THE SOUTHERN GATEWAY. 479 



74- • 

 It is now quite clear that our record shows the existence of a natu series 

 parallel with ati. Therefore the forms which exhibit an initial nasal with 

 the final i are to be regarded as natu attracted by ati. In this examination 

 of the New Guinea material it is safe to assume natu as a Polynesian stem. 

 Mekeo ngau-nga is readily establishable upon Galoma nan with a nasal 

 termination, and the latter may perhaps be seen in Malekula netin. 



76. 



Particular interest attaches to the discovery of the ainu type in Mabuiag 

 and Miriam, western and eastern islands of the straits and remote from 

 the New Guinea coast, for we shall make opportunity to discuss the paucity 

 of Polynesian material in these islands. The existence of amu in Fotuna 

 affords us reason to regard the type as ancient Proto-Samoan, and that 

 Mabuiag and Miriam received it directly and not on secondary loan from 

 Motu. 



109. 



We note here two series, yet reference to the several phonetic tables will 

 show that the essential mutations, while in general feasible, are in par- 

 ticular not well supported. 



A. Retentive of the initial consonant: Galoma, Wedau; and with 



obliteration of the inner labial, Sinaugoro, Sariba, Rubi. 



B. Abrading the initial, but retentive of some representative of the 



inner labial : Pokau, Mekeo, Hula, Motu, Panaieti, Misima. 



Nada pwau may be regarded as apu under 213 metathesis, and the then 

 initial labial under such maltreatment as marks the Melanesian q, cf. 285. 

 But from pwau Kiriwina pwak conforms to no phonetic law as yet found 

 in this region. 



122. 



Kabadi vat and Roro bai may be taken as simple variants upon the 

 vei type which has effected a lodgment in Nuclear Polynesia in Viti and 

 Rotuma. Kabakada wara has already been under comment in the principal 

 note on this item. The other forms are clear variants upon the Proto- 

 Samoan fe. 



149. 



The Motu and Kabadi forms, referred to the simpler stem in Nifilole 

 laki, may be adjudged composites of a liki variant. The liki type is found 

 in New Guinea only in a metathetic variant : Keapara, Hula, Galoma, Suau, 

 and Sariba in series. The other forms are of the kiki type with various 

 foreign elements in composition. 



190. 



In the principal note on this item we have established an elemental pe; 

 in this district it is recoverable from composites in Awalama and Tavara, 

 in which the u suggests the intimately associable ula component elsewhere 

 widely extended. The duplicated stem pepe is found in Hula, Keapara, 

 Galoma, and in composition in Dobu. The ready mutant bebe is found 

 simply in Rubi, Suau, Sariba; Taupota and Wedau are consimilars of »- 



