MAY, 1921. AMERICAN MARSUPIAL, C^ENOLESTES OSGOOD. 135 



all of which it should be noted are forms having certain primitive char- 

 acters not retained by the didelphids. Double-rooted canines are not 

 unknown in other groups than marsupials but are very rare except in 

 insectivores, another primitive group. Among these, premolariform 

 canines show no evidences of retrogression. Double-rooted canines are 

 said to occur also in the Miocene pig, Hyotherium (Beddard). In 

 Antechinomys laniger, one of the most primitive polyprotodonts, the 

 canine is single-rooted but has its crown distinctly premolariform. To 

 interpret this as primitive seems quite as reasonable as to call it retro- 

 gressive (see p. 119). Canines with lateral or antero-posterior cusps are 

 found also in certain bats (e. g. Cynopterus, Harpionycteris). In the 

 Miocene caenolestids the canine is mostly unknown, but in Parae- 

 panorthus, as figured by Ameghino (1903, fig. 62), the canines are small 

 and single-rooted, not unlike those of many phalangers. These teeth 

 may have been derived from double-rooted ones but indicate no tendency 

 toward them. Such a form as Paraepanorthus, however, is evidently 

 farther advanced in some respects than C&nolestes, for its incisors are 

 more reduced in number. 



On the whole, it appears that double-rooted, groove-rooted, or 

 premolariform canines are found almost exclusively among forms 

 exhibiting various primitive characters. The Peramelidae are primitive 

 in certain other respects and, although it is not unlikely that their 

 canines may have passed through various changes prior to recent times, 

 it seems quite as possible that they may have retained an early condi- 

 tion like that shown in the Jurassic Triconodon. The same may be true 

 of Myrmecobius and of C&nolestes. 



The objections to a diprotodont evolution from the peramelids 

 therefore, are not insuperable. In fact it is clear that many of the 

 changes necessary to a transition from a didelphid to a phalanger or 

 caenolestid are already present in Perameles and it does not seem demon- 

 strable that these changes took place subsequent to the separation of the 

 peramelid and diprotodont stems. In other words, the early peramelids 

 may have become considerably differentiated from the generalized 

 didelphoid marsupial and later divided into one line leading to the 

 modern peramelids and another to the diprotodonts. At least, the con- 

 clusion is unavoidable that among living forms those most suggestive of 

 what the ancestor of C&nolestes and other diprotodonts was like, are the 

 Peramelidae. Moreover, if the early Tertiary form Myrmecoboides is 

 in reality a "properamelid," as suggested on another page (p. ! i42), belief 

 in the origin of diprotodonty from this group is greatly strengthened. 

 The following resemblances between Ccenolestes and Perameles are to 

 be noted: Six or more exclusive cranial characters (see p. no), tooth 



