POLYNESIAN RELICS IN MELANESIA. 135 



1. Involving the aspiration : 



k-h h-v x-h h / i-h p-h 



Superficially these are extra-serial mutations. If, however, my expla- 

 nation be valid that the aspiration should be regarded as close to each 

 series it will be permissible to regard it as the decay stage of each of these 

 columns. The f-h change, too, is frequent in Polynesia. 



K-h. This rests upon five instances, one triply and two doubly sup- 

 ported. In mataku (258) we find Ambry m matehag Malekula 

 metoh Vaturanga matahuni. In kafika (193) Malekula Pangkumu 

 has havih, two instances in one word. In ika (300) we find New 

 Georgia thani Malekula na-ih. In feast (301), noting the existence 

 of a second stem hai, we find New Georgia hai Vaturanga hai. 

 There are two widely separated foci of the mutation, Ambrym- 

 Malekula, leeward islands in the central New Hebrides ; Vaturanga 

 and New Georgia, leeward in the central Solomons. 



H-/. For this we have the single instance of hala (339) path Aneityum 

 ne-falaig. The language is not very satisfactory and no great 

 value may attach to this unique instance. 



H-v. For this we have but a single example, Sesake vinaga (169); 

 and in this there is uncertainty as to whether the Proto-Samoan 

 was aspirate or sibilant. 



\-h. This, the converse of the next preceding item, rests upon a 

 single instance. In lava (307) we find Saa laha and Wango raha. 

 The occurrence of this mutation is in the southern Solomons on 

 either side of the straits which part San Cristoval and Malanta, 

 not far from the northern k-h focus. 



P-h. This rests upon the single instance of vula (284) in which we 

 find Wango hura. This lies within the focus of the preceding 

 item. 



2. We have, then, two distinct and distant foci in which there is a ten- 

 dency to reproduce certain of the Proto-Samoan consonants by the aspira- 

 tion. The islands on which this occurs are large islands, with the possible 

 exception of Ambrym, which is near the dividing line between the large and 

 the small. 



By far the larger group of the anomalies in mutation is that in 

 which there is clearly a passage from one series to the next in order. 



3. Lingual to labial: 



L-ra. This rests insufficiently upon the single instance of mala ma 

 (322) light Bierian ma ma ma. 



N-m. This rests on a single doubtful instance, arms (351) to spit, in 

 which we find Nggao misu and New Georgia kamisu. It occurs 

 in the central islands of the Solomons. 



T-m. If this be a valid mutation it rests upon but a solitary word, 

 tama (217), which in Omba, Gog, Alite and New Georgia becomes 

 mama, and in Merlav, Lakon, Pak, Sasar, Vuras, Mosin, Alo 

 Teqel, Motlav, Volow and Norbarbar mam. The argument for 

 this mutation will be found in the note upon this item in the 

 systematic study of the data. This again has two foci: one, the 

 Banks Group (omitting Mota), dipping down to Omba in the 

 northern New Hebrides; the other in the central Solomons. 



