MELANESIAN ANNOTATIONS ON THE VOCABULARY. 75 
At this point I interject class F for consideration. It has the con- 
sonant structure of boro—bor, but between the consonants appear two 
vowels producing an extra syllable. The suggestion occurs that bo 
of 61-63 might be a variant expression of the slurred b more commonly 
written bw; this is negatived by the presence of the accent in 64-70. 
It is clear that the forms included in this class are in interdependence, 
but to attempt to associate them with bor involves a phonetic prin- 
ciple for which we have not as yet confirmation. Yet it seems quite 
likely, and, subject to correction, I assign this group provisionally 
to the bor type. 
Class J, items 88-91, form a group in which 88 bosu and the lisping 
gi botho show such relation to boro as 89-90 bos to bor. ‘This involves 
the r-s mutation, and my studies in these languages have not dis- 
closed this movement. In view of the fact that we can support the 
mutation It by several instances, and that l-s is mutation in the same 
direction but to a point nearer in the same series, we may give con- 
siderable weight to the Ith mutation occurring in lango—thango from 
Bugotu (The Polynesian Wanderings, 361). 
Class K, 1tems 92-98, is principally found in the northern Solomons, 
but with an interesting instance 96 from Papuan New Guinea, which 
is of peculiar significance because Rossel Island, where the language 
is spoken, is far advanced on the course toward the northern Solomons 
which would be followed by a canoe coasting the south shore of New 
Britain. The seven forms are clearly interdependent; they hint 
somewhat strongly at an association with bor. ‘The forms of class 
G 71-80 seem to come within the same system. We have no warrant 
for the r-m mutation, and in the absence thereof this class remains a 
puzzle. 
Class L, ttems 99-114, falls readily under the type form bo, and 
that is clearly a regular devolution form in the series boro—bor—-bo. 
The initial consonant b is found in 99-109, p in 110-114; in each case 
we have examples of the Melanesian inability to enunciate labials 
distinctly in 102-106 and 111-113. ‘The stem o is found in 99-106 
and 110-113. ‘The vowel changes to u in 109 and 114, to a in 107, 
to e in 108; the stem is here so far reduced that we may not venture 
to affirm that the two latter are really bo variants except in so far as 
these mutations are not unknown, and the persistence of the b points 
in that direction. 
Class M, items 115-118.—The form 115 bobo suggests duplication 
of the bo stem. To pass by mutation to bobu involves a change of 
stem vowel which is quite rare in the mechanics of duplication, yet 
not impossible. The same note, applied to the consonant, holds in 
bawa as a duplicant of 107 ba; but having once admitted this form, baa 
naturally follows. The latter pair seem associable with 130 bawe and 
134 boo. 
