POLYNESIAN RELICS IN MELANESIA. 137 



K-v. For this we find the single instance of kidl (194) clog Malo vuria 

 Tangoan Santo vuriu. It will be observed that the area of this 

 leaping mutation is included in the but scantily more extended 

 area of the similar ng-m mutation. 



K-w. This very interesting mutation, not only a leaping mutation 

 but as well from a consonant to a semivowel, while most firmly 

 validated, rests upon the single instance of kutu (306) louse Omba 

 (etc.) wutu Merlav (etc.) wut Pak (etc.) wu. The area of its 

 occurrence includes all of the ng-m k-v area of the Banks Group 

 and the northern New Hebrides, with a yet further extension in 

 Sesake to the central tract of the latter archipelago. 



8. Labial to palatal : 



Ni-ng. In a single instance, malum (316) Lo melunglung Alo Teqel 

 mulunglung. It lies in the same Banks Group area of anomalies. 



F-ng. For this, too, we have but one example, fiha (215) how many 

 Nggao ngiha Vaturanga ngisa. It lies in an area of marked 

 anomalies, in fact with the k-h and n-m areas it completes a 

 triangle in southern Ysabel and New Georgia and northern 

 Guadalcanal' of the Solomons, within which lies Savo of a far 

 different linguistic character. 



F-k. This rests as yet uncertainly on ufi (273) yam Tanna nuk, and 

 then only if n function as article with a substantive uk, n-uk. 

 We note in uncertainty ifo (206) down Aneityum suko Gog sug. 

 The mutation in nojo (201) to sit Nggao nokro is wholly abnormal. 



9. Finally we have left for consideration a half dozen variants which 



elude the foregoing effort to find order in irregularity. 

 NQ-nj. Noted in matangi (274) the wind Aneityum ni-mtinjop. It 



is possible that m(i)tin is the survival of the stem matangi and 



jop an accretion of some sort ; I have been unable to identify jop 



as an independent word or in other composition in Aneityum. 

 N-ny. This occurs in namu (328) mosquito Aneityum inyum Moanus 



njam. Buka and Bugotu have already been noted as varying 



this n to gn. 

 S-y. In sulu (182) torch Baki yidu. 

 T-dr. Found in to'a (375) to subside Ambrym dro dru to abide; 



tutulu (367) to leak Baki drudruli. Associated herewith is t-ndr 



talinga (350) ear Moanus ndrilinga, tahi (352) sea Moanus ndras. 



In this reinforced r we have elsewhere found evidence of the 



effort to reproduce r grasseye. 



The result of this inspection of the anomalies in mutation is that 

 we identify two distinct areas in which Polynesian material was 

 rudely subjected to purely Melanesian methods, one area somewhat 

 diffuse in the northern New Hebrides archipelago, including the 

 Banks Group and Torres Islands, the other sharply defined on the 

 larger islands in the mid Solomons; and that in at least the two 

 more numerous of these anomalous mutations there seems an inter- 

 relation between the two areas. 



If the result of all this painful examination were no more than the 

 circumscription of these two small areas, interesting as that result 



