110 SISSANO. 
stood as an aspiration of 25 atin but the assumption of an initial labial 
in 23 baun does not comport with the recorded history of this group 
of stems; 11 kanua may be linked with 34 and 6 kauna, for we find 
the speech of Rotuma to be strongly characterized by metathesis, 
which in this case would take the form of the transfer of the initial n 
of the syllable na to a new position as initial of the syllable u, a meta- 
thesis included in my discussion of this mutation method (Easter 
Island, 29); 10 kénye has resemblance to kanua, it may be that such 
metathesis has been effective in Ingros and Entsau at the entrance 
into Melanesia as in Rotuma at the distant exit therefrom. 
In the examination of the stems afuna and auna, which are possibly 
but a single stem, we have passed under review all the forms which are 
characterized by the vowel pair a-u associated with a succeeding n. 
Now we consider residual forms which preserve the a-u but lack the 
nasal element. As a formal exercise, probably a mere juggling with 
phonetic rules, we should start from kaun postulated as a product of 
34 and 6 kauna, though not recorded, and through the principle of 
final abrasion develop 1 kau. We might deal with 2 kaua as a weaken- 
ing of 8 kawuna by loss of the nasal, but the history of the stems pro- 
vides us no instance of such weakening and in general it must be 
regarded as highly anomalous. It is easier to regard the final a of 
kaua as an assumption which in turn has picked up the semivowel 
preface exhibited in 3-4 and 28-29. In 4 kauwek we should have no 
difficulty in comprehending the assumption of a final k, for that is 
not an infrequent occurrence in several Melanesian tongues; the uku 
of 5 kauuku may be associable herewith, or it may be a distinct stem 
gathered up in order to give precision to the diffuse signification of 
kau, that is to say, a determinant compound has been formed. 
In this discussion reference has been made more than once to an 
academic problem of phonetics, and in so far as this term may seem 
to cast a shadow of doubt over the course of the argument, it will be 
held to have served the end which it was employed. To a certain, 
in fact to a very large, extent all problems of phonetics are academic; 
from amassed facts of observation it is proper to deduce rules which 
may be held to govern the usage of any speech. Here we mention 
Grimm’s law. It may properly be impleaded in the trial of any of 
the linguistic possessions of the Indo-European languages, for it is 
the resultant of innumerable instances carefully codified and made 
applicable. But the validity of Grimm’s law does not in itself estab- 
lish the equal validity of other phonetic laws which seem to derive 
from a linguistic group of far different constitution. Because Grimm’s 
law holds for Indo-European speech it does not necessarily follow 
that in these primitive tongues of Melanesia and the but slightly 
advanced tongues of Indonesia we shall find inherent validity in a 
law of progressive devolution. In the foregoing examination of this 
