The second family of South American marsupials which seem to have a distinctly Australian appearance are the caenolestids. These little animals were long known from a few fragmentary fossils (Epanorthus, Abderites, etc). In 1895 Oldfield Thomas announced that this supposedly extinct family was actually repre sented in the living mammalian fauna of Ecuador and Bolivia by a small shrew-like animal, which he called Caenolestes and pointed out that it resembled certain of the extinct Pata gonian epanorthids in its dentition, also that it resembled in other respects certain of the Australian phalangers, with the important exception that its hind feet were devoid of the syn dactyl specialization of the second and third toes. Since that time a number of species of Caenolestes have been found in the moun tains of the Andean parts of South America and the anatomy of the animal has been thoroughly studied, especially in the recent monograph by W. H. Osgood. From all this it appears that Caenolestes exhibits a highly confusing mixture of resemblances with the Australian bandicoots and diprotodonts, but that it con trasts widely with the opossums except in a few details, such as the construction of Jacobson's organ in the interior of the nose; in this it shows strong resemblances both to the opossums and the bandicoots. The general impression left by a careful considera tion of the evidence is that, in spite of the somewhat phalanger like features of its dentition, Caenolestes is not a true diprotodont but merely a phalanger-like offshoot of a primitive marsupial stock that was perhaps more nearly allied with the bandicoots than with the American opossums.
How then did the extinct borhyaenids and the extinct and recent caenolestids, both apparently more nearly related to Australian stocks than to the American opossums, ever get into South America? Are they derived from two as yet undiscovered northern ancestral stocks which also gave rise respectively to the Australian dasyuroids and to the Australian bandicoots and diprotodonts, or did they come into Patagonia from the south, after crossing some long-sunken archipelago connecting Patagonia with the once flourishing continent of Antarctica? Matthew has shown that in a great many other cases relict forms of the southern land masses have been derived from northern ancestors and this may even tually prove true in the case of the borhyaenids and caenolestids. Nevertheless, the more direct evidence so far brought forward by recent authors leaves the impression that the American opossums are not closely related either to the borhyaenids or to the caeno lestids and that the latter two are more nearly related respectively with the dasyuroid and perameloid stocks of Australia in spite of their wide geographic separation.
for example, appear to be more nearly related to the Australian true diprotodonts than to the American polyprotodont opossums, while the functionally "diprotodont" caenolestids appear to be more nearly related to the polyprotodont bandicoots than to the true diprotodonts. No less objectionable is the old grouping into Diadactyla and Syndactyla recently revived by Wood Jones, which is based solely upon the presence or absence of the syndac tylous specialization of the second and third toes of the hind foot. However, the bandicoots, which resemble the kangaroos in pos sessing the syndactylous type of foot, differ profoundly from them in many other respects. From a comparative study of the den tition we may even affirm with B. A. Bensley that both bandicoots and kangaroos, along with all other Australian families, must be derived from primitive Australian polyprotodonts much resem tiling the pouched mice (Pliascogale). In short, in the present state of knowledge it seems advisable to abandon the old larger groups and to group the families into the following series of superfamilies: 1. Superfamily Didelphoidea. North America, Cretaceous to Recent; Europe, Oligocene; South America, Oligocene to Recent. Primarily arboreal with strongly grasping hind feet, the great toe (hallux) large and divergent. Second and third digits not syn dactylous. Snout pointed. Insectivorous-carnivorous, with un reduced dentition; polyprotodont. Dental formula typically P M4, seldom reduced. Upper molars triangular, with strongly developed metacones (main posteroexternal cusps) and various developed accessory marginal cusps; internal cusps large. Lower molars primitive, tuberculo-sectorial. Auditory process of alisphenoid small or moderate, not covering tympanic cavity below. Jacobson's organ well developed. Nipples numerous (5-27). Pouch variable, if present opening forward or downward. Paired vaginal canals separate or partly united in a variable median cul-de-sac. Allantoic sack abortive, not forming a placenta.
Family Didelphiidae. North America, Cretaceous to Oligo cene, Recent; South America, Oligocene (?), Lower Mio cene, Recent.
2. Superfamily Dasyuroidea. Australian Region,-Pliocene to Recent ; One family in South America,-Miocene. Terrestrial or slightly arboreal, with narrow or variously elongated hind feet, the hallux more or less reduced. Snout pointed to blunt. Insec tivorous to extremely carnivorous dentition, polyprotodont, often with reduced premolars. Upper molars triangular to shearing. Internal cusps often small. Lower molars tuberculo-sectorial to shearing. Auditory process of alisphenoid much inflated. Nipples usually numerous (4-1o). Pouch variable, often reduced or ab sent ; if present, marsupial bones sometimes absent ; typically opening backward. Median vaginal canal temporarily pierced at parturition. Allantoic sack abortive.