of the posterior end of the stomach, and the whole alimentary canal appears like an almost simple narrow tube greatly exceeded in calibre by the oviducts.

The intestine has a well-marked duodenal section or bursa Entiana (b.e) into the left lateral wall of which the stomach opens by a small pylorus (py) guarded by a well-marked annular pyloric valve. The rest of the intestine is somewhat narrower than the stomach, and of tolerably uniform diameter except at its posterior end, where it narrows considerably before entering the cloaca; with the dorsal wall of this posterior portion or rectum (r) is connected the large rectal gland (r. gl).

The stomach is supported by a mesogaster attached along the anterior two-thirds of its dorsal side: the intestine is free save for a mesorectum attached to the rectal gland, and to the dorsal wall of the rectum posterior to that structure.

The spiral valve (sp.v) is the most perfect apparatus of the kind I have yet examined. It belongs to what I have elsewhere^* described as “type C,” that is, the width of the valve is greater than the semi-diameter of the gut, and the plane of any part of it is inclined, from its attachment to the intestinal wall, forwards or towards the duodenal end. There are twenty-seven turns to the valve, the total length of the intestine being inches. The muscular wall of the intestine (w) is greatly thickened, the thickening being often especially well marked between the turns of the spiral valve. Thus the absorbent surface of the mucous membrane is further increased, an additional obstacle is offered to the passage of the intestinal contents, and great muscular power is obtained for their propulsion towards the cloaca. This great development of the intestinal musculature is an exaggeration of what I described, in the paper just referred to, in Scyllium canicula.

The liver (fig. 1, lr) is of immense size, its two lobes reaching quite to the posterior end of the abdominal cavity; it weighed 9 lbs. in the fresh state. There is no gall-bladder; the wide bile-duct (figs. 2 and 3, b. d) passes from the liver in the gastro-hepatic omentum (fig. 1, g. h. o) to the right side of the stomach, and then proceeds directly backwards to open into the anterior wall of the bursa Entiana (fig. 2).

The pancreas (figs. 1 and 2, pn) consists of two lobes: one (fig. 2, pn) closely applied to the ventral surface of the intestine, just beyond the bursa Entiana; the other (pn′) passing backwards and outwards to the left side of the spleen (spl), and surrounding the right mesenteric vein (r.m.v). The spleen (spl) is large, compact, scarcely at all lobulated, and very distensible, swelling to two or three times its orginal size when injected through the arteries.

walls, several small pancreatic veins from the right lobe of the pancreas, and a large anterior splenic vein (a. sp. v) from the spleen, as well as the intra-intestinal vein mentioned below: the mesenteric receiving transverse veins from the intestinal walls, and a large lieno-gastric vein (l.g.v) which is formed mainly by a longitudinal vein from the left side of the stomach and receives also veins from the spleen. After receiving the lieno-gastric, the mesenteric vein runs through the left lobe of the pancreas, receiving veinlets from it, and unites with the duodenal immediately anterior to that gland. The common portal vein passes dorsal to the bursa Entiana and along the right side of the stomach, parallel with the cœliac artery and bile-duct, receiving as it goes the veins for the right side of the stomach.

In the paper already referred to on the spiral valve of the skate, I described that structure as being supplied entirely by the mesenteric arteries and veins, but stated that the bursa Entiana received a special blood-supply in the duodenal vessels. Owing, however, to imperfect injections, I missed one important point. The spiral valve of Elasmobranchs has, in fact, a double blood-supply: vessels from the transverse branches of the mesenteric—and in Scynnus, Mustelus, etc., of the duodenal—arteries and veins pass inward to it, but in addition to these its free edge encloses an artery and vein which may be traced forwards into the duodenal artery and veins respectively. The vein in question has been shown by Balfour^* to be formed from part of the sub-intestinal vein of the embryo. As far as I know no name has been given to it as it occurs in the adult, and as it corresponds to part only of the sub-intestinal vein, and is known to persist only in the spiral valve of Cyclostomata and Elasmobranchs, I propose to call it the intra-intestinal vein, and the artery accompanying it the intra-intestinal artery. It attains its greatest dimensions in those sharks which possess a scroll-valve, such as Zygœna, Carcharias, and Galeocerdo,^† but is almost equally conspicuous, as I have lately found, in Mustelus antarcticus and in Callorhynchus antarcticus, both of which have an ordinary spiral valve, although that of the Holocephali shows transitional characters to the scroll-valve.^‡ In Scymnus, as in the skate, the intra-intestinal vein is quite small and easily missed in injecting.

