tive, nominative, and accusative are connected together much more closely than the remaining cases ; they coincide in the neuter gender, and no one of them ever interchanges with or becomes equivalent to any one of the other group. On the other hand, in Sanskrit the ablative often coincides with the genitive, and the locative (in the dual) with the genitive or dative, while in Greek the instrumental is re placed by the dative, in Latin by the ablative ; dative and genitive coincide in the Greek dual, dative and ablative in the Latin plural, and the locative always in Latin coincides in form with genitive, dative, or ablative. The vocative may be regarded as a relic of the preceding uninflected stage. The nominative and accusative are closely connected with theme-formation, and seem to have been but a new develop ment of the same principle. From a root svap, " sleep," came, as has been seen, at an early stage svap-na, "sleep ing"; from kar, "make," came kar-ta, "made." It was only an extension of the same method when the pro nominal sa and ma were added to the themes thus formed. Nominal inflexion was created as soon as it came to be re cognized that the last additions were movable, and that the same stem might, according to circumstances, appear with one or the other or with neither. The fact that -TO is found as the suffix of the nominative in some pronouns (e.g., Sanskrit aha-m = e yw-i/, tva-m = Twrj, &c.) seems to point to a time when this was used as a determinative for nominative and accusative alike ; but it soon became specialized as a characteristic of the latter. There is reason to believe that this process was facilitated, if not occasioned, by the use of the m-suffix to denote gender, or more strictly the absence of gender, in neuter nouns. It was only natural that the same suffix which distinguished the theme as a living being should be applied to mark it out as the subject or source of an action, while, conversely, that which denoted the absence of life should be used to mark the object. It is no improbable conjecture which finds in this accusative character of the sign of the neuter the ex planation of the ordinary Greek idiom which constructs a neuter plural substantive with a singular verb ; ra a>a Tpe ^ei, " the animals are running." Further, the wide and varied usage of the accusative case in Greek appears to point to a time when it was the only oblique case. At a later period the second group of cases made its appearance ; this includes at least the genitive, ablative, dative, locative, instrumental, and sociative. Whether we are also to regard the various terminations which appear in some adverbs, which cannot be referred to any one of these, as originally case-suffixes is a question not easy to determine, and one which is, after all, rather one of terminology than of any real importance. The theory of the purely local force of the cases, attractive as it is at first sight from its simplicity, and its apparent conformity with the sound theory which bids us, in dealing with language, proceed from the concrete to the abstract, and not vice versa, breaks down when we come to apply it in detail. For the genitive, at any rate, it is much safer to postulate an original adjectival force, a view borne out both by striking similarity of formation in some instances (cf., e.g., Sry/xo-o-io, the earlier form of the Homeric S-^aoio, the Attic 8-^/j.ov, and S^d-o-io-?, "belong ing to the people") and by numerous analogies from various languages. It has even been conjectured, though perhaps on inadequate grounds, that the genitive had originally the final s, which was dropped only when the sense of its origin became obscured. In the ablative we have apparently a use of the pronominal element -ta corresponding to that of -sa in the genitive, and originally in the nominative, the a being afterwards dropped, so that vdk-a-s = vocis is to vdk-a-t = voce(d) as ja-s = is is to ja-t = i-d. The syntactic force of the ablative may often be represented as adjectival ; and the differentiation of the two cases may well be a product of later times. The earliest forms of the other cases, the formation of which has not hitherto been satisfactorily explained, will be pointed out below.
In the seventh period assumed by Curtius we have the Advert petrifaction of some forms of particular themes with case- an d P r < suffixes, which were no longer declined throughout, and P osltio: thus gave rise to adverbs and prepositions. The adverbial force was undoubtedly the earlier, as we can see from in dications in the Homeric poems ; the prepositional force came later, first perhaps in connexion with verbs, and afterwards as governing cases. To the same period pro- bably belongs the singularly interesting form of petrified cases presented by infinitives. These have long been re cognized as cases of verbal nouns (nomina actionis) no longer inflected throughout. The agreement of the cognate languages in the use of this device for extending the range of language seems to be a sufficient indication that it had been introduced before the original unity broke up. At the same time the great variety of the forms actually selected by different languages as the basis of this con struction is a clear proof that no well-defined system of infinitives had then been brought into use.
Such were the stages by which, according to our greatest living authority, that language grew which was destined to be the mother, not only of Greek and Latin, but of al- most all the tongues in which human culture has found an utterance. It is by no means impossible to reconstruct it, at least in outline, as it must have been spoken before the original unity broke up. This task has been attempted, so far as its phonetic laws and inflexional forms are concerned, by Schleicher in his well-known Compendium der ver- gleichenden Grammatik der Indogermanischen Sprachen (3d edition, Weimar, 1871) ; and its vocabulary has been recon structed by Fick in his Vergleichendes Worterbuch already referred to, Schleicher indeed ventured to narrate a brief story in this primitive language (Kuhn s Beitrdge, vol. v. pp. 206 sqq.). 1 On particular points he may well have been mistaken. The tendency of modern philology is to admit within the period of the united national life a fuller develop ment than that assumed by Schleicher. Several scholars, working along different lines of research and entirely independently, have established the great probability of a bifurcation of the gutturals ; and it is by no means certain that the vowel system was not already becoming more rich and varied. We have probably to admit that dialectic dif ferences already existed, such as could hardly have failed to arise, even before the nation broke xip completely, so soon as it attained any considerable magnitude. And above all it must never be forgotten that we are dealing with the pro ducts of a period to which chronological limits cannot well be fixed, but which language gives us strong reasons to be lieve must have been at least as long as that to which the data of other branches of anthropology appear to point. It is impossible to be sure that all the elements which are introduced were ever strictly contemporaneous. Our review of the history of the language thus far is enough to show that one form may have begun to show traces of phonetic decay at a time when another form was not yet created. Hence M. Bre"al (Melanges, p. 376) does well to warn us against the common error of philologists in endeavouring to get more out of the reconstructed " primi tive speech " than the facts on which it is based will warrant. But used with discretion it affords a highly con venient means for stating the results to which the com parison of languages brings us. For our present purpose it will be well to mark one intermediate stage between the source of the Greek lan-
1 Mr J. P. Postgate has published a similar composition (Academy, June 14, 1879), re-written by Mr T. C. Snow (ib., June 28) on the principles of Bruginaa and De Saussure.
