OF THE AUSTRALIAX MARSUPIALTA. 183 



Tliat the Dasyurichr should present more primitive characters than tlie Didelphyidse, a 

 family ^vhich is otherwise so completely prototyjial, appears at the outset improbable. In 

 addition it may be shown that the three cusps figured by Winge in Marmosa cinerea 

 and Phascogale are not homologous, the third cnsp of the latter animal not being present 

 in Ma7'mosa, whose third cusp is equivalent to the second of Phascogale, and whose 

 second cusp represents a new developmeut. 



In PI. 5. fig. 1 b will be found a carefully drawn external profile view of the inner 

 molar of Pei-atherium., in which the outer cingular ridge will be seen to bear in all six 

 elevations. The latter may be designated in the order of their occurrence from before 

 backwards as a, b, b-^, c^, c, c.,*, the letters indicating those elements which are well 

 developed and have, for the most part, definite homologues in other polyprotodont forms, 

 while the letters to which numerals are appended indicate smaller and more subsidiary 

 elements. The cus]is b ( or ab) — see account of Dasyui'idte, — <?, and c.^ represent the three 

 outer cusps figured by Winge in Phascogale peniciUala. On examining the arrange- 

 ment of these structures in the various species of Marmosa and Peramys, wdth reference 

 to the type presented by Perathermm, there is found to be not a single character of 

 them which will serve to separate one genus from the other. The stylar formula varies 

 to a certain extent in different individuals of a species, and even in different teeth of 

 the same animals, the range being from a formula of a, b, c to one of a, b, ij, c^, c, c.^- 

 In Peramys the more subsidiary cusps are on the whole jjoorly represented ; c^ 

 (Winge's 3) may be present or absent. In Marmosa c, is nearly always present and 

 comparatively well developed, the only notable exception being in 31. elegaiis, in which 

 only two of some twelve or fourteen specimens presented signs of intermediate cusps. 

 In Marmosa murina, M. cinerea, and 31. rapjjosa, c^ is better developed, as compared 

 with the otlier cusps, than in the remaining species of the genus, and a similar condition 

 is found in some cases in Caluromys. This cusp is that designated as 2 in Winge's 

 figure of 3Iarmosa cinerea. Its developmental stages, as here interpreted, are repre- 

 sented in PI. 5. figs. 1, 28, 29, 30 of Peratherium, Peramys Iheringi, Peramys 

 americana, and 3Iarmosa cinerea. 



The question here ai'ises — How can it be shown that the cusp c, is a subsidiary 

 clement which is becoming enlarged in such forms as M. cinerea rather than an original 

 cusp which is becoming reduced ? According to Winge's view, the molars of 31. murina 

 are more primitive than those of the smaller forms M. microtarsus and 31. pusilla, 

 because in the former the median outer cusp is better developed and the teeth are less 

 compressed in an antero-posterior direction. But in reality they are much more specialized, 

 because not only 31. cinerea but also 31, murina and M. rapposa are closely allied to 

 Caluromys, of w'hich they represent an ancesti-al type. Their teeth are transitional in 

 pattern between those of Caluromys and those of the smaller forms, and this is true in 

 respect to the peculiarly elongated canines and the reduced anterior premolars as to the 

 molars. The molars of Caluromys are noticeable for their lateral compression and the 

 obsolete character of the external cingulum {cf. PI. 5. fig. 27). The apparently weU- 



• Cf. text-fig. 1, p. 89. 



