tion and become the mere exponents of grammatical ideas. Professor Earle divides all words into presentive and sym bolic, the former denoting objects and conceptions, the latter the relations which exist between these. Symbolic words, therefore, are what the Chinese grammarians call " empty words," words, that is, which have been divested of their proper signification and serve a grammatical purpose only. Many of the classificatory and some of the flexional suffixes of Aryan speech can be shown to have had this origin. Thus the suffix tar, which denotes names of kinship and agency, seems to come from the same root as the Latin terminus and trans, our through, the Sanskrit tar-ami, " I pass over," and to have primarily signified " one that goes through " a thing. Thus, too, the English head or hood, in words like godhead and brotherhood, is the Anglo-Saxon had, " character" or " rank ;" dom, in kingdom, the Anglo-Saxon dom, "judgment ;" and lock or ledge, in wedlock and know ledge, the Anglo-Saxon fdc, "sport" or " gift." In all these cases the " empty words," after first losing every trace of their original significance, have followed the general analogy of the language and assumed- the form and functions of the suffixes with which they had been confused. A third mode of representing the relations of grammar is by the symbolic use of vowels and diphthongs. In Greek, for instance, the distinction between the reduplicated present Si($o)/u and the reduplicated perfect Se Sw/<a is indicated by a distinction of vowel, and in primitive Aryan grammar the vowel a seems to have been set apart to denote the subjunc tive mood just as ya or i was set apart to denote the poten tial. So, too, according to M. Hovelacque, the change of a into i or u in the parent-Aryan symbolized a change of meaning from passive to active. This symbolic use of the vowels, which is the purest application of the principle of flexion, is far less extensively carried out in the Aryan than in the Semitic languages. The Semitic family of speech is therefore a much more characteristic type of the inflexional languages than is the Aryan. The primitive Aryan noun possessed at least eight cases, nominative, accusative, vocative, instrumental, dative, genitive, ablative, and locative. M. Bergaigne has attempted to show that the first three of these, thj "strong cases" as they are termed, are really abstracts formed by the suffixes -as (-s), -an-, -m, -t, -i-, -a, and -ya (-i), the plural being nothing more than an abstract singular, as may be readily seen by comparing words like the Greek CTTO-S and o7re-9, which mean precisely the same. The remaining ; weak " c ises, formed by the suffixes -sma, -sya, -syd, -yd, -i, -an, -t, -bhi, -sit, -i, -a, and -a, are really adjectives and adverbs. No distinction, for exunple, can be drawn between "a cup of gold " and "a golden cup," and the instrumental, the dative, the ablative, and the locative are, when closely ex amined, merely adverbs attached to a verb. The termin ations of the strong cases do not displace the accent of the stem to which they are suffixed ; the suffixes of the weak cises, on the other hand, generally draw the accent upon themselves. According to Hiibschmann, the nominative, accusative, and genitive cases are purely grammatical, distinguished from one another through the exigencies of the sentence only, whereas the locative, ablative, and instrumental have a logical origin and determine the logical relation which the three other cases bear to each other and the verb. The nature of the dative is left undecided. The locative primarily denotes rest in a place, the ablative motion from a place, and the instrumental the means or concomitance of an action. The dative Hiibschmann regards as "the cise of the participant object." Like Hiibschmann, Holz- weissig divides the cases into two classes, the one gram matical and the other logical ; and his analysis of their primitive meaning is the same as that of Hiibschmann, except as regards the dative, the primary sense of which he thinks to have been motion towards a place. This is also the view of Delbriick, who makes it denote tendency towards an object. Delbriick, however, holds that the primary sense of the ablative was that of separation, the instrumental originally indicating concomitance, while there was a double locative, one used like the ablative absolute in Latin, the other being a locative of the object. The dual was older than the plural, and after the develop ment of the latter survived as a merely useless encumbrance, of which most of the Aryan languages contrived in time to get rid. There are still many savage idioms in which the conception of plurality has not advanced beyond that of duality. In the Bushman dialects, for instance, the plural, or rather that which is more than one, is expressed by repeating the word; thus tu is " mouth," tutu "mouths." It may be shown that most of the suffixes of the Aryan dual are the longer and more primitive forms of those of the plural which have grown out of them by the help of phonetic decay. The plural of the weak cases, on the other hand (the accusative alone excepted), was identical with the singular of abstract nouns ; so far as both form and meaning are concerned, no distinction can be drawn between OTTCS and C TTOS. Similarly, humanity and men signify one and the same thing, and the use of English words like sheep or fish for both singular and plural shows to what an extent our appreciation of number is determined by the context rather than by the form of the noun. The so-called " broken plurals " of Arabic and Ethiopic are really singular collectives employed to denote the plural. Gender is the product partly of analogy, partly of pho netic decay. In many languages such as Eskimo and Chock- taw, its place is taken by a division of objects into animate and inanimate, while in other languages they are separated into rational and irrational. There are many indications that the parent-Aryan in an early stage of its existence had no signs of gender at all. The terminations of the names ol father and mother, pater and mater, for example, are exactly the same, and in Latin and Greek many diph thongal stems, as well as stems in i or ya and u (like vavs and VCKUS, TroAis and Xts), may be indifferently masculine and feminine. Even stems in o and a (of the second and first declensions), though the first are generally masculine [ and the second generally feminine, by no means invariably maintain the rule ; and feminines like humus and 680?, or I masculines like advena and TroXtVr;?, show that there was a time when these stems also indicated no particular gender, but owed their subsequent adaptation, the one to mark the masculine and the other to mark the feminine, to the in fluence of analogy. The idea of gender was first suggested by the difference between man and woman, male and female, | and, as in so many languages at the present day. was represented not by any outward sign but by the meaning of the words themselves. When once arrived at, the concep tion of gender was extended to other objects besides those to which it properly belonged. The primitive Aryan did not distinguish between subject and object, but personified objects by ascribing to them the motives and powers of living beings. Accordingly they were referred to by different pronouns, one class denoting the masculine and another class the feminine, and the distinction that existed between these two classes of pronouns was after a time transferred to the nouns. As soon as the preponderant number of stems in o in daily use had come to be regarded as masculine on account of their meaning, other stems in o, whatever might be their signification, were made to follow the general ana logy, and were similarly classed as masculines. In the same way, the suffix i or ya acquired a feminine sense, and was set apart to represent the feminine gender. Unlike the
Semites, the Aryans were not satisfied with these two
