THE YEAR'S VERTEBRATE PALAEONTOLOGY 507 



characters of the skull and skeleton confirming the evidence 

 afforded b}' the dentition. The more typical South American 

 Canidae appear, on the other hand, to be the nearest living 

 representatives of the middle Tertiary Noihocyon and Cyiio- 

 dcsmiis ; Mesocyon being slightly off the line in the direction of 

 TcvDiocyoii. 



The generally accepted classification of fossil Canidae is, 

 moreover, in the same author's opinion, misleading ; more 

 especially the association of the genera CcpJialogalc, Siniocyon, 

 Oligobiiuis, E)ihydrocyou^ and Hydrocyon in the single subfamily 

 Simoc3'oninae. The two last-named genera are, for instance, 

 probably not separable, and together with Pliilotrox appear to 

 be related to Tcuinocyon^ and so with the modern Cyon ; while 

 Oligobiuds is probably referable to the Mustelidse. Again, 

 Cephalogale, which has been regarded as ancestral to the bears, 

 seems rather, judging from dental characters, to be related to 

 the typical members of the family (Caninae) ; such aberrant 

 features as it possesses in this respect being to some extent 

 paralleled in Cynodesmus. 



Here it may be well to quote a very important note on 

 classification generally appended ib}^ Dr. Matthew to his obser- 

 vations on extinct dogs. It appears, he writes, " obviously 

 improper to separate two closely allied genera or species because 

 they are destined to give rise in the future to two distinct 

 families, in the opinion of the investigator. So far as classifica- 

 tion and nomenclature are based upon the facts of structural 

 resemblance or difference, they are permanent, but the attempt 

 to make families, genera, or species correspond with hypothetical 

 phyla is not only misleading but destructive of any approach 

 to uniformity or permanency of arrangement." 



These remarks the present writer may take the opportunity 

 of endorsing. An attempt has been made to introduce such a 

 phylogenetic classification in the case of the early Perissodactyla, 

 and the genera Hyracotheriuni and Systemodon^ which are 

 obviously near relatives, have been referred to distinct families 

 on account of the presumption that while the one is the ancestral 

 form of the horse tribe, the other has given origin to the tapirs. 



Dr. Matthew, as already mentioned, transfers Oligobiiuis, which 

 had been regarded as ancestral to SpcotJws, to the Mustelidae. The 

 short muzzle lacks the constriction in front of the carnassials 

 distinctive of the early Canidae, and the other dental features, 



