68 I. ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. (Iv. 18²) for durgáham etát, and it is never analysed by the Padapāṭha in this way. It may perhaps have started from the analogy of the doublet túbhya beside túbhyam 'to thee'¹. 2. Before mutes, final m is regularly assimilated, becoming the corresponding nasal, and before n similarly becoming n. The Mss. and printed texts, however, represent this assimilated m by the Anusvāra sign; e. g. bhadrám karisyási (1.16) for bhadrán karisyási; bhadrám no (x. 20¹) for bhadrán no. This actual change of m to n before dentals led to some errors in the Pada text; e. g. yán ni-pási (IV. 116), analysed as yát instead of yám; ávi-venan tám (Iv.246), analysed as ávi-venan instead of ávi-venam (cp. IV. 25³). 3. Before rsss and h³, final m becomes Anusvāra (m); e. g. hótāram ratnadhátamam (1. 1¹); várdhamānam své (1. 19); mitrám huve (1. 27). From its original use before sibilants and h, Anusvāra came to be employed before the semivowel also 4. A compound like sam-ráj- 'overlord' shows that m originally remained unchanged in Sandhi before r. - ♥ 4. Before y / v, final mn is assimilated as nasalized i. The TPr.º, ĭ y however, allows Anusvāra beside these, while the APr. even requires Anunāsika before y and v; and the printed texts regularly use Anusvāra; e. g. sám yudhí (1. 83); yajñám vastu (1. 30). Forms like yamyámana- 'extended' and ápa- mlukta- 'concealed' show that final m originally remained unchanged in Sandhi before y and 75; and forms like jagamván, from gam- 'go', point to its having at one time become n before v in Sandhi. 76. Euphonic combination of final f. - 1. Before , final 6 becomes fully assimilated as l; e. g. ángal lómnaḥ (x. 1636) for ángät lómnaḥ. 2. Before palatals (including )7 final ¹ becomes palatal8; e. g. tác cákṣuḥ (VII. 66¹6) for tád cákṣuḥ; rohic chyavá (1. 100%) for rohit syavá; and in a compound yatayáj-jana- 'marshalling men', for yatayát-jana-. 77. Euphonic combination of final n. - 1. Before vowels. Final nº, a. if preceded by a short vowel, is doubled"; e. g. áhann áhim (II. 115) for áhan. The final n is here chiefly based on original ns or nt. Though it is always written double, the evidence of metre shows that this rule was only partially applied in the RV.¹² 2 b. If preceded by a long vowel, it becomes within ¹3 a Pada in the RV. m after a ¹4, but mr after iu 7¹5; e. g. sárgam iva, for sárgan; paridhímr áti, for paridhin; abhísumr iva, for abhísun; nŕňr abhi, for nŕn. This Sandhi was caused by then having originally been followed by s¹6; e. g. výkan originally ¹¹ For various explanations of this doubling, see WACKERNAGEL I, 279 a (p. 330). 12 Cp. OLDENBERG 424 f., 429 ff. 13 At the end of a Pāda, -ān -īn -ũn remain unchanged (as being in pausa) before a vowel. For passages in which -ān is un- changed within a Pada, see below and OLDENBERG 428. 5 Cp. WACKERNAGEL I, 283 c. , 6 Representing th d dh also, if they occur. 7 In the MS. t anomalously becomes instead of c, before s; see L. v. SCHROEDER, ZDMG. 33, 185; ed. of MS. I, p. XXIX. On the aspiration of in this Sandhi, see below 80 a. 14 In the MS. and K., -am is shortened to -am; c. g. asmám ašnotu for asmán. Cp. v. SCHROEDER, ed. of the MS. 1, p. xxIx. 8 Final dentals never come into contact with initial cerebrals in the Samhitas. 15 Fn becomes mr only once (v. 54¹5), remaining unchanged elsewhere because two sounds are avoided in the same syllable (see below 79). 16 For examples of this Sandhi applied to nominatives in -an, see LANMAN 506 A, note (mahán), 512 (-van), 514 (-yan), 517 (-man); for accusatives, 346. ¹ Cp. WACKERNAGEL I, 283 a, note. 2 Op. cit. I, 283, b a. 3 Op. cit. 1, 283 d. 4 In the post-Vedic language Anusvāra came to be allowed before mutes and nasals also. 9 This rule applies to final guttural n also; e. g. kidŕnn í- (x. 1083), cp. LANMAN 490; hinn akrnot (1. 16428) for hin a-. IO The compound vysan-asvá- 'having stallions as steeds', forms an exception.
