166 THE POLYNESIAN WANDERINGS. 



The extinction of n, found once in Marina, is highly doubtful in 

 the single Indonesian case in fa fine (290) woman Ilocan babai. 



The n-l in Kayan, n-nj in Dyak and Macassar, and n-y in Pam- 

 pangas are found only in the vocable (328) which in Macassar has 

 also afforded an instance of n-l and gives n-ny in Melanesia. The 

 single instance of k-kn is of uncertain nature. 



H. 



The foregoing tables record a considerable number of mutation 

 forms for the aspiration in Indonesia. For the mutation h-s, no less 

 than twenty-two languages afford examples; yet on closer inspec- 

 tion it will appear that in the latter case but three vocables are at 

 all involved, and that six vocables only are involved in the whole 

 range of mutation of the aspirate. We feel justified, therefore, in 

 regarding this as partaking more of word quality than of speech 

 endowment. In Polynesia the aspiration has all but vanished; 

 comparison of mutant forms, therefore, would be futile, the only 

 forms common to the three families being h-s and extinction. Mela- 

 nesia, in stiong contrast with each family, shows a marked ten- 

 dency to employ the aspiration in its proper function ; over an equal 

 and interlacing area it follows the h-s mutation. Common to Indo- 

 nesia and Melanesia are these, h-t, h-d, and h-j, the two latter being 

 variants upon the primal h-t mutation; the vocables involved in 

 these changes in Indonesia are similarly involved in Melanesia, which 

 has a few in addition of its own. Mutations peculiar to Indonesia 

 are the following: h-l, h — , h-ra, an upward movement in the series, 

 discoverable only in the two vocables recorded in our data under 

 numbers 47 and 278; regarding ndr, as so frequently, in the light 

 of an effort to compass the r grasseye, h-ndr (Malagasy 47) falls into 

 this group. In Kisa 278 h-ng is explicable as the upward shift h-n 

 (which is nowhere discoverable), upon which is added the rearward 

 shift n-ng; confirmation being found in our record of the latter 

 mutation in that speech. 



S. 

 The sibilant is better retained in Indonesia than in Polynesia, a 

 facility which it shares with Melanesia. The mutation common to 

 the three families is s-h, normal to at least eight Polynesian lan- 

 guages, far less frequent in Melanesia, and in Indonesia affecting 

 but seven vocables in six languages. Of the mutations common to 

 Indonesia and Melanesia, s-r (found in Malagasy and Java 298) 

 rests doubtfully on Tanna 337; and s-t is found in Alite and Wango 

 in the same 337. Wholly Indonesian are s-ch, s-j, s-d, but they 

 group naturally with s-t. s-n (Malagasy 344) lacks confirmation. 

 s-ng (Togean 298), however, seems quite feasible, s-g (Malay 132) 

 is very doubtful. 



