ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. 78. Euphonic combination of final s. I. Before voiced sounds. a. Finals after all vowels except ă, assumes its voiced form before all vowels and voiced consonants ¹; e. g. sibhir idyo (1. 1²) for ŕşibhis; agnir hótā (1. 15) for agnís; paribhár ási (1. 14) for paribhús; nútanair utá (1. 1²) for nútanais. 70 The s doubtless became through an older transitional voiced cerebral , as is shown by the Avesta, e. g. in dužita- dur-ita-3. b. Finals after à must originally have become before voiced sounds. But äz drops the sibilant before vowels and voiced consonants, while az drops it before vowels except a4, but becomes o5 before voiced con- sonants and a; thus sutá imé (1. 3+) for sutás; víśvā ví (1. 3¹2), for visvas; khya á for khyas (1.4³); no áti (1.4³) for nas; indavo vam (1.2¹) for indavas. 2. Before voiceless consonants. a. Before the palatals c, ch, s final s becomes the palatal sibilant 7; e. g. devas cakrmá (x. 37¹²). b. Before the dental tº, final s following à always remains; e. g. yás te (1. 4¹); médhirūs téṣām (1. 117). After i ¹, it remains as a rule; e. g. ánvībhis tánā (1. 34). But s becomes s, which cerebralizes the following t to t: a. regularly in compounds in all the Samhitãs; e. g. diş-tara- 'insuperable' for dis-tara- (but rajas-túr- 'traversing the air'); f. often in external Sandhi in the RV. This occurs chiefly, and in the independent passages of the other Samhitas ¹ 12 only, before pronouns; e. g. agnis te, krátus tám; otherwise it occurs occasionally only in the RV.; thus nis-tatakṣúr (x. 317); gobhiş țarema (x. 42"); nákiş tanúşu (VIII. 20¹2) ¹3. C. Before k kh p ph, final s as a rule becomes Visarjanīya (its pause = ¹ When final s becomes before 7, it is | a voc. in -tar (from a stem in -tr), which treated like an original , being dropped would become -tā before 7. after lengthening the preceding vowel. 2 This would be the voiced cerebral corresponding to the voiceless s which appears before voiceless consonants (e. g. in duskha). The starting point of this Sandhi was probably the treatment of az before voiced dentals, where the sibilant was dropped and the preceding vowel lengthened to e or u. 3 This would account for the Sandhi of The latter finally carried the day, e surviving some Vedic compounds formed with dus-only in sắre duhitá (1. 345); cp. BARTHOLO- ‘ill’: dù-ḍábha-, dū-ḍás-, dū-ḍhí-, dū-ņáša-, dū- MAE, BB. 15, 1f.; WACKERNAGEL I, p. 338. náśa-, beside dur-niyántu-. It may possibly 7 Só cit does not stand for sah cit, but also account for the forms svádhitīva (V. 87), | for sá u cit. for svádhitiz va (Pp. svádhitir-iva), and urv iva (IX. 9615), where the metre requires uri va (Pp. uri iva); the cerebral being dropped after lengthening the preceding vowel. 8 If initial cerebral mutes had existed in the Samhitãs, finals would doubtless have become the cerebral sibilant ș before them. 4 In the MS. unaccented a for -as is lengthened before an accented vowel; e. g. látā indraḥ. In TS. 11.4.7¹ ar for a in jinvár avýt is merely a bad reading (MS. 11. 47| jinva rāvát); cp. above p. 33, note 13. 5 In the compound ánar-vis- (1. 1217) 'having a wain as his abode', ar instead of o (cp. anas-vant- 'possessed of a wain)' is perhaps due to the influence of vanar-, beside vanas and vana-. On a still more anomalous compound of the same word, anad-váh- drawer of a cart', 'bull', cp. WACKERNAGEL I, p. 339, top. The vowel à appears instead of o in pracetā | ráian (1. 24¹4) ‘O wise king' (praceto rájan in the same verse, TS. 1. 5. 113), probably owing to the voc. pracetaḥ, as it would have been pronounced at the end of a Päda in the original text, having been misunderstood as 9 No example of initial th occurs in the Samhitas; but the internal Sandhi of sthā- 'stand', in ti-şthati (for ti-stha-ti) shows that initial th would have been treated in the same way as t. 10 Final 's never occurs; s, occurring only once in the RV., remains unchanged in mātṛs trin (1. 164¹). ¹¹ The only exception in the RV. is cátus-trimsat "thirty-four', doubtless due to the avoidance of the combination str. 1² The TS. also has niş tap- 'heat'. On the usage of the SV., see BENFEY, SV. p. XLIII; on that of the AV., see WHITNEY, APr. II. 84. 13 Owing to the far more numerous occurrences of as before t, combined with the disinclination to change the following initial, the retention of s after i ž gradually gained ground and finally prevailed in the post-Vedic language, even in compounds.
