The next step is to be sought in Bopp's Comparative Grammar. Here we find the following extract:—"The old Slavonic dakh 'I gave,' and analogous formations remind us, through their guttural, which takes the place of a sibilant, of the Greek aorists [a]ἔθηκα], [a]ἔδωκα], [a]ἧκα]. That which in the old Slavonic has become a rule in the first person of the three numbers, viz. the gutturalization of an original s, may have occasionally taken place in the Greek, but carried throughout all numbers. No conjecture lies closer at hand than that of regarding [a]ἔδωκα] as a corruption of [a]ἔδωσα]," &c.... "The Lithuanian also presents a form which is akin to the Greek and Sanscrit aorist, in which, as it appears to me, k assumes the place of an original s." (vol. ii. p. 791, Eastwick's and Wilson's translation.) The italics indicate the words that most demand attention.

The old Slavonic inflection alluded to is as follows:—

SINGULAR.DUAL.PLURAL.
1.Nes-ochNes-ochowaNes-ochom.
2.Nes-eNes-ostaNes-oste.
3.Nes-eNes-ostaNes-osza.

Now it is clear that the doctrine to which these extracts commit the author is that of the secondary or derivative character of the form of [a]κ] and the primary or fundamental character of the forms in [a]σ]. The former is deduced from the latter. And this is the doctrine which the present writer would reverse. He would just reverse it, agreeing with the distinguished scholar whom he quotes in the identification of the Greek form with the Slavonic. So much more common is the change from k, g and the allied sounds, to s, z, &c., than that from s, z, &c. to k, g, that the à priori probabilities are strongly against Bopp's view. Again, the languages that preeminently encourage the change are the Slavonic; yet it is just in these languages that the form in k is assumed to be secondary. For s to become h, and for h to become k (or g), is no improbable change: still, as compared with the transition from k to s, it is exceedingly rare.

As few writers are better aware of the phænomena connected with the direction of letter-changes than the philologist before us, it may be worth while to ask, why he has ignored them in the present instances. He has probably done so because the Sanscrit forms were in s; the habit of considering whatever is the more Sanscrit of two forms to be the older being well-nigh universal. Nevertheless, the difference between a language which is old because it is represented by old samples of its literature, and a language which is old because it contains primary forms, is manifest upon a very little reflection. The positive argument, however, in favour of the k being the older form, lies in the well-known phænomenon connected with the vowels e and i, as opposed to a, o, and u. All the world over, e and i have a tendency to convert a k or g, when it precedes them, into s, z, sh, zh, ksh, gzh, tsh, and dzh, or some similar sibilant. Hence, as often as a sign of tense consisting of k, is followed by a sign of person beginning with e or i, an s has chance of being evolved. In this case such a form as [a]ἐφίλησα], [a]ἐφίλησας], [a]ἐφίλησε], may have originally run [a]ἐφίληκα], [a]ἐφίληκας], [a]ἐφίληκε]. The modified form in [a]σ] afterwards extends itself to the other persons and numbers. Such is the illustration of the hypothesis. An objection against it lies in the fact of the person which ends in a small vowel, being only one out of seven. On the other hand, however the third person singular is used more than all the others put together. With this influence of the small vowel other causes may have cooperated. Thus, when the root ended in [a]κ] or [a]γ], the combination [a]κ] radical, and [a]κ] inflexional would be awkward. It would give us such words as [a]ἔλεκ-κα], &c.; words like [a]τέτυπ-κα], [a]ἔγραπ-κα], being but little better, at least in a language like the Greek.

The suggestions that now follow lead into a wide field of inquiry; and they may be considered, either on their merits as part of a separate question, or as part of the proof of the present doctrine. In this latter respect they are not altogether essential, i. e. they are more confirmatory if admitted than derogatory if denied. What if the future be derived from the aorist, instead of the aorist from the future? In this case we should increase what may be called our dynamics, by increasing the points of contact between a k and a small vowel; this being the influence that determines the evolution of an s. All the persons of the future, except the first, have [a]ε] for one (at least) of these vowels—

[a]τύψ-σ-ω], [a]τύψ-σ-εις], [a]τύψ-σ-ει], [a]τύψ-ε-τον], &c.

The moods are equally efficient in the supply of small vowels.

The doctrine, then, now stands that k is the older form, but that, through the influence of third persons singular, future forms, and conjunctive forms, so many s-es became developed, as to supersede it except in a few instances. The Latin language favours this view. There, the old future like cap-s-o, and the preterites like vixi (vic-si) exhibit a small vowel in all their persons, e. g. vic-s-i, vic-s-isti, vic-s-it, &c. Still the doctrine respecting this influence of the small vowel in the way of the developement of sibilants out of gutturals is defective until we find a real instance of the change assumed. As if, for the very purpose of illustrating the occasional value of obscure dialects, the interesting language of the Serbs of Lusatia and Cotbus supplies one. Here the form of the preterite is as follows; the Serb of Illyria and the Lithuanic being placed in juxtaposition and contrast with the Serb of Lusatia. Where a small vowel follows the characteristic of the tense the sound is that of sz; in other cases it is that of ch (kh)

LUSATIAN.ILLYRIAN.LITHUANIC.LETTISH.
Sing.1.noszachdoneso, donijenesziaunessu.
2.noszeszedonese, donijeneszieinessi.
3.noszeszedonese, donijeneszieinesse.
Dual1.noszachwe nesziewa
2.noszestaj neszieta
3.noszestaj neszie
Plur.1.noszachmydonesosmo, donijesmonesziemenessam.
2.nosześćedonesoste, donijesteneszietenessat.
3.noszachudonesosze, donijeszeneszienesse.