GEOGRAPHY OF THE MIGRATIONS. 159 
judgment, itself resting upon the detailed study of speech material 
along the Melanesian chain, that in ‘The Polynesian Wanderings”’ I 
rested my conception of the two courses of migration, the Samoa 
stream and Viti stream, as denominated by their respective terminal 
points. It is possible that at a few points there may have been some 
currents leading from one current to the other; one such we shall have 
shortly to examine, but in the main the reexamination in this work of 
the earlier material with considerable additions tends to confirm my 
view as to this divarication of migration streams. 
It should be made clear at the outset that our linguistic material 
from New Guinea is as yet scanty, but as other ethnical material is 
still less available, we are forced to set our greatest reliance upon the 
comparison of languages. As between the north coast and the south 
we have about the same number of stations of speech record; yet 
there are two factors of difference which are somewhat influential 
upon the results which we may derive. It is only the north coast 
series of tongues which comes to us in any larger bulk than mere word 
lists, since for the Bongu we have a rather considerable volume of 
vocabulary. The recording stations along the north coast are so 
spaced that we are able to obtain a somewhat comprehensive view 
from Geelvink Bay to and into the Bismarck Archipelago. ‘The 
recording stations along the south coast are compactly grouped in a 
comparatively restricted area from the eastern shore of the Gulf of 
Papua to the Louisiades. This leaves us a great gap between the 
Gulf of Papua and the Arafura Sea and the southern tier of Malayan 
islands from which we have no speech records whatever. In the 
inconvenience of this fact there is nothing which argues against the 
passage of the Viti stream through Torres Strait; equally there is 
nothing confirmatory. A still more important difference between 
the speech records of the north coast and those of the south coast is 
that in the latter the Polynesian element is appreciably nearer to the 
type normal to that speech family. We are not in possession of 
sufficient material to admit of a comparison by quantity, but the 
superior quality of the southern material is immediately manifest. 
We shall best begin our examination of migration courses relative 
to New Guinea as a wedge-conduit by looking at the particular area 
in which migration tracks following along the north shore might be 
led into reunion with such as may have followed the south shore out 
from Indonesia. We may regard the Sissano-Arép lagoons as a some- 
what critical point in the northern coasting voyages. At some point 
in this vicinity the seamanship of such voyaging would lead to a 
point of departure for those fleets which under the actuation of any 
of several causes might relinquish the mere skirting of land in sight 
and put out upon the northern sea. On some such theory of diversion 
we incline to account for a certain, though not complete, element of 
