482 THE POLYNESIAN WANDERINGS. 



of transition phases which may establish wo as a mutant of ma. In the 

 former consonant of the stem the mutation n-ng-g-k to extinction is steady 

 and well established ; so, too, the series of the latter consonant f-v-p-w. 



258. 



In the Proto-Samoan stem matakut we find four consonants to engage 

 our attention. The m, in fact the whole conditional preformative ma with 

 the exception of Kabadi me, exists unaltered through our New Guinea 

 identifications. The t in second place undergoes no greater change than 

 the kappa tion in Pokau and Kabadi and the change to r in Tagula, on 

 which see the principal note. In Tagula mar ode we shall arrive at the 

 most satisfactory conclusion by assuming that the stem syllable ku has 

 vanished, thus leaving d as representative of stem t. The stem k has van- 

 ished completely in all these New Guinea identifications. The final t is 

 preserved in Dobu, Oiun, Awalama ; becomes d in Tagula, s in Sariba ; has 

 vanished from Pokau and Kabadi. 



259- 

 The development of these widely diverse forms is to be studied in the 

 mutations of the stem consonants. The former remains n in Mukawa, 

 Awalama, Ta vara, Roro, Kabadi, Mekeo, Pokau, Panaieti.Tubetube, Misima, 

 Tagula; it is no more than a breathing in Motu; is lost from Taupota, 

 Wedau, Doura, Uni, Galavi, Boniki. The / becomes v in Taupota and 

 Wedau, b in Mukawa, w in Awalama and Tavara, h in Roro, and vanishes 

 in Mekeo. From this stage we reach the extirpation of the second syllable 

 in Panaieti, Tubetube, and Misima; and Tagula I regard as a devolution 

 form of the latter. Returning to Roro we find the h passing to ^ in Motu 

 and Kabadi. Here comes a lacuna. If we imagine a passage to t, prac- 

 ticable but nowhere recorded, the not infrequent kappation would account 

 for Pokau, Doura, Galavi and Boniki. Thence to Uni is an easy passage. 



265. 



The forms in the Torres Straits area run in the series ruma-luma-numa- 

 numi-yuma-uma. With the exception of numi we find Melanesian instances 

 of each form. 



272. 



The probably more elemental stem e/uis traceable in Keapara and Mekeo. 

 The lefit stem appears in Motu, Doura, Kabadi, Roro, Uni, Wedau, Galavi, 

 Boniki. In the principal note I pronounced, on the material then acces- 

 sible, against kabu of the Duke of York. We now recognize a kefu stem, 

 either coordinate with lefu and nefu or possibly derivative from the latter. 

 This is found in Taupota, Wedau, Motu, Tavara, Awalama, and Sariba. 



273- 

 The addition of this material removes the objection noted in the principal 

 note (b) against the form kuvi. The assumption of initial k is of wide 

 extent in New Guinea ; in fact but two identifications lack it, Motu uhe, 

 which is not the common yam name, and the variant in Pokau, which is 

 metathetic of the 231 type. 



