164 THE POLYNESIAN WANDERINGS. 



Marquesas, and Sikayana, sporadic in Rotuma. In Melanesia it is 

 not normal (with a possible exception in the case of Santo); it is 

 traceable through the languages of that area rather as a quality 

 with which a certain few loan words are endowed. In Indonesia it 

 is seen to be a quality of talinga (350) ; in Malagasy only is it suffi- 

 ciently frequent to be considered as a speech endowment ; the Mala- 

 gasy words in which this mutation is found are all identifiable as 

 Polynesian material, and several of them are carried with the same 

 modification through Melanesia. 



In Polynesia the mutation ng-£ is normal in the eastern Mar- 

 quesas, sporadic in one instance in Viti. In Melanesia its place is 

 taken by ng-g, a word quality. In Indonesia it is found but once. 



ng-n; is highly problematical in the sole instance in which it is 

 found in Aneityum. Its presence in the Indonesian rests upon 

 Tregear's citation talinga (350) ear Chamorri talanja, against which 

 we set talanga from Fritz's Chamorro dictionary. 



ng-n/*, resting on longo (336) sound Malagasy rohona, is not cer- 

 tainly Sawaiori. 



ng-nh in talinga (350) ear Satawal talinhe is a good identification 

 but the mutation is without parallel. 



K. 



In Polynesia the principal variation of k is to extinction, this 

 being normal in Samoa, Hawaii, and Tahiti, and not unknown in 

 several other languages. In Melanesia it is found widely extended 

 as a word quality; in Bululaha, Saa, Ugi, and Ulawa it appears 

 as a speech endowment, these being in the region of crop colonies 

 in the Solomons. In Indonesia it appears as a word quality of but 

 two vocables. 



Of the k-g mutation we have record of but a single instance, 

 sporadic in the Paumotu. In Melanesia this is of frequent occur- 

 rence and in many languages it is normal. In Indonesia it is infre- 

 quent outside of the Malay, a speech endowment none the less 

 though sluggish; the Malay and Javanese words in which it occurs 

 are Sawaiori, those in other languages doubtful. 



k-h is of rare occurrence in Melanesia. In Indonesia it is satis- 

 factorily identified in Teor and in some, though not in all, of its 

 occurrences in Malagasy. 



k-ng is of rare occurrence in Polynesia, equally rare in Melanesia, 

 and its single appearance in the Malay is not a wholly satisfactory 

 identification. 



k-ngk practically corresponds to the k-ngg found four times in 

 Viti, rarely in Melanesia. In Indonesia it is found in but two 

 vocables, each in two languages. 



k-ch and k-j are not satisfactorily identified. 



