ON PRE-JUKASSIC TETRAPOBS. 167 



10. A. Sketch Classification o£ the Pre- Jurassic Tetrapod 

 Vertebrates. By D. M. S. Watson, M.Sc, F.Z.S., 

 Lecturer in Vertebrate Palaeontology in University 

 College, London. 



[Received February 20, 1917 : Read Marcli 20, 1917.] 



(Text-figures 1 & 2.) 



Index. 



SrsTEMATic : — Page 



Amphibia 168 



Reptilia 171 



The enormous expansion of our knowledge of early Tetrapods 

 during the last twenty years, which we owe especially to the 

 work of Broili, Broom, Case, v. Huene, Moodie, and Williston, 

 has led to the general realisation of the inadequacy of our existing 

 scheme of classification. 



It is the purpose of this paper to produce a classification of 

 these animals which, whilst including all existing information 

 and paying attention to the taxonomic views of other students, 

 shall be so far as possible a consistent whole expressing my own 

 view of the relationships of the forms which fall within its 

 scope. 



The difiiculties of classification of eai'ly Tetrapods are identical 

 with those which lead to divergence between those classifications 

 of the early Eocene placental mammals characteristic of the 

 American and European schools. 



Most American authors, for example, follow Osborn in dividini^ 

 the Lower Eocene Perissodactyls. which are all very much alike 

 in structure, between the families of Horses, Tapirs, Rhino- 

 ceroses, Lophiodonts, Calicothei-es, and Titanotheres, whilst 

 European authorities include them all in the one family Lophio- 

 dontidfe, ancesti'al to all other families of Perissodactyls. 



Both methods are quite legitimate, expressing as they do 

 different aspects of the subject. 



Prof. Osborn's method has the great merit of forcing atten- 

 tion to the consideration of the small details wdiicli persist 

 throughout families, and of bringing out clearly our knowledge 

 of actual lines of descent. Its drawbacks are that, without a 

 very considerable knowledge not onlj^ of one animal, but of its 

 descendants, it is impossible to be certain of its position in the 

 system, and that the families are with difficulty, if at all, 

 definable. 



The other method, of having large primitive families ancestral 

 to all later lines of an order, has the advantage of emphasising 

 the great resemblances between all members of an order in its 

 early youth and of giving readily definable families into which 

 any relatively well-known type can be securely placed. It 



