206 ANNALS NEW YORK ACADEMY OF SCIENCES 



occurrence during tlie few centuries of scientific record by the almost 

 infinite duration of geological epochs and periods, we obtain a finite and 

 quite probable chance, which it is perfectly fair to invoke, where the 

 evidence against land invasion is so strong. Furthermore, the fact that 

 continents have not in general been peopled in this way one from another 

 is well accounted for by the fact that spefci^ already existed there which 

 filled the place in the environment and by their competition prevented 

 the new form from obtaining a foothold, or greatly reduced the chances 

 thereof. In oceanic islands, however, the favorable environment existed 

 without the animal to fill it. Very often, on account of this lack, some 

 other type was evolved to fill its place; birds being widely distributed on 

 account of their powers of flight have in many oceanic islands developed 

 large terrestrial adaptations to take the place of the absent or scanty 

 mammals. 



KATUP.AL RAFTS AXD THE PROBABILITIES OF OVER-SEA MIGRATION 



THEREBY 



The following series of facts and assumptions may serve to give some 

 idea of the degree of probability that attaches to the hypothesis of over- 

 sea transportation to account for the population of oceanic islands. 



1) Natural rafts have been several times I'eported as seen over a hundred 

 miles off the mouths of the great tropical rivers such as the Ganges, Amazon, 

 Congo and Orinoco.^^ For one such raft observed, a hundred have probably- 

 drifted out that far uuseen or unrecorded before breaking up. 



2) The time of svieh observations covers about three centuries (I set aside 

 the period of rare and occasional exploring voyages). The duration of Ceno- 

 zoic time may be assumed at three million years (Walcott's estimate). 



3) Living mammals have been occasionally observed in such records of nat- 

 ural rafts. Assume the chance of their occurrence (much greater than of their 

 presence being noticed) at one in a hiindred. 



4) Three hundred miles drift would readily reach any of the larger oceanic 

 islands except New Zealand. Assume as one in ten the probability that the 

 raft drifted in such a direction as to reach dry land within three hundred 

 miles. 



")) Tn case such animals reached the island shores and the environment 

 afforded them a favorable opening, the propagation of the race would require 

 either two individuals of different sex or a gravid female. Assume the proba- 

 bility of any of the passengers surviving the dangers of landing as one in 

 three (by being drawn in at the mouth of some tidal river or pi-otected inlet), 

 of landing at a point where the environment was sufliciently favorable as 

 one in ten. the chances of two individuals of different sexes being together 



88 A recent number of the Popular Science Monthly (Sept., 1911, vol. Ixxix, pp. 303-307) 

 gives the recorded observations of the drift of a natural raft of- this sort, covering over 

 a tlioiisand miles of travel. 



