POLYNESIAN AND MALAYAN. 139 
The most frequent of these forms is the zu of all three Oceanic 
areas and its immediate derivatives. A nu type is found in Mangaia 
alone of Polynesia, in Nengone and Nifilole of Melanesia, and in Java 
of Indonesia. In each of the latter the derivatives of mu may readily be 
recognized. In 2, ascribed to New Caledonia, where a multiplicity of 
languages exists, we may have no more than a variant of the Nengone 
nu, U-i being well recognized among the vocalic mutations. From niu 
to nu we find a feasible series of transition forms in Torres Strait, nzura 
to diura to rura, save that at the last we find no explanation for the loss 
of iany more than we can comprehend the same loss as between niu and 
nu in their simple form. 
Reverting to the 1z type, for which I have already suggested the 
possibility of vocalic mutation, it is worthy of closer examination. If 
nt were primal and carried such sense as to the logical powers of these 
primitive folk suggested a genus, then it might be possible to regard niu 
as composite of uz and u, the latter being too formless for us now to 
venture to interpret. In Mangareva we have an interesting pair of 
coconut words; mi—u is used of the young palm, uz—kau when it has 
grown up. But we are able to identify kau; it is the general term for 
tree (cf. item 30); therefore uzkau is clearly the mi-tree; by analogy niu 
should be the 1z—aliquid, an indefiniteness which we can not yet resolve. 
Disregarding a terminal consonant or even additional syllable, we find 
for this second member of the word, u, hu, yu, 0, yo, eé, we, wr. Of these 
the y-forms are susceptible of ready explanation: ~yu amounts to no 
more than variety of transliteration of nzu; in niyog the y represents the 
glide of the vowel from the palatal position of 1 to the labial position of 
0. Conversely the aspiration in mzhu represents the purposeful inter- 
ruption of such glide. When we pass through 0 to e, a mutation series 
which is well established, we find a group of forms in Ceram which 
exhibit marked changes. For niwel we have abundant support in 
several languages. From niwel to nimel is supported by Lariko in 
Amboyna; since the semivowel w is close to the lips and m is the labial 
nasal the mutation lies within the same series. But we are left without 
any accounting for the intrusive palatal mute in mzkwel. At present 
the resolution of this tangle eludes us. 
62. pe, po interrogative particles; Subanu ba id. 
pe Samoa, Futuna, Hawaii. 
be Tonga. be Efaté. 
pee Uvea. 
po Samoa, Futuna, Niué. ba Subanu, Visayan. 
This particle is in wider Polynesian use, but with a variation in 
sense. As interrogative it is restricted to Nuclear Polynesia and the 
Proto-Samoan; in the Tongafiti languages it is but a disjunctive particle. 
The Efaté is an interesting link between the Philippines and Nuclear 
Polynesia. 
