MELANESIAN ANNOTATIONS ON THE VOCABULARY. 83 
I am not prepared to state that closed syllables are primordial. 
My hypothesis of the evolution of speech by the application of con- 
sonantal modulants leads more and more distinctly to the belief 
that the earliest employment of consonants was their initial appli- 
cation to the vowels. Of course it is not beyond the bounds of possi- 
bility that a later stage of development hit upon the device of adding 
a final consonant. At present it is not feasible to determine that 
point. However, we have many instances in which a stem now open 
in Polynesian appears closed in many Melanesian languages through 
the presence of a final consonant and that, in its turn, is clearly a 
development through abrasion from a former open stem. For an 
example of this we need look no further than the next preceding item 
(76 pul) where, in items 3-11, we have the series pakasi—mpokas— 
puka. Much study must be put upon this problem before we can 
arrive at a satisfactory determination; here I do no more than note the 
possibility that closed stems have arisen through final abrasion of 
the vowel of an added open syllable. Along this line of reasoning 
danum is not to be represented by BACEc but as BACED(1) and the meta- 
thetic damun is not BAcEC but BADEC(I); this signifies that all conso- 
nantal metathesis consists solely of the interchange of initial conso- 
nants of succeeding syllables. 
Having cleared away the complications introduced by the meta- 
thesis the series runs for some distance with satisfactory smoothness. 
Items 1-5 exhibit the closed ranum type, and in 2, 4, and 5 we note 
the mutation u-i in the latter syllable, and to this series belongs 8 
ramtn. Items 6, 7, 9-11 exhibit the open ranu type with the u-i muta- 
tion in 7. From 12 to 17 we find readily comprehensible variants 
of the ran type, merely ranu which has undergone final abrasion of 
vowel to be adjusted to speech in which the closed type is preferred. 
This is evidential that the stem was introduced to these regions by 
folk who employed ranu and from whose memory had passed the recol- 
lection of an earlier, if indeed earlier, stem danum or ranum (?). The 
rd mutation is frequent in these languages; in this series it is found 
in 1-5, 15-17. At this point I have inserted the distinctly northern 
New Guinea type 18-22, in which we recognize a ran type with that 
duplication of vowels which we note in speech of this particular group 
and whose explanation we are not yet far enough advanced in the 
knowledge of the linguistics of the region to comprehend. ‘The items 
23-27 may stand as of the ran type after yet further final abrasion. 
The Namarodu 27 tach is anomalous in the assumption of a final 
palatal; we note that Friederici reports it as less positive than the 
palatal of German nach. ‘The form 26 ta-va, somewhat widely dif- 
fused on the Gazelle Peninsula in New Britain, is explained by this 
author as a compaction of the water words ranu and vai. In this 
explanation it corresponds to the Viti wai-dranu already mentioned, 
but the elements are set in different order. 
