INDONESIAN ANNOTATIONS ON THE VOCABULARY. 147 
In this Indonesian assemblage we discover the Proto-Polynesian 
uha in practically unmodified form in 46 oha, the reformed meta- 
thetic Rotuma was in 47 usan, the common Polynesian ua in 48 
uan, and 49 huya is but lightly differenced therefrom. This Proto- 
Polynesian type may be traced without much confusion in a com- 
prehensive chain throughout Melanesia into Nuclear Polynesia and 
thence eastward by purely Polynesian migration movements. ‘The 
remaining Indonesian forms are almost wholly variant forms of a con- 
sistent series of development by the mutation of the central consonant 
from h onward. From 50 to 69 the change is a mutation to the liquid 
semivowels I-r; this mutant occurs sparsely in Melanesia (14-17) on 
the north shore of New Guinea at Mandam, somewhat inland on the 
south shore of the great southeastern peninsula of that island, off the 
south shore of Neu-Pommern, at Buka and Bougainville at the north 
of the Solomons, and widely removed at Gog in the northernmost 
New Hebridean complex. It suggests the traverse through the 
Dampier-Vitiaz exit with a rearward coastal movement to the Rubi 
locus. A strengthening mutation to the mute of the lingual series d 
is found in 74-77, a variant which is absent from Melanesia. ‘The 
intermediate mutation to dj (70-72) appears only in 13 Guadalcanar 
utha, and that may be a secondary development from another 
mutant form. We are therefore justified in the decision that in the 
general migration through Melanesia it was only the uha type that 
was transported, that the Polynesians left Indonesia before mutation 
had begun, and that a much later partial migration in the former 
track after corruption had set in is responsible for the slight trace of 
the ura type. I have included the ut from Mille as showing that 
the migration eastward along the equator carried to Micronesia a 
later form of evolution and thus indicates the relatively modern 
period of Tongafiti migration. 
115. wesch paddle. 
75. fosa East Ceram. 77. bogsay Visayan. 
76. bosoi Celebes Alfuro. 78. fohe Subanu. 
We are still left with a paucity of Indonesian material, but these 
forms show that Polynesians left behind them here, as all along their 
Melanesian traverse, evidence of the early fohe type of stem. 
116. wok boat. 
42. waga Wayapo, Massaratty, Buru. | 47. bangca Visayan, Pampanga. 
43. waha Tobo. 48. fangka Bontoc Igorot. 
44. waa Kayeli, Amblaw. 49. bunka Bouton. 
45. haka Liang, Morella, Batume- | 50. wanga Bolaang, Mongondou. 
rah, Bahasa. 51. wangkang Malay, Macassar. 
46. banca Tagalog. 52. wog Gani. 
This not only preserves the history of the stem in Indonesia, but 
enables us to reconstruct the primitive form. This offers an unsus- 
pected glimpse at the beginning of Polynesian phonetics. 
