DATA AND NOTES. 315 



tion after sunrise, when, in the clear light of day, the primal sense is lost. 

 The vSikayana rehu lime, a specialized powder, finds a far-eastern parallel 

 in Tahiti rehu, which is distinctly applicable to any powder. 



In Efate we find efu {abu, afu, au) and lefu (libu, lifu); mafu is not to be 

 correlated with any of the Polynesian stems. 



Viti ndravu, both dust and ashes, recalls the vowel found in Efate' abu, 

 and at the same time the ndr is evidential that the Proto-Samoan stem 

 was ref u with the r grasseye (Samoan Phonetics, 1 7 Journal of the Poly- 

 nesian Society, 152). 



In Melanesia we shall institute our search for the simple stem. I have 

 already remarked that this was efus. This enables us at once to pick up 

 theVanua Lava forms, the uwus of Pak, Sasar, and Alo Teqel; the s, in fact 

 the us, is clear at first sight, the f-w mutation is confirmed by a widely 

 extended range of examples. Mosin tuwus is this identified uwus with 

 frontal accretion of some sort. Vuras is but a few miles from Sasar along 

 the same beach, therefore wowo is readily acceptable as a dialectic variant ; 

 the terminal ^ has dropped off and the then final wu has undergone modifi- 

 cation to wo and later duplication. Motlav has the same form, and Lo 

 wowa but slightly differs. Having now established wo as the vestige of efu 

 we find it a composition member of puwo of Norbarbar and tarowo of Gog 

 and Mota. Starting afresh with efu, we find it in Efate afu, abu; in Marina 

 avuavu; as a composition member in Arag, Lakon and Maewo forms ; having 

 lost its labial we have au in Efate and Sesake, this identification being 

 open to such doubt as our lack of cordiality to internal loss may warrant. 



We next examine the lefu stem. In Efat6 lifu and libu are clear. Ravu 

 is well distributed, both independently and in composition, rafu, rahu, lavu, 

 and in Mota rav. 



The Duke of York kabu ashes is to be rejected, but I have left it with 

 this material because of its reappearance in Viti kambu vapor, which is one 

 of the senses pertaining to this protean efu stem. 



Our Indonesian material is scanty, but efu and lefu stems are recognizable. 



In the Semitic offered for our consideration it must be acknowledged 

 that there is a certain resemblance, but the disclosure of a final t where our 

 Polynesian has the ^-terminal is enough to stamp this as not a family 

 resemblance. 



273- 



(a) 5fa ki, 5fa ki, to bury. 



Samoa : iifi, a lid, a cover, to cover, to conceal ; ufita'i, to cover with. 

 Tonga : uufi, a cover, to cover, to overspread ; ufiufi, to cover, 

 to conceal. Futuna: uufi, to cover, to hide, to conceal, to 

 disguise ; ufia, covered ; ufiufi, to cover things without envelop- 

 ing them. Uvea: ufi, uufi, to cover. Niue: iifi, a covering, 

 to cover up, to conceal. Maori: uhi, uwhi, to cover, a cover- 

 ing. Hawaii : uhi, a covering, a veil, to cover over a thing so as 

 to hide it. Mangareva : uhiuhi, to hide, to cover. 



Viti: umbia, to cover over; umbi, the top, lid, cover, a quilt. 



Malagasy : afina, to be concealed ; manafina, to conceal, to bury. 



Arabic: u aba, to conceal, to be concealed, to bury. 



