CH. VI.

perception of the distinction of sounds was nice and cultivated. Yet it is quite clear that the Indo-European race before its separation did possess a rising scale of all the vowel-sounds. This scale has been mentioned before': it may be repeated here.

The inten

sified vow el forms differ in different languages.

Most recog

nisable in Sanskrit; and most regularly employed there.

u,

These new sounds were employed by the different peoples of the original stock under different forms according to their various phonetic laws, and with more or less of system and precision according to their different gifts. A complete list of all the substitutes is given by Schleicher2. Those employed by the Greek and Latin will come immediately under our fuller consideration. We may glance for a moment at those of some of the other nations, so far as they employed them.

First, then, Sanskrit remained the closest to the system of the original speech, only varying indeed from it by substituting é and ô for the first steps of the I and U scales, respectively: in the a-scale it has not attained to any means of distinguishing the first or second steps; indeed the Indian grammarians say that there is no Guna of a, only Vriddhi, that is, no first step, only a second one3. One of the most important uses of the scales is the formation of nominal bases primary and secondary: thus from vid, "to know," comes by regular ascent the wellknown word Veda: and the second step (together with the

1 See p. 48.

2 See Comp. p. 160.

3 These terms Guna and Vriddhi have become to a certain extent familiar to those who are not Sanskrit scholars, by their occurrence in treatises on comparative philology-more especially in England from their being used by the late Dr Donaldson. They have been, with good reason, generally rejected by later philologists: they are purely Indian, and do not express satisfactorily all the uses which can be made of the vowel-scales, by languages possessed of a fuller vowel-system than the Indo-European and Sanskrit.

CH. VI.

suffix -ika, which I take to be formal) gives us Vaidika, "belonging to the Vedas," an adjective which (minus its final a) is now commonly used by English Sanskritists instead of the commoner "Vedic." A more full, indeed redundant, list of derivatives than the Sanskrit possesses by this method of vowel-intensification with formal suffixes, cannot well be conceived. The Greek and Latin have similar examples, as we shall see; but nothing like the fulness of the Sanskrit vocabulary. I have already said that it is in this power of forming bases, both nominal and verbal, and its marvellous facility in combining nominal bases thus formed, that the genius of the Sanskrit is especially manifested, as compared with the classical languages. It is not equally manifested in conjugation. Here we find what is perhaps its oldest application, namely to strengthen some verbal bases in those persons whose terminations are technically called weak. Among these are the three persons singular of the present. Thus from i "to go" is formed, émi, “I go;" éshi, "thou goest;" éti, "he goes:" but in the plural, imús, "we go." This is exactly analogous to the εἰμι, εἰ, εἰσι...ἴμεν: compare also δίδωμι with δίδομεν. This phenomenon has been explained in different ways, into which it is not here the place to enter fully: the subject. most satisfactory, if it could be fully applied, would be Professor Benfey's, who makes it the result of accentuation'. He lays down that the accent naturally falls on the modifying syllable of a word. Thus in i-más, "we go," the strong termination mas modifies and restricts the general idea of going to the particular going of some persons, more than two, and spoken of by themselves; therefore it is accented. But when certain terminations became weakened, e. g. mi from ma, they were unable longer to bear the accent; which then fell back either on the radical syllable and strengthened it, as émi, or on some additional modifying element, if such existed, as the reduplicated syllable in dádámi (plur. 1, dadmás), or a formative suffix 1 See his Kurze Sanskrit Grammatik, §§ 153, 154.

Theories

on the

CH. VI.

as nu: thus from chi, chi-nó-mi, but plural chi-nu-más
(compare Seík-vu-μι, deiк-vi-μev). Here, however, the
Greek is not in accordance with the Sanskrit, for Sídoμev
and ev throw their accent back in accordance with the
common Greek rule: though in many minute respects
the accentuation is the same in the two languages. It
is of course possible that here also it was originally the
same, and that in course of time, as the reason of the
variety became forgotten, the distinction in accent passed
also out of use1. Such an explanation is very possible
and quite in accordance with analogy: but the usage
even of the best preserved languages of the Indo-Euro-
pean family is too confused to allow us to regard the
theory as proved. The laws of accent are very different
for different branches, e.g. for the Sanskrit, the Greek,
the Latin, and the Teutonic: very often there is so
much difference in the same language as to lead us to
the conjecture that there must have been an older law in
the language different from that which we find prevalent.
Now this fact undoubtedly gives reason to suppose that
there was one common law of accent for the Indo-Euro-
pean race, from which all the nations have deviated in
different ways.
But even if this was so, as is quite
probable, it is by no means necessary that there should
be any connection between accent and vowel-intensifi-
cation. Each is a possible method of expressing that the
speaker regards a certain syllable of a word as important:
it is quite likely that the two originally existed side by
side, acting to the same end, but in quite distinct ways2.
The tendency at last, no doubt, is towards the destruction
of quantity by accent: this we shall see later on in Latin.
But that quantity and accent may be found in the same

1 The accent is still found on the last syllable in paμév and coμév, but these two verbs are again discordant with the Sanskrit in having cluí and pnul; even though the latter has the radical vowel increased in the singular.

