RAPANUI SOURCES AND VARIETY. 41 



speech material which in time had been lost by the separated branch 

 of the family, or which had been acquired along the Proto-Samoan way, 

 the latter hypothesis on many accounts being scarcely tenable. 



Centuries later arrives upon Samoa the Tongafiti swarm, speaking a 

 language representable by abcd-ijkl, the respective symbols bearing 

 similar explanation. Now if we find our present Samoan to consist of 

 abcd-EFGH-i, and our present Maori to consist of abcd-e-ijkl, we 

 have no hesitation in ascribing i and E respectively to accumulation 

 during contact of the two swarms when convection movement was pos- 

 sible. The greater share of such contact is to be attributed to those 

 centuries of association, even though violent, in Samoa; a lesser share is 

 to be attributed to contact at distal points of migration in which the 

 later comers found an earlier settlement of the older swarm, in support 

 of which we have not only the deductions of philological analysis but 

 the consenting record of history when we learn to interpret annals of 

 the genealogy of this race. 



Accordingly, if in this province of Southeast Polynesia we encounter 

 a speech element of the type abcd-eh we shall be justified in assigning 

 it to a direct migration from Nuclear Polynesia of Proto-Samoans to 

 this natural limit of all successful migration. If, similarly, we find a 

 speech element abcd-ik, we shall assign it to a Tongafiti migration. 

 This it is which we shall now examine. The first group in the table 

 (items 293-728) represents abcd, the common element. These are all 

 satisfactory form identifications; the inner content of sense will point 

 the more definite assignment of deviation forms to one or to the other 

 branch of the family. The material here grouped is of very uniform 

 concord. Where variety superficially appears the notes appended in 

 the vocabulary to each such item point out the substantial agreement. 

 In a few cases, where the reduction to uniformity of signification is 

 found impracticable, the compared data show that in general these 

 instances upon closer study are more properly to be assigned to one 

 or other of the separate migration streams. 



When we turn to the list of identifications which are chargeable to the 

 Proto-Samoan source and which show no contamination along the way 

 of the sea or in this distant terminus of migration, we find, however, a 

 marked difference. The table shows that we are dealing in this class 

 of data with 116 items (729-839 of the finding table). Of these no less 

 than 3 1 show such sense deviations as call for the particular study which 

 has been recorded in the notes appended to each such item in the vocab- 

 ulary. The variant stems are thus listed: 



From this list we are obliged to remove those items in which our dic- 

 tionary material is either insufficient in sum or else lacking in precision 



