DETERMINATION OF THE PLACE OF RAPANUI. 



153 



element of speech which is recognizable backward along the migration 

 course. Mangareva shows the three minima of this element, 4 and 3 

 and 3; the percentages of the three items are compassed within the 

 range of a single per cent ; that one added point is given to the general 

 Polynesian item and therefore ceases to be of value in our research ; the 

 lesser figure applies alike to the Proto-Samoan and the Tongafiti 

 streams. 



Tahiti comes next in the advancing order. Its general Polynesian 

 content remains the same as in Mangareva; its Proto-Samoan content 

 remains the same; its Tongafiti content is double that of Mangareva. 



Third in order is the Marquesas. Retaining the same general Poly- 

 nesian figure, it owes one point more to the Proto-Samoan, the highest 

 figure which that element attains in the province ; it falls but one point 

 below Tahiti in the higher ratio of the Tongafiti. 



But when we reach, last of all, the Paumotu record we find the ord- 

 erly table with its flat curves thrown into marked confusion. With 

 Mangareva and Tahiti it has the minimum showing of Proto-Samoan 

 influence ; its general Polynesian content is the highest, yet not strongly 



Table 32. 



removed from the general average wherein the three other archipelagoes 

 inosculate. But in the Tongafiti columns it peaks remarkably; it has 

 more than three times as much Tongafiti material as Mangareva, its 

 southern and next neighbor; twice as much as the Marquesas, very 

 nearly twice as much as Tahiti. 



Now apply these records of figures to the known geographical base. 

 The Paumotu are a chain of which Mangareva is a loose link at the 

 south, Tahiti is the nearer neighbor to leeward of the northern part of 

 the chain, the Marquesas the nearer neighbor to windward. Assume a 

 migration from the west on the wind along the course which would 

 bring the fleet toward the northern Paumotu. The chain in its great 

 extent would capture the largest numbers of the fleet, the leeward 

 neighbor would hold the next largest, those that by too much northing 

 overran Tahiti would show in the Marquesas the next largest number, 

 while Mangareva, away to the south, would get but the stragglers and 

 the dull sailers. Thus our knowledge of winds and currents and canoe 

 seacraft provides a probable explanation. But this variety is note- 



