pockets. This, as has been pointed out by other writers, is the representative of the circular canal of the Hydromedusæ. Practically the only difference between the structure of the gastro-vascular system of the Cubomedusæ and that of a form such as Liriope, is that in the latter the fused areas have broadened out at the expense of the radial canals, while in the Cubomedusæ on the contrary they have become long and narrow.

One is strongly tempted by the foregoing comparison to speculate a little as to whether the reproductive organs of the Cubomedusæ, which lie _in_ the stomach pockets and are generally supposed to be endodermal, may not bear some closer relation to those of the Trachomedusæ, which lie “in the course of” the radial canals (Lang’s Text-book) and by common consent are ectodermal. And while we are being led by facts such as those just mentioned above to wonder just a little whether after all the position of the Cubomedusæ among the Acraspeda is so firmly assured--doubting some, yet in the frame of mind of one who “fears a doubt as wrong”--the velarium suggests itself as another point in question. Haeckel does not hesitate to state emphatically that the velarium of the Cubomedusæ and the velum of the Craspedote medusæ are only analogous, but the reasons that he gives (sie sind unabhängig von einander entstanden, und ihre Structur ist zwar ähnlich, aber keineswegs identisch; namentlich das Verhalten zum Nervenring ist wesentlich verschieden: System, p. 426) somehow do not produce so much impression upon one as the very velum-like appearance of the velarium itself. The origin from the fusion of marginal lobes is not as yet a matter of observation, and the relation to the nerve ring is not essentially different from that of the velum to the lower (_i. e._ inner) nerve ring in the Craspedotæ. The four frenula and the diverticula from the gastro-vascular system seem to be the chief differences in structure after all, and these Haeckel evidently did not think worth mentioning. This speculation, as to the possible relation of the Cubomedusæ to such forms of the veiled medusæ as Liriope, though it may be very tempting, is scarcely fruitful enough to repay much effort on the part of either reader or writer. The whole subject must remain uncertain until the facts of the development of the Cubomedusæ are known.

If the structure of the vascular lamellæ of the internal system has been made clear, the appearances of the vertical and horizontal components in the figures will be understood without much further explanation. The four vertical strips in the interradii (_ivl_) have been already referred to in the figures of the cross-sections of both Charybdea and Tripedalia. In the longitudinal sections of the two jelly-fish through the interradii, the vertical lamellæ are cut throughout their entire length from stomach to connecting canals (Figs. 5-20, _ivl_). The horizontal cross-pieces at the tops of the vertical lamellæ also appear in several of the figures. Fig. 36 represents the appearance that would be given by a longitudinal section taken through any portion of the upper part of the bell except in the interradii, or in the perradii, through the gastric ostia. The horizontal vascular lamella (_hvl_) is shown connecting the endoderm of the stomach (_ens_) with that of the stomach pocket (_enp_). In a longitudinal section directly through an interradius (Fig. 5 or 20) the horizontal lamella is cut just at the point where it joins the vertical, so that the two are not differentiated. In a section through the region of a perradius (Fig. 4 or 19) the horizontal lamella is of course not cut, since the section passes through the gastric ostium, whose existence is conditional upon fusion not having taken place between the endodermal surfaces.

The first figure in each of the series of cross-sections (Figs. 6 and 21) also shows the horizontal vascular lamella, cut across slantingly twice in each quadrant as it passes between the gelatine of the ex- and of the subumbrella to connect the epithelium of the stomach with that of the stomach pocket. The fact that more of the lamella does not appear in such a cross-section only shows that its course is not perfectly horizontal.

The region in which the same lamella lies is indicated in the surface view of the top of the bell of Charybdea (Fig. 2) by the bent line _hvl_ in each quadrant. The figure manifests the appropriateness of Claus’s name for the horizontal lamella--“bogenförmige Verwachsungs-Streifen.” Haeckel calls the same structures “Pylorus-Klappen,” and in his account of Charybdea Murrayana in the Challenger Report, speaking of the three divisions of the stomach (buccal, central and basal) which he traces upwards from the stalked forms of Scyphomedusæ, he says: “The central stomach in this Charybdea, as in most Charybdea, is joined to the basal stomach, as the pyloric stricture between the two is not developed and only faintly indicated by the slightly projecting pyloric valves.” Again, in speaking of the valves of the gastric ostia, he says: “These four perradial ‘pouch valves’ alternate with the interradial pyloric valves.” It is difficult to understand, however, how the “bogenförmige Verwachsungs-Streifen” of Claus, which are undoubtedly the same structures as those which I have called the horizontal lamellæ, and are only strips of endodermal fusion, can be “projecting pyloric valves,” or indeed can properly be spoken of as valves at all. Possibly Haeckel was not quite able to understand Claus’s description, and in his desire to find something in the stomach of Charybdea which would serve to set off a central from a basal part, such as is found in the Lucernaridæ, hit upon Claus’s “Verwachsungs-Streifen.” I have elsewhere given it as my opinion that in such of the Cubomedusæ as I have studied there is no structure in evidence that would properly serve to mark a limit between a basal and a central portion of the stomach.

We have next to describe the marginal system. The vascular lamellæ mentioned above in every case connected endoderm of one cavity with endoderm of another; those of the margin have the noteworthy difference that they run from endoderms of some part of the gastro-vascular system to _ectoderm of the surface_. The outermost cells of the endodermal lamellæ make direct connection with the ectodermal cells, without the usual intervention of a layer of gelatine.

