DATA AND NOTES. 389 



futi forms. The vnndi of Vaturanga, Nggela, and Bugotu, the same as Viti, 

 is also futi. We have, therefore, a perfect identification of futi in the 

 Solomon Islands, a series of crop colonies on the Samoa track. Baravon 

 vundu is easily associable with vundi and therefore with futi, and Kalil 

 'huddu represents the same secondary stem with the d not reinforced. I 

 have already commented on the borrowing at second hand of a reinforced 

 consonant by speakers to whom a double consonant was objectionable and 

 who, in their ignorance of the primal form, have ignorantly retained the 

 reinforcement and have dropped the radical. Thus from vundu of Baravon 

 and wundu of the Duke of York we pass to the 'hiin of Lambell, the hun 

 of Laur, the wan of Lamassa, and the un of Carteret Harbor in the Lamassa 

 region of New Ireland. These form a second, a more remote, stage on the 

 Samoa track ; they lie within the eastern gateway. 



We are next to examine a group of forms having a for the earlier vowel, 

 a for u. These have all lost the initial consonant. Our studies in the east- 

 ern gateway have shown us the vanishing of v in progressive stages. Here 

 we have no record of progress in the dilapidation, yet we may accept the 

 result and view this group as representing an original fati variant of futi. 

 We find this in Efate ate, Sesake audi; and Maewo undi may be taken as 

 retaining somewhat of the transition phase which links fati and futi. Look- 

 ing back at Efate we find dtse. This is no unusual /-variant. In Polynesia 

 it is regular in Niue and Tongarewa; in this stem we find a blunter form 

 of it in Aniwa hutshi and Fila butsh. Then by reduction of the doubled 

 consonant and selection of the wrong member we come to ase. 



We next find a group in which primal futi has become fiti. We do not 

 find fiti itself any more than we found fati. But its immediately secondary 

 phase is found in Malekula Pangkumu ne-vij and Uripiv na-vits. From 

 na-vits we pass as before to na-vis, and thence it is a normal step in one 

 direction to villi which we find in Bierian of Epi, in the other to hisi which 

 is preserved in Paama ahisi. Through vihi we are led simply to Arag ihi. 



Paama ahisi has shown us a futi derivative assuming a prefixed a. We 

 have already become acquainted with the change of / to s. Therefore we 

 need have no hesitation in accepting Malekula abus as a futi derivative; 

 still less in joining therewith Eromanga no-bos, for its no- is readily seen 

 to be the Melanesian n-article chameleon-colored by the nearest radical 

 vowel. Twice already have we seen the primal / develop into h; therefore 

 the no-hos of Ambrym and Aneityum takes its place as a futi derivative. 

 Moanus mbur is the only form which for the present seems irreducible. 



We find somewhat parallel variations in Indonesia. Primal futi is repre- 

 sented immediately by Ceram fudi, Matabello phiidi, Gah fudia, Timor hudi, 

 Malagasy hutsi, Tobo fud, Caimarian tiki, Malagasy untsi, Batak unchi, 

 Macassar unti, Samba wa punti, Pangasinan ponti. Sanguir busa introduces 

 the ^ which we have twice already developed. Massaratty fuati, Mayapo 

 fuat may represent the transition phase of futi to fati. The fiti form is 

 seen in Ceram and Ahtiago phitim, and with the ^--change may pass into 

 Malay pisang. 



We have passed in review many variants of the initial consonant of futi, 

 v, p, b, w, h, and its complete vanishing. Not one case suggests the m 

 which this Semitic requires to bring it into kinship. 



