800 ARYAN RACE AND LANGUAGE grammatical forms made and parts of speech distinguished. The addition of pronominal endings to verbal roots made a verbal tense in three numbers (the dual perhaps of later origin than singular and plural, and mostly lost again in the later languages), with three per- sons in each. The prefixion of an " augment " (doubtless a pronominal adverb, meaning "then") made of this a past tense; but this augment-preterite has left only scanty and doubtful relics, except in Indo-Persian and Greek. Another past tense, or perfect, was formed by reduplicating the roots, apparently to signify completed action. This is the original of the Greek and Latin perfects, our (" strong " or irregular) preterite, &c. Futures were made later, with auxiliary verbs; one, from t, "go," apparently passed over into a modal use, as an optative, and was succeeded by another, from as, "be." A subjunctive mood, of more doubtful derivation, was added ; and an imperative, probably limited at first to the second person. This, along with parti- ciples or verbal adjectives (for the develop- ment of distinct infinitives, verbal nouns, was probably later), appears to have been the whole primitive structure of the simple verb ; a cau- sative conjugation, besides, has had important developments in the derived tongues. The de- clensional inflection (of nouns, adjectives, and pronouns) distinguished also three numbers, and (including the vocative) eight cases in each number ; or, besides the six we know in Latin, an instrumental, denoting with or by, and a locative, denoting in. Into this inflection was introduced further a distinction of sex : first, by the special characterization of a feminine ; later, by the additional separation of a neuter (which in general differs from the masculine only in nominative, accusative, and vocative). From the original basis of sex, however, there was a very wide departure even in the primi- tive period, the system of grammatical gender becoming very complicated and artificial. The declension of pronouns was in many points ir- regular ; there were as yet no relative pro- nouns, that order having grown later out of the demonstratives or interrogatives. Of the other parts of speech, the adverbs alone were j a fully formed class ; prepositions were still only adverbial prefixes to verbs ; conjunctions were very few, and only the merest connec- tives, the construction of sentences being of the simplest character ; articles did not come j into existence till comparatively modern times. Numerals had been produced at least up to a hundred ; as to thousand, the case is very doubtful. The apparatus of noun and adjec- tive derivation, in both primary and secondary suffixes, was already elaborated in its principal features ; conspicuous examples are the endings of comparative and superlative, and of par- ticiples. This primitive structure of Aryan language has been variously modified, reduced, and added to, in the later history of the family ; it was most fully and distinctly preserved in the Sanskrit, which on that account casts most light upon the common history of all ; but there are points in which each branch leads the rest. On the whole, the tendency has been toward a reduction of earlier synthetic structure, and a prevalence of analytic methods of expression, by the substitution of prepositions for case- endings, and of auxiliaries for verb inflections, and by the multiplication of relational words. Of all the languages of the family, the English has gone furthest in this direction. Aryan language is called distinctively " inflective " (as is also Semitic, although its inflectiveness is of a very different character). By this is meant that it not merely forms combinations from elements originally independent, reducing one of them to a subordinate position, as " forma- tive element," indicating a modification or re- lation of the other or radical element ; but that it also peculiarly integrates or unifies the com- bined elements, losing sight of and disguising their separate individuality, and even allowing the radical part to become modified within itself by the addition of the other : thus, San- skrit rid, veda, vaidika; Greek fatrru, Ihnrov, Ittoura ; Latin fid, fido, fcedus ; English (where the endings have disappeared) sing, sang, sung, song. It is matter of dispute among linguists whether such changes are by origin purely phonetic, or symbolically significant : the former appears to be the better opinion. The unity of Aryan language necessarily im- plies the former existence of a unitary Aryan people ; that is, at some time in the past there must have existed somewhere in the world a single limited community, in whose use the language above described grew up and took shape, and by whose extension and separation it became so widely spread and so diversified as we find it actually to be ; but when and where, it is impossible to say with any definite- ness. The greatest antiquity we can attain in the history of the family is 8000-2000 B. C., when the Indians and Persians formed together one people. The oldest parts of the Vedas may be as old as 2000 B. C. ; of the Avesta, considerably less. We have no trustworthy scale as yet with which to measure chrono- logically the changes of language; and our opinions of the time of Aryan unity must be governed greatly by our opinions as to the wider question of the antiquity of man ; the former is variously estimated at from three up to five or ten thousand years before Christ. As to the place, current opinion is inclined to fix it on the highland of central Asia, near the head waters of the Oxus and Jaxartes ; and this situation is claimed to be pointed out by the evidence of the language itself. That, however, is by no means the case; the mere fact that Indo-Persian language is less changed from original Aryan than is the speech of any other branr.fr' which is all the linguistic evidence that can iV^lleged) does not at all prove that the Indo-P8rsian common abode is nearest to the original abode of the family ; the changes
