¹ AnotLer way of determining from linguistic observations that migrations have occurred is by vocabulary studies, as exemplified in H. Bender, The home of the Indo-Europeans (Princeton, 1922).

² I wish to acknowledge my indebtedness to the following for their helpful comments : A. Spoehr, F. G. Lounsbury, and K. Reichardt.

³ Hereafter referred to as TP. It first appeared as Memoir 90, Anthropological Series No. 13, Geological Survey, Department of Mines, Canada (Ottawa, 1916). It was reprinted in Selected writings of Edward Sapir (ed. David G. Mandelbaum) 389–462 (Berkeley and Los Angeles, 1949). Page references are to the latter.

⁴ TP 456.

^4a Linguistic time depth results so far, IJAL 21.103 (1955).

⁵ As presented for example in his article Lexico-statistic dating of prehistoric ethnic contacts, PAPS 96.452–63 (1952).

⁶ Sapir uses the term ‘equivalent’, TP passim.

⁷ TP 455.

⁸ Provided of course that there is no language boundary between them. Cf. J. Vendryes and E. Benveniste in Les langues du monde (ed. Meillet and Cohen) 58 (Paris, 1952) : ‘Le nordique comprenait, dès le XI^e s., quatre dialectes principaux: le danois, le suédois, le norvégien et l'islandais, celui-ci évoluant indépendamment du norvégien ... Le norvégien (norsk), à peine moins ancien que l'islandais, meurt à la fin du moyen âge en tant que langue littéraire, tandis que le suédois (svensk) et le danois (dansk), écrits à partir du xiii^e s., ont vécu brillamment jusqu'à nos jours.’ The critical word here is ‘dialecte’ : is it being used descriptively as well as historically, or only historically? Conversations with those who know the languages indicate that there is probably no language boundary between any of those on the continent. Icelandic, on the other hand, presents a peculiar problem. It has certainly developed differences which mark it off from what is spoken on the continent, but are these differences enough to let us now call Icelandic a different language?

⁹ If in fact there is a language boundary between them. For a full discussion bearing on this point, see J. Endzelin, Lettische Grammatik 6–8 (Heidelberg, 1923). He says (6), after indicating that the dialect of the Kurs has disappeared : ‘Von den baltischen Sprachen sind untereinander näher verwandt das Litauische und Lettische; einen Übergangsdialekt zwischen beiden haben die Kuren gesprochen ...‘ It seems quite likely that the loss of the transitional dialect would produce a language boundary. O. Weidemann, in his Handbuch der Litauischen Sprache (Strassburg, 1897) shows no hesitation (3) in dealing with the boundary between Lithuanian and Lettish.

¹⁰ Vendryes and Benveniste, Les langes du monde 66: ‘Aujourd'hui le slave méridional comprend, de l'Adriatique à la mer Noire, une succession de parlers dont les limites propres sont souvent très difficiles à déterminer ... En Macédoine, quoique la limite entre les parlers de type bulgare et de type serbe ne puisse être fixée avec précision, la majorité des parlers est nettement de type bulgare, avec forte serbisation’.

¹¹ According to R. Thurneysen, Handbuch des Allirischen 2 (Heidelberg, 1909).

¹² TP 456.

¹³ TP 456 f.

¹⁴ For a demonstration of this point, see the appendix.