The marginal lamella of Charybdea lies, as the name implies, just on the bell margin where the edge is curving round into the velarium. All around the whole circumference of the bell it is found (in Charybdea) at this same horizontal bend, except in the eight principal radii, where the tentacles and the sensory clubs have brought about modifications. In any place except these a vertical section through the margin will show the marginal lamella connecting the endoderm of the marginal pocket with the ectoderms of the surface, as represented by _vlm_ in Fig. 38, which is a vertical section through the sensory niche a little to one side of the perradial axis.

In the interradii the marginal lamella undergoes modifications due to the fact that the bases of the pedalia are situated a little upwards from the exact margin, and that the lamella follows the outline of the bases. Fig. 1 shows one of the interradial corners of the bell margin looked at directly from the surface, so that the curved outline of the junction of the base of the pedalium with the exumbrella is seen. The trace made by the lamella where it meets the surface ectoderm follows this outline. The lamella is also shown in the vertical section through the interradius (Fig. 5 or 20, _vlm_), where it is seen running from the connecting canals (_cc_), which joins the two adjacent marginal pockets, upwards and outwards to meet the surface ectoderm. Its course from canal to surface is not in a direct line, but curved with the concavity upwards. Hence, in cross-sections at certain levels through the interradial corner it is met more than once and gives rise to appearances that seem at first sight too complicated for it to be just the same structure as the simple marginal lamella described above. That it is the same, and that the complication is only due to the insertion of the pedalia above the margin, can be determined by following through a series of cross-sections, the essential ones of which, as I hope, are given in Figs. 40-43. The levels of these are shown on Fig. 5 by the letters _w_, _x_, _y_ and _z_, respectively. Fig. 40 shows the lamella cut but once, just below its highest part. The section is above the level of the connecting canal and hence still shows the vertical interradial lamella _ivl_. Fig. 41, at the next lower level (_x_), shows the same portion of the lamella intersected a little nearer the interior, while the junction with the endoderm of the connecting canal is shown still further inside. Fig. 42 is at level _y_, just through the bend of the loop, so that in part of its course the lamella is cut almost horizontally, _i. e._ in its own plane. Fig. 43 finally shows the lamella as it appears below the level of the connecting canal, cut twice, each portion joining endoderm of marginal pocket with ectoderm of surface. It thus bears exactly the same relations that it had when we first met it in Fig. 38 (_vlm_), except that here in Fig. 43 one finds that a cross-section cuts it at right angles instead of a vertical as in Fig. 38, as a result of its being pushed upwards from its former position on the margin by the insertion of the pedalium above the margin.

The vascular lamella of the sensory niche has already been alluded to as part of the marginal system, and brief reference has been made to it in the section on the sensory clubs. Like the rest of the marginal lamella, it connects endoderm with ectoderm. The line that its fusion with the ectoderm traces on the surface frames in a shield-shaped area at the bottom of the sensory niche, which is seen in the drawing of the outlines of the niche, Fig. 44 (_vls_). This lamella was observed by Claus, and was figured by him both in surface view and in cross-section through the niche. Apparently, however, he omitted vertical sections through the niche, so that he supposed that the outline traced by the lamella was not continuous above, _i. e._ over the stalk of the sensory club (’78, Fig. 41; text, p. 28). That the outline is closed above, though masked in surface view by the roof of the sensory niche, is seen at once in vertical sections, such as Figs. 37 and 38, one of which is directly through the perradius, the other a little to one side. Both show the vascular lamella of the sensory niche (_vls_) intersected twice, above and below the sensory club, and completely cutting off the exumbrella from any share in the bottom (or inner wall) of the sensory niche. Fig. 39, which is a cross-section through the upper part of the niche, and is essentially like the similar figure of Claus, shows in like manner that the bottom of the sensory niche belongs to the subumbrella. H. V. Wilson was the first to point out, in his unpublished notes, that the lamella of the niche is complete all round.

In the adult structure of Charybdea and Tripedalia the lamella of the niche is connected with that of the margin by a vertical strip of endodermal fusion that does not come to the surface like the rest of the marginal system, but remains just internal to the gelatine of the exumbrella, connecting the two adjacent marginal pockets. In the cross-sections of Charybdea it is seen in Fig. 16 (_vlc_); in those of Tripedalia it is seen in Figs. 28 and 29. In vertical section it is found in Figs. 4, 19 and 37. In Fig. 44, which represents the bell margin and velarium of Tripedalia arranged as if the velarium were vertical and pendant from the margin (instead of suspended by the frenulum so as to be at right angles to the vertical plane), the connecting lamella is shown as a dotted line (_vlc_)--dotted because it does not come to the surface--joining the lamella of the niche with that of the margin (_vlm_).

The same figure (No. 44) shows a characteristic difference between the marginal lamella of Tripedalia and that of Charybdea. While in Charybdea, as Claus points out, the marginal lamella keeps at one level, just a little above the bell margin, all the way round (except where disturbed by the special modifications of the tentacles and the sensory clubs), and never descends into the velarium, in Tripedalia on the other hand it describes a sinuous course, following the outlines of the marginal pockets, as is indicated in the figure by the light parallel line _vlm_. The course as it would be seen in a surface view is obscured just at each side of the interradius by the overhanging of the bases of the two lateral pedalia. This is why the lamella is not indicated at these points in the diagram. The course is seen to lie almost wholly on the velarium, that is, in the figure below the line which represents the bell margin proper, the line at which the angle comes when the velarium is in its normal position, horizontal to the vertical side of the bell.