2 See Corssen, 1. 626, &c.: where a good and clear summary is given of the views which have been held on this matter.

word on different syllables is obvious from the classical CE. VI. Greek.

Not one-tenth of the Sanskrit verbs belong to the second conjugation, which distinguishes the strong and weak forms. The first class of the first conjugation, which alone comprises more that half of the verbs in the language-probably however the latest in time, like the Greek verbs in w-agrees with the Greek in raising the vowel (of i or u, but not a) one step to form the present stem. Thus from budh we have bodh-a-mi, by the side of πυθ, πεύθομαι. The first and second increase are found in the formation of the Sanskrit perfect, without regard to the conjugations, in accordance with special rules which may be found in any Sanskrit grammar: thus the perfect of budh is bubodha, where there is no further increase from the present stem bodh; compare the Greek puy, deúyw, πéþeʊya: but verbs ending in vowels generally exhibit the full scale: thus dru, "to run," present drav-d-mi (for dro-d-mi, by a regular euphonic law), perfect du-drâv-a (for du-drâu-a). Further illustrations from Sanskrit would be out of place here. I have thought it necessary to say so much, because the extent to which the principal vowel-change pervades the whole system of language is more fully seen in the Sanskrit than in any other of the sisters; because, when Sanskrit was at its fullest development, that principle remained living and productive in all the base-formations; whilst in other languages its regular action can with difficulty be traced except in the formation of verbal bases; in some languages, as the Latin, scarcely there.

The substitutes in Gothic for the different steps of the vowel-scales may be interesting to students of English. We find in the i-scale i, ei, ai, in the u-scale u, iu, au. The reason of this variation is that the Gothic has no ā to employ in the second step; and i for ǎ in the first steps. Indo-European STIGH, Greek

and therefore substitutes e
Thus, from √stig, to climb,
στιχ (whence στείχω, στοῖ-

Vowel

scales

in the

Gothic;

CH. VI,

and in
the Lithu-
anian;

in Greek and Latin.

(i) the A-scale.

Difficulty of distinguishing the two

steps.

Xos), we have present steiga, I climb, perfect staig, I clomb; from bug, to bend, or "bow," Indo-European BHUG, to bend (Greek puy, and Latin fug, to fly, i.e. to bend out of the original course: compare the Greek трожη in the same sense, and Tрóπatov1), we have biuga, I bend, and baug, I bent. In the a-scale the Gothic (like the Greek, as we shall see below) distinguishes the first from the second step by employing é for the first—the O. H. G. keeps a—and ô for the second; thus lat (German lassen = to "let"), léta, lai-lôt.

Lithuanian, which possesses e besides a as a radical vowel in the a-scale, has a for the first step and ô for the second, like the Teutonic family. In the i-scale it has ei or e, the last a modification of e produced by sounding a immediately after it, and consequently always long. The second step is ai, like the Gothic. In the u-scale we find first au or a, which is a long o with the same parasitic a as above, second au. Au and au resemble the Sanskrit in becoming before vowels av and ov (Sanskrit av) respectively.

It is not necessary to go further into the different languages to establish the principle. Sufficient examples have been given from the most important (except the Greek and Latin) to shew that the power of intensifying ideas in this way is as old as Indo-European times; and we may now pass to a fuller consideration of the principle among the Greeks and Italians".

In the a-class an obvious difficulty occurs. How is it possible to distinguish the two steps when a + a becomes ā, and a +ā has no further symbol to represent it? The effects of this difficulty in the Latin will be obvious when we look at the irregularity of the cases where the ǎ has

1. Gr. Et. no. 163.

2 See Schleicher, Comp. p. 135, or his Lithuanian grammar, where the excessively difficult and numerous vowel-sounds are fully explained.

3 Examples are taken principally from Leo Meyer, 1. 131-162, Cors. sen, 1. 348-628, of the 2nd edition, in which this part has been enormously expanded.