MATTHEW, CLIMATE AND EVOLUTION 207 



might be assumed as one in ten, tlie alternate of a gravid female as one in 

 five. Tlie clianee of one of tlie two liappening would be 1/10 + 1/5 = 3/10. 

 The chance of the species obtaining a foothold would then be .3/10 X 1/3 X 

 1/10 = one ill a hundred. 



If then we allow that ten such cases of natural rafts far out at sea have 

 been reported, we may concede that 1000 have probal)l.v occurred in three 

 centuries and 30,000,000 durinoj the Cenozoic. Of these rafts, only 

 3,000,000 will have had livincr maramals^^ upon them, of these only 30,000 

 will have reached land, and in only 300 of these cases will the species have 

 established a foothold. Tliis is quite sufficient to cover the dozen or two 

 cases of Mammalia on the larger oceanic islands. 



Few of these assumptions can be statistically verified. Yet I think 

 that, on the whole, they do not overstate the probabilities in each case. 

 They are intended only as a rough index of the degree of probability that 

 attaches to the method, and to show that the populating of the oceanic 

 islands through over-sea transportation, especially upon natural rafts, is 

 not an explanation to be set aside as too unlikely for consideration. 



I have considered the case only in relation to small mammals. With 

 reptiles and invertebrates, the probabilities in the case vary widely in 

 different groups, but in ahnost every instance they would be consider- 

 ably greater than with mammals. The chance of transportation and sur- 

 vival would be larger and the geologic time limit in many instances much 

 longer. "Wind, birds, small floating drift and other methods of acci- 

 dental transportation may have played a more important part with in- 

 vertebrates, although they cannot be invoked to account for the distribu- 

 tion of vertebrates. The much larger variety and wider distribution of 

 infra-mammalian life in oceanic islands is thus quite to be expected. And 

 the extent and limits of such distribution are in obviously direct accord 

 witii the opportunities for over-sea transportation in different groups. 



On tlie other hand, the transportation of very large animals in this way 

 ^ may fairly be regarded as a physical impossibility, which could not be 

 multiplied into a probability by any duration of time. The only methods 

 of accounting for such animals would be by evolution in loco from small 

 ancestors, by swimming, by introduction through the agency of man and 

 by actual continental union. 



The first hypothesis would involve evolution in an isolated and more 

 or less altered environment and would result in wide structural differ- 

 ences from any continental relatives. The second applies with greater 

 probability to large than to small animals, but, except for animals of 



™ Small reptiles and invertebrates would only rarely be observed, if present. 



