GEOGRAPHY OF THE MIGRATIONS. 163 
Gulf coast (Charts VI and VII).—In each language stock we have 
no difficulty in observing three groups of association. The northern 
group is based upon Mekeo, Roro, Uni, Pokau, and Doura. On the 
Melanesian chart we find two strongly characterized pairs, Uni-Roro 
and Doura-Pokau, and these pairs are linked by the frequency of the 
Pokau-Roro association, a line of such strength as appears but twice 
elsewhere on this chart. On the Polynesian chart we find the same 
pairs and the same linking of pairs, but with a marked reduction in the 
weight of the lines. Here we encounter a diagonal linking of the pairs, 
Uni-Pokau, which does not show itself on the Melanesian chart. 
From Roro both charts exhibit lines of affiliation, exclusive of Pokau, 
which extend to Kabadi, Motu, and the Hula-Keapara pair individ- 
ually. ‘These lines are by no means heavy, and the only difference 
between the Melanesian and the Polynesian charts is that the lines 
Roro-Kabadi and Roro-Motu are one unit heavier in the former. 
The next group along this coast comprises Pokau, Kabadi, and 
Motu, with scanty interrelations of the inner Doura with Kabadi and 
Motu. Each chart exhibits the same relative strength of the lines 
Pokau-Kabadi and Kabadi-Motu, so that we may regard the three 
sites as under the same sweep of speech movement. ‘The line Pokau- 
Motu is relatively stronger in the Polynesian chart. Between Pokau 
and Hula-Keapara exclusive of Motu we find traces of interrelation, 
but so scanty as not to call for particular remark. ‘The third group 
associates Motu with the particularly strong association of Hula- 
Keapara-Galoma and brings it at that southern region into contact 
with the interrelation of the inland Sinaugoro and Rubi with the same 
coastal communities. The Melanesian chart shows particularly 
strong association of Hula with Keapara and but a single unit lower 
between Keapara and Galoma; the line Galoma-Rubi-Sinaugoro is 
practically of the same weight as the line Hula-Motu. The Poly- 
nesian chart exhibits a line of the same weight from Motu by way of 
Hula and Keapara to Galoma, a manifestation of a more even dis- 
tribution of the influence which has been at work; the inner associa- 
tion with Rubi and Sinaugoro is distinctly less prominent. ‘The final 
note upon the two charts is that the eastward extension of these stems 
into other provinces is greater in the Polynesian material than in the 
Melanesian, and that implies far greater extension relatively to the 
varying amounts of material. 
South Cape (Charts VIII and IX).—The examination of the two 
language stocks in the matter of the draft of material from the Gulf 
Coast province makes clear a difference, even though the significance 
thereof must for the present elude our study. The Melanesian chart 
presents the lines of this draft for each of the terminal points. We 
find Suau the principal terminus, Mugula and Sariba at the ends of 
lines of equal weight, Tubetube somewhat inferior, while the material 
