48 THE POLYNESIAN WANDERINGS. 



ward drift our comparison of the Polynesian of Efate with that of 

 Nuclear Polynesia gives us warrant to say that it is highly improb- 

 able that Efate" received its Polynesian from the east; it gives us 

 no warrant to say that it is impossible. 



C. 



It may be difficult and it may remain difficult to prove that the 

 Polynesian in Melanesia did not come there from Sawaiori settle- 

 ments in the central and eastern Pacific ; the proof of a negative by 

 no means loses any of its logical difficulties when stated in terms 

 of an unlettered people. But there is all the proof that consistent 

 tradition can give to the migration theory which introduces the 

 Melanesian chain as the line along which the migrant fleets passed 

 from the known sojourn in Indonesia to the known occupation of 

 Polynesia. The presence of so large a share of Sawaiori vocables 

 in Efate" does not in itself prove that this island lay on the fairway 

 from Indonesia to Savai'i, but it is hugely confirmatory of all other 

 evidence upon which this theory is based. We are here really not 

 so much concerned with the proof of this migration theory as we 

 are with the determination of which swarm it was that included 

 Efate in its voyaging. 



In the Proto-Samoan we have a distinct, and surely an early, 

 phase of Sawaiori speech which differs from the Tongafiti in the 

 possession of certain vocables which the latter has not — in the 

 absence of certain vocables which the latter has preserved or has 

 acquired since the separation of the two stems. Above all else the 

 Proto-Samoan is distinguished from the Tongafiti by its maintain- 

 ing superficially, or recoverable under the surface by the merest 

 scratching, the remnants of the original speech in which closed 

 syllables were possible. 



Everywhere in this Polynesian of Efate we find the closed syllables 

 of the Proto-Samoan, not the Tongafiti softening away of the final 

 consonant. This is so common that it has not seemed necessary 

 to collate for this feature, it has been sufficient to insert the Proto- 

 Samoan radical in each item of the collected data where it has been 

 possible to recover it. On this score we have no hesitation in asso- 

 ciating the Polynesian content of Efate with the Proto-Samoan. 



In consideration of the vocabulary we are dealing with 324 items. 

 Of these there are but 1 2 which are not identified in Nuclear Poly- 

 nesia; if we include Viti in Nuclear Polynesia, as seems quite 

 justifiable, we shall reduce the reckoning to 9. Now 9 in 324 is 

 readily explicable by the known habit of Polynesians to cast out 

 certain words under the manifold working of their tabu.* Against 

 these 9 is to be set the fact that of our 324 identifications 20 per 



*See note 255. 