In this sinuous course of the marginal lamella we have another point of resemblance between Tripedalia and the Chirodropidæ. H. V. Wilson worked it out in his sections of Chiropsalmus, and the reconstruction which I have given in the figure under discussion is in all essentials similar to his for Chiropsalmus. The differences lie only in the fact that Chiropsalmus has more velar canals, and that the chief marginal pocket in each quadrant is not forked peripherally, as is that of Tripedalia (_mp_), but presents its distal margin parallel to the edge of the velarium. The two smaller marginal pockets in the perradii (_mp´_) are on identically the same plan in both.

Tripedalia, having three tentacles joining the umbrella in each interradius, shows a disturbance of the course of the marginal lamella in these regions by just so much the more complicated than in Charybdea. The plan, however, is exactly the same. The lamella is pushed upwards from the margin by each of the bases of the three pedalia just as is done by the base of the single pedalium of Charybdea. Fig. 29 shows the lamella in the same relation to the canal of the central tentacle (_ct_) that it has in the similar sections of Charybdea (Figs. 16 and 43); and in addition the first appearances (as the series is traced downwards) of the arches of the lamella over the two lateral tentacles (_ct´_), which are inserted a little lower down than the middle one of the group. As concerns these lateral tentacles, the relations of the vascular lamella at this level are the same as that in the level of Fig. 40 for Charybdea.

It has been stated more than once already that the vascular lamella of the sensory niche is a part of the lamella that runs round the margin, and so far the only evidence given has been the strip of endodermal fusion running from the marginal lamella to that of the niche. This strip, however, as has been described, does not come to the surface and consequently seems at first sight to be a different structure from the lamella of the margin. That it is not, however, I found very prettily shown in a series of sections of one of my youngest Tripedalia. In this the lamella of the niche as it was traced in successive sections downwards, was found not to form a closed ring at the bottom of the niche, but each side was continued directly and separately downwards to the margin, where it passed into the corresponding part of the marginal lamella. A reconstruction of the condition, similar to that of Fig. 44, is given in Fig. 45, and I think explains itself at a glance. Evidently the vascular lamellæ that connect the lamella of the sensory niche with that of the margin at first come to the surface, like the rest of the marginal system, but as the animal grows older come to lie within the gelatine. In this way the condition found in cross-sections just through the margin of my very small Tripedalia, and represented in Fig. 46, becomes that of the adult seen in the corresponding portion of Fig. 29. It is as complete a demonstration as could be required that the lamella of the sensory niche is at first only a loop of the marginal lamella, a conclusion that had been already reached by H. V. Wilson on theoretical considerations, based upon the facts of the adult structure as he found them in Chiropsalmus.

As Wilson pointed out in his notes, these facts have a close bearing upon the question of the origin of the velarium. Sixteen marginal pockets are found in both Chiropsalmus and Tripedalia, and all of them extend into the velarium. It is not unnatural to suppose that these belong to sixteen marginal lobes, and that these lobes have fused together to form the velarium. In the Chirodropus figured by Haeckel (Taf. XXVI) in his “System” gelatinous lobe-like thickenings are shown in the velarium, corresponding to the sixteen marginal pockets. In Tripedalia no special gelatinous thickenings are found, but the arrangement of the marginal pockets is the same as that of the Chirodropidæ, and perhaps I ought, when treating of the systematic relations of Tripedalia (p. 5, Fam. III), to have recognized the analogy to the extent of saying that marginal lobes may not be completely absent from the velarium of Tripedalia. At any rate the gelatinous lobes in the case of Chirodropus on the one hand, and on the other hand the sinuous outline of the margin still mapped out by the lamella in Chirodropus, Chiropsalmus and Tripedalia, are certainly very suggestive of an ancestral Cubomedusa in which there was no velarium, but sixteen free marginal lobes instead. Two more indications favor slightly the same view. In both Charybdea and Tripedalia a small notch is seen in the edge of the velarium in the perradius (Fig. 44). Its constancy suggests that it may not be a chance or meaningless feature. The second point is the small size of the two marginal pockets adjoining the perradius. These are in the position of the ephyra lobes of the Discomedusæ, which always lie on either side of each sensory club, and which do not keep pace with the other marginal lobes in development. In the Rhizostome jelly-fish especially they are found much smaller than the other lobes, as will be seen by a glance at such figures as Haeckel’s for Lychnorhiza (System, Taf. XXXIV Fig. 2), or for Archirhiza (Taf. XXXVI, Fig. 5), or Hesse’s figure of the margin of Rhizostoma Cuvieri (’95, Taf. XXII, Fig. 22). The resemblance between such margins and that of Tripedalia (Fig. 44), with its simple, unbranched velar canals, is very suggestive. On the other hand it must be remembered that in considering the vascular lamellæ of the internal system we found the indication pointing rather more to Hydromedusan affinities than to any other. Charybdea throws no light on the question, since it has no marginal lobes on the velarium and the marginal pockets end strictly at the margin, so that the only diverticula of the gastro-vascular system in the velarium are the velar canals.

