86 MR. D. M. S. WATSON ON THE 



the evolution of that group, and the known forms are so few that 

 no classification is possible. 



It is, I think, reasonable to assume that the Therocephalia 

 have sprung from the Gorgonopsid stock, and that they represent 

 a series of distinct branches which display a much more rapid 

 advance in structure than the conservative main stem. These 

 advances are, on the whole, along the trends of Gorgonopsid evo- 

 lution, quite early Therocephalians thus agreeing in certain 

 features with Oynognathids. It is, as I have already shown, 

 probable that Bauria and its allies are descendants of Thero- 

 cephalia, representing the product of a parallel series of changes 

 to that which resulted in Cynognathuus imposed on a different 

 ancestor. 



In the preceding part of this paper, I have dealt only with a 

 selected series of Gorgonopsids which present resemblances to the 

 Oynognathids, and have tacitly assumed that these forms are 

 the main stock. There are, however, many other Gorgonopsids 

 which appear to represent side-branches, displaying either 

 accelerated evolution of certain features or else individual 

 specialisations. 



Of these forms the earliest and one of the best known is jEluro- 

 scmrus felinus Owen. This form was first described by Owen, 

 Q. J. G.S. vol. xxxvii.p. 261, pi. ix., Seeley subsequently figuring 

 an incompletely prepared palate. Broom later corrected certain 

 features of Seeley's description of the side of the skull. 



The British Museum includes, in addition to the type, two 

 snouts which were regarded by Lydekker as M. felinus. Broom 

 left manuscript- labels concurring in the identification, and 

 a detailed examination which I made of them showed that the 

 external surface and dentition are in complete agreement. One 

 of the specimens had no lower jaw attached and the palate has 

 been developed, with the remarkable result that it is shown to 

 differ very considerably from the type, being probably generically 

 distinct. The whole circumstance is of importance, because it 

 shows that a specimen showing only the outer surface of the 

 snout and dentition of a Gorgonopsid may be an inadequate type. 

 In text-fig. 25 I give three slightly reconstructed views of the 

 snout of B.M.N.H. R. 855 from the Undothiodon-zonel of Beau- 

 fort West. This type is very advanced in the depth and rounded 

 section of the snout, in the supression of a step in the maxilla 

 before the canine, and in the relatively slight overhang of the 

 anterior border of the nasal. It retains a very large septo- 

 maxillary foramen and a large facial exposure of the septomaxilla. 

 The palate unfortunately shows no sutures, but gives a good view 

 of the general form. In general form this palate differs very 

 considerably from that of Arctojjs, Gorgonops, Scymnognathus,etc. 

 The median region is excavated into a very narrow groove 

 bounded laterally by massive processes, whose palatal surfaces 

 bear small teeth in sockets. Lateral to the process is a small 

 fenestra or possibly a very deep pit with a well-defined margin ; 



