I. ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. Anusvāra and Anunāsika are commonest when final. As in that position they belong to the sphere of Sandhi, the conditions under which they then appear, will be stated below'. Medially Anusvāra and Anunāsika are found under the following conditions: 54 a. regularly before sibilants and h; e. g. vamšá- ‘reed'; havímși ‘offerings'; māmsá- 'flesh'; simhá- 'lion'. This is generally recognized as the sphere of Anusvāra, except by the APr.; but mamscatú 'light yellow', is stated ³ to have Anunāsika, while pumścali- (AV. VS.) is said to have both Anunāsika and Anusvāra4. The Anusvāra usually appears before s, and all forms with ms are shown to be based on original ns or ms by the evidence of cognate forms corroborated by that of allied languages 5; thus mámsate, 3. sing. subj. aor. of man- 'think'; jighamsati, des. from han- 'strike'; pimşánti beside pinásti, from pis- ‘crush'; the neuter plurals in -āmsi, -īmṣi, -umsi from stems in -as, -is, -us, with nasal after the analogy of -anti, nom. acc. pl. from neuters innt; stems in -yāms, -vams beside nom. sing. in-yan, -van; kramsyáte (AV.) from kram- 'stride'. When Anusvāra appears before and h derived from an IE. palatal or guttural, it represents the corresponding class nasal. The reduction of these old nasals to Anusvāra is probably IIr. after long vowels (am, -imr, -umr); probably later after short vowels, for it does not appear in forms like áhan (for áhans), though s was not dropped after an in the IIr. period; but the s must have been dropped early in the Indian period, before -ans became -ams Anusvāra before h (= Av. j) must have arisen in the Indian period. b. before any consonant in the intensive reduplicative syllable, the final of which is treated like that of the prior member of a compound; e. g. námnam- īti (v. 835) from nam- 'bend'. c. before any consonant which is the initial of secondary suffixes; thus from sam 'blessing', is formed sam-yu- 'beneficent'; sám-täti- or śántiti- 'blessing'. 61. Voiceless Spirants.-a. The sound called Visarjanīya8 in the Prātiśākhyas⁹, must in their time have been pronounced as a voiceless breathing, since they describe it as an uşman, the common term for breathings and sibilants. This conclusion is borne out by the fact that, in the RPr., Jihvāmūlīya and Upadhmānīya, the sounds which are most nearly related to and may be replaced by Visarjanīya, are regarded as forming the second half of the voiceless aspirates kh and ph respectively, just as forms the second half of the voiced aspirates gh, bh, etc. At the present day Visarjanīya is still pronounced in India as a voiceless breathing, which is, however, followed by a weak echo of the preceding vowel. According to the TPr.", Visarjanīya has the same place of articulation as the end of the preceding vowel. The proper function of this sound is to represent final s and r in pausā; it may, however, also appear before certain voiceless initial sounds, viz. k, kh, p, ph, and the sibilants (occasionally also in compounds) ** b. Beside and instead of Visarjanīya, the Pratiśākhyas recognise two voiceless breathings as appropriate before initial voiceless gutturals (k, kh)¹² ¹ See below, Sandhi, 75. 2 APr. 1. 67; 11. 33f. 3 Cp. RPr. IV. 35. 4 Cp. VPr. 1V. 7 f. 5 Cp. WACKERNAGEL. I, 224 a. 6 Loc. cit.; cp. BRUGMANN, KG. 354, 17. 7 In AUFRECHT's edition printed nánna- mīti. 8 The erm visarga is not found in the Prātiśākhyas (nor in Pāņini). 9 RPr. 1. 5 etc., VPr. 1. 41 etc.; APr. I. 5 etc.; TPr. I. 12 etc. 10 See WHITNEY on TPr. II. 48. ¹¹ See below, Sandhi, 78, 79. 12 See RPr. 1. 8; VPr. III. 11 etc.
