232 PROCEEDINGS OF THE ACADEMY OF [1889. 



to be closely related. Although characters which belong to these 

 two categories are by no means valueless, those which are derived 

 from differentiations of a similar kind are much more important ; 

 that is to say, positive characters are of more value than negative 

 ones. With these principles before us, we may proceed to the con- 

 sideration of the question. 



The arrangement of the fissipede Carnivora into Arctoidea, 

 Cynoidea, and Aeluroidea proposed by Flower (No. 5), founded 

 chiefly upon the characters of the basis cranii, has been very gen- 

 erally accepted among English speaking zoologists, but as Schlosser 

 (No. 14) remarks, it receives little support from paleontology, 

 as in the various early fossil forms the characteristics are mingled in 

 a very confusing way, and in a way not intelligible if Flower's 

 arrangement be accepted. Schlosser regards the dogs and bears as 

 being very closely related, while the mustelines, civets and hyenas 

 are placed together as another series. The Procyonidce (Subursidce) 

 "endlich stellen einen Formenkreis dar, der sich von den primitivsten 

 Typen der Carnivoren, ausser im Zahnbau, fast gar nicht entfernt 

 hat" (No. 14, p. 237). The cats, on the other hand, Schlosser 

 believes to be widely separated from all others groups of Carnivora 

 and to have been independently derived from creodonts, though 

 admitting that as yet we know nothing of their history. 



In most respects I regard these views as justified by paleontologi- 

 cal evidence. Thus Gaudry (No. 6, ch. IX) and Lydekker (No. 9, 

 j). 202 ; No. 10, p. 106) have satisfactorily shown the derivation of 

 the bears from Amphicyon through Dinocyon and Hycenardos. The 

 connection between the viverrines and mustelines, though not so 

 clear, is at all events very probable, as is shown by such forms as 

 Stenopleskiis. Through the kindness of M. Filhol, I had an oppor- 

 tunity of examining a fine skull of S. cayluxi in his collection and 

 found that it exhibited a remarkable combination of viverrine and 

 musteline characters. The skull is like that of the civets in shape; 

 the auditory bulla is divided by a septum and the separation into two 

 chambers is plainly indicated by an external constriction; the bulla 

 is in contact with the paroccipital process, though the latter is small 

 and does not enclose the former as in a capsule, a striking difference 

 from the viverrines. The condylar foramen is entir ly free from the 

 foramen lacerum posterius and ] erforates a ridge passing from the 

 paroccipital process to the condyle. The alisphenoid canal, if 

 present at all, is very minute, but the glenoid foramen is large and 



