104 SISSANO. 
. I. ai wood. 
73. kayu) Malay, Baju. 80. aow’ Buru Alfuro. 
74. kaju) Tombulu. 81. hazo Malagasy. 
75. kayao Bontoc Igorot. 82. kai Tombulu, Teor. 
76. gayo Subanu. 83. ai Bahasa. 
77. cahoy Subanu, Visayan. 84. ai-ie Ceram Alfuro. 
78. kau Buru Alfuro. 85. ail Bahasa. 
79. kao Buru Alfuro. 
Here also we find the three stems which have been distinguished in 
all our former inquisition into this vocable. ‘These are ai (82-85), 
au (73-80), asu (81). We have noted that stem ai is numerically the 
most frequent in New Guinea and Melanesia; in Indonesia we find 
but four occurrences of its use, three in the eastern subdivision close 
to the northwestern tip of New Guinea, and Tombulu not far remote 
in Celebes; in the area of Polynesian settlement it occurs only in Fiji 
and Rotuma. ‘This material adds nothing to the comprehension of 
the position of the asu stem; 81 hazo, of which we note a variant hazu, 
approximates 44 kasu and 45 gazu and in the initial aspiration cor- 
responds with 46 hasie; it is possible that 77 cahoy from the southern 
Philippines is associable with this stem, for thus can we better account 
for the presence of the inner aspirate as a mutation product of the 
sibilant, and from the fact that this is common to Subanu and Visayan 
we may regard it as later than the Subanu 76 gayo. In the paucity 
of the material we can do no more than to note the sporadic char- 
acter of the occurrence of the stem, Madagascar, Philippines, Solomon 
Islands, New Hebrides, a single instance in each particular locus, and 
(47-49) two very doubtful instances on the Torres Straits shore of 
New Guinea. ‘The austem appears unmistakably in the three citations 
(78-80) from the Alfuro of Buru, and is there observed in the nude 
form (80), but also with the modulant k, which is largely found in 
the New Hebrides and is the dominant form in Polynesia. ‘The 
group involving the semivowel y in the interior position (73-77) seems 
best associable with au, in which case we regard the semivowel as no 
more than a light septum phonated in consequence of the recognized 
need of preventing the coalescence of u with the preceding a; if, on the 
other hand, the phonetic value of the semivowel is regarded as vocalic, 
we should have the suggestion of a connection with stem ai seriously 
complicated by the problem of explaining the final vowel or vowels. 
If these forms are regarded as part of the au stem we shall set apart 
this stem in the same dominant position in Indonesia as in Polynesia. 
4. ain to eat. 
43. cana Subanu. 51. caon Visayan. 
44. kane Bahasa. 52. ijénan Massaratty. 
45. mo-konie Togean. 53. anee Ceram. 
46. kan Tontemboan. 54. ane Bahasa. 
47. ma-kan Malay, Bontoc Igorot. 55. niane Ceram. 
48. mangan  Bontoc Igorot. 56. en Ceram. 
49. kaan Ceram-Bonfia. 57. aa Bahasa. 
50. gaan Subanu, Solor. 58. a Bahasa. 
