484 THE POLYNESIAN WANDERINGS. 



fa we find fai in Suau, confirming its occurrence at three Solomon Island 

 stations : 



Former/: to h, to w, to p, to b, to v, to extinction. 

 Latter/: to h, to s, to w, to p, to 6, to -y, to extinction. 

 Diversely mutant: w-v, Taupota, Wedau, Galavi, Boniki; w-s, Mu- 

 kawa ; w-h, Suau ; w — , Dobu ; — h, Redscar Bay. 

 The only changes of n are to k in Mukawa, to / in Kiriwina, to extinction 

 in Mekeo. In Rubi gaiine I regard ga as a foreign composition member 

 and iine as the remnant of fifine after extirpation of/, as also in Nada. 



291. 



The final liquid hitherto recognized in Indonesia is of wide extent in 

 New Guinea. It occurs in these languages: Tubetube, Massim, Taupota, 

 Wedau, Misima, Panaieti, Mugula, Suau, Awalama, Kwagila, Tavara. It 

 will tend toward simplicity to include therewith Kiriwina waia and Kabadi 

 veina. The p-b forms discovered in Melanesia account for Roro bei. The 

 k forms, sparsely found in Melanesia, are here represented by goila in five 

 languages. The vowel change from vai to vei has already been discovered 

 and accepted. In Nada and Murua rai offers a form without parallel, 

 unless we are prepared to establish the \-r mutation upon this instance 

 and a single form in Misima and Panaieti in item 109. 



292. 



These identifications are very satisfactory. We note two which are 

 unusual. In Tubetube and Suau the initial labial is replaced by a light 

 vowel sound, evidently transitional to its entire absence in Rubi and 

 Roro. In Sinaugoro, Hula, and Keapara we lack explanation of the inter- 

 jected palatal. The loss of the final vowel in Rubi is rare, but not without 

 parallel in Melanesia. 



297. 



The Proto-Samoan stem nowhere exactly appears in any of these New 

 Guinea languages save Sariba and Kiriwina. The closest approximation 

 is stem gila recoverable from Keapara and Misima. Next we find a gala 

 stem with its range of consonant mutation. At the last we find a karo 

 stem in Kwagila. I have collated only for the consonantal elements and 

 with no great insistence on the validity of these groups. 



298. 



A very good series is here exposed. The only unusual element is the 

 preface in Awalama, Taupota, and Mukawa; and this recurs in Nggao. 

 In the principal note I expressed doubt as to the Malay and Javanese ; we 

 come nearer to them with this New Guinea material, where idu and ilu 

 are regular in their mutation. 



300. 



The prevalence of n in these New Guinea identifications inclines me now 

 to the establishment of the Proto-Samoan stem ikan, the more particularly 

 as this region of the great island seems to have been little visited by Indo- 

 nesian rovers. The central k is altogether absent, except perhaps that the 



