¹ James W. Marchand, Was there ever a uniform Proto-Indo-European?, Orbis 4.428 (1955), referring to my article: Family tree, wave theory, and dialectology, Orbis 2.67-72 (1953).

² F. W. Householder Jr., in his review of the Studies ... Whatmough (The Hague, 1957), Lg. 34.401 (1958), referring to my article in these Studies 239-52: Linguistic expansion and diversification.

³ N. S. Trubetzkoy, Gedanken über das Indogermanenproblem, Acta linguistica 1.81-9 (1939). Similarly Franz Altheim, Sein und Werden in der Geschichte, Die Gestalt 21.50 ff. (1950).

⁴ This is denied by, among others, Paul Thieme, Die Heimat der indogermanischen Gemeinsprache 59 ff. (Abhandlungen der Akademie der Wissenschaften und der Literatur, Mainz, 1953.11). See also Ernst Pulgram, The tongues of Italy: Prehistory and history 140 ff. (Cambridge, Mass., 1958).

⁵ Roland G. Kent, in his review of V. Pisani, Linguistica generale e indeuropea (Milan, 1947), Lg. 24.194 (1948).

⁶ W. F. Twaddell, The prehistoric Germanic short syllables, Lg. 24.139 (1948).

⁷ The illegitimacy of equating method with fact applies also to some procedures of descriptive linguistics. Persons preoccupied with the design and construction of linguistic computers (translation machines, for example), have been heard to analyze languages in very peculiar ways, quite different from what most would normally call standard linguistic analysis. Well and good, and no harm done—unless we are asked to believe that this is the only and the best way of operating. The truth is that such an ad-hoc analysis, conditioned by the requirements of the computer, describes not the language but the machine.

⁸ It would be interesting to undertake a reconstruction of Proto-Romanic, that is (if I continue in terms of my present context that deals with the closest predivisional parent language), of Latin, to be accomplished by means of a comparison of the Romanic languages, but under the pretense that Latin is as unknown an article as Proto-Indo-European. I am certain that many of the resulting asterisked forms would correspond to no kind of known Latin, and that many other lexical items of this uniform reconstructed Proto-Romanic, looked at from the Latinist's point of view, would belong to different local and social dialects of different eras. Let us not forget that no amount of reconstruction would deliver, for example, more than two cases for the noun and four for the pronoun. This means that this reconstructed Proto-Romanic would, at best, correspond in this regard to some unknown and unattested koine of a late date, and not at all to the predivisional proto-language of the Proto-Romanic Urheimat—that is, in the sense in which one invariably speaks of the Proto-Indo-European Urheimat, to prehistoric Latium of perhaps the end of the second millennium b.c. Similarly, how can one be sure that the Proto-Indo-European actually spoken in the authentic Urheimat had, not the eight cases which reconstruction provides, but rather nine, or fifteen?

⁹ Although I favor a certain degree of terminological hygiene, I trust I shall not be charged with engaging here in an essentially fruitless terminological discussion of ‘Proto-Indo-European’. A great deal more is at stake. In particular, problems outside linguistics have come to be involved, dealing with prehistoric ‘Indo-Europeans’ and their culture, which are incapable of solution on the basis of Reconstructed Proto-Indo-European.