HISTORY OF THE fASTRAPOTHERIA 513 



main so pending the discovery of much more complete speci- 

 mens of the various genera which made up the series. I am 

 incUned to the opinion, however, that all of the indigenous 

 groups of South American ungulates, which inhabited that 

 continent before the great immigration from the north, were 

 derivatives of the same stock and more nearly related to one 

 another than to any of the orders which Uved in other regions. 



In looking over the labyrinth of ungulate history, as re- 

 corded by the fossils, certain facts stand out clearly, while 

 others are still very obscure. It is Uke trying to trace the 

 plan of vast and complicated ruins, which here are deeply 

 buried in their own d6bris, there are fully exposed and in 

 another place are swept away so completely that hardly a 

 trace remains. But the problem is far more complex than any 

 which can be presented by buildings, for the factor of repeated 

 migrations from continent to continent comes in to obscure 

 the evidence. Had each of the great land areas received its 

 original stock of early mammals and then been shut ofif from 

 communication with any other, many of the difficulties would 

 be removed, but the story would lose half its interest. 



Within the Umits of the family, giving to that group the 

 broad and elastic definition which has hitherto been employed, 

 we have repeatedly found it feasible to construct a phylogenetic 

 series which very nearly represents the steps of structural 

 modification as they occurred in time. Much less frequently 

 is it possible to trace alUed families to their common starting 

 point, and, so far as the hoofed animals are concerned, in no 

 case have we yet succeeded in doing this for the separate orders. 

 The obstacle hes in the fact that the ordinal groups were al- 

 ready distinct, when they made their first appearance in the 

 known and accessible records, and the hypothetical ancestors 

 common to them all, or to any two of them, are to be sought 

 in regions of which we know little or nothing. Nevertheless, 

 certain legitimate inferences may be drawn from the available 



2l 



