DATA AND NOTES. 323 



Bierian : nihua. Arag, Omba : uhe. Aneityum, Eromanga : ehe. 

 Guadalcanar : utha. Alite : uta. Santo, Vate, Malekula : us. 

 Lakon: uh. Ambrym: o. New Ireland (Duffield) : eyus, eus. 

 Rubi : ura. Buka : urata, uroto, uruotta. Bougainville : urata. 

 Gog: urei. Arag, Merlav: reu. Pak: wat. Sasar, Alo Teqel: 

 wet. Lo : weta. Volow : wend. Norbarbar, Vuras, Mosin : wen. 

 Mota: wena. Wango, Fagani: rangi. Nggao: hani. 

 Kayan: usan. Togean Islands: udjan. Rotti: udan. Matabello: 

 udama. Kaili : uda. Malay, Sandol : hujan. Java : hudan. 

 Gani, Wahai, Salu, Timor, Visayas: ulan. Tobo: u'lan. 



Cajeli, Caimarian: ulani. Bual: ulanu. Awaiya: uldne. 

 Liang, Morella: hulan. Batumerah: hulani. Amblaw : ulah. 

 Ende : ura. Tidore, Pampangas : uran. Ceram : urana. Sali- 

 babo: urong. Kisa: ungang. Gah: uan. Malagasy: orana. 

 Galela: hura. Baju: kron. Teor: hurani. Lariko: haran. 

 Sula: huya. Bolanghitam : oka. Mysot : golim. 

 Arabic: ba'a, to rain continuously; ba'a'a, rain, rain water. 

 The Proto-Samoan stem is uha. This is shown by Viti utha, which might 

 derive with equal facility from uha or usa, but if the latter were radical 

 Samoa would have preserved it as usa, and it would have appeared as uha 

 in the long list of sister languages which represent a Proto-Samoan sibilant 

 by the aspiration. It does appear as uha in Tonga and Niue, which nor- 

 mally preserve the Proto-Samoan h. In Rotuma it has become usa and 

 has then undergone metathesis to comply with the local idiosyncrasy for 

 closed stems. In Efate the prefixing of b is unexplained. 



The Proto-Samoan uha is preserved in Vaturanga, Bugotu, Nggela; with 

 one vowel change in Arag and Omba, with alteration of both vowels in 

 Aneityum, and with terminal abrasion in Lakon. Guadalcanar repeats the 

 Viti utha. The mutation to s is seen in the usa of Sesake, Marina, Tangoan 

 Santo, and Maewo; the us by abrasion in Santo, Vate, and Malekula; prob- 

 ably in Duffield's New Ireland eyus, eus. The ua of the modern phase of 

 Polynesian is found in Deni and Epi, and in the latter island both Baki 

 and Bierian preface it with weak sounds. The uta of Alite is a degeneration 

 form of Guadalcanar utha arising from a second borrowing of loan material 

 and without knowledge of the true form of the original. This is no merely 

 theoretical deduction. In Fiji I have observed that Melanesians in learning 

 the Viti almost uniformly reproduce the dh of that speech with t, d, nt, nd, 

 where the sound (though the Fijians had no knowledge of the fact) was the 

 preservation of Polynesian stems containing the aspiration. The o of Am- 

 brym is explicable as a degeneration form to which we are led by Lakon uh. 

 We must postpone further consideration of the Melanesian material until 

 we have examined the Indonesian record. We shall compare these forms 

 with the primal and successive forms hereinbefore developed. 

 uha. This appears in but a single instance, Bolanghitam oha. 

 usa. This, too, is found but once, Kayan usan, in which appears the 

 final Indonesian n which is so prevalent in this area and will not 

 be further noticed here. 

 ua. In Gah uan and in Sula huya with the prefix of the aspiration 

 which is found in Bierian ni-hua and freely in Indonesia. 



