n. Euphonic Combination. Finals. Rules of Sandhi.
6i
for vi-pri'is. These are the only examples occurring in the RV. and AV. In the only two examples in which s occurs in the RV. as a final in the com- pound form ks, it is dropped: andk 'eyeless', from an-dks-; d-myak, 3. sing. aor. of myaks- 'be situated' (?).
y. The palatal s becomes either k or t'; e. g. -^r^ fox drs- (m. 61 9); vtpai, N. of a river, for vip3,s-.
8. According as it is guttural or palatal in origin, h become k ar i; thus d-dhok, 3. sing. impf. of i/z^,^- 'milk'; hvX d-vdt, 3. sing. aor. of z/a/^- 'carry'.
c. Th'i rule is, that only a single consonant may be final. Hence all but the first of a group of consonants are dropped; e. g. dbhavan for
- dbhavant tAn for *idns; tuddn for iuddnts; prAh for *prAhks (= *prancs);
acckaii for *ackantst, 3, sing. aor. of chand- 'be pleasing'.
a. k t or t, when they follow an f and belong to the root, are allowed to remain; e. g. vdrk, 2. 3. sing. aor. of vrj- 'bend'; urkj^S), nom. of a;;;- 'strength'; d-m3,rt, 3. sing, impf. of mrj- 'wipe'; a-vart, 3. sing. aor. of vrt- 'turn'; stthdrt (AV.), nom. of suhard- 'friend'. The only instance of a suffix remaining after ris dar-t, 3 sing. aor. of dr- 'cleave', used also for 2. sing, beside d-dah = d-dar (for *d-dars).
8. Some half-dozen instances have been noted, in the Samhitas, in which a suffixal s or f seems to have been retained instead of the preceding consonant; but they are probably all to be explained as due to analogical influence. They are :
1. the nominatives sadka-mas (beside sadha-mdd) 'companion of the feast'; ava-yds 'sacrificial share', and piro-das (ace. purodasam) 'sacrificial cake'. Sadha-mds may be due to the influence of nom. with phonetic j- like -mas 'moon', beside inst. pi. mJd-bhis (44 a 3). Ava-yas, in the only passage m which it occurs in the RV., has to be read as quadiisyllabic (also in AV.), i. e. as ava-yajah, and is probably to be explained as a con- traction which retains the living -s of the nom. (and not the prehistoric s of *-yaj-s). Puro-das (from das- 'worship'), occurring only twice in the RV., may be due to the in- fluence of a frequent nom. like draviiw-djs 'wealth-giver'. That the prehistoric nom. -j- should in these three forms have survived in the linguistic consciousness of the Vedic poets, and as such have ousted the preceding consonant, which in all other analogous nominatives alone remains, is hardly conceivable =. The only reasonable explanation is to assume the analogical influence of the nom. -s which was in living use after vowels.
2. The four verbal preterite forms [a-yas (for *a-yaj-s) beside a-y3f, 2. sing. aor. olyaj- 'sacrifice'; srSs {KV.) = *a-srdj-s, 2. sing. aor. oi srj- 'emit'; a-bhanas[KS>j ^*a-bhanak-s,
2. sing. impf. of bhanj- 'break' ; and a-srat (VS.) = *a-sras-t, 3. sing. aor. of sras- 'fall') are the beginnings of the tendency (of which there are several other examjDles in the Brahmanas)3, to normalize the terminations, so as to have -s in 2. sing, and -/in
3. sing. This tendency is extended in the RV. from the s and i of 2. 3. sing^ even to i. sing, in the forms a-kramlm (beside a-kramisani) owing to a-kram-u, a-k7-am-it; and vam (for var-am) owing to 2. sing, vak (for var), aor. of vr- 'cover'.
67. Rules of Sandhi. — The body of euphonic rules by which final consonants are assimilated to following initials and hiatus is avoided between final and initial vowels is called Sandhi in the Pratisakhyas +. The editors of the Samhita of the RV. have greatly obscured the true condition of the text vidth which they dealt by applying to it rules of euphonic combination which did not prevail at the time when the text was composed. Thus though the Sandhi between the verses of a hemistich is (excepting a few survivals from the older form of the text) 5 appHed with greater stringency than elsewhere, the metre clearly shows that the end of the first verse of a hemistich con- stitutes a pause as much as the end of the last. Within the verse, moreover, Sandhi is, according to metrical evidence, not applied where the caesura occurs; nd, when it means 'like' (as opposed to nd 'not'), is never contracted with any following vowel, nor (?^ 'then' with a preceding a^; t and w before
1 Cp. 43 a.
2 Cp. Whitney 146 a; Bt.oomfield, AJP. 3, 28 ff.; Bartholomae, KZ. 29, 578 ff.
3 Cp. Whitney 555 a.
4 RPr. 11. 2. 7. 13; VII. l; VPr. III. 2; APr. IV. 414.
5 E. g. manisa agnih (l. 701).
6 In these instances there is a pause in
