guage thus revealed to us and the language itself as given in its earliest records, by noting the common Graeco-Italic modifications of the primitive speech.
The original sounds of Indo-Germanic speech may be conveniently tabulated thus :—
unds. Consonants. Vowels. Momentary. Continuous. Unaspirated. Aspirated. Spirant. Xasal. Guttural Palatal .. Surd. k Sonant. 9 Sonant. gh Surd. Sonant. j(y] r n m a i ai u au Lingual . Dental ... Labial ... t P d &(?) dh Hi S v (w)
That the later surd aspirates (kh, th, ph) were developed from the sonant by the influence of the aspiration seems to have been clearly proved by the very careful researches of Curtius. The fact that while Greek has surd (^, 0, <) the Latin representatives of the same are initially the pure aspirate (k) or the spirant /and medially the correspond ing sonants (b, d, g) is enough to show that they had not lost their sonant character in Grseco-Italic times. On the whole the mutes must have remained unchanged. The numerous modifications of the k found in Greek (K, y, TT, T) are not to be traced in Latin, and although p often replaces it in Oscan and Umbrian, the fact that it rarely if ever does in Latin proves that the guttural was unchanged at the time of the separation except in the way of generating a parasitic w (v) or y (j). In the same way Latin shows no trace of the change of g to /?, though it has the parasitic v which sometimes causes the loss of the g ; thus /2ios and vivus point to a Gneco-Italic gvigva-s. It is doubtful whether the primitive language ever used b unaccompanied by the aspiration ; but Greek and Latin furnish us with sufficient words agreeing in this respect Fick quotes 25 to make it clear that this letter was so used before the separation. If Schleicher is right in denying I to the primitive lan guage and this seems very doubtful in face of the facts collected by Curtius (Principles, ii. 174) there can be no doubt that this was abundantly developed by the European unity ; and there seems to be no single instance of an r retained in Latin where Greek has A, while villus as com pared with epiov is the only case of the converse. The spirants undoubtedly remained in their full vigour. But while the Grseco-Italic consonants are on the whole the same as those of the primitive tongue, there is a highly important and significant change in the vowel-system. The original a, retained for the most part in Sanskrit, and modified in Zend only under conditions which make it plain that this is not a phenomenon of very ancient date there, has in Europe undergone a change in two directions. The very valuable paper by Curtius previously mentioned con tains five tables, from which it clearly appears that a (1) is retained in 106 Greek and Latin words ; (2) becomes e (i) in 102 ; (3) becomes o (u) in 56 ; (4) (a) is retained in Greek, becoming e (i) in Latin in 21; (b) ditto, becoming o (u) in Latin in 18 ; (c) remains in Latin, but becomes e iu Greek in 18 ; (d) ditto, becoming o in Greek in 11; while there are (e) 19 words in which Greek e (t) answers to Latin o (u), and (/) 10 in which Greek o answers to Latin e (i). Hence it is abundantly plain that the " thinning " of the primitive a to e and its " dulling " to o must have taken place in the great majority of instances during the Gneco-Italic period. The instances of agreement are three times as numerous as those of disagreement, and most of the latter are to be ascribed to the operation of well-known phonetic tendencies distinctive of the two stocks after their separation. It is worth noticing as to the other members of the European stock that, while there is a striking agree ment in the cases of the retention of the a or of its weak ening into e, this is not found with the third process, the dulling of a into o ; it is therefore legitimate to assume that the first was common to the European family, while the second was specifically Greece-Italic. Thus the numeral odo, the roots gno, "know," mo? , "die," od, " smell," olc (op), " see," and the words ovi-s, poii-s, porko-s, ovo-m, are all Grasco-Italic but not European.
The inflexion of nouns was complete before the time of inflex: the separation of languages. We have no reason to believe of nou that any new case-form was developed either in the Euro pean unity or in any individual nation after this date. The changes are wholly in the direction of loss. The cases which can be shown to have existed, and the terminations by which they were denoted, were as follows :
Singular. Plural. Dual. Nominative -s (s)a-s -(s)d(s}. Accusative -am am-s Ablative -at bhjam-s bhjdm-s. Genitive Locative -as (asja) -i (s)dm-s sva(s) aus(T). Dative -ai -bhjam-s bhjdm-s. Instrumental (i. ) Instrumental (ii. ) . . . (Sociative) Vocative . . -d bhi (no sign). bhi-s
In some cases these were modified according to the ter mination of the stem to which they were suffixed ; and the stems themselves suffered phonetic adaptation to the ter mination. Otherwise there was no distinction of declen sion, except that the fuller form of the genitive was used for the most part in the case of a-stems.
If we examine the changes which may be assumed for Grace the Grseco-Italic period we find (1) the first instrumental Italic case is retained only in a few Greek and (possibly) Latin ad- C 1W1 8 verbs, so that this may be supposed to have dropped out of ordinary flexion ; (2) the ablative is retained in Latin, and hence it was a Grosco-Italic case, though it appears in his toric Greek only in adverbs (Ka/ws, a>?, &c.) ; (3) the loss of the dual in Latin makes it impossible for us to determine exactly the form of its inflexions at this period ; probably they had already become worn down to something like the form in which we find them in Greek ; (4) the existence of a final s in the nom. plur. of o-stems in some Italian dia lects (Old Latin equis, Oscan -os, Umbrian -as, Oscan fern, -as, Umbrian -as, -ar) shows that the analogy of the pro nominal declension had notyet established exclusively the -oi, -ai terminations, though these were doubtless already in use. In the flexion of adjective pronouns there is an agree ment in the nom. plur. (cf. TOI, rat, is-ti, is-ta?) which may be a Graeco-Italic development, the origin of the termin ation being obscure. In the declension of the personal pronouns it is to be noticed that the complete distinction of the stems used iu the first and second persons plural (a^,/xe- y^e-, v /x/x,e- v/j-f-, as compared with nos, ws) proves that the parallel forms asma-, nas, and jusma-, vas to which Sanskrit points as concur rently existing, were still used side by side. The comparison of adjectives was made by the employ ment of the same stem-suffixes (jans or tara, and ta, tama, &c.), though a different selection became the normal one in Greek and in Latin. The inflexion of verbs underwent far greater changes than that of nouns, after the separation, but mainly in the
