404 NOTES ON THE CAKTD.E OF THE WHITE RIVER OLIGOCENE. 



upper Eocene of Europe, occurring in the same horizons as those in which Cynodictis 

 first appears. For similar reasons, it is very difficult to believe that Amphicyon can be 

 the ancestor of the thooids, for that genus has already begun to become differentiated in 

 the direction of the bears and is contemporary with or even younger than certain Ameri- 

 can genera, such as Temnocyon and Cynodesmus, which are undeniable thooids. 



M. Boule's hypothesis involves some rather startling consequences ; if true, we shall 

 be forced to conclude that the two series of modern Canidce have been separated ever 

 since the close of Eocene times and that they had no common ancestor nearer than the 

 middle Eocene or Bridger stage. This conclusion would imply such an extreme and 

 remarkable degree of parallelism or convergence as has hardly been believed possible, 

 an exact parallelism in all parts of the dentition, skeleton and soft parts, terminating in 

 almost complete identity of structure. Indeed, many systematists regard most of the 

 modern foxes and wolves as belonging to the single genus Canis, and Huxley speaks of 

 the differences between them as being so slight, that a generic separation can be justi- 

 fied only on the grounds of convenience. Is it conceivable that two series of mam- 

 mals which were already separated in the Eocene should have converged into what is 

 practically a single genus ? 



Unlikely as it may appear, 1 am inclined to believe M. Boule's hypothesis concern- 

 ing the relationship of Cynodictis to the alopecoids is not to be summarily dismissed, but 

 that it may eventually prove to be well founded. It is certainly a suggestive fact that 

 Cynodictis, like the foxes, is devoid of any frontal sinus, while all of the other Ameri- 

 can genera, from Daphcenm onward, have well-marked sinuses, as in the wolves. Fur- 

 thermore, whatever conclusion we may reach with regard to the single or dual origin of 

 tin' Canidce, there is much reason to believe that such extreme eases of parallelism and 

 convergence have occurred among mammalian phyla and that they may be more fre- 

 quent than is commonly supposed. One very striking example is that of the true cats 

 (Fclina') and the sabre-tooth series (Machairodontince) originally pointed out by Cope 

 and elaborated in much detail by Adams (''•"'>). 



Unfortunately, complete demonstration is lacking in this very extraordinary case of 

 parallel development, because the early stages in the phylogeny of the true cats have not 

 yet been recovered, but the successive genera of the Maehairodonts are fairly well known, 

 and they form a connected series. None of these machairodont genera, not even the ear- 

 liest and most primitive of them, can lie regarded as ancestral to the true cats, for with- 

 out exception they all display the characteristic and unmistakable features which place 

 them in the sabre-tooth series. The more primitive genera, such as DinicUs, possess a 

 dentition which is hut slightly modified in the direction of the cats, and cranial foramina 

 resembling those of the early dogs in the presence of an alisphenoid canal, the separa- 



