In the latter part of 1868, Mr. Flower, as he was then styled, communicated to the Zoological Society a most important paper entitled, “On the Value of the Characters of the Base of the Cranium in the Classification of the Order Carnivora,” which was published in the first part of the Society’s _Proceedings_ for the following year. Working on the lines suggested twenty years previously by Mr. H. N. Turner, who had pointed out the importance of certain peculiarities of the base of the skull in the Mammalia, and especially demonstrated their constancy in the different groups of the Carnivora, Flower felt himself justified in dividing, on these characters, the existing terrestrial representatives of that order into three groups. These were—1st, the Æluroidea, comprising the cats (_Felidæ_), the fossa (_Cryptoproctidæ_), civets and mongooses (_Viverridæ_), the aard-wolf (_Proteleidæ_), and hyænas (_Hyænidæ_); 2nd, the Cynoidea, including only the dogs, wolves, and foxes; and 3rd, the Arctoidea, embracing the bears (_Ursidæ_), the raccoons and pandas (_Procyonidæ_ and _Æluridæ_), and the weasels, badgers, otters, etc. (_Mustelidæ_). One result of this classification from cranial characteristics was to determine definitely the position of the American cacomistle (_Bassaris_ or _Bassariscus_), which had been previously uncertain. The genus, as might have been expected from distributional considerations, turned out to belong to the raccoon family (_Procyonidæ_). As regards the relationship of the three main groups, subsequent palæontological discoveries have fully confirmed Flower’s view that the _Canidæ_ (Cynoidea) occupy a central, or perhaps rather a basal, position. Palæontology has, however, also shown that the bears (_Ursidæ_) are a direct offshoot from the _Canidæ_, and accordingly that, if extinct forms be taken into consideration, there is no justification for the separation of the two families into distinct primary groups (Arctoidea and Cynoidea). On the other hand, fossil forms from the Lower Tertiaries of France and of North America seem to demonstrate the existence of a complete gradation between the primitive dogs (_Canidæ_) and the ancestral civets (_Viverridæ_), thus breaking up the distinction between the Cynoidea and the Æluroidea. Nor is this all, for according to the French palæontologists, there exists a transition between the primitive civets and the early weasels (_Mustelidæ_); which, with what has been already stated in connection with the bears, indicates that the Arctoidea is a more or less artificial group, the members of which have come to resemble one another to a certain degree in regard to the characters of the base of the skull, owing to “parallelism.” In this connection it is somewhat curious to note that a certain resemblance, which had been pointed out by Turner as existing between the mongooses or ichneumons (_Viverridæ_) and the weasels, was regarded by Flower as of no importance.