But the most interesting point in the vascular system of Scymnus is the presence of a large lateral vein, having the same essential relations as the vein I described in the skate.^§ Posteriorly it is connected with the veins from the pelvic fin, from the anterior border of which it passes forwards

In front of the mid-brain comes the greatly elongated thalamencephalon or ‘twixt-brain (the); it is best described as a shallow trough, roofed over only by a small band of nervous matter at its hinder end and for the rest by pia mater; the third venticle or thalamocœle is thus widely open above, and there are no lateral thickenings answering to thalami optici.

In most Elasmobranchs,—indeed, according to Günther, in all,—there exist on the ventral surface of the thalamencephalon just behind the optic chiasma (o.c), paired ovoidal bodies, the lobi inferiores: there is no trace of them in Scymnus; the thalamencephalon is merely produced ventrally into a thin-walled tubular infundibulum (inf), which extends backwards over the ventral surface of the mid-brain and is continued directly into the hollow thin-walled pituitary body (pty). Extending along the middle ventral line of the infundibulum and pituitary body is a flattened one-lobed saccus vasculosus (s.v).

The prosencephalon is very interesting: instead of forming a transversely elongated mass, either solid, as in Raja, or containing small ventricles, as in Scyllium, it consists of a small unpaired hinder portion (prc) continuous with and passing insensibly into the thalamencephalon, and of paired, divergent, pyriform bodies, the cerebral hemisphere (c. h). Similarly, the cavity of the prosencephalon consists of an unpaired posterior portion (prc) which may be conveniently called the prosocœle, and is perfectly continuous with the third ventricle, and of paired lateral ventricles (l. v). The walls of the whole fore-brain are very thin, and there is no constriction between the lateral ventricles and the prosocœle, or between the prosocœle and the third ventricle.

The olfactory lobes (olf) are comparatively short, dilated at their ends, and contain olfactory ventricles (olf. v) continuous with the lateral ventricles.

The brain of Scymnus is thus seen to exemplify with diagrammatic clearness the typical structure of the vertebrate encephalon. We have the large fourth ventricle; the cerebellum retaining its primitive character of a hollow out-pushing of the roof of the fourth ventricle; the mid-ventricle showing no distinction into aqueduct of Sylvius and optic ventricles, and of approximately equal calibre with the third and fourth ventricles; the prosencephalon, or cerebral rudiment of the embryo, composed of an unpaired hinder portion which bifurcates in front to form the paired hemispheres, and these again continued insensibly into the olfactory lobes; the fore-ventricle or cavity of the fore-brain, in the form of a Y-shaped space, the stem of the Y being represented by the prosocœle and third ventricle, the arms by the lateral and olfactory ventricles; finally all the cavities are large, and their walls but little thickened: this is especially noticeable in the case of the

In other characters, however, this embryo had advanced considerably beyond “I,” and was indeed as far advanced as “M.” The mouth (m) is large, and its thickened edge due to the presence of the pterygo-quadrate (pt.q) and Mechelian (mn) bars which afterwards become the upper and lower jaws, is very obvious. There is already the full number of six visceral clefts, of which the first (sp) has completely taken on the character of a spiracle: on the anterior edge of the second (first branchial) cleft (br.1) are minute denticulations, which appear to be the rudiments of the external gills. Lastly the pectoral fins (pc) are well developed, occurring in the form of small outgrowths a short distance behind the last gill-cleft (br.5).

