DATA AND NOTES. 357 



The Polynesian word varies but slightly, and that normally, from kilt. 

 As it is in use solely with the value of a noun and hence needs no formative 

 affixes which might protect a radical terminal consonant as in verb-deriva- 

 tives, we have no means of proving a kilit stem. Indonesia exhibits a stem 

 which differs in the former vowel and preserves a final consonant, kulit. 

 Melanesia sometimes approaches the Polynesian, other times the Indonesian. 



Of the Polynesian type we note Ambrym and Sesake, both of which have 

 abraded the initial k, and in neither is it usual to drop that letter, Lo pre- 

 serves the vowel in gilit and at the same time has the Indonesian /-final. 

 Codrington (Melanesian Languages 90, note) suggests as a possibility that 

 Mota wil to peel is homogenetic ; in his Mota dictionary, however, he defines 

 with great precision "to turn round horizontally, to peel, turning the fruit 

 over in peeling," this in sense and in form bringing the word into close 

 association with vili to turn round ; the k-w mutation in Mota is suggested 

 but once, kutu (306) a louse Mota wut. 



The two forms in Efate are not wholly satisfactory as transition phases, 

 for they would involve the unsupported k-w mutation. Still we do find 

 kuli, a frequent Melanesian and altogether Indonesian form, alternative 

 with uili, which seems to offer some possibility of a connection with the 

 Polynesian kill. 



The kuli form without consonant termination is found in Viti, Rotuma, 

 Epi (Bierian and possibly Baki), Murray Island, Nggao, Bugotu, Nggela, 

 New Georgia, Savo, Vaturanga, Wango, Malo, and Santo. There is a con- 

 siderable variation in these forms, but they are to be connected with this 

 stem. In Baki kulukuti we may have a composite of which kulu alone 

 pertains to this group. In Moanus we have a distinctly Indonesian form; 

 geographically it is within the feasible limit of Post- Polynesian Malayan 

 intercourse ; yet Lo, almost too far south for this explanation to be consid- 

 ered valid, has the /-final in gilit; and Eromanga no-kohit-an is suggestive, 

 although we have no other instance in that language of l-h mutation. 



All the Indonesian identifications are of the kulit type, except that 

 Madura and Savu have either lost or have not acquired the /-final. 



The Arabic shows resemblance to the Indonesian type as to consonants, 

 to the Polynesian as to vowels. 



306. 

 kutu, the louse. 



Tonga, Futuna, Uvea, Fotuna, Marquesas, Mangareva, Rarotonga, 

 Rapanui, Nukuoro: kutu, the louse. Niue: kutu, the louse, 

 the flea. Maori: kutu, the louse, human vermin in general. 

 Samoa: 'w/w, the louse; 'utufiti, the flea. Paumotu: gutu, the 

 louse. Tahiti: utu, id. Hawaii: uku, a small insect. 



Viti: kutu, the louse. 



Deni, Nada, Kiriwina, Mukawa, Raqa: kutu, the louse. Murua: 

 kuti, id. Arag, Nggela, Bugotu, Sinaugoro: gutu, id. Kiviri, 

 Oiun: guta, id. Vaturanga: ngotu, id. Marina, Lo, Malekula 

 Pangkumu: gut, id. Hula, Keapara: gu, id. Aneityum: 

 get, id. Tanna : kiget, id. Motlav, Volow: git, id. Makura: 

 na-kit, id. Nengone: ote, id. Buka: autu, id. Motu, 

 New Britain, Malo, Rubi, Dobu, Awalama, Taupota, Wedau, 



