No. 3.] MODE OF EVOLUTION IN THE MAMMALIA. 369 



and reached the other by immigration. Other genera form 

 abortive side branches which lead to no permanent results, and 

 thus it seems altogether probable that there is only one series, 

 at least as late as the Pliocene. That some species of Equits 

 should be derived from Protohippus and others from Hippo- 

 therium, as Cope has suggested, it is by no means impossible, 

 but cannot be admitted until the ancestral history of these 

 species has been followed out step by step. 



These considerations show how profoundly important it is 

 that in all phylogenetic discussions the possible effects of 

 parallelism and convergence should never be lost sight of. 

 One of the commonest fallacies to which neglect of these 

 factors leads is the assumption that because two allied groups 

 possess a certain character, their common ancestor must also 

 have possessed it. This may or may not be true in any given 

 case. For example, it has been shown that the spout-like 

 odontoid process of the axis has been independently developed 

 several times, yet Boas concludes that the ancestor common to 

 all ungulates must have possessed it : " Auch beim Pferd, beim 

 Anchitherium (nach ' Kowalevsky) und (nach Flower) beim 

 Tapir verhalt sich der Proc. odont. ahnlich wie bei den typischen 

 Wiederkauern, und es ist somit wahrscheinlich dass die gemein- 

 samen Vorfahren aller Ungnlaten diesen Charakter besassen und 

 dass auch die konische Form, etc., desselben beim Schwein se- 

 kundar erworben ist" (No. 3, p. 518). But even this conclusion 

 involves the admission of a parallel development of the peg-like 

 odontoid in the pigs, tragulines, rhinoceroses, etc. Flower's 

 view is more consonant with the facts of palaeontology. " The 

 form of the odontoid process in the Tylopoda might lead to the 

 idea that they were segregated from the ruminant stock after 

 the Tragulina had been given off ; but as it is also found in the 

 horse, it is probably adaptive, as are the hypsodont molars " 

 (No. 17, p. 190). 



It is further evident that Huxley's dictum to the effect that 

 in forming natural groups "it is more important that similarities 

 should not be neglected than that differences should be over- 

 looked," cannot be maintained. On the contrary, it is only a 

 due consideration of the differences that enables us to distin- 

 guish artificial from natural groups. It is likewise clear, if 

 these conclusions be sound, that no general rules can be laid 



