116 THE SUBANU. 
not phonetic but social; it is best explained as an adoption from the 
Paumotu fetzka under the influence of the word-tabu known as te 7; 
wholly anomalous in Polynesia (and it must be recalled that the Pau- 
motu is filled with intricate problems of speech) we find no affiliate of 
fettka except bituek of Silong in Indonesia. The vowel alteration to 
fatou in Aniwa is paralleled by fatwi of the Sulu. The vowel change to 
hoku in Hawaiian occurs again in that speech in to‘elau—koolau. So far 
as our Polynesian material extends, we have no evidence that the stem 
is other than open; the incidence of the accent upon the ultima, how- 
ever, suggests a device of some compensation. But in Indonesia a 
final consonant is so common and in general so uniform as to preclude 
the interpretation of local accretion. In nineteen forms there are but 
three which lack a final consonant, of which Sulu and Menado retain 
the second vowel characteristic of the Polynesian, one station at the 
threshold of the Philippines, the other in the Celebes subprovince. In 
ten forms the final consonant is n and in four more it isng, which we 
know to be a most frequent mutation product of n. In the Silong 
bituek, with which is associable the Paumotu /etika, the k may be 
regarded as an ng mutation once removed. In Gani betol we are at no 
loss to consider | as a frequent mutation product of n upward in the 
lingual series. We find such an agreement upon final n or recognizable 
n—products that I am willing to propose fetun as the original stem of 
the word. In the general absence of the labial spirants in the languages 
of Indonesia we find two instances in which the initial f is weakened in 
borrowing and passes vowelward to win Sanguir and Bugis. In eleven 
instances it is strengthened to the ultimate labial possibility, the mute 
b, and these instances are smoothly distributed over the whole archi- 
pelago. The second consonant t remains unaltered except in the soli- 
tary instance of Teor toku, and this t-k mutation, so generalin Poly- 
nesian, may well have begun to be felt before the exit from Indonesia; 
mention of this has been made in item 4. We next direct attention 
upon a special group of three forms, making a series by themselves: 
bitang is readily to be established in the Malayan series; bintang follows 
with the preface of the mute by the nasal of its proper series; lintang 
shows an anomalous mutation b-I extra seriem, but the agreement with 
bintang in other particulars is sufficient to place it in the group. ‘The 
characteristic former vowel e but once appears in Indonesia. Thea 
which we have found in Aniwa is also in Sulu and Pampangas, both 
Philippine languages and archetypal. In twelve instances evenly dis- 
tributed over the region the vowel is 1, and without complicating the 
record by citation of examples I note that this is the characteristic 
vowel throughout the Melanesian traverse. The characteristic latter 
vowel tis well preserved. Last of all we find a group of interrelated 
forms in which the stem has abraded its former syllable; these are found 
in the Ceram subprovince. 
