212 THE POLYNESIAN WANDERINGS. 



Arag, Nggela, Vaturanga, Bugotu: vet, reflexive prefix. Wango, 

 Saa: hei. Ulawa: hai. Fagani: fai. Koro :bai. Kabadi: 

 vai. Sinaugoro, Keapara, Kabadi, Hula : ve. Galoma : be. 

 Pokau: vi. Duke of York : we. Motu: he. Nengone: e. 

 Mota, Merlav, Retan: var. Pala: har. New Britain, Bara- 

 von: war. Kabakada: wara. Lakon: va'. Motlav, Pak, 

 Leon, Vuras, Mosin, Alo Teqel, Gog, Norbarbar, Lo : ver. Volow : 

 vear. Omba: vui. Maewo: vagal. 

 We have several elements in this problem which detailed examination 

 will tend to simplify. 



(i) The Nuclear Polynesian type, fe. In Melanesia this is found in 

 Efate' hi, fi; in Duke of York we; in several variants in Torres Straits; 

 and as an attrition fragment probably in Nengone e. 



(2) The Viti type, vei. This I regard as the same as fe but representing 

 a later development, for the t-suffix is well marked as verb-formative when 

 the language is beginning to feel the need of specification in the use of the 

 much including attributive (27 American Journal of Philology, 378). 



Its fullest form is in Viti, Arag, Nggela, Vaturanga, Bugotu, vei; thence 

 to Wango, Saa, hei. 



A second series is Fagani fai, Ulawa hai. 



With the single exception of Arag in the New Hebrides to serve as a 

 link these forms are all from the Solomons, with New Guinea offshoots of 

 Kabadi vai, and Roro bai. 



Rotuma hoi is the second element in yet a third series, the spirant form 

 with this vowel nowhere appearing. 



(3) We now have a widespread suite with a second consonant. For 

 this reason we must dissociate it from the foregoing. It does not come 

 over into Polynesia with the slightest trace of the second consonant. Yet 

 it is possible that it represents an original from which the foregoing derive 

 by abrasion. 



Its fullest form is in New Guinea in Kabakada wara. 



With abrasion of the final vowel we have Mota, Merlav, Retan var; New 

 Britain, Baravon war; Pala har. In Volow it is a characteristic peculiarity 

 to introduce e before a in a closed syllable; its vear serves as a transition 

 form to Motlav, Pak, Leon, Vuras, Mosin, Alo Teqel, Gog, Norbarbar, Lo, 

 ver. The Lakon va' is a degeneration of var, dependent on the language 

 peculiarity which Dr. Codrington records "at the end of words r is not trilled 

 and sometimes with abrupt pronunciation is not heard;" it may be taken 

 as the transition stage from the forms with two consonants to the Poly- 

 nesian type. This suite is found in the Banks Group, New Britain and 

 New Guinea. 



Omba vui is not explained in our vocabularies. If it be permissible to 

 regard the u as functioning as semivowel w, then vwi might be regarded as 

 associable with Efate bi, fi, which also shows an uncertainty in striking the 

 exact sound of the labial. From its close neighborhood to Arag, which has 

 vei, Omba might be expected to show some form of this type. 



The Maewo vagal seems an utter anomaly. Dr. Codrington says "it is 

 not clear what vagal may be." It will be seen that va-ga-l suggests the 

 common Banks Group type var with an infix; but we have no authority 



