Siamese wan ‘day (A2)’ and waan ‘yesterday (A2)' present some phonological as well as etymological problems in Proto-Tai (PT). They represent a rare type of initial which has to be assumed in the Proto-Tai system. The first word has often been assumed to go back to the PT initial w-; thus Haudricourt lists it unhesitatingly under his PT *w-, and Benedict compares it with Indonesian wari. Wulff has noted, however, that the word appears in Dioi as ŋon, thus showing a curious correspondence of Dioi ŋ- to Siamese w-; but this observation of Wulff has been neglected by both Benedict and Haudricourt.

In an article entitled Dempwolff's *R (Lg. 29.359–366), I referred (362 f.) to a difference between Dempwolff and myself in the interpretation of the Ngaju-Dayak material which he assigns to a posited ‘alte Sprachschicht’. What was there characterized as a difference in interpretation can now be shown to involve a fundamental procedure in the application of the comparative method. For this reason it appears worth while to specify the procedure.

This paper seeks to derive historical information from linguistic material by means of statistical techniques. The material is drawn from the Melanesian languages of southeastern Papua. The principal technique employed may be designated the Czekanowski-Klimek technique, after Jan Czekanowski, the Polish anthropologist who first applied it to linguistics, and his pupil Stanislaw Klimek, who introduced it into this country, and from whom the present writer learned it.

The examination of any map of linguistic distributions for an extended area will show some regions of great diversity (e.g. New Guinea, the Nuba Hills in the Anglo-Egyptian Sudan) and others of relative uniformity (e.g. the aboriginal Eastern Woodlands area of North America, the contemporary United States), while still others seem to be intermediate between these extremes. The problem considered here is that of developing quantitative measures of this diversity in order to render such impressions more objective, allow the comparing of disparate geographical areas, and eventually to correlate varying degrees of linguistic diversity with political, economic, geographic, historic, and other non-linguistic factors.

From one point of view, the modern output of phonemic papers may be pretty sharply divided into two kinds, those which consider alternative possibilities and those which do not; the latter are hereinafter termed assertive. The difference reflects neither schools, in the sense of theoretical biases, nor linguistic sophistication, which has not been lacking in the English-language and Americanist fields since these have come to serve as testing grounds for advances in methodology. For assertive phonemicization our examples are from Hoijer, Whorf, Bloomfield, Trager and Bloch and Smith, Swadesh, Harris and Voegelin. Alternative solutions are considered by Wonderly, Harris, Kluckhohn and McLeish, Wolff, Robinett, Trager, and Pike. Recognition of the two contrasting approaches is all that is asked for in these introductory remarks to a study concerned with an alternative dialect phonemicization. The remarks which now follow are intended to be neither an exposition of the weakness of assertive phonemic treatments nor a plea for alternative presentations. (The former includes some of the best linguistic work—where the field worker makes experimental tries at various solutions but presents only that result which turns out to be an experimental success. The latter includes some trivial work—where the alternative presentation is little more than a critique of a previously published solution.)