When corroborative evidences are practically defective, one turns one's eyes to the hypotheses of a very general character. In fact, when the word or language is spoken of as «common,» it is presumed that the word is transmitted by the mechanism of direct transmission, and it is presumed that the people and the language are intimately connected. For this reason, the linguists are often looking for people, and when linguistical evidences are supported by other cultural evidences, and especially anthropological evidences, then the former are regarded as definite solids. However, this approach of the problem is erroneous in principle. First of all, the transmission of words from one generation to another is nothing but a form of «loan»; moreover, the language is not transmitted as a whole, but transmitted starter by starter, increased with the new elements or decreased, and very often changing from one generation to another. The complex is thus the same, so long as it is spoken without any change, i.e., perhaps less than during one generation. Second, the influx of alien elements through the adoption (intermarriage with neighbouring groups, migration, etc.) may result in a complete substitution of one physical population by another genetically distinct, but the language may persist as well. The same is true with reference to the change of other elements of the cultural complex, which may be substituted by an alien complex, the language being preserved. It is, of course, easier when the ethnical unit, or a group of them, spreads over the territory and brings with it the linguistical complex transmitted from one generation to another. This is exactly the case which is before the eyes of most linguists when they speak about the common origin of words and languages. But this case is only one of the possibilities of «common words» and «common languages.» This has been observed during the periods of great migrations of large masses, as happened in Europe and Asia during the centuries about the beginning of the present era, and as happened with the migrations of some groups, like the English-speaking people to North America, the Spanish- and Portugese-speaking people to South America, the Russian-speaking people to Asia, etc. The life of the ethnical units who transmit their languages from one generation to another is not always disturbed by migrations, but the languages, as more or less similar complexes, may spread through the mechanism of a total or partial imitation by the neighbouring alien groups formerly speaking entirely different languages. So there is nothing which connects the language with the physical bearers, except, perhaps, a certain physical adaptation to the phonetics and a certain psycho-mental complex in which the language forms but an element.
When a group of words and a certain complex of phonetics and methods of showing relationship between the sounding symbols is called by a certain name of a people, it is mere convention, useful merely as a method of classification, but not implying any conclusion regarding «origin» and an actual connexion between the people and the language. On the other hand, the language considered as an entity is also a convention, for no language may be considered as an isolated phenomenon beyond the whole ethnographical and even ethnical complex; but if it is so, it becomes a mere abstraction. Even when the basic contents of a language is well established and it is cleared of the elements enumerated in the first five sources of common words and «loan-words» (a condition absolutely theoretical, for it cannot be practically achieved), it is evident that the projection of a language into the past is an adventurous enterprise which cannot go further than an abstraction, which is dangerous when used as a scientific tool. If another step is made, namely, to connect a hypothetic language with the physical bearers, the chance of mistake is still greater. The search for a language which had been spoken by a certain definite people is certainly easier; for in the equation, out of two unknown, there is at least one known element. It is not surprising, therefore, that if the attempt to establish the place, people, and epoch at which hypothetic pra-languages existed can ever be restored without error, it might never have existed as a phenomenon connected with some physical bearers located in a definite territory. These attempts may be understood as by-products of the theory of the organic character of language and its evolution in the sense known from a modern European complex.
This does not mean, however, that no classification of languages and no attempts at finding hypothetic stems should be made. The classification is needed for further simplification of studies and memorizing facts. Yet the restoration of stems is also helpful, especially when they are used as any other ethrographical elements and with the necessary caution.