DATA AND NOTES. 241 



Nowhere in Polynesia can I find a meaning of till which would shed any 

 light on this composite, and I am forced to the conclusion that this element 

 has passed out of independent existence. The presence of the composite 

 in Aniwa, Fotuna, and Fila argues its great age as a compound. I am very 

 loth to accept an Indonesian identification except it be supported by a 

 satisfactory chain of evidence. To present such a resemblance based upon 

 a single instance goes against my practice in the present work. Yet with 

 this deprecation I note the Macassar djili the lightning flash. Faitili then 

 might mean that which makes the lightning. It may be that we shall find 

 a trifle in geographical support of this suggestion. Macassar is the region 

 in Celebes which lies in closest proximity to the island of Salayer, a name 

 which seems not remotely preserved in the honorific fa'alupenga of Samoa 

 in the high phrase for all Tutuila. 



Tuloutia a 'oe le motu o Salaia. Saving thy grace, island of Salaia. 



167. 

 ulumwa, a pillow for the head. 



Samoa : alunga, alulunga, a soft pillow. Nuguria : aruna, headrest. 

 Tonga : olunga, pillow. Futuna, Uvea, Niue : ulunga, id . Maori, 

 Mangareva: urunga, id. Hawaii: uluna, a pillow, to sleep on 

 a pillow, to tie up a bundle for a pillow. Paumotu : r urunga, a 

 pillow. Tahiti: urua, turua, id. Rapanui: rangua, id. 

 Dufaure Island (New Guinea) : unua, a pillow. New Britain : ulula- 

 lag, id. Mota Maligo : ilinga, head-rest, pillow. Mota Veverau : 

 ulunga, id. 

 Mahri: here, haroh, hare, eres' , the head. 

 Dr. Macdonald repeats and again repeats the hint that this is a derivative 

 of ulu the head. Tregear notes that the word is probably connected with 

 uru the head, turn, and runga. 



This can only apply in those instances which have initial u, Futuna, Uvea, 

 Niue, Maori, Mangareva, Hawaii, Tahiti, Mota Veverau. It leaves unsatis- 

 fied Samoa, Tonga, Paumotu and Mota Maligo. The Samoan plural is 

 proof that the word is a composite of which the latter element is lunga. 

 The former element is most commonly u, but a, o, and possibly i occur. The 

 Paumotu rurunga is accordingly a preduplication of the second element and 

 the common first element does not appear in it at all. If, then, lunga inde- 

 pendently or in composition carries the sense of headrest it will be seen 

 that the stem can not be ulu the head. The alternative Tahiti turua is 

 anomalous on either theory of the word. The Rapanui form is susceptible 

 of explanation by double metathesis. 



Mota Veverau ulunga is of the general type. I incline to regard Mota 

 Maligo ilinga as a mere variant of local dialect. The only way by which 

 Dufaure unua can be brought into correlation is by establishing the loss 

 of n and the l-n mutation. Each of these changes is abundantly manifested 

 in Melanesia and Polynesia; whether they are current in Dufaure Island 

 the only other word we have from that speech (ama (340) outrigger Dufaure 

 sarima — unsatisfactory) does not allow us to judge. The New Britain 

 ululalag can be correlated only as a compound of ulu head and lalag equiv- 

 alent to lunga with reduplication, and there is no evidence forthcoming to 

 support the latter, while the former is distinctly contraindicated. 



