MODES OF MAMMALIAN EVOLUTION 659 



pose. To give one instance, the very early camels of the upper 

 Eocene and lower Oligocene had small canines, which though 

 not at all functionless or vestigial, were yet little larger than 

 incisors. Though the ancestral camels of the middle and lower 

 Eocene are not yet definitely known, there is strong reason to 

 believe that in them, as in all of their contemporaries among the 

 ungulates, the canines were enlarged and fang-like. If so, the 

 canine teeth in the camels underwent decided fluctuations in 

 size, being first larger, then smaller and again enlarging. If 

 Dr. Matthew's interesting theory as to the origin of the true 

 felines from primitive fsabre-tooth cats (see p. 540) should be 

 confirmed, it would furnish a very striking example of fluctuat- 

 ing development. The acceptance of the theory involves the 

 admission of the following changes : (1) The upper canine was 

 enlarged and changed into a thin, recurved, scimitar-hke tusk ; 

 (2) the lower canine was much reduced, becoming little larger 

 than the incisors; (3) the lower jaw developed a flange on 

 each side from its inferior border, against which the inner side 

 of the upper canine rested, when the mouth was closed, and the 

 chin was nearly flat, meeting the outer surface of the jaw at a 

 right angle. After these peculiarities had been fully established, 

 the stock divided into two series ; in one, the fmachairodonts, 

 the specialization continued along the same lines, assuming 

 more and more exaggerated forms, while in the true cats it 

 was reversed. The upper canine grew shorter and thicker, 

 the lower canine was very greatly enlarged, the lower jaw lost 

 its flange, and its external and anterior surfaces no longer met at 

 a right angle, but curved gradually into each other. As pre- 

 viously stated, such a reversal strikes me as improbable and 

 not to be accepted without very much more complete evidence 

 than we now have, but it is perfectly possible that such evi- 

 dence may be forthcoming. 



Making the fullest allowance for all such cases of fluctua- 

 tion, it remains true that in the great majority of the phyla 

 whose history may be followed in some detail, development 



