480 THE POLYNESIAN WANDERINGS. 



composites in Awalama and Tavara ; Motu and Sinaugoro of this stem are 

 initial composites in a class by themselves; Mukawa and Raqa are com- 

 posite initial and final, the latter not identifiable but probably not mata, the 

 former showing the ula component ; the doubly composite Pokau involves 

 the same liquid component, which has elsewhere been identified in each 

 position; and from this we pass easily to Roro. The slightly variant bebi 

 stem appears simply in Mugula, Tubetube, Tagula, Murua; the Panaieti 

 and Misima form is most readily explicable as preduplicated. The next 

 stem variant is beba, simply in Nada and Kiriwina, conduplicated in Boniki. 

 In Kwagila we release stem bibi with an initial kara suggestive of the 

 common liquid component. The stem feje appears simply in Mekeo, and 

 with a termination of unascertained value in Kiviri and Oiun. Thus is 

 established a very satisfactory series of evolution forms. 



206. 

 The few New Guinea forms exhibit harsh but not wholly improbable 

 mutations of the consonants and, to a less degree, of the vowels of Proto- 

 Samoan hifo, the i being the most consistent element. The aspirate is 

 found only in Tavara ; it passes to the sibilant in Tubetube, Mukawa, and 

 Roro (zt) ; the mutation to d is abundantly established in Motu and Pokau, 

 passes easily to / in Roro, and thence is caught in the general kappation 

 movement to Mekeo kipo; these have all been downward mutations, the 

 change to r in Keapara and Galoma, upward in the lingual column, has 

 been discussed in the supplementary note 47 as diagnostic of secondary 

 borrowing; in Wedau the aspirate is extinct. The /-mutations to v, p, h, b, 

 are abundantly familiar ; the ^-mutant in Keapara looks like a secondary 

 borrowing from Galoma neighbors; in other items it appears in Kiriwina, 

 Rubi, and Uni; the peculiar siio of Tubetube is no doubt due to the fact 

 that this language has acquired neither v nor/. The vowel changes call for 

 no note except the occurrence of \-o in Tavara and Wedau; elsewhere in 

 these data this mutation is identified in Awalama, Boniki, Galavi, and Motu. 



207. 

 These forms entail no difficulty until we reach ahu, which Motu uses 

 along with tapu. Inasmuch as the t is abundantly determined in this 

 language, ahu looks like a secondary borrowing from Mekeo, where t is 

 commonly elided. 



212. 



The New Guinea forms are all/w/w derivatives, except possibly Suau uru. 

 As that language has no/, this may derive from fulu; equally it may derive 

 from the ulu which we have well established as in Proto-Samoan posses- 

 sion during the Melanesian transit. 



214. 



As in the principal note on this item, so in the Torres Straits littoral we 

 have no difficulty in recognizing the two stems faka and fa. 



217. 

 The presence of the form ama in Awalama, Taupota, Tavara and Wedau, 

 languages which possess the t, forms a connecting link for this state of the 

 stem between its occurrences in Melanesia and those in Indonesia. 



