Now the Samoan Archipelago is very nearly the point from which we should have derived the _proper_ Polynesian population, had we taken the course of the Papuan islands; so that it constitutes a point wherein the two lines meet. Hence, if upon historical, philological, or any other points of external evidence, we gave a preference to the Samoan Archipelago, over the Tonga group, as the source of the population for other parts of Polynesia Proper, we should reduce the _general_ question as to the original of South Pacific islanders to that of the origin of the Samoans. This, however, is a matter of detail, of less importance than the recognition of the necessity of making the geographical continuity of the chain which connects the Polynesians with the Continent, agree with the ethnological. This can only be done by deriving the Polynesian population from Micronesia. In this case the stream of migration goes _round_ the Kelænonesian area, and not _across_ it.
The rule of taking, as lines of insular migration, those series where the _maximum_ interspaces of ocean are the smallest, has already been twice insisted on, and in both cases it has been qualified by the indication of particular reasons, which might, in certain cases, lead us to depart from it. These reasons have not been exhibited in detail. Two sorts, however, of them have occurred, as it were spontaneously, _i.e._, in the natural course of our investigations. These showed themselves, first in the preference given to Timor over New Guinea, as the origin of the Australian population; and next, in the case of Polynesia, just discussed. A _third_ sort will now present itself, _i.e._, _the effect of winds and currents_; since it is clear that it is easier to pass over a large interspace of sea with wind and current (one or both) in your favour, than over a small one with either one or both against you.
_The prevailing winds in the Pacific are against a line of insular migration, being from west to east, at all; since for three fourths of the year they blow from America towards Amphinesia rather than from Amphinesia to America._
_Valeat quantum._ All that can possibly be got would be a chance of three to one in favour of an American origin for the Polynesians, _provided that all other conditions were equal_. But this is not the case; the _a priori_ probabilities are neutralized by a vast difference in the maximum interspaces of ocean, and by the non-American character of both Micronesia and Polynesia.