MODES OF MAMMALIAN EVOLUTION 653 



American hoofed animals known as the fLitopterna (Chap. 

 XIII). The many remarkable resemblances between these 

 ungulates and the perissodactyls and, more specifically, be- 

 tween the family jProterotheriidse and the horses, have been 

 very differently interpreted by palaeontologists. Some have 

 insisted that the fLitopterna should be merged in the Peris- 

 sodactyla, on the ground that such a degree of likeness could 

 not have been independently acquired. Others hold that this 

 is a remarkable case of parallelism or convergence, and the 

 latter is, in my opinion, much the more probable view. Until 

 the ancestry of both groups, Perissodactyla and fLitopterna, 

 shall have been definitely ascertained, it will not be practicable 

 to make a final decision between these alternatives, nor, if 

 the similarities were really independently acquired, to deter- 

 mine whether parallel or convergent evolution is involved. 

 It is quite possible that both groups were rooted in the common 

 ground of the f Condylarthra, and, if so, their relation is one 

 of parallelism ; but no such common ancestry has been proved, 

 and it is equally possible that their ancestry was totally dis- 

 tinct. In the latter case the resemblances were due to con- 

 vergence. 



Assuming that the remarkable resemblances between the 

 fProterotheriidae and the horses were separately acquired, 

 it should be emphasized that these similarities nowhere amount 

 to identity. The likenesses are not confined to a few structures, 

 but are general throughout the skeleton and may be noted in 

 the teeth, skull, trunk, limbs and feet, but in every single one 

 of these parts the similarities are offset by differences of great 

 significance. No competent anatomist would mistake any of 

 the bones of the fproterotheres for the corresponding parts 

 of the horses, whatever view he might hold as to the relation- 

 ship between the two groups. The case is thus one of a very 

 instructive kind, as tending to show that identity of structure 

 in so highly complex creatures as mammals is not independently 

 attained by widely separated or entirely unrelated forms. 



