72 THE MAMMALIA OF THE DEEP EIVEK BEDS^ 



bekannt, die jedenfalls in naherer oder entfernter Beziehung zu dieser Gruppe stehen, 

 allein da von denselben das Skelet entweder noch nicht gefunden ist oder doch von 

 jenem der Hunde sehr bedeutend abweicht, so bleiben wir noch immer iiber die 

 eigentlichen Ahnen des Hundegeschlechtes fast ganz nnd gar im Ungewissen" (No. 

 29, p. 247). The gradual recovery of the dogs of the American Miocene formations 

 is bringing us nearer to a satisfactory solution of this difficult problem. As Schlosser 

 has pointed out, the numerous cynoids of the European Oligocene, with the possible 

 exception of some species of Cynodictis, can be of little phylogenetic significance, 

 and in the lower Miocene of Europe the dogs disappear completely ; they are repre- 

 sented in the upper Miocene and Pliocene by only a few remains, and do not attain 

 great importance till the Pleistocene (N"o. 30, p. 488). Throughout the American 

 Miocenes, however, from the White River to the Loup Fork, they play a very impor- 

 tant part, and are not only abundant in individuals, but very varied in type, no less 

 than nine genera of Miocene dogs, most of them containing several species, having 

 been described from the different American horizons. This fact of itself would indi- 

 cate the greater probability of an American rather than a European origin of the 

 family. 



In the Loup Fork beds, aside from the aberrant JElurodon, several species of 

 cynoids occur which are indistinguishable from Cams, and, so far as the remains at 

 present known are concerned, must be referred to that genus, though complete 

 material will probably require their separation from it. One of these species, 

 C. brachypus Cope, is very probably of phylogenetic importance and is significantly 

 like Cynodesmus. This is a microdont species, which retains many primitive char- 

 acters, such as the small sectorials, short face, long cranium, elevated sagittal crest 

 and weak feet. The lower Loup Fork strata of the Deep River valley (Cyclopidius 

 beds) contain a possible species of Canis, the C. f anceps, which will be hereafter 

 described. So far as this form is known, it is intermediate between the C. braclvypus 

 and Cynodesmus. The latter genus is found in the lower Deep River beds, which we 

 have already referred to the summit of the John Day horizon, and its connection 

 with C. brachypus is a fairly close one, as is apparent not only in the dentition but in 

 the characters of the skull as well ; as, for examj)le, in the characteristic shape and con- 

 nections of the paroccipital processes, length of the zygomatic arches, size and shape 

 of the coronoid process of the mandible, etc. We may also fairly assume that " the 

 elevated sagittal crest and the small feet " are shared by the older genus. The 

 White River type, Daplicenus, is separated from Cynodesmus by a wider interval, 

 the typical John Day horizon, in Oregon, not having as yet yielded any form which 

 can be placed in the series, unless we are to find the missing link in the species 



