DETERMINATION OF THE PLACE OF RAPANUI. 151 



Why, then, should the Paumotu differ in so marked a degree from 

 the other archipelagoes of the province? The answer is simple. The 

 Paumotu are spread over their central sea that is to say, inner with 

 respect of the limiting points of the province in a loose chain which 

 facilitates sailing from island to island, no difficulty at all to such navi- 

 gators as the Polynesians. At its southern links it lies close enough to 

 Mangareva for interchange of visits ; at its northern end it lies in close 

 touch with Tahiti and the Marquesas. Individually these three archi- 

 pelagoes stand at the two-thirds mark of unrecognizability ; the Paumotu, 

 enjoying intercourse with all three and then convectively diffusing its 

 better education throughout its own extent, stands at the highest point 

 of recognizability, as nearly as possible at the half-way mark. 



In conclusion of this disquisition upon the unidentifiable element 

 within the province, let me repeat my belief that it is as purely Poly- 

 nesian as any that we know. We lack data, of course, whereupon to 

 consider its assignment to Proto-Samoan or Tongafiti source ; I have no 

 hesitation, however, in holding a personal opinion, on grounds wholly 

 a priori, that we should expect to find in such remote lurking-places 

 material carried away from their old home by Proto-Samoans making 

 their escape from Tongafiti highhandedness. This is not put forward 

 dogmatically; really it is no more susceptible of disproof than of evi- 

 dential establishment. In fact, after all these pages of figures and 

 painful proof I may fairly claim my reward in the happy expression of 

 an opinion for which I am well aware no proof could ever be adduced. 



Unto this end, through it, indeed, and to the very end of this end, I 

 still maintain my division of the identifiable material into the two 

 classes of that which has Rapanui affiliates and that which is extra- 

 Rapanui. Proceeding now with our examination of the identifiables, I 

 shall direct the attention first upon the latter. 



At this point we shall find it convenient to introduce Table 30 showing 

 the extra-Rapanui element of the province in strict parallelism with 

 the earlier table (28) of the Rapanui affiliates. 



It will be understood that these identifiables are not of equal value, 

 that a vocable of the Paumotu recognizable in Mangarevan alone does 

 not properly function as of the same power as one found in all four of 

 the languages under this study. To this point we shall presently recur, 

 but first we must present in Table 31 the record of the sums and the 

 percentages of the extra-Rapanui identifications in the four languages. 

 The percentage is based on the remnant figure of the gross vocabulary 

 after subtraction of the sum of the Rapanui affiliates, which we are to 

 keep rigidly apart as a distinct element. 



When we compare this with the undivided results in Table 29, we find 

 that the same relative order obtains ; the percentages have been reduced 

 by an amount lying between 7 and 10. The larger figure for the Pau- 

 motu shows that in this particular group of figures the archipelago 



