ON THE CLASSIFICATION OF THE THERIODONTIA. 35 



3. The Bases of Classification of the Theriodontia. 

 By D. M. S.Watson, F.Z.S., University College, London. 



[Received October 19, 1920 : Read February 8, 1921.] 

 (Text-figures 1-29.) 



Among the first series of reptiles from the Karroo system 

 of South Africa sent home by Andrew Geddes Bain were a few 

 poor fragments of animals with a heterodont carnivorous denti- 

 tion. Later collections from the same rocks included more 

 satisfactory remains of these animals, which were described by 

 Owen, who recognised their mammalian appearance and despite 

 his ante-evolutionary views even suggested that they were mam- 

 malian ancestors. Prof. Seeley's visit to South Africa marked 

 a turning-point in our knowledge of these reptiles, because he 

 showed that their remains were found in rocks of widely different 

 ages, and that the latest assemblage — Diademodon, Gynognathus, 

 and Trirachodon, — were more mammal-like in their dentition than 

 were their earlier forerunners. He showed also that they pos- 

 sessed a mammal-like secondary palate, but failed to arrive at a 

 satisfactory interpretation of that region in the less complete 

 remains of the earlier forms known to him. Neither Owen, 

 Seeley, nor Lydekker was able to draw up any useful classification 

 of these reptiles on account of the paucity of material, and the 

 first definite step in so doing was made by Broom, when in 1904 

 he showed that Scylacosaurus sclateri, a form from the lowest 

 zone of the Beaufort beds, differed from the " Oynodonts " of 

 the highest zone of that formation in lacking any trace of a 

 secondary palate. 



Subsequent work by Broom added many new generic types to 

 those included with Scylacosaurus in that primitive division of 

 the carnivorous Therapsids whose members lacked a secondary 

 palate and had uncusped molar teeth. This division Broom 

 made into an order and called Therocephalia. 



No further important additions were made to our knowledge 

 of the skull of any of these reptiles till, in 1911, the writer gave 

 a very detailed account of the skull of Diademodon and Broom 

 a more general description of the skull in all the Oynodonts. 

 The first important addition to our knowledge of the earlier 

 Theriodonts was the description by the present author of the 

 posterior half of a skull from the Gistece]}haht,s- zone, which 

 agreed with Gorgonops in having a broad parietal region, the 

 parietal bone being excluded from the margin of the temporal 

 fossa. In the same paper some of the more salient features of 

 the palate of Goi-gonops were described, and it was indicated 

 that the form showed the beginnings of the Cynodont secondary 

 palate, the skull known as Arctognathus curvimola showing an 

 intermediate condition. Whilst I was writing this paper in 



3* 



