INDONESIAN ANNOTATIONS ON THE VOCABULARY. 107 
We have hitherto noted that the alo stem fails to reach the Poly- 
nesian languages of the present, but we must regard it as certain that 
this stem was included in the speech of migrants out of Indonesia 
for some distance along the track followed by a Polynesian swarm. 
In the Melanesian collation we find the stem unmistakably in the 
New Hebrides, along Torres Straits, in the Dampier-Vitiaz Straits, 
in the northern Bismarck Archipelago, and probably in the Arop-Sér 
lagoons on the north coast of New Guinea. In Indonesia the alo 
stem is clearly recognizable only among the Alfuros of northern 
Celebes. Concurrently with a clear alo runs a stem which in its 
simplest form appears as lea of Massaratty, and with a nasal suffix 
of some sort, as lean, this enters the alo region of Celebes. In the 
two Bahasa forms leamata and riamatai we encounter a second sun 
word of the type form mata which is of wide occurrence in the Indo- 
nesian family, its presence in determinant compaction with this lea 
and with other stems showing that neither lea nor the others carried 
the specific sun sense with sufficient strength to survive in all cases. 
We are unable to establish any community of source for lea and alo; 
furthermore, we can not establish either as having the same origin as 
the general Polynesian la; the only element common to the three is 
the liquid consonant, and this single point of agreement is far too 
slender to serve as a link. For the alo forms we have no difficulty 
in setting forth a migration track from northern central Indonesia 
along the north coast of New Guinea, on a course offshore from that 
island to the northern members of the Bismarck Archipelago, on a 
course alongshore through the Dampier-Vitiaz exit, and by reverse 
coastwise sailing to the south shore of New Guinea and by remoter 
voyaging to a yet more distant landfall in the New Hebrides on the 
course which I have proposed as the Viti stream. ‘The lea stem is 
found in the eastern division of Indonesia in a position which might 
lead to transmission along either course about New Guinea. 
12. at stone. 
60. fatu) Massaratty. 65. watu) Magindano, Savo, Maronene, 
61. fahou Satawal. Kolon, Kei, Banda, Celebes 
62. vato Malagasy. Alfuro. 
63. batu) Malay, Kayan, Silong, Macas- | 66. wadu Bima. 
sar, Togean, Ceram, Rumbia, | 67. hatu Ceram. 
Mengkoka, Bouton, Bahasa, | 68. hatul Bahasa. 
Celebes Alfuro. 69. haul Bahasa. 
64. bato Ilocano, Subanu, Visayan, Bon- 
toc Igorot. 
We observe here the remarkable closeness with which the Indo- 
nesian forms cling to the Polynesian original; indeed, there is far less 
variety than has been observed in the intricacies of Melanesian speech. 
Bahasa 68 hatul offers an assumed consonant for which we find else- 
where no parallel. The form is important, however, for it establishes 
