40 EASTER ISLAND. 



tageous to maintain the segregation of the data by the classes of the 

 occurrence of the identification in Polynesia exterior to this southeastern 

 province. These are three: (i) identification in both migrations; (2) 

 identification in the Proto-Samoan exclusively; (3) identification in the 

 Tongafiti exclusively. 



The first of these, much the largest, we shall pass first under review. 

 But before we can make much headway it will be necessary to give some 

 preliminary consideration to the method by which speech-elements are 

 assigned to these three classes. Nuclear Polynesia was the meeting- 

 place of the two migration streams, and in that central province Samoa 

 is most distinctly the scene of the reunion of the long and widely sun- 

 dered branches of this most errant race. We have most conclusively 

 established that the early, or Proto-Samoan, migration swarmed out 

 from Indonesia through two gateways at, or slightly prior to, the Chris- 

 tian era. That it pursued leisurely courses of voyaging, in the Samoa 

 stream by way of New Britain, the Solomons, Santa Cruz, and thus to 

 the new home in Samoa; in the Viti stream by way of Torres Straits, 

 the New Hebrides, and Viti. That in a movement of convection within 

 Nuclear Polynesia these two streams rejoined and created a settle- 

 ment quite homogeneous save for an anterior Melanesian element in 

 Viti and perhaps in Rotuma. Upon this Proto-Samoan colony of 

 Nuclear Polynesia arrived (an uncertain number of centuries later and 

 by a course which we must positively exclude from the Melanesian 

 traverse, but which otherwise we are wholly unable to identify) a 

 second migration of the same race, the Tongafiti swarm. 



This had so long been sundered from the earlier and isolated colony 

 that independent and divergent development of language had taken 

 place. This half-alien swarm, whencesoever it came, rested upon 

 Samoa for a period whose beginning we have no present means of 

 establishing with accuracy upon our calendar, but which there seems 

 somewhat good reason to assign to about 600 A. D. We have excel- 

 lent agreement of many Samoan annals to adjust the expulsion of the 

 intruders to a period in or about the eleventh century. The Tongafiti 

 conquerors of Proto-Samoan Samoa have left such a record of cruelty 

 that the wise and brave youth who expelled them in the running fight 

 of Matamatame became a national hero and the first of the Malietoas, 

 Savea. Yet there was opportunity during these overbearing centuries 

 for the two stages of the mother tongue to meet and to some extent to 

 mix. This it is which we are to investigate. 



Since the mother tongue was common, a certain and assuredly a 

 large proportion in Samoa of the vocables of the two migrations must 

 be common property. Let us represent that element by symbols abcd. 

 The Proto-Samoan colony, then, would be discovered in the home of its 

 remote isolation speaking a language representable by abcd-EFgh, in 

 which the latter group of symbols may represent ancient and common 



