HISTORY OP THE fTOXODONTIA 469 



whole appearance of the anterior part of the jaw was totally 

 changed. 



This mode of forming the tusks by the enlargement of the 

 second upper and third lower incisor is an unusual one, though 

 it was repeated in another South American ungulate order, the 

 fLitopterna, and nearly so in the Proboscidea, in which both 

 upper and lower tusks were the second of the three original 

 incisors. 



In both jaws, the canines of ^Nesodon were insignificant and 

 sometimes absent. The premolars, which were smaller and 

 simpler than the molars, had quite high crowns, but early 

 ceased to grow and formed long roots. The molars were truly 

 hypsodont and formed no roots till late in life ; they were con- 

 structed on the same plan as those of ^Toxodon, but were de- 

 cidedly more complex, the upper ones having several spurs 

 and crests given off inward from the external wall, in addition 

 to the two principal transverse crests, and they had a certain 

 superficial Ukeness to the teeth of a rhinoceros. As in ^Toxo- 

 don, these upper molars were curved inward, so as almost to 

 meet those of the opposite side above the palate. The lower 

 molars had the same bicrescentic plan as in ^Toxodon, but 

 were more comphcated, and in the concavity of the hinder 

 crescent was a vertical pillar, which was well-nigh universal 

 among the indigenous South American ungulates. 



If \Nesodon was really the ancestor of ]Toxodon, then the 

 development of the grinding teeth must have been a process 

 of completing the hjrpsodontism, until the teeth grew per- 

 sistently, never forming roots, and, at the same time, of sim- 

 plifying the pattern. This is contrary to the usual course of 

 evolution, in which the pattern grew more complex in the suc- 

 cessive stages ; but such steadily increasing complexity was not 

 invariable, and several instances of undoubted simphfication 

 are known among mammals, though not yet in other ungulates. 

 Only the recovery of the intermediate genera will enable us to 

 determine whether ^Nesodon was the actual ancestor of 



