78 On the Structure of the Therocephalian Skull. 



Tlie Gorgonopsids appear to have generally five upper 

 incisors, one canine, and a variable number of cheek-teetli ; 

 in the lower jaw they may have tlnee incisors, one canine, 

 and cheek-teeth. Arthognathus has four upper incisors, one 

 canine, and five cheek-teeth. The Cynognathids appear 

 to have | i., \ c, cheek-teeth variable, so that as far as 

 dentition goes these types agree. 



One curious special feature of the skull of Baun'a, the 

 deep keel on the under surface of the basisphenoid, can be 

 exactly matched by a large Therocephalian skull from tlie 

 Puriasaurus-zone in the British Museum. 



The relation of the family Gorgonopsidje to the other 

 Therocephalia is not easily decided. It agrees closely with 

 them in the general structure of the snout, palate (except 

 for the absence of suborbital vacuities), and occiput, and, what 

 is perhaps more important, in the size and general relations 

 of the quadrate. It differs in the broad and extremely short 

 parietal region, but this difference is only due to the retention 

 of a primitive feature. It also differs in the presence of a 

 preparietal bone, but this may also be a primitive feature, 

 and, moreover, it is possible that there is a small preparietal 

 in Syclacoaauras, 



Altogether the final conclusion seems to be that the earliest 

 Therapsids had a skull resembling that of Moscops with its 

 obvious specializations removed, and that from such a type 

 different but parallel lines of descent gave rise to the Anomo- 

 dotits and several families of Therocephalia, some of which 

 subsequently developed into Cynodoutia, Baaria and Diade- 

 vwdon being the terminal types of two such lines. Another 

 of these lines became mammals. In any case, the passage 

 from the Therocephalia to the Cynodoutia seems to be so 

 completely covered even by known types, that those two 

 groups can scarcely be given ordinal separation from one 

 another. 



The view of Oynodont derivation outlined above differs 

 from that shown by Dr. Broom's phylogenetic tree (Proc. 

 Zool. ISoc. lyil, p. 923) in the conception that the boundary 

 between the Tlierocephalia and the Cynodoutia has been 

 repeatedly crossed, and that the actual common ancestor of 

 the various Cynodont families (and the mammals) was most 

 probably in a pre-Tlierocephalian condition when the first 

 cleavage between the families took place. 



