bones in the anal." In the Palæoniscid fish I am describing there is pretty conclusive evidence that there is only one set, unless the other be extremely rudimentary.

Appendicular Skeleton.—In the type specimen the shoulder girdle is only represented by the upper part of the supraclavicle and the post-temporal, and these are very badly preserved. In the second specimen much of the shoulder girdle is seen in good condition. The supraclavicle is of large size, and the lateral line passes over the upper part of it. The postclavicle is small, but distinct. The clavicle is well developed. The rays of the pectoral fin are distinctly articulated distally, but it is not clear whether the proximal halves are also articulated.

The pelvic fin is seen in the type specimen, but it is very imperfectly preserved.

Unpaired Fins.-The dorsal fin begins a little in front of and ends opposite the middle of the anal fin. It is large and triangular, and is formed by about 40 rays, of which the 8th is the longest and the first 3 very short. They are articulated throughout their whole. length. Most of the rays divide near their middle and again at the beginning of their distal fourth. A series of small fulcra support the front of the fin, and one or two enlarged scales lie in front of the anterior end.

The anal fin has a longer attachment than the dorsal, and the rays are rather longer. They are about 52 in number, and the 8th and 9th are the longest. Like those of the dorsal fin, they are articulated and branch dichotomously.

The caudal fin is somewhat imperfectly preserved. It is probably fairly deeply bilobate, but neither the upper nor the lower lobe is complete, and the two portions are artificially separated from each other. The lower lobe is well developed, the rays being supported by large hæmal spines. The upper caudal lobe is well developed.

Squamation.-The scales of the body are small and rather thin. Those of the anterior part, which are better developed, are rhombohedral, and ornamented with 8 or 9 little irregular ridges which pass backwards and slightly downwards. The scales of the abdominal region are very narrow and closely imbricated. Those of the posterior part of the body are thinner and smaller. On the upper lobe of the tail the scales are again thicker and form long, narrow rhomboids ornamented by a few oblong ridges passing backwards.

A few enlarged scales lie in front of the dorsal fin, but apparently

there are none before the anal or caudal. Above the tail are, however, a row of very large fulcral scales.

While this large Palæoniscid seems different from any previously described form, it is a little difficult to be sure in what genus it ought to be placed. It resembles to some extent the New South Wales genus Apateolepis and to a less extent the genus Myriolepis, but it seems to agree so much more closely with the English Liassic genus Oxygnathus, that I think it better to place it, at least provisionally, in that genus. The only points in which the South African fish does not agree with Smith-Woodward's definition of the genus Oxygnathus are in having the ridge scales of the upper caudal lobe large and in the body scales being rather thin.

DICTYOPYGE FORMOSA, n. sp.

This new species is represented by a single specimen in the collection obtained by Dr. Kannemeyer. It is a small species, measuring only about 70 mm. in length, and the head 14 mm. The body has a greatest depth of 13 mm.

The head is not sufficiently well preserved to enable one to say much about its structure. The orbit is placed well forward, and both the maxilla and the dentary are long, and each has a row of small uniform teeth.

The

The pectoral fin is composed of about 11 rays, of which the 4th is longest and the first 3 short. There are well-developed fulcra in front. The fin rays are not manifestly branched, and it is doubtful if they are articulated. The pelvic fin is not well preserved. The dorsal fin is only very slightly in front of the anal. It is composed of 24 rays, of which the 8th is longest. There are no fulcra. anal fin is composed of about 36 rays, of which the 8th is the longest. A few fulcral rays are present in front of the distal part of the fin. The caudal fin is slightly imperfect, but enough is preserved to show that it is a variety of the abbreviate heterocercal. The axial portion of the upper lobe is longer than in typical Dictyopyge, but distinctly not Palæoniscid.

The scales are well developed and rhomboidal. They are characterised by the presence of two prominent ridges, which are directed backwards. The dorsal scales are perhaps slightly enlarged, and two large scales lie in front of the dorsal and at least one in front of the anal.

SEMIONOTUS CAPENSIS, Smith-Woodward.

In 1888 Smith-Woodward (9) described, under the above name, some fossil fishes from the Stormberg beds of the Drakensberg Range.

Though the specimens were in fair preservation, so many very much finer specimens have since then been discovered that it is now possible to add a good deal to Smith-Woodward's description, and also to modify one or two details.

The finest specimens are believed to have come from near Ficksburg in the Orange River Colony, and good examples are to be found in most of the South African museums. The finest I have

seen are those in the Cape Town collection.

The majority of specimens measure from 160 to 210 mm. in length. In the example which is 210 mm. long, the body is 42 mm. in depth at the deepest part, and the head measures 48 mm. to the back of the operculum.

E. Schellwien (10) has recently described a number of specimens from the Orange River Colony, and has shown the more important features of the skull structure. The specimens I have examined confirm most of his observations, but in one or two points I am inclined to differ from him. Schell wien's specimens show one or two features not seen by me in those I have examined, while I am able to add a number of characters not observed by him.

Almost every detail of the skull is now known except the basicranial region. The frontals are large, and extend from the nasal region to behind the plane passing through the back of the orbit. The back part of the bone is about twice as wide as the middle portion. Behind it is a large oblong parietal. Below the parietal is a slightly narrower squamosal. My specimens do not satisfactorily show the supratemporal region, but Schellwien finds a narrow supratemporal and a post-temporal.

