180 



THE MAMMALIA OE THE DEEP RIVER BEDS. 



Summary. 



It will be convenient to sum up briefly here the principal results of this investi- 

 gation. 



(1) The beds of the Deep River valley belong to two horizons, as originally 

 pointed out by Grinnell and Dana. These horizons differ widely in lithological 

 character and even more markedly in their contained fossils, and are almost certainly 

 separated by an unconformity of erosion, which represents a considerable lapse of 

 time. The lower series should be placed at the summit of the John Day and the 

 upper at the base of the Loup Fork, where they form a well-marked subdivision (the 

 Ticholeptus beds of Cope). This subdivision is not certainly known in other regions 

 than the present one, and the deposits in Oregon, Nebraska and Wyoming which 

 have been referred to it most probably belong to the Loup Fork proper. 



MAP OF UPPER SMITH BIVER VALLEY, MONTANA. 

 Drawn by W. B. Harris from a sketch by 0. C. Mortson. 



(2) The nearest European equivalent of the upper Deep Eiver beds appears to 

 be the upper Miocene of Sansan and Simorre. 



(3) In the genus Gynodesmus, which has the dentition of Canis combined with 

 the skull and brain of the more ancient genera of the phylum, we find an important 

 link in the genealogy of the dogs, leading back to the White River form, Daphczmis, 

 through some as yet unknown genus of the lower John Day, which, however, must 

 have been not unlike the so-called Temnocyon josejjhi. The abundance of Miocene 



