Statistical phenomenon of convergence is common. In order to show this, let us suppose that we have stems consisting of one consonant and one vowel, and for producing a new starter, for a new conception, there is a choice of thirty consonants and ten vowels. The combination of these elements and their permutation offer six hundred different cases; so in this condition, the six hundred first case must be the same as one of the previous. However, the occurrence of similar syllables will take place much earlier, for certain combinations are not convenient for the given complex of articulation. Furthermore, the alternation of consonants is so frequent a phenomenon that they are usually compared by groups. If we take the instance of the so-called Ural-Altaic language, the vowels, being uncertain, are omitted, and the consonants are actually grouped into labials, dentals, glottals, and also liquids and nasals. So the chance of occurrence of the same initial in the monosyllabic words used for the same «ideas» in two «genetically» distinct languages will be as one against five. The comparison of three languages reduces the chance of coincidence, but it will still be high. The combination of two syllables naturally will be more numerous than monosyllabic ones when compared in two «genetically» distinct languages. However, since, in the combination, there are only five elements, the chance of coincidence will be high, for there will be only twenty-five combinations. The chance of coincidence may be increased by the extending o the «meanings» of the stems — the larger the meaning, the greater the chance. When one gives a broad definition of the «meaning of stems» in a group of languages, there is no chance but to find phonetically similar words. Yet, even the fundamental five consonants and vowels are sometimes still reduced, as in the case, for instance, of G. Ramstedt's «law,» treated in Part II of the present work, which formulates the alteration in different languages of labials→glottals→zero. The operation with monosyllabic stems in these conditions implies a meeting of similar words with an extended meaning in almost every case, at least in two of the several groups compared. Although the theory of probability and the theory of combinations and permutations might give a good warning against the comparison of monosyllabic words of distant langupges, yet the experiments in finding them were very often made and are still practised in spite of the fact that the positive result, i.e., the finding of «common» words, proves nothing as to the common origin. Indeed, the same is true in reference to di-syllabic ones.
The spreading of ethnographical phenomena, together with the names over a certain territory, is quite a common phenomenon. They may belong to the so-called cultural phenomena; e.g., araq and its modifications are known from the Atlantic Ocean in Africa up to the Okhotsk Sea. It designates different kinds of alcoholic drinks made of various raw products. It is impossible to establish whether or not the element was originally migrating as in the case of «tea,» or only the method of manufacturing the alcoholic drink was migrating. The word in question spread over different groups living in different conditions, e.g., the agricultural groups of Asia Minor and the reindeer hunters of Siberia. The age of this word and its exact origin cannot be established. It is different with the binding of the phenomenon «tobacco,» which spread over an enormous territory, together with its name (B. Lauf er, «Tobacco and Its Use in Asia»). The number of such words is naturally numerous, for the elements of culture very often are spread from the centre of their invention; and if there is no great phonetic difficulties in adopting the name, it is adopted. In case the word is modified under the influence of particular phonetic conditions, it sometimes cannot be recognized at all. One cannot naturally say how many words have been modified and lost, and how numerous are the words the origin of which cannot be established, owing to their great age. Naturally, the finding of similar words proves nothing as to the origin of the different languages in which they are found.
There are large groups of words denoting local phenomena, such, for instance, as «tundra,» «taiga,» names for local winds and storms, names of local plants and animals, and also special methods and implements of hunting and fishing used for local animals in the given territory. These common words belong rather to the ethnographical complexes of geographical areas than to the ethnographical complexes of ethnical groups. The overlapping of these complexes is a well-known fact. In certain favourable conditions, the complex of the local terms may persist in spite of the changes in culture and even ethnical groups (bearers) and languages. As an indication of the genetic connexion of languages, these words show nothing.
The class of words of onomatopoetic origin is not, perhaps, as numerous as it was sometimes thought, but some cases are evident, as, for instance, the name «coocoo» and the like. Some languages are particularly inclined to produce words of this type. The Manchu language is very rich in different expressions (cf. I. Zaxarov, «Grammar of the Manchu Language») which in many cases are responsible for the origin of derivatives not found in other Tungus languages. However, sometimes the onomatopoetic interjection is not borrowed by the Northern Tungus, but the derivative (verbs and nouns) is borrowed. So in Tungus their origin is very confused. The same external sounding phenomena may produce similar sound-expressions, yet they may be absolutely misleading as to an indication of «common origin.»
The class of words transmitted through imitation is too well known to be discussed here in detail. If there is no great phonetic difficulty, any word which belongs to the influential ethnical unit may spread over other units substituting the old equivalents. The mechanism of their transmission is subject to a great variety, and different reasons for it may be found; e.g., simple fashions, imitation of «superiors,» etc., not to speak of the needs of transactions. The imitation of single words may proceed from one group to another, so that the source of the origin may be altogether lost. This fact may be easily established in the case of neighbouring groups and the adoption of the voluminous complexes of words (e.g., anatomical terms in Latin) in some European languages; but in the case of single words it is sometimes very risky, even in the case where the origin is known but the transmitting links are lost. This class of words is not indicative of the «common origin,» so these words may also become misleading.
The only group of words which can be called «common words» in the sense of the origin from a common language are words directly transmitted from one to another generation within the same group of people. How difficult it is to establish the common words and the operation with them is shown by A. Meillet (cf. «La Methode,» op. cit., pp. 33-42), who says: «Le risque qu'un mot soit emprunte est toujours grand, et l'etymologiste, d'une langue ancienne ou recente, qui raisonne comme si les mots a expliquer avaient a priori toutes chances d'etre indigenes s'expose a des erreurs frequentes» (id., p. 35). The situation with the Ural-Altaic languages is handicapped by another difficulty — the appearance of short stems and a lack of flexion, which «excluent, par leur structure, les demonstrations etymologiques rigoureuses. Il y aura la une methode nouvelle a trouver si l'on veut parvenir a de veritables demonstrations» (id., p. 39). Indeed, in the Indo-European languages the position of the linguist is rather simple, for he has the history of several «languages» (as complex) and the detailed history of populations; while, for most of the Ural-Altaic languages, the only linguistical corroborative evidence is a not-yet-completed dialectology and other cultural and anthropological evidences, the value of which is doubtful from a strict linguistical point of view, as that of A. Meillet and other theoreticians.