56 I. ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. sound; from stan- 'thunder', stanayitni-, beside tanayitnú- 'thunder', tanyatú- (also AV.) 'thunder', tanyú- 'thundering', tanyati 'resounds'; stená- 'thief', stay-ánt- (AV.) 'stealing', stayi- (VS.) 'thief', stéya- 'theft', beside täyi- 'thief'; sty, beside t- 'star'; spás 'spy' and spas 'see' in verbal forms á-spas-ta aor., paspas perf., spāśáyati caus., -spasta- part., 'seen', beside pásyati 'sees'¹. The loss of initials may be inferred in the following words from the evidence of cognate languages in which it has been preserved: tij- 'sharpen'; tuj- ‘strike'; tud- beat'; narmá- (VS.) ‘jesť; nṛt- ‘dance'; parṇá- ‘wing’; piká- (VS.) 'Indian cuckoo'; plíhán- (VS. AV.) 'spleen'; phéna- 'foam'; mrd 'crush', mydů- (VS.) 'soft'; vip- 'tremble'; supti- 'shoulder'. A few examples occur of the loss of the semivowels y or v² as the last element of an initial consonant group. Thus y disappears in derivatives of roots in i formed with suffixes beginning with consonats: -mūta- ‘moveď', mū-rá- 'impetuous', mi-tra- (AV. VS.) 'urine', from mīv- 'push'; suci- 'needle', sū-ná- plaited basket', sú-tra- (AV.) 'thread', beside syú-man- 'thong', syū-tá- 'sewn', from siv- 'sew'. Loss of v seems to have taken place in siti- 'white' (only at the beginning of compounds), beside svit- 'be bright', śvity-áñc- ‘brilliant', švitrá- (AV.) ‘white', śvitrya-, perhaps 'white'; and, on the evidence of the Avesta, in kṣip- 'throw' (Av. hšiw), sáș- (Av. hšvaš). 3. When the group is medial, the loss usually taken takes place between single consonants. a. The sibilants and s thus regularly disappear between mutes; e. g. á-bhak-ta, 3. sing. aor., for *ábhak-s-ta beside á-bhak-ş-i, from bhaj-, 'share'; cas-te for caks-te (original *cas-s-te) 3. Similarly a-gdha- (TS.) 'un- eaten', for *a-ghs-ta-, from ghas- 'eat'4. b. The dental t has disappeared between a sibilant and c in pas-cá and pas-cát 'behind' (— IE. post-që, post-qét); and between pand s in *nap-su which must have been the loc. pl. (AV. nafšu), from *napt-, ‘weak stem of nápät-, beside the dat. nád-bhyas 'to the grandsons', for napt-bhyas, where on the other hand the p has been dropped. c. A mute may disappear between a nasal and a mute; e. g. panti- for pańkti-, yundhi for yungdhi. This spelling is common in Vedic Mss.; it is prescribed in APr. II. 20, and, as regards the mediae, in VPr. vI. 30. 4. The only example of the loss of an initial 5 mute in a medial group of consonants seems to be that of b before dbh in nádbhyas for *nabdbhyas from napt for nápat-. The semivowel seems to be lost before a consonant when another follows in cakr-át (Pp. cakrán) and cakr-iyās6 for *carkr- beside carkar-mi, from kr- 'commemorate'; and, on the evidence of the Avesta, in tuástr- (cp. pwaras 'cut', pwaršta- 'created'), 3 See above 56, b. şkṛta-, sám-skrta-; also after nis in nir askrta (Pp. akrta), perhaps owing to Sandhi forms 4 On ápnas- property', dámpati- 'lord of with sk such as niskuru (AV.). There can | the house', see WACKERNAGEL I, 233 c, note. be little doubt that the s here was not ori- ginal but was due to analogy; cp. WACKER- NAGEL I, 230 a, , note. ¹ WACKERNAGEL I, 230 a 7, note, discusses several uncertain or erroneous etymologies based on loss of initial s (including maryas interpreted as 2. sing. opt. of smr-). The evi- dence of Prakrit seems to point to the loss of initial s in sépa- 'tail', krīd- 'leap', kruś 'cry', parasú- 'axe', parusá- 'knotty'; cp. WACKERNAGEL I, 230 b. 2 On the possible loss of in bhanj (Lat. frango) and of k in savyá- (Lat. scaevus), see WACKERNAGEL I, 232 c, note. 5 The guttural mute only seems to be lost when ks+t becomes st: see above, 56, 6. The loss of m in the inst. sing. -ni of nouns in -man is only a seeming one; for dānā, prāthiná, preṇā, bhūná, mahină, variņá (TS.) appear beside the stems dáman- etc., because some of them had a stem in - without m, so that - seemed an alternative from of -mnă: see BENFEY, GGA. 1846, 702. 880; Göttinger Abhandlungen 19, 234; BLOOM- FIELD, JAOS. 17, 3; cp. IF. 8, Anzeiger, p. 17. 6 But see GELDNER, VS. I, 279.