Before leaving the subject of marginal lobes and pockets I must answer a possible objection that may occur to some careful reader. It may seem that I am wrong in holding that there are two marginal pockets in each octant instead of three, that just as there is one velar canal from each of the smaller perradial pockets (_mp´_, Fig. 44), so each prong of the forked larger pocket (_mp_), since it is continued into a velar canal, ought to be called a marginal pocket likewise, the whole number of marginal pockets then being twenty-four instead of sixteen. Such a revision of the terminology would not be without some reason in its favor, and perhaps a study of more forms would show it to be correct. But for the present, at any rate, it seemed to me best to abide by the analogy of Chiropsalmus, in which the peripheral edge of the larger marginal pocket in each octant is not bow-shaped, but runs parallel to the edge of the velarium. A revision of the terminology of the marginal pockets such as implied in the suggestion above would also give rise to complications when applied to Charybdea, since the latter has no marginal pockets in the velarium.

As to the functions of the vascular lamellæ, there is too little known to say much. It is rather improbable that structures retained so definitely should be mere scaffolding left over from a previous stage of usefulness. Claus has found in Chrysaora that the lamellæ form a kind of capillary network in communication with the gastro-vascular system, and he with others supports the view that they perform an accessory function in the nutrition of the tissues they penetrate. Upon this point I have no observations of my own to add.

The marginal vascular lamella is regarded by Claus as perhaps the vestige of a circular canal around the bell margin. On this subject, too, I have nothing to add. A lamella of endoderm that connects directly with the ectoderm of the surface along its whole course is a structure whose meaning I am wholly unable to understand or even to guess at. A similar lamella is described by Hesse (’95, p. 430) as occurring in the ephyra lobes of his Rhizostoma, and he mentions Eimer as the first to discover this structure, probably meaning the first to discover it in the Discomedusæ. Whether the lamella is found all around the margin is not stated. Hesse refers it to the ephyra, and remarks that the investigation of it in the ephyra would undoubtedly give interesting results.

I will close this part upon the vascular lamellæ with a very pertinent suggestion made by Professor Brooks to the effect that the usual way of speaking of the sensory clubs as having moved up from the margin is looking at the matter in the wrong way. The level of the sensory clubs undoubtedly represents the original margin, which elsewhere has grown down and away from its former level, leaving the sensory clubs like floatage stranded at high-tide mark. Only in this way can the lamella of the sensory niche have any meaning.

B: THE NERVOUS SYSTEM.

The nervous system of the Cubomedusæ is the most highly developed that is found in any of the jelly-fishes. If the position of the group among the Acraspeda is established, it alone is ample to prove that the Hertwigs had not sufficient evidence when they stated in their monograph on the nervous system of the Medusæ (’78) that the Acraspeda show a much lower nervous organization than the Craspedota.

The system naturally groups itself under three heads, the nerve ring, the sensory clubs, and the motor plexus of fibres and ganglia that underlies the epithelium of the subumbrella. The general relations of the nerve ring and of the sensory clubs have been given before in the description of Charybdea Xaymacana, so that we may pass at once to the consideration of the finer details of the nervous tissues.

In the structure of the nerve ring I have found myself unable to come to the same results as those given by Claus, who so far as I know is the only one that has studied the nerve with special reference to its histology. Our difference amounts to this, that he finds two distinct types of cells in the epithelium of the nerve, sensory and supporting, which would make it a receiving as well as transmitting organ, while I have not been able to demonstrate satisfactorily the sensory cells, and, therefore, so far as my own observation is concerned, I am disposed to attribute to the nerve simply the function of conducting impulses. I do not know just how much weight to assign to my inability to find evidence in my sections of the sensory type of cells. Eimer (mentioned by Hesse, ’95, p. 420), the Hertwigs (’78) and Claus (’78) have independently discovered the two types in one medusa or another, and the Hertwigs, at least, have demonstrated them by macerated preparations. So far as Charybdea is concerned, however, Claus had only preserved material and had to rely upon sections, as have I, since the material which I had preserved with especial reference to maceration did not turn out well. The results that we get from sections vary enough for me to believe that Claus interpreted his sections very much by analogy with other forms--as indeed, is suggested by his own words (’78, p. 22): “Da es mir nicht geglückt ist die durch die längere Conservirung in Weingeist fest vereinigten Elemente zu isoliren, habe ich das muthmassliche Verhältniss beider Elemente nach Analogie der mir für die Acalephen bekannt gewordenen Verhältnisse, welche O. und R. Hertwig so schön auch am Nervenring der Carmarina zur Darstellung gebracht haben, zu ergänzen versucht.” There can be no doubt of our having the same structures to deal with, for C. Xaymacana is so much like C. marsupialis as to be perhaps more worthy of being called a variety of the latter than a distinct species.

