160 EASTER ISLAND. 



code which carried the message, "Let us sail in company for mutual 

 protection." 



It is only reasonable to suppose that the Proto-Samoan refugees 

 clung together in their flight. We see by these indices of diffusion that, 

 small a party as they were in the Paumotu, they were no larger in the 

 three sister archipelagoes, evidence that when once they had found 

 their asylum they clung close to it. But, if flight concentrates, pursuit 

 is diffuse. We have evolved the idea of the angered Tongafiti in hot 

 haste to overtake the party which has escaped their oppression. There 

 is none to tell them where they may come upon their prey, they must 

 scatter in search, they must go from island to island, from archipelago 

 to archipelago. What else can be read in the greater diffusion of the 

 Tongafiti three times on one scale, three times on the other than the 

 result of this scattering of set purpose of wrath or recovery into the very 

 nooks of the sea? 



We next examine briefly the restrictively provincial member of the 

 table. In this we can indeed be brief. If my theory of the signification 

 of this speech element be tenable we are dealing with the relics of a 

 migration movement so ancient that silent history may not be induced 

 to speak. Time has removed the asperities of the curves, the indices of 

 diffusion do not show any such sharpness of group distribution as is 

 fallaciously indicated by the numerical percentages based on the uncon- 

 ditioned arithmetic of word count. Yet here, as in the other member, 

 we find the Paumotu and the Marquesas in summit association, and 

 Tahiti and Mangareva at the bottom. The same explanation holds. 



In this prolonged study of interpretation of philological data into 

 terms of geography, seamanship, and folk movement we have laid a 

 comparable foundation whereupon we may better adjust the record of 

 Rapanui speech and assign to it its position in the history of this remote 

 province, of which it is the most remote outlier, the extreme limit of 

 Polynesian settlement, the Ultima Thule of the great sea of Kiwa, te 

 moana nui o Kiwa, forbidding, arid, unhospitable, yet the home of Poly- 

 nesians, and naught beyond but sky to the death horizon in the very 

 eye of the wind. 



It will be recalled that throughout this research we have had to bear 

 in mind a large component of each language as to which we lack data 

 enabling us to coordinate it with the elements which we have been able 

 to establish ; I refer to the unidentifiable component. It has seemed to 

 me best to deal with this in two ways. In the specific chapters upon 

 each language I have laid it aside from the computations ; in this final 

 chapter I have included it, thus reducing the size of the severally deter- 

 mined figures. In the preceding pages of this chapter in the discussion 

 of the extra-Rapanui element I adopted as a net stock that element with 

 the unidentified element. Now for the computation of the Rapanui 

 affiliates I include once more this unidentified element, deriving a new 



