MAY, 1921. AMERICAN MARSUPIAL, CENOLESTES OSGOOD. 137 



mediate form must be excluded, if for no other reason, on account of its 

 large size. . . . All the forms which approach the hypothetical inter- 

 mediate type are of small size. Not only this, but the diprotodont 

 modification itself may, as already explained, be shown to represent an 

 insectivorous adaptation which could only have taken place in com- 

 paratively small animals. The relations of Wynyardia are more probably 

 with one of the advanced genera, such as Pseudochirus, Phascolarctos, 

 or Phalanger." 



Osborn (1907) says: "Among fossil forms the gap between the two 

 suborders is largely bridged over by the extraordinary genus Wynyardia 

 of Baldwin Spencer, which presents a perfect melange of characters seen 

 elsewhere only in the Opossums and Dasyures (Polyprotodonts), and 

 in the Phalangers and Kangaroos (Diprotodonts)." Gregory (1910, 

 p. 214) speaks of Wynyardia as "another form that helps to bridge over 

 the structural gap between the Poly protodontia and theDiprotodontia." 

 Later (Osborn, 1910) it fails to receive formal classification. 



Spencer, in summarizing the important characters of Wynyardia 

 finds five which he interprets as indicating polyprotodont affinities. 

 These, as numbered and stated by him are as follows: 



"(i) Proportionate length to breadth of the skull, 100:67. This 

 approximates most nearly to Dasyurus, and shows a decidedly 

 greater width than in the Phalangeridae. 



" (2) Lambdoidal crest well developed, as in Dasyurus. 



" (3) Sagittal crest strongly developed, resembling that of Dasyuridae 

 and species of Didelphys. 



" (5) The wide sweep and upward curvature of the zygomatic arches, 

 as in Dasyuridae. 



" (7) The transverse elongation of the glenoid cavity, the downward 

 produced plate of bone which forms the boundary is not con- 

 nected with any structure forming part of the auditory passage. 

 In this respect it agrees with Dasyuridae and Perameles, and 

 differs markedly from the Phalangeridae, amongst which it 

 forms the anterior part of a bony auditory canal." 



At least the first three of these characters cannot be regarded as 

 exclusively polyprotodont. Taking them up numerically, it is found as 

 to (i) that, although Dasyurus and Sarcophilus have very wide skulls, 

 polyprotodonts in general have relatively narrow ones. The ratio runs 

 from only 39.7 in Perameles to 81.2 in Sarcophilus (see p. 99). On the 

 other hand the average relative width is greater in diprotodonts than 

 in polyprotodonts, the only very narrow type seen among diprotodonts 

 being in the prhpiitive Ccenolestes. 



As to (2), it is to be noted that a high lambdoid crest is no more than 

 a generic character among polyprotodonts. Within the limits of the 



