No. 3.] MODE OF EVOLUTION IN THE MAMMALIA. 373 



parts is a very common teratological occurrence, and there is no 

 a priori reason to assume that such reduplication might not 

 occasionally be selected and perpetuated. The fourth molar of 

 Otocyon appears to me to be probably either a case of this sort, 

 or of permanent reversion, as any other assumption with regard 

 to it seems very unlikely, a) That all the known fossil cynoids 

 have no direct connection with the existing dogs, can hardly be 

 believed by any one who has examined the material, and yet all 

 of these, with occasional individual exceptions (which still occur 

 among the recent Canidce, see Filhol, No. 16), have but three 

 molars at most, b) That the Canidce are of dual origin, one line 

 through Miacis, Daphceniis, Cynodictis, etc., to Cams, and the 

 other through unknown ancestors to Otocyon, involves a degree 

 of convergence which known facts do not justify us in assuming. 

 Between the third and fourth hypotheses, that the additional 

 molar is atavistic, or that it has been added de novo, it is very diffi- 

 cult to decide ; for in the fairly complete mammalian series which 

 we have been considering in the present and preceding papers, 

 neither process can be shown to have taken place. Yet both are 

 conceivable and do take place in individual cases ; and if so, 

 they might, under favorable circumstances, and provided they 

 recurred frequently enough, be rendered permanent. In Otocyon 

 the dentition is much more primitive (aside from the number of 

 the molars) than in any other known member of the family, 

 recent or fossil. Judging from the state of development of the 

 skull and skeleton, this condition almost certainly implies retro- 

 gression, which may possibly have stimulated atavistic tenden- 

 cies. The problem can be definitely solved, however, only by a 

 much more complete phyletic series than we can at present com- 

 mand. At all events, this reacquisition or addition de novo can 

 only be very exceptional in the history of the higher mammals. 

 7. In large groups of mammals the method which is followed 

 in the reduction of parts is generally a very uniform one. Thus 

 Owen's laws of the reduction of digits hold good for all artio- 

 dactyls and perissodactyls. Great uniformity is also exhibited 

 in the order of disappearance of the teeth, but this is not without 

 exceptions ; e.g. in artiodactyls the first premolar is usually the 

 first one to disappear, but in the camel this tooth is retained, 

 while the second is lost. The Creodonta and Carnivora follow 

 another law, but in these groups there is less uniformity, and 



