DATA AND NOTES. 273 



Salayer, Liang, Lariko, Saparua, Awaiya, Caimarian, Wahai, Teor: 

 ama. Morella :a'ma. Cajeli: a'mam. Ahtiago : amdi. Teluti: 

 amaeolo. Amblaw : amao. Bouton : amana. Menado : iama. 

 Sanguir : yaman. Tobo : jaman. Bolanghitam : kiamat. Gah: 

 mama. Mysot: mam. Mayapo: nama. Massaratty: ndama. 



In the Polynesian this is distinguished from tdma child by the accent 

 iama or by the addition of a final syllable which automatically secures the 

 same incidence of the accent, tamdi, tamdna. Diacritical marks have been 

 but sparingly used in our Melanesian vocabularies and for that reason we 

 lack a sure guide as to the accenting of the western affiliates. We find but 

 the cases of New Britain, and Pala in which the accent is printed, and Mota, 

 Nifilole, Redscar Bay, and Aneityum in which it is inferential. In not one 

 of these languages have we any evidence of the use of tdma child, therefore 

 the accent is not a differential in their own material, but has carried its 

 ictus from the source whence the tarn a father has been borrowed. 



We shall first examine the languages which retain the t-m consonant 

 skeleton. There is a long series in which the two vowels remain as in 

 Polynesian ; they are therefore identical words except that the accent may 

 vary, and on that point we are without information. The final a is almost 

 wholly permanent, not only in the t-m series but in the m series which will 

 come up for our later consideration. The solitary exceptions are Tanna 

 timiin) and Eromanga temi. The former a in this case becomes e in Ero- 

 manga and i in Tanna. The i- mutation also appears in Santo lima. In 

 Nifilole we find u in tumai. 



Our next variant of tama involves frontal abrasion affecting the t. This 

 we find in two discrete areas in the Solomons, respectively, north and south 

 of Malanta and separated by an area of greater degradation. With the 

 ama of this abraded type I include Fagani wama. 



This middle type gives us a suite of occurrences of the transition form 

 by which we arrive at forms in which not only the initial t but the vowel a 

 thus become initial have been subjected to frontal abrasion. The simplest 

 form is ma in Nggao and Lo. By reduplication of ma we may more logically 

 account for mama than by attempting to establish a t-m mutation. By final 

 abrasion of mama we arrive at mam, a form very widespread in the Banks 

 Group, the northern subdivision of the New Hebrides. 



Few forms lie outside this chain. Aneityum etman is only superficially 

 irregular, for by punctuating apart its formative elements we find in 

 e-tma-n our tama theme with loss of the former vowel, which is character- 

 istic of many of the Polynesian homogenies in that language. The lata of 

 Malekula and Tangoan Santo and the chacha of Nengone appear to be the 

 result of a terminal abrasion involving the final syllable and then a redup- 

 lication. There is a priori no reason why such a course should not have 

 been followed by the former syllable as well as the later; we must note, 

 however, that we find no evidence such as transition forms would afford in 

 support thereof. If Baki karama stems with the ma type of tama it 

 involves a component kara as to which we lack information. 



The Indonesian homogenetic forms are most largely of the transitional 

 ama type, only Tobo jaman suggesting a tama possibility and in our igno- 

 rance of the source of this record jaman may be but a variant transcription 



