14 ANNALS NEW YORK ACADEMY OF SCIENCES 



are descended from a single stock, a single ancestral type, and as the 

 gi'oup is peculiar to the island the inference would be that its differentia- 

 tion took place after it reached the island and not before. The arrival of 

 that type represents therefore one invasion and not nine. 



2) The Cricetines are the older group, prevalent in Holarctica since 

 the beginning of the Oligocene and presumably widely spread in the 

 Ethiopian and Oriental regions during the Middle and later Tertiary. 

 The Murines are a much later development and comparatively recent 

 immigrants into the tropical Old World, where the}'' have supplanted the 

 older Cricetines more completely than in the less accessible tropical New 

 World. The Cricetines have therefore had a much longer time than the 

 Murines to reach Madagascar through accidental transportation; it is to 

 be expected that the single stock which arrived should be a Cricetine. 



3) The Malagasy lemuroids, Gregory has recently shown, all belong to 

 a single group, distinct from any of the continental lemuroids, and may 

 equally be supposed to represent a single invasion. But the wide differ- 

 entiation indicates that this invasion must be a much earlier one than 

 that of the Cricetines. 



4) The Carnivora — all viverrines — are not so clearly derivable from 

 a single stock, but they may be so {Cry ^Jto pro eta, although highly special- 

 ized, is fundamentally a Middle Tertiary viverrine). The date of arrival 

 can hardly be much later than Oligocene, and certainly not earlier. 



5) The insectivores — all Centetidse — are likewise derived from a single 

 stock, peculiar to the islands, its only continental relative being the 

 Potamogalids. The diversity of the Malagasy Centetids indicates an 

 early Tertiary invasion. 



6) The pigmy hippopotamus must be a late arrival, for the Hippo- 

 potami were not evolved until the late Tertia.ry, and no adaptive radiation 

 has occurred in Madagascar. 



7) An Eocene land connection and subsequent isolation might account 

 for the Lemurs and Centetids, but would not allow of land invasion by 

 Viverrids, Cricetines or Hippopotami. We should expect also that at 

 least some remains of the Eocene ungulates and creodonts would survive, 

 since they were not displaced by higher types as on the mainland. 



8 ) xAlu Oligocene land connection with isolation before and after might 

 explain Lemurs, Centetids and Viverrids, but not the Cricetine group 

 and certainly not the Hippopotami, and would be difficult to reconcile 

 with the absence of monkeys and of the primitive ruminants and pro- 

 boscideans (all already in Africa in Lower Oligocene). 



9) A Miocene or Pliocene land connection with isolation before and 

 after would explain the Cricetines, and either the Yiverrines (if early) 