The structure of the nerve as I conceive it is given in Figs. 47 and 48. The former represents a cross-section, and shows, as others have pointed out, that the layer of circular muscle fibres (_cm_) is interrupted by the nerve. It is evident that the tissues which elsewhere on the subumbrella were differentiated into muscle epithelium and muscle fibre have here become nerve epithelium and nerve fibre, a point that has not been remarked upon before, so far as I remember, and that may be of interest in connection with the neuro-muscular theory. The epithelium of the nerve (_scn_) is seen to be made up of cells whose inner ends narrow down into a kind of stalk or process that runs to the gelatine of the supporting lamella (_gs_) and there joins a little cone of the gelatine that juts out to meet it. The cells are smaller in general than those that overlie the muscle layer, especially on the two lateral margins of the nerve, where they are more crowded together and overarch the nerve-fibres. The fibres are seen in cross-section between the processes of the cells. They apparently must lie imbedded in some clear, watery fluid that does not show in the preserved material. The processes of the epithelial cells give the fibres the appearance of lying in alveoli, or being divided into strands, and one of these strands (_ax_) is always discernible among the others by reason of its more numerous or finer or more compactly massed fibres. This is the “axis” of Claus. Here and there in its course appear ganglion cells having their long axis in the longitudinal direction of the nerve. Elsewhere, in the nerve as well, and usually nearer to the surface, are found other ganglion cells, mostly bipolar, some multipolar, which are readily distinguishable from those of the axis by the fact that their long axis lies across the nerve. One of these cells is shown in the figure (_gc_). Here and there in the epithelium alongside the nerve are found mucous cells (_mc_), distinguished by their clear contents and by the small exhausted-appearing nucleus at the base with a few threads of protoplasm.

In Fig. 48 I have tried to represent the structure of the nerve by means of a series of five different views such as would be given by focusing at five successive levels. In the first (1) we have the epithelium of the nerve (_scn_ in Fig. 47) in surface view, the cells appearing polygonal in outline, with here and there a mucous cell. In (2) we find a very slight layer of ganglion cells and fibres having a transverse direction (_gc_ and _fp_ in Fig. 47). These are continuous with the plexus of fibres and ganglion cells which lie above the muscle layer all over the subumbrella, and which represent the motor part of the nervous system. This connection with the nerve shows how co-ordination is effected. At the same level are found fibres of the axis also having a longitudinal direction. In (3) is seen the main body of fibres, divided in the osmic preparation from which the drawing was made into irregular wavy strands which are in all probability largely the result of preservation, but are in part also due to the separation by processes of the epithelial cells, as was seen in Fig. 47. The axis is seen with one of its longitudinally directed bipolar ganglion cells; and at the sides the fibres of the circular muscle of the subumbrella. These show a slanting direction to the nerve, due to the fact that the nerve, as mentioned before, has a sinuous course from the margin in interradius to the level of sensory club in perradius. At the next focus (4) we come to the gelatine of the subumbrella (_gs_ in Fig. 47), and below this (5) to the larger polygonal outlines of the endodermal cells of the stomach pocket (_enp_, Fig. 47), which like the ectoderm show mucous cells at irregular intervals.

A comparison, now, with Claus’s figures (’78, Taf. II, Figs. 19-21) will show that, except for the rather unimportant matter of the mucous cells, which he finds regularly and thickly disposed on each side of the nerve (’78, Fig. 21), our only essential difference lies in the matter of sensory cells in the epithelium. His figures show a multitude of spindle-shaped sensory cells whose central ends are continued in processes that bend around into the mass of fibres of the nerve. In his Fig. 20 a relatively small number of nuclei, just one-third as many, are seen attached nearer to the surface, which represent the supporting cells. The plan of structure (as shown in his Fig. 20) is an alternation of (1) supporting cells offering a broad peripheral end to the surface and having the central end continued as a supporting fibre to the gelatinous lamella, and (2) spindle-shaped sensory cells with nuclei at a lower level, which send their peripheral process up between the supporting cells to the surface, while the central process becomes continuous with the nerve fibres, often branching into two processes. In my sections I have not been able to see either a regular alternation of nuclei at different levels, or central processes which unmistakably bend round into the nerve fibres. In every case in which I could trace the central process of a cell clearly it ran to the supporting lamella, and this whether the nucleus of the cell lay near the surface of the nerve or deeper down, as in the somewhat spindle-shaped cell seen on the left of the centre of the nerve in Fig. 47. Of course in many cases the central process could not be traced in a section, and this leaves room for the supposition that such were always the sensory cells. From my inability to demonstrate sensory cells in the nerves of Charybdea, I by no means wish to deny their existence; for that remains to be proved, or disproved, by macerations. At any rate, they cannot be so numerous as has been supposed. The position of the nuclei shows that.

The epithelium of the nerve is said by Claus to be ciliated. It has been suggested by Schewiakoff that probably in such cases the sensory cells bear one long cilium, while the supporting cells have many smaller cilia. Unfortunately, I made no observations upon the ciliation of the nervous structures of the living animal, and the traces of cilia that are shown in preparations of preserved material are a poor basis to speculate much on. Claus considers the sensory cells of the epithelium of the nerve a special seat of tactile sensation.

The way in which the nerve reaches the sensory clubs is interesting. Under the topic of the vascular lamellæ it was explained that the sensory clubs and the bottom of the sensory niche from which they spring are parts of the subumbrella. Fig. 37 reminds at a glance better than any other one drawing how the bottom or inner wall of the niche is completely cut off from the exumbrella by vascular lamellæ above and below the stalk of the club. From this figure, now, it will readily be understood that the nerve in order to pass to the base of the stalk has simply to traverse the gelatine of the subumbrella. This fact, which seems surprising enough at first sight in view of the position of the clubs on the external surface of the umbrella, was correctly pointed out and explained by Claus, but one or two figures will serve perhaps to give a clearer idea of it.

