268 Annals of the South African Museum. 



be suborbitals. The preopercular is narrow and overlapped by the 

 suborbitals. The opercular is broad, but the division between it 

 and the subopercular cannot be made out. A number of branchio- 

 stegal rays can be seen. 



The dorsal fin is composed of apparently 16 rays with a few very 

 small fulcra in front. The anal fin has at least 11 rays. The 

 pectoral fin is badly preserved in both specimens, but the pelvic is 

 well seen in the type. It has at least 6 rays, and there is a series 

 of well-developed fulcra in front. The caudal fin is bifurcate but not 

 deeply, and is nearly homocercal. It is formed of about 21 rays, and 

 there are a number of small fulcra at least in front of the lower lobe. 



The scales are in about 37 series and about 10 rows. The middle 

 3 rows in the front two-thirds of the body are roughly about 

 twice as deep as the others, and these scales are distinctly serrated 

 behind with about 10 teeth. From these teeth feeble ridges run 

 forward on the scales a short distance, but otherwise the scales are 

 smooth. There is practically no ornamentation on the bones of 

 the head. 



CONCLUSION. 



With the exception of Seminotus capensis, which is believed to be 

 of Stormberg Age, all the fishes here described belong to Upper 

 Beaufort beds. The fishes collected by Dr. Kannemeyer and Mr. 

 Brown are pretty certainly all of one horizon, and form one of the 

 most interesting collections of fossil fish ever obtained. The Upper 

 Beaufort beds with little doubt correspond to the Keuper of Europe. 

 A species of Cyclotosaurus occurs at Eouxville, and a species of 

 Capitosaurus at Burghersdorp, and there are many other facts which 

 seem to confirm this determination of the age. 



A comparison of this Karroo collection of fishes with those of the 

 Hawkesbury of Australia is very interesting. In both faunas we 

 have the genera Dictyopyge, Gleitlirolepis, and Pholidoplwrus, and 

 possibly when the Australian and African beds are more fully 

 explored other common genera may be found. The Australian 

 Palseoniscid genus Nyriolepis may be regarded as the representative 

 of the South African Oxygnathus. Still, in spite of the resemblances, 

 the differences between the two faunas is more striking. In the 

 Hawkesbury beds Belonorliynclius and Pristisomus are very pro- 

 minent genera, yet quite unknown in the Karroo beds. On the 

 other hand, the two most prominent genera in the South African 

 beds Hybodus and Helichthys are unknown in the Australian 

 beds. 



