OF THE AUSTRAIJAX MAESUPIALIA. 161 



'away from what would be considered a primitive type in comparison with the Pha- 

 langerida?, and also from one another, the PhascolomyidiB and DiprotodontidjB appear to 

 be allied in several features, namely ; (a) the (presumably) quadrituberculate bunodont 

 origin of their mohirs ; (b) the rodent moditications of the iacisors ; (c) the similarity in 

 size and arrangement of the cusps of the lower posterior premolars ; (d) the obliteration 

 of the anterior and median premolars. To these may be added : (e) reduction of the 

 deciduous premolars. 



It is a question not wholly answerable, however, whetlier or not some of these 

 characters may be the result of convergent developaient. Phascolarotus presents points 

 of general resemblance with. Phascolomys, 'And as regards its dentition shows three of the 

 characters above referred to, namely, i, d, and e. There is good reason for believing all 

 of the latter to be the result of convergent development. Fhascolarctus is isolated by 

 its advanced selenoid molar patterns; and while there is overwhelming evidence, both in 

 Marsupials and Placentals, of the conversion of the bunodont type of molar into a 

 lophodont one, no case is known of the conversion of the selenodont tyjie into a similar 

 lophodont one. As already pointed out, although the molars of 'Fhascolarctus and the 

 other Phascolarctinae have retained, with but little modifications, certain of the cusp- 

 characters of the polyprotodont forms, which are alike ancestral to them and to the 

 PhalangeriuLe, they are much too advanced in a special selenodont direction to have 

 given rise to those of the last-named division. Purthermore, Phascolarctus occupies a 

 derived relation to Pseudochlnis, a genus lacking characters b and d. Finally, the 

 smaller members of the Piialangerinte, such as Acrobates, are far more primitive in their 

 general dental characters than the Phascolarctinfe. "We have, however, no such proofs 

 of convergence between the Diprotodontidae and Phascolomyidse as we have between both 

 families and Fhascolarctus; so that if it is permissible to generalize on the evidence 

 available, we may assume that tlieir common features are characters of affinity — on the 

 whole, that they have diverged from common ancestors possessing a reduced premolar 

 formula, a reduced tooth-change, bmiodont quadrituberculate molar patterns, and 

 incipiently rodent incisors. As mentioned below, an analogous case of a combination of 

 fundamental characters of resemblance with special characters of divergence is observable 

 in the foot-structure of the two families. 



THYLACOLEONTID.E. 



Tiie recent discussion by Broom (1898) of the controversial question of the habits of 

 the extinct form Thylacoleo appears to allow of no further doubt as to its predaceous 

 carnivorous character. The questiun now arises as to how the relationsbips of the animal 

 with the Phalangeridie are to be explained. Throngliout tlie present paper the effort has 

 been made to show that the Australian radiation l)egan with insectivorous prototypes and 

 proceeded along two primary lines, one of them carnivorous, the other omnivorous and 

 finally herbivorous. In the second line all the advanced forms are diprotodont, and all 

 the typical terminal forms are highly specialized hcrbivora. Tlnjlacoleo is clearly a 