Fig. 49 is a diagram of the nervous structures in the region of the sensory niche, as they would be seen on the surface of the subumbrella turned toward the bell cavity. The outline of the sensory niche as it is seen through the tissue of the animal is represented by the line _osn_. The sensory club (_scl_), and its stalk with a conical basal portion are given by the lightly dotted outline and are also imagined as seen through the animal. The nerve (_n_), being on the surface of the subumbrella, is shown as a heavy line describing an arch over the outline of the niche. In the middle point of the arch is a slight thickening of the nervous tissue (_rg_) which shows in section a large increase in the number of ganglion cells, and is the radial ganglion of Claus. The same is seen, exaggerated in size, in Fig. 12. From it there extends upward a slender strand of nervous tissue (_rn_), the radial nerve of Claus. In Charybdea this can be traced but a very short distance. In Tripedalia it is much more distinct and traceable for a longer distance, and I might say in passing that this and the sensory organs in the proboscis are the only differences I have noted between the nervous systems of Tripedalia and Charybdea.

Nerve ring, radial ganglion and radial nerve all lie on the bell cavity surface of the subumbrella. The way, now, in which the nerve ring reaches the base of the stalk is simply by sending two roots through the gelatine of the subumbrella to the conical base of the stalk. These roots are seen in the diagram at _rns_. After passing through the gelatine the roots come together on the inner side of the base--that is, the side turned toward the bell cavity--and then pass downwards (_nst_) on the inner side of the stalk of the club to the mass of nervous tissue at its end.

This passage of nervous tissue through the gelatine in order to reach the sensory club is a little hard to grasp at the first, and I have tried to render it more intelligible by a couple of drawings of sections. Fig. 50 is a transverse section through the upper part of the region of the sensory niche, not quite horizontal (_i. e._ parallel with the bell margin), but slanting so as to lie on the plane of the reference arrow _x-y_ in Fig. 49. The plane passes just through the top of the niche, and in two areas has cut through the roof with its epithelium of ectoderm (_ece_, _ecs_) so that the space of the sensory niche (_sn_) appears. The vascular lamella of the sensory niche (_vls_) is shown, as in Figs. 13 and 14, running on each side from the endoderm that lines the canal of the sensory club (_enc_) to the endoderm of the adjacent stomach pocket (_enp_). By it the gelatine of the exumbrella is separated from that of the subumbrella, and one sees that it is only through the latter that the nerve has to pass in order to reach the base of the sensory club. It is also seen that one part of the roof of the niche which is cut through lies outside of the ring of lamella and is therefore lined with ectoderm of the exumbrella (_ece_) while the other lies within the ring and is lined with ectoderm of the subumbrella (_ecs_). Owing to the slanting direction of the cut only the root on one side is cut through. The other is indicated, however, on the right side of the drawing. In this method of passage of nerve fibres, together with the accompanying ganglion cells, directly through the gelatine to the stalk of the sensory club my work is only confirmation and explanation of Claus.

Fig. 51 is a vertical section through the base of the stalk in the plane of the reference arrow _w-z_ in Fig. 49, and therefore passing through one of the roots of the nerve of the stalk. Here again the region is seen to be cut off from the exumbrella by the vascular lamella of the sensory niche (_vls_), and the nerve is seen passing through the gelatine of the subumbrella from the surface of the bell cavity (_sc_) to the base of the stalk hanging in the sensory niche (_sn_). One of the ganglion cells (_gc_) that accompany the nerve is seen to have two nuclei, a not infrequent occurrence which has been pointed out by others.

The same figure shows that the axis (_ax_) of the nerve has penetrated the gelatine with the other fibres. Here at the base of the stalk it takes a horizontal course and becomes directly continuous with the similar structure of the other root, as Wilson, I believe, first pointed out. This part of the nervous tract which runs horizontally along the base of the stalk between the two roots (Fig. 49, _rns_) has been considered by Claus the representative in Charybdea of the upper nerve ring of the Craspedota, which therefore exists in Charybdea in four separate portions. Seeing, however, that the region in which it is found belongs to the subumbrella, the homology seems very doubtful. Moreover, the fact that the axis of the nerve ring runs through this outer portion, instead of remaining on the inner surface of the subumbrella and passing to the radial ganglion, rather indicates that the outer portion is part of the original course of the nerve ring, while the portion that remains on the inner surface is perhaps a later formation.

A very interesting feature of the nervous system occurs in the same region in the form of a tract of fibres underlying the endoderm, and separated from the other fibres by the gelatine of the supporting lamella. It is seen in vertical section in Fig. 52 (_enf_), which is a section through the base of the stalk in just about its median plane, and, therefore, to one side of the arrow _w-z_ in Fig. 49 and the corresponding drawing, Fig. 51. In cross-section it is represented also in Fig. 50 (_enf_). It varies in size and prominence very much in different specimens. Fig. 52 is a camera drawing of it in the case that showed it most developed. Ganglion cells are found in it, but comparatively infrequently. In some cases the tract itself can hardly be found with certainty. Hesse has described in a Rhizostome a much more highly developed tract in a corresponding position on the base of the marginal body. Fibres from the “outer sensory pit” pass through the gelatine to the sub-endodermal tract, which is described as surrounding the epithelium of the canal of the marginal body like a collar and is most thickly developed on the under surface of the canal, at the place that just corresponds with the point where, and where only, I find the tract in Charybdea. Hesse thinks that fibres then pass from this region to the nervous epithelium of the “inner sensory pit” lying underneath the base of the marginal body, which contains a rich supply of ganglion cells and is considered by him to be the centre of the nervous system of the medusa. A close comparison cannot be drawn with Charybdea in this matter, however, since Charybdea has nothing to correspond with the “outer” and “inner” sensory pits. Moreover, the endodermal tract is not found encircling the canal of the sensory club, nor could I trace fibres passing from it through the supporting lamella into the fibres of the nerves.

