324 THE POLYNESIAN WANDERINGS. 



We now enter upon a great group of forms in which the primal aspiration 

 has undergone greater mutation. The majority of these mutations are 

 in the lingual series, a proof that the original is the h near the lingual 

 closures. In this series we shall consider the weakening mutations, those 

 upward in the vowel direction, disregarding the initial accretion of h and 

 the formative terminal n. 



ula. Gani, Wahai, Salu, Timor, Visayas, Tobo, Cajeli, Caimarian, 

 Bual, Awaiya, Liang, Morella, Batumerah, Amblaw. Mysot 

 golim may be considered in this series, but with no great insistence. 

 ura. Ende, Tidore, Pampangas, Ceram, Salibabo, Malagasy, Galela, 

 Baju, Teor, Lariko. 

 We next examine the strengthening mutations, the downward movement. 

 The first of these is usa, already presented as a normal variation. 

 udja. Togean Islands, Malay, Sandol. 

 uda. Rotti, Matabello, Kaili, Java. 

 We have a single example in which occurs h-ng mutation ; this represents 

 a resultant of the first weakening variation to ula and then the horizontal 

 strengthening by movement to the palate, l-ng being a mutation of which 

 the tabular view will supply examples in Melanesia. These are to be 

 regarded as specifically Indonesian mutations; they are not found at all 

 in Polynesia, and h-j and h-d very sparingly in Melanesia outside this stem. 

 We shall now resume the consideration of the Melanesian material. The 

 ula-ura which has occupied so large a space in Indonesia may with consider- 

 able probability be identified in Buka and Bougainville urata and variants. 

 In case this identification be considered acceptable I would ascribe it to a 

 Post-Polynesian period, a raid along the ancient track conducted by Indo- 

 nesian peoples, for these two northern outliers of the Solomons are within 

 reach of the prahus. In note 277lhave already introduced this explanation. 

 But Gog urei and Arag-Merlav reu, metathetically derived therefrom, lie 

 geographically too remote to be susceptible of this explanation in view of 

 the fact that no intervening record appears. But is such the case? 



In Buka uruotta, readily seen to be a variant of urata, we find the pos- 

 sibility of a transition form, a stem in which the final / is established for 

 this area. It is possible that in further borrowings uruotta, losing its first 

 two letters (not a true syllable), might, as uotta, lead us to weta of Lo, wet 

 of Sasar and Alo Teqel, wat of Pak. From wet to Volow wend is easier than 

 it might seem, for wet varying toward wed would encounter the idiosyncrasy 

 that Volow and several other Melanesian languages, as well as Viti, can not 

 essay the d without the supporting preface of the nasal of the same series. 

 From wend to wen and wena is very simple on the explanation of secondary 

 borrowing, as already pointed out in the case of h-th-t, the reinforced con- 

 sonant being objectionable to later borrowers, and on that score they reject 

 the final one of two linked sounds in their ignorance that this it is which 

 represents the primal sound. Thus we may bring the second group of 

 Melanesian identifications into line, representative of a Post- Polynesian 

 movement. 



The third Melanesian group, Wango-Fagani rangi and Nggao hangi, is 

 the Polynesian langi the sky, which we have already seen (274) pressed into 



