CLASSIFICATION OF THE THERIODONTIA. 79 



upwards, and a little forwards. The postfi'ontal is veiy 

 small, the preparietal is absent. The maxilla is very short and 

 deep, there is no step in the upper jaw. The prefrontal is short 

 and probably small. I have already shown that the basicra.nial 

 and otic regions are very advanced (Proc. Zool. Soc. 1914, 

 p. 1028). The epipteiygoids are widened and flat, very much 

 as in Diademodon. The posterior eiid of the quadrate ramus of 

 the pterygoid no longer reaches the quadrate. The pterygo- 

 parasphenoidal girder, instead of having a flat ventral surface 

 from which a deep median crest rises, is rounded, with the crest 

 represented by a low fillet. 



The palate is most conveniently compared with that of 

 Gorgonops, which represents an earlier stage leading to it. 



The great median groove is deepened and its roof is entirely 

 formed by a median vomer, which presumably represents a 

 forward growth of the posterior median bone of tlie Gorgonops 

 palate. Anteriorly the ridge rising from the vomer in Arcto- 

 gnathus suggests that there was a soft secondary palate into 

 which a secondary bony plate may have grown out in the 

 concealed part of the palate. 



Tlie tooth-bearing roughened area of the anterior ramus of 

 the pterygoid lies much further back than in Gorgono'ps, and the 

 pterygoid no longer I'eaches the posterior nares. 



The quadrate of Arctognathus is much smaller than that of 

 Hcymnognatlius. 



The conversion of Arctognatlius into a "Cynodont" like Cijno- 

 giuUhihS demands only the following changes : — Htill further thin- 

 ning of the basis cranii, further reduction of the quadrate wing of 

 tlie pterygoid, the development of a connection between the 

 quadrate wing of the epipterygoid and the paroccijjital ; further 

 retraction of the anterior ramus of the pterygoid, so as to reduce 

 the roughened areas to a pair of small knobs on each side of the 

 posterior end of the median groove ; a little reduction of the 

 ectopterygoid ; the development of secondary plates from the 

 maxiila3 and palatines in the existing soft secondary palate ; 

 the conversion of the narrow intertemporal ai-ea into a sagittal 

 crest, to increase the length of the temporal muscles ; the loss 

 of the postfi'ontal, and a farther reduction of the frontal and 

 prefrontal, leading to an inci'ease in size of the posterior part of 

 the nasals. These changes are all in the same direction as those 

 which convert a Pelycosaur like VaranosauriCs into a Theriodont 

 like Gorgonops, and an animal like Gorgonops into a form like 

 Arctognathus, and are, on the whole, smaller than those which are 

 necessai-y to carry out the earlier improvements; in fact, Arcto- 

 giiathas, which is technically a Gorgonopsid, is structurally closer 

 to Cynognathus than it is to Gorgonops. 



Amongst other advanced forms allied to the Gorgonopsids and 

 coming from the Cistecephahis-7,one are Cynosuchus and Whaitsia, 

 which have both been excellently described, tliough not com- 

 pletely figured, by Haughton. 



