I. ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. Pl. N. m. 1. ágrbhita-socisas, ajirá-socişas, váta-dhrajişas (Kh. 1. 3³) 'having the impulse of wind', vrktá-barhisas, V. vrkta-barhişas, sádma-barhisas 'preparing the litter', V. