THE SOUTHERN GATEWAY. 481 



225. 

 In the Pacific we find as stems kill, kali, kalo, and kalu. These are found 

 in New Guinea as follows, but because of this consonantal fixity we shall 

 not list mutant vowels. 



kili: Galavi, Nada, and perhaps Raqa. 



kalo:Kubiri, Kiviri, Oiun, Motu, Taupota, Wedau and Boniki. 



kali: Kiriwina. 



To the above New Guinea contributes : 

 kuli: Kiriwina and Panaieti. 



Wedau and Mukawa giai scarcely seems associable, the resemblance to 

 kili being only in the first syllable. 



239- 



Interesting results appear in the comparison of these New Guinea forms 

 with the variants already noted from the Proto-Samoan asu. Take first 

 the initial elements. The palatal is found in Moanus, the first landmark 

 of the Samoa Stream ; in Torres Straits, at the first landmark of the Viti 

 Stream, in Sariba, Panaieti, Mukawa and Awalama; the whole island bars 

 these two streams. The labial preface of Malekula Aulua is found in Galavi, 

 Taupota, Wedau, and Boniki. Ambrym walehi seems paralleled with Motu 

 kwalahu. The asu stem is identified in Suau and Dobu without alteration. 

 Roro hiavu is strikingly like Lo hiev, in which we distinguish hi as the asu 

 derivative and ev-av-avu as the a/i-fire derivative. Nada and Murua museu 

 (and the invert umseu in Kiriwina) seem like asu with the wz-preface not 

 elsewhere seen. In the phonetic tables regard has been had only of the 

 recoverable asu and its mutants. 



247. 



Tagula bibido is our sole New Guinea occurrence of the pito stem, all the 

 other examples in this area deriving from puto. These have been freely 

 dealt with by the New Guinea tribesmen, but close analysis will show 

 several series of variants, each consonant being strongly affected in the 

 mutation. For p we have p, b,f, m, and extinction; for t we have /, d, k, s, 

 h, r, and extinction. The ^-forms find distant support in Buka. The muta- 

 tion to / has a satisfactory extension in this area and affords valuable 

 support to the note on the subject under item 258. 



249. 

 In the absence of t in the languages, Hula and Keapara gia and Galoma 

 ia derive from Sariba gita. As to Kabadi is' ana, Dr. Ray states distinctly 

 that s' is found only once in New Guinea and that once is not Kabadi; 

 however, it hangs quite properly upon ihana which he presents as a variant 

 of itana in Roro, and the latter connects through ita and gita with kita 

 and kite. 



254- 



For the most part these New Guinea identifications exhibit the wt/i-stem 



before the acquisition of the conditional ma, which, however, appears in 



Mekeo, Galoma, and Keapara. In Taupota and Awalama we have an 



initial element which is probably formative, but we are without evidence 



