366 THE POLYNESIAN WANDERINGS. 



We shall then examine these eight-and-twenty languages. Four lan- 

 guages seem to point one way, two dozen point the other. The twenty-four 

 congruent languages have the five- lima in its clearest form, and hand-lima 

 has undergone some form of secondary mutation, the nature of which is 

 indicated in the remarks. 



The four opposing languages are now to be examined for the ascertain- 

 ment of the value of their testimony. Tanna is an island remote from either 

 of our migration streams, the quality of its Polynesian content is absolutely 

 the lowest in my scale. The five-lima and the hand-/iwa are alike obscured 

 by alien admixture, so much so that they are barely recognizable. The 

 testimony of Tanna, therefore, should not for one moment count against 

 the agreement of twenty-four. 



The Duke of York has not only Urn for five, but also lima. The posses- 

 sion of the latter is sufficient to remove this speech from this consideration ; 

 it belongs in the outer ring in which the five word and the hand word are 

 the same lima. 



Laur and Lamassa (and King, despite another variation element, may be 

 associated therewith) are really the only evidence in opposition. What do 

 these witnesses amount to against a principle established in our triple area, 

 in Polynesia, in Melanesia, in Indonesia? They are three languages, dia- 

 lectic variety at best, spoken on the New Ireland coast in the very jaws of 

 the eastern gateway. The source from which I derive the vocabularies con- 

 tains also a social register, a city directory, of two of these abodes, or huts, 

 of culture : Lamassa has a population of 23 men, 28 women married to them, 

 and 40 single persons including children ; King is peopled by 38 women and 

 girls, 30 men and boys. Not many in the census are these who reverse the 

 system of the whole Pacific. 



On the whole we may disregard such exceptions entirely. Therefore I 

 am willing to aver that in every case where five-/tmo and hand-lima differ 

 it is the latter that is secondary in form ; therefore five it is which is the 

 primordial sense; the hand is lima simply by virtue of its possession of 

 five enumerable fingers. 



The following notes have to do with the varieties in form of the five-lima. 



In Baki and in some other unnamed dialect of Epi the o of jimo, limo, is 

 explicable on my theory of the neutral vowel; this vowel change is found 

 in Basakrama, Saru, and the Togean Islands in Indonesia. The l-j muta- 

 tion occurs also in Aneityum ni-jman. The m-n mutation in Marina and 

 Tangoan Santo Una is found again in manu (317) bird Marina nanu, and 

 mata (324) eye Marina nata. Eromanga shows a /iw-composite with the 

 element sik-suk-suo. Motu ima finds its only congener in the extreme 

 east, not in Tahiti this time but in theadjacent archipelago of the Marquesas. 

 Aneityum, with n-article and attraction of the nearest vowel, yields two 

 forms, jiman and kiman, each with the n suffix. The former has already 

 been discussed ; the l-k mutation is found only once again in lima (313) hand 

 Nggao kame Vaturanga kima. We now find in the northern New Hebrides 

 a group of lima forms prefaced by tava in no less than four vowel variants 

 and a fifth in which the initial t has been abraded. Not a word of expla- 

 nation is offered in any of Codrington's disquisitions upon these numerals. 

 I venture the suggestion that the added element may find some illumination 