Claus has figured (’78, Taf. V, Fig. 45, _Fb_) a small bundle of fibres in the stock of the sensory club lying between the endoderm cells of the canal and the supporting lamella. The same bundle is found in both Charybdea and Tripedalia and can be traced in cross-sections up the stalk to a point which must correspond with that at which the endodermal tract is seen in Fig. 52. Downwards it can be traced only as far as the entrance of the stalk into the knob of the club where it invariably becomes lost to view. According to Hesse (’95, p. 427) Schäfer found under the endoderm cells of the whole stalk of the marginal body a fibrous layer like that under the endoderm cells which he refers to slender processes from the cells of the crystalline sac. Although Hesse, as we have seen, finds the layer more limited in extent than Schäfer gives it, and does not trace it to the same source, the observation of Schäfer seems to me worthy of mention here, inasmuch as the trend of the fibrous bundle under the endoderm cells of the stalk in Charybdea and Tripedalia suggests quite strongly that the fibres come from the crystalline sac, as Schäfer thought to be the case in his medusa.

Besides the radial ganglion situated in the course of the nerve ring at its four perradial points there are four other similar ganglia on the subumbrella. These lie in the interradii, at the four lowermost points of the nerve’s course, and undoubtedly send off nerves into the pedalia at whose bases they are situated. F. Müller (’59), whose work was not accessible to me, is quoted by Claus as recording two ganglia opposite the base of each pedalium which gave off a great number of nerves partly into the velarium, partly into the tentacles. Claus observed nothing of the kind in Charybdea and states that even the interradial ganglia do not exist.

That they do, however, is shown without doubt in sections of both C. Xaymacana and Tripedalia, but nerves to the velarium or to the tentacles I was unable to find.

On the two sides of each frenulum and of each suspensorium are found subepithelial ganglion cells in greater numbers than elsewhere on the subumbrella, and I am inclined to ascribe to them also the importance of special ganglia controlling the musculature of the frenula and suspensoria. Certainly such ganglia would not be out of place.

It has been mentioned that the greater prominence of the radial nerve and the possession of special sensory organs in the proboscis were the only points of difference I had noted between the nervous systems of Charybdea and Tripedalia. These sensory organs remain to be described. They are simple ciliated cysts containing a concretionary mass, and are situated in the gelatine of the proboscis, irregularly disposed of at any level, from the lips to the beginning of the stomach, and in any radius. In one series of the adult animal fifteen were counted, of which seven were situated about interradially, four perradially, two adradially and two subradially. In another, twenty-one were counted, twelve in the perradii and nine situated between the sub-and perradii. The one shown in Fig. 24 is in the perradial position, often seen. In the sections of the very young Tripedalia in which the vascular lamella had not reached the adult condition the sensory organs of the proboscis were not found, although the sensory clubs showed practically no difference from the adult. Their structure is very simple--merely a round or oval sac lined with ciliated cells which bear up and keep in constant motion an irregular coarsely granular concretion. Fig. 53 is a sketch made in Jamaica from the living specimen. Sections were somewhat disappointing in that they added but little. Fig. 55 was drawn to show that now and then a mucous cell (_mc_) is found among the other cells of the sensory epithelium. An irregular-shaped mass (_rc_) was always found inside the cysts as the organic remains of the concretion. It gave no trace of cellular structure and offered no evidence whether the concretion was the product of one or few or of all the cells of the cyst. The latter would be unique among the medusæ. Even if the otocyst is the result of the activity of only one or a few cells, it is, so far as I know, the only case known for the jelly-fish of a free, unsuspended concretion.

As to whether the cysts are of ectodermal or endodermal origin could not be determined, but there was some evidence in favor of the latter. Fig. 56 is a drawing of one seen in optical section in a whole mount of part of a proboscis, and shows a definite connection with the endoderm of the proboscis. This was the only case when such connection was satisfactorily established, but in sections it was not uncommon to find what seemed to be the remains of the broken stalk, as in Fig. 54 (_rs?_). No connection could be traced between the cysts and any other part of the nervous system. As to function, the idea that they serve to give perception of space relations suggests itself as readily as any other hypothesis.

