54 SISSANO. 
a form in use in Viti and the loss of the k characterizes modern er 
and thus,we arrive at ‘ali without a break in the chain. 
Following the study of the prefaced forms, we find two which are 
somewhat anomalous, 30 pilingén and 31 vatlingén ‘The latter seems 
to be a compaction of vat, a word-stem of sense not yet determined, 
and the remnant of 21 ulingén after frontal abrasion. But the p pre- 
face in ptiliingén offers an interesting suggestion. In Paumotu rurunga, 
28 luluna, and 29 lula we find a group of liquid prefaces. But in 3 guli- 
ngen, and 1 kélinga, in Tahiti turua, in 30 pilungén, we find a complete 
series of mute prefaces, palatal g-k, lingual t, labialp. We have already 
pointed out another series of mutation which suggests progression of the 
mutes. In the isolating languagesof the Pacific we can establish no such 
simple rule as the Grimm’s law of the inflected tongues of the Indo- 
European family, but we shall find a great interest in the few, perhaps 
fortuitous, instances in which we may trace the beginnings of such a 
system. It serves here to set against this apparent suggestion of 
mute progression that the method of mutation normal to these isolat- 
ing languages is most distinctly serial and that no frequency of inter- 
serial mutation is discovered save among the nasals, and that only 
partially. 
In the Polynesian series we find an a-o—u range of the initial 
vowel, but with a considerable preponderance of u,. The same inclina- 
tion holds in Melanesian; we have 17 forms.in u, 7 in a, 6 in o, and a 
single instance of i in 13 ilinga, a vowel which does not appear at 
all in the Polynesian series. Two widely separated forms of close 
resemblance are 24 uilinga of the central Carolines and therefore 
properly Micronesian, and 25 tilana of the northern Solomons. I have 
provisionally counted these among the u forms, but there seems reason 
to assign them to theigroup. ‘The ictus in ifland suggests that the 
preceding u has value as compensatory for the loss of a vanishing 
consonant, a suggestion which I have advanced in the discussion 
of 39 eanua in item 8, and 20 wamain item 27. The stem 1 retains 
its character throughout Melanesia, with the single exception of 
27 unua, and this I1-n mutation is quite common and involves but 
the slightest possible movement in progress of the lingual possi- 
bilities. The second stem vowel remains u without change in the 
Polynesian series after resolving the metathesis of Rapanui rangua, 
and is constant through 19 Melanesian forms. It passes to a in 
14-16 and 25, a group of closely related languages; 29 lila is prob- 
ably of this a series, but the form is so much reduced below criti- 
cal character that we include it only on tolerance. It is represented 
by o in but one form 12 alék. The passage to i possesses great 
critical importance for us in the establishment of the unity of 
alunga and ‘ali; we find it in 8, 9, 13, 17, 19, and 24. ‘The mutation 
undergone by the ng element is typical of Oceanic speech; it passes to k 
in 5-9 and 12, to g in 11, to n interserially in 16, 25, and 28, and to 