It is thus seen that the mouth, the gill-clefts, and the pectoral fins develope far more rapidly in Scymnus than in either of the genera (Scyllium and Pristiurus) studied by Balfour. A Scymnus embryo of stage “I” has its pectoral fins as far advanced as a Scyllium or Pristiurus of stage “L,” while its gill-clefts are in the condition of those of the same genera in stage “M.” Further observation will be necessary to show whether this is a constant family difference, Scymnus belonging to the Spinacidœ, Scyllium and Pristiurus to the Scyllidœ, or whether the embryo I have just described is abnormal. I have noticed more than once in Mustelus antarcticus one fœtus out of the whole number in a single gravid female in a far more backward stage of development than the rest, and such arrest of development is not unlikely to be accompanied by deformities of some sort.

Second stage.—To this, as to the first stage, only one of the embryos belonged; it was about 18 mm. long, and is shown in fig. 12.

It is, on the whole, intermediate between Balfour's stages “M” and “N,” inclining in most respects to “M.” The head has a remarkably square outline in side view, the cerebral flexure having proceeded just far enough to bring the fore- and mid-brains (f.b, m.b) into the same transverse vertical plane. The eye (e) is large, and the nostril (na) well formed. The distal end of Mechel's cartilage has rotated forwards to such an extent that the axis of the mandible (mn) is nearly vertical. The rudiments of external gills are visible in all the branchial clefts but the last: none have as yet appeared in the spiracle (sp).

All the fins (pc, pv, d 1, d 2) are now formed, and occur in the form of flattened crests, mostly with evenly curved free edges: the caudal fin (c) is perfectly diphycercal. The somatic stalk has undergone great relative reduction.

Third stage.—The remaining eight embryos correspond pretty nearly with stage “O” of Balfour, although in correspondence with the fact that the brain is less advanced, in comparison with other organs, than that of Scyllium, the cerebral flexure and general features of the head correspond

rather with “L.” The mid-brain (fig. 13, m.b) forms the anterior termination of the head, and the ventral surface of the fore-brain (f.b) looks directly backwards.

The mouth (fig. 14, m) is greatly reduced, and has in fact almost precisely the form, relative size, etc., as in Balfour's “O.” So also have the branchial apertures (by. 1, br. 5), from which as well as from the spiracle (sp) the long external gills now emerge. The dorsal (d 1, d 2) and pectoral (pc) fins are beginning to assume their adult form, their line of attachment being no longer their greatest dimension. The caudal fin (c) shows the first indication of the change from diphy- to hetero-cercality: on its ventral edge, near the tip, is a slight emargination, evidently the commencement of the very marked notch in the corresponding position in the adult (fig. 15). In front of the anterior end of the pelvic fin (pv) the lateral ridge is now visible and has the same relations and proportions as in the adult (fig. 15, l.r). The length of the embryo in this stage is about 40 mm.

Up to the present time I have been able to do very little towards the further examination of these embryos, and all I propose to bring forward in the present paper is the fact that in the third stage the lateral vein (figs. 5 and 6, l.v) is well developed, and is indeed nearly as large as the cardinal vein (cd), and considerably larger than the dorsal aorta (d.ao). Of the two sections figured, fig. 5 is taken along the line xy in fig. 13, or just in front of the pelvic fins, fig. 6 along x′y′ or through the pelvic fins. In both figures the lateral vein (l.v) is seen to be a very obvious structure, a fact which makes its absence in Scyllium and Pristiurus—since it is not shown in any of Balfour's figures—somewhat remarkable. Having, unfortunately, no spirit specimen of either of these genera, I have not been able to ascertain whether it exists in the adult: as mentioned above, however, it is present in the allied Chiloscyllium.

A few weeks ago I received a letter from Mr. Balfour,^* in which the following passage occurs:—“I was very much interested in your paper on the skate's venous system. The lateral veins you describe are very peculiar, and I should not hesitate to consider them as confirming my view of the fins, were it not for the specialized character of the skate, which you yourself urge in your paper. One would like to find them either in the embryo or in some less specialized form.”

The necessary confirmation is afforded by the facts detailed in the present paper. The lateral vein exists in every Selachian I have yet had the opportunity of examining: in all it follows the direction of Balfour's lateral ridge, from the anterior border of the pelvic fin forwards and upwards to