490 Annals of the South African Museum. 



appear that, in general, they are intermediate between the true 

 Cave Sandstone of the South and the Forest Sandstone and Nya- 

 inandlilovu Sandstone of Rhodesia - - the more southerly outcrops 

 partaking more of the nature of the former, the westerly and nor- 

 therly ones more of the character of the Rhodesian rocks. Thus we 

 can picture the loess-like formation of the south giving place gra- 

 dually to the true desert sands of the north, the fragments of the 

 Transvaal deposits which now remain to us representing the tran- 

 sition stage between the two types and shewing an intermingling of 

 the line-grained angular fragments from the south with the rounded 

 desert sands of the north. 



Age of the Stormberg Series. 



Until fairly recently the age of the Stormberg Series was accepted 

 as ranging from Rhaetic to Lower Jurassic; within the last few 

 vears. however, there has he-en a tendency to throw back the age 

 along the time-scale and to consider the Cave Sandstone as no later 

 than Rhaetic. 



The evidence in favour of the older view was mainly two-fold, 

 hased on the nature of the plants of the Molteno Beds and that of 

 Notochampsa from the Cave Sandstone. The latter was thought by 

 Broom to be a true crocodile allied to Pelagosaurus \ and as croco- 

 diles did not appear until the Jurassic in other parts of the world 

 the Cave Sandstone was deemed Jurassic. The Molteno plants were 

 considered by Seward to be of Rhaetic age; and the Series was thus 

 spaced between those limits. 



Detailed examination of the evidence leads, however, rather to the 

 acceptance of the later view. In the first place the Rhaetic age 

 assigned to the Molteno plant forms is based upon what is by no 

 means clear evidence and in view of recent discoveries the conclu- 

 sion may need considerable revision. Dr. du Toit is studying a large 

 amount of material mainly collected by himself, and although his 

 results are not ready for publication he has permitted me to say 

 that the undoubted association of Glossopteris, CJiiropteris, Pterophi/l- 

 Itnn, Cirlli/rferiditntt and apparently Rhexoxylon with the typical Mol- 

 teno genera Baiera, Thinnfeldia, and Tae.ninf)leris give a very pro- 

 nounced Keuper appearance to the flora. 



The vertebrates which it should be remembered come from still 

 higher horizons, when studied in detail, also bear a Triassic aspect. 



The three Cynodont forms are specialised relics from the Upper 

 Beaufort Beds. These latter, which underlie the Molteno Beds, have 