We come now to the consideration of the terminal knob of the clubs, the sensory portion proper. A complete and detailed account of the complex structure of these organs would fill many pages and involve much useless repetition. Claus (’78) has described them with accuracy, but not in great detail, and since then Schewiakoff (’89) has given a careful general description and has supplemented Claus’s work by observations upon the finer structure made with the aid of more recent technique. It seems in place for me, therefore, to give in the briefest possible way a general idea of their structure, and to pass then at once to the points in which my work has led me to different conclusions from those of Claus and Schewiakoff. In brief, then, the knob of the sensory club consists of a thick, complex mass of nerve fibres, more or less imbedded in which lie the special sensory organs, surrounding the ampulla-like terminal enlargement of the canal. The surface between the special organs is covered with less specialized sensory epithelium. The sensory organs are seven in number. Of these, four are simple invaginations of the surface epithelium arranged in two pairs symmetrically to the median line in the proximal end of the knob (the end where the stalk enters) and having pigment developed in the cells so invaginated, while the space of the invagination is filled with a gelatinous refracting secretion. These are considered simple eyes. Two more of the organs are complex eyes situated on the median line of the inner surface of the knob, the upper one smaller than the lower, but having almost exactly the same structure. Each has a cellular lens over which extends a superficial, corneal layer of cells; below the lens a refractive “vitreous body”; and below this a retina with pigmented cells. The seventh organ is the crystalline sac, which lies almost at the end of the knob opposite to the stalk and contains a large concretion. In view of the fact that the sensory clubs _in toto_ have been abundantly figured by Claus and Schewiakoff, it is my intention to give but one simple figure of the general relations, and I justify that one in that it was made from the fresh material. Fig. 57 is a camera sketch of the outlines given by a sensory club seen in optical section from the side. The smaller upper and the larger lower complex eyes which are situated on the mid-line, are seen in profile, while the two small simple eyes give the outlines that they would in a surface view of their side of the knob. Of course it is understood that two similar ones would appear on the other side, since the four simple eyes are symmetrically paired on either side of the mid-line. The sketch seems to show at least this much, that even in the living state the lens of the larger eye projects out beyond the other contours of the surface, so that the marked convexity ascribed to it in descriptions is not to be attributed to the preservation.

It is in reference to the structure of the retina and vitreous body of the complex eyes that I have found myself unable to come to the same conclusions as Claus and Schewiakoff. Since the work of the latter goes much further into the detail of the subject than does Claus’s paper, it will be sufficient for me to compare my results simply with those of Schewiakoff.

The latter finds that the retina is composed of two kinds of cells, corresponding to the supporting and sensory cells referred to in the description of the nerve ring. These he figures (’89, Taf. II, Figs. 12 and 13) as alternating regularly. The two kinds of cells differ as follows:

(1) Shape. The supporting cells like those referred to before, are cone-shaped, having a proximal fibrous process that runs into the underlying stratum of nerve fibres, and on the surface of the retina a broad distal pigmented termination. The sensory cells are spindle-shaped, the proximal processes becoming continuous with fibres of the underlying nervous mass, while the distal process runs up to the surface of the retina (the part toward the lens) in between the ends of the supporting cell. The two kinds of cells are accordingly designated as pigment and visual.

(2) Position of nucleus. This comes in as a corollary of the shape. The nuclei of the visual cells lie in the enlarged central part of the spindle-shape, and, therefore, at a lower level than the nuclei of the alternating pigment cells.

(3) Processes in the vitreous body. The distal processes of the spindle-shaped visual cells are continued through the vitreous body to the cells of the lens as rod-like visual fibres which lie in canals in the (supposedly) homogeneous vitreous body. The pigment cells on the other hand have no fibres passing from them through the vitreous body, but in the latter are situated cone-shaped masses of pigment whose bases rest upon the broad ends of the pigment cells without, however, being a part of the cell.

(4) Pigment. The distal ends of the pigment cells in the retina are strongly pigmented, as the name implies. The processes of the visual cells, which alternate with these, are pigmented likewise, but the pigment is not so abundant and lies in the periphery of the cell body, leaving free a highly refracting central axis.

If the relation of these cells to each other has been made sufficiently clear, it will be understood that, in accordance with Schewiakoff’s scheme of the structure, sections that cut the retinal cells transversely give very different appearances at different levels. A section through the very tops of the retinal cells, that is, the last section of the retina before striking the vitreous body, would show large polygonal areas of heavy pigment (the ends of the pigment cells), in between which would lie the much smaller, less pigmented, highly refracting ends of the visual cells (’89, Taf. II, Fig. 19). A section lower down in the retina, that is, more toward the centre of the club, would strike the low-lying enlarged central portion of the visual cells with their contained nuclei, and the smaller, proximal ends of the pigment cells. It would, therefore, give the reverse appearance from the preceding section, namely, that of large unpigmented (or but slightly pigmented) areas (the swollen bodies and nuclei of the spindle-shaped cells), and in between them smaller pigmented areas, the ends of the proximally tapering pigment cells (’89, Taf. II, Fig. 20). A section on the other side of the one first described, that is, one of the first through the vitreous body, would show pigment areas of the same size as the large ends of the pigment cells (the cone-shaped streaks of pigment in the vitreous body which according to Schewiakoff are associated with the pigment cell), and in between them the cross-sections of the rod-like processes from the visual cells, lying in canals in the clear homogeneous ground-substance of the vitreous body (’89, Taf. II, Fig. 18).

Let me give a resumé of Schewiakoff’s conception of the structure of the retina.

a. There is an alternation of pigment and visual cells, the nuclei of the spindle-shaped visual cells lying at a lower level than those of the cone-shaped pigment cells.

b. From the visual cells extend rod-like processes into the vitreous body, lying in canals in the latter.