406 NOTES ON THE CANID# OF THE WHITE RIVER OLIGOCENE. 
hypothesis, I think it should not be assumed in a given case except upon the clearest 
evidence. Whichever of these alternatives be true, it is, in any event, probable that the 
alopecoids are not of American origin. 
Still a third possible solution of the problem concerning the mutual relationships of 
the wolves and foxes is that Cynodictis, or some similar form, is the common ancestor of 
both lines, and that the supposed early thooids, such as Daphenus and Cynodesmus, are 
devoid of permanent phylogenetic significance. This is decidedly the least probable of 
the three alternatives, for the thooids of the American Oligocene and Miocene seem to 
form a truly connected series, in which Cynodictis has no place. Further, this view 
involves the assumption that the supposed thooids have independently run a course par- 
allel to that of the true thooids and thus encounters the very difficulty which it was 
intended to avoid. The conclusion which we reach is, therefore, that the thooids are 
probably of American origin and that the alopecoids are a branch which the wolf stem 
gave off after certain of its representatives had established themselves in the Old World. 
The thooid genealogy itself is by no means free from difficulties. In a former paper 
(94), I suggested that the line begins in Daphanus of the White River, and is con- 
tinued by the John Day Cynodesmus, but now that we haye learned the remarkable char- 
acters of the skeleton, especially of the limbs and feet, of the former genus, this view no 
longer appears so simple and natural, and its acceptance carries with it some far-reaching 
and unexpected consequences. In particular, it might be objected to this view that the 
peculiar differentiation of the feet in Daphenus would exclude that form from any place 
in the direct canine phylum, for it seems @ prior? unlikely that the dogs should first have 
acquired the power of retracting the claws and should then have subsequently lost it. 
Indeed, many morphologists are inclined to deny altogether the possibility of this method 
of evolution. In the present state of knowledge, however, such a denial is at least prema- 
ture, and there is a considerable body of evidence which goes to show that it does not 
properly apply in the case of the canine phylum. 
In the first place, the John Day genus Zemnocyon, the osteology of which has been 
very fully described by Eyerman (’96), appears to be a direct descendant of Daphnuse, 
with which it agrees in the essentials of structure, though, at the same time, it displays 
many marked changes and advances. One of the most striking of these changes in the 
later form is in the great elongation of the limbs and the assumption of a digitigrade 
gait, both limbs and feet quite closely approximating those of the modern Canide. Yet 
even in Temnocyon a reminiscence, as it were, of the partially retractile claws of Daphe- 
nus may be observed in a certain asymmetry of the second phalanges of both manus and 
pes, which are slightly excayated on the ulnar and fibular sides respectively. While 
Daphenus was a short-limbed, plantigrade or semi-plantigrade form, which, in all 
