DATA AND NOTES. 347 



The Indonesian identifications are satisfactory. 



In the Arabic offering there is far too much, and of what does exhibit 

 some resemblance, 'd-win-at, the alternative form sacrifices a half. Further, 

 the change is not in series, which we have come to regard as essential in our 

 speech family, but in w-y, a two-column leap from labial proximity to palatal 

 proximity. No matter were the superficial resemblance even greater, we 

 could not accept an identification which so far violates the instinct of 

 Polynesian phonetics. 



297. 



ngil i, kil i, kili, to dig. 



Tonga: keli, to dig, to sink, a dyke, a trench, a ditch; jekeli, to 

 paddle quickly. Niue, Uvea: keli, to dig. Samoa: 'eli, to 

 dig, to pull hard in paddling. Hawaii: eli, to dig in the 

 ground. Maori: keri, to dig, to rush along violently. Manga - 

 reva, Nuguria, Rapanui, Paumotu: keri, to dig. Marquesas: 

 kei, id. Nukuoro: keni, id. Tahiti: eri, to undermine; fieri, 

 to dig a hole, as a rat or a crab ; ari, to scoop out the earth 

 with both hands. 



Viti: keli, a ditch; kelia, to dig a hole. 



Santo: keli, to dig. Kiriwina: kelikeli, id. Sariba: keri, id. 

 Malo: cele, id. Nggela: gelt, id. Motu : gei, id. Solomon 

 Islands: eli, id. Makura: nggili, id. New Britain: kir, 

 kire, id. Baki, Bierian: mkili, id. Mota:^z7, id. Malekula: 

 kiri, id. Keapara: gia, id. Misima: giar, id. Ambrym: 

 gali, id. Kwagila: karo, id. Taupota : garai, id . Wedau, 

 Galavi: g'arai, id. Awalama: haraia, id. Tavara: halaia, id. 

 Dobu, Kiviri, Oiun: sara, id. 



Malay: #a/t, to dig. Malagasy: hadi, id. 



Arabic: kara', karw', to dig. 



In the Polynesian section Nukuoro keni (l-n mutation) is a common 

 variation. 



In the broader aspect the variants in the three island areas are the usual 

 changes of k and /. The former vowel is somewhat critical ; in Polynesia 

 and Viti it is e throughout, in Indonesia (one sound identification, one 

 debatable) it is a; in Melanesia the characteristic form is in i. Variants 

 from the Melanesian kili stem are : in the Polynesian direction Nggela, Santo, 

 Malo and the Solomon Islands (?); in the Indonesian direction Ambrym, 

 Efat6, and several New Guinea languages. 



It would be better if we had transition forms to assist in the Malagasy 

 identification, but its changes are in line with Polynesian phonetics and are 

 no bar to acceptance of the form. 



The Semitic here is at least to be accepted as a resemblance, for in the klh 

 triliteron the final consonant might readily pass from a Polynesian lan- 

 guage, yet it must be said that it would more probably be retained as s. 

 There is no evidence to show that the Proto-Samoan stem was ever other 

 than keli, that is the root kel with the suffix of the verb-formative i. Even 

 when reduced to kel the Semitic resemblance is not abolished. 



Item 303 should be included with this. 