The opercular bones are very like those of Lepidotus. The operculum differs in being relatively considerably wider in its lower half. Inferiorly it joins the subopercular in a manner very similar to that in the better known genus. The subopercular in Semionotus is only about one-third the size of the operculum instead of half as large as in Lepidotus, while the interopercular is less than half the size of that in Lepidotus. In front of these three opercular bones is a narrow curved preopercular, along which there runs a mucous canal.

In Schellwien's diagrammatic restoration the postorbital seems to me to be rather too small, while the interopercular is much too large. Above the anterior end of the long preopercular is an elongated suborbital smaller in size than the postorbital. The portion of the figure dealing with this region is, in my opinion, erroneous.

I should be inclined to restore the anterior portion of the skull

somewhat differently from Schell wien, but my material is too unsatisfactory to decide the matter.

The lower jaw has an elongated triangular dentary and a powerful angular.

The palato-pterygo-quadrate arch is fully ossified, but the exact limits of the different element cannot be made out with certainty. There is a long narrow bone below the quadrate stretching from the articular region to the lower end of the hyomandibular. This would seem to be the symplectic. The hyomandibular is a powerful bone and fairly similar to that of the ordinary Teleosteans. In addition to supporting the opercular bones and the quadrate arch, it supports the hyoid arch. There is a large quadrangular epihyal and an elongated triangular ceratohyal. The interhyal has probably been cartilaginous, as has also probably been the hypohyal and the urohyal. Under the subopercular are six branchiostegals.

The clavicular arch consists of the clavicle, supraclavicle, postclavicle and post-temporal, but there seems to be no trace of an infraclavicle. A mucous canal crosses the supraclavicle obliquely as in the Palæoniscids. There is a small ossification which possibly may he the coracoid as is thought by Schellwien.

The pectoral fin consists of 14 rays with 5 or 6 fulcra in front. The rays are much flattened distally, but apparently not branched. The pelvic fin consists of 7 rays which are branched distally. The fulcra are powerful.

The dorsal fin begins exactly in the middle of the back of the fish and consists of 13 rays, of which the last 3 are very small. All the rays are branched distally and articulated. In front are a row of very powerful fulcra, 9 in number. The anal fin consists of 9 rays with 9 powerful fulcra in front.

The caudal fin consists of 16 rays, all of which are branched and articulated. Below and in front of the first ray are 14 fulcra, and 14 fulera also lie above the tail, gradually passing into dorsal scales in front. The rays of the dorsal, anal, and caudal fins are double. Though the tail is in a sense brevi-heterocercal the upper portion is really continued as a long, slender process bearing small rhombic scales about 20 mm. beyond the end of the middle of the tail.

All the specimens of Semionotus capensis are believed to come from Stormberg beds, and to be thus of lower Jurassic Age.

CLEITHROLEPIS EXTONI, Smith- Woodward.

This remarkable deep-bodied Ganoid was described in 1888 by Smith-Woodward from specimens found at Rouxville. Though the

ype is a fairly well-preserved specimen there were a number of points left in doubt which can now be settled. I have only seen three specimens of this species and only one approaching perfection. Though the head is fairly well preserved it is very difficult to make out the sutures owing to the large numbers of ganoine tubercles scattered over the bones. In the numerous specimens of the following allied species the cranial details are somewhat better seen.

The operculum and subopercular are well preserved in a number of specimens, the latter being nearly twice the size of the former. In front of these lies a large curved bone about half the width of the operculum and extending from the level of the upper border of the operculum to the middle of the subopercular. Smith-Woodward in characterising the genus Cleithrolepis (B. M. Cat., vol. iii., p. 155) states, "the preoperculum narrow and almost covered by the suborbitals." I feel satisfied this is not the case in Cleithrolepis minor at least, for in it down the relatively broad preopercular runs the sensory canal in almost exactly the same way as is seen in the preopercular of Semionotus. The circumoculars are small bones which form a ring round the orbit, the one at the upper and posterior corner of the orbit being the largest. A canal seems to run up over the head in the supratemporal region and another branch down the frontal and round the front of the orbit. Above the operculum there seems to be a fairly large squamosal traversed by the anterior continuation of the lateral line. And there is some evidence of a second canal running parallel to the other above it. In one of the specimens of Cleithrolepis extoni the parasphenoid is well seen. The maxilla and mandible are very short. If there is a distinct interoperculum it is very small. Three or four branchiostegal rays can be detected in one of the specimens of C. minor.

The pectorals are not preserved in any of the specimens of C. extoni I have seen though the place of attachment is manifest, and only a trace of the pelvic fin is preserved. In all three specimens the dorsal fin is nearly perfectly preserved. It consists of 20 rays with a series of small fulcra in front. The anal fin is shorter than the dorsal and consists only of 15 rays. Both fins extend to near the base of the tail. The caudal fin is large and symmetrical, but not very deeply cleft, and the longest rays are only a little longer than the longest of the dorsal and anal fins. There are altogether 30 rays in the caudal fin and a series of small fulcra both above and below, and all the rays are articulated and bifurcated distally.

The greatest length of the complete fish from the snout to the line