II.] CELEBES. 49 



Miocene ; this, however, is obviously too early a date, since the 

 only known allies of the anoa are met with (in common with the 

 earliest of all the oxen) in the Siwalik Pliocene of northern 

 India. 



In regard to the Moluccas, Dr Wallace 1 observes that the 

 absence of many characteristic groups of Papuan birds, and 

 likewise of kangaroos and the smaller marsupials of New Guinea, 

 leads to the conclusion that these islands " cannot be mere frag- 

 ments of the old Papuan land, or they would certainly, in some 

 one or other of their large and fertile islands, have preserved a 

 more complete representation of the parent fauna. Most of the 

 Moluccan birds are very distinct from the allied species of New 

 Guinea; and this would imply that the entrance of the original 

 forms took place at a remote period. The two peculiar genera 

 with clearly Papuan affinities, show the same thing. The casso- 

 wary, found only in the large island of Ceram and distinct from 

 any Papuan species, would however seem to have required a land 

 connection for its introduction, almost as much as any of the 

 larger mammalia." 



In another work 2 , summing up the general conclusions with 

 regard to the fauna of Celebes, the same writer observes that " we 

 are fully justified in classing it as an ' anomalous island,' since it 

 possesses a small but very remarkable mammalian fauna, without 

 ever having been directly united [during Tertiary times] with any 

 continent or extensive land ; and, both by what it has, and what 

 it wants, occupies such an exactly intermediate position between 

 the Oriental and Australian regions that it will perhaps ever 

 remain a mere matter of opinion with which it should properly 

 be associated. Forming, as it does, the western limit of such 

 typical Australian groups as the marsupials among mammalia, 

 and the Trichogfassida* and Mdiphngidct* among birds, and being 

 so strongly deficient in all the more characteristic Oriental families 

 and genera of both classes, I have always placed it in the Austra- 



1 Geographical Distribution of Animals, Vol. I. p. 419. 



2 Island Life, p. 432. 



3 Equal LoriidfB ; includes the brush-tongued lories and loriquets. 



4 Honey-suckers. 



L. 4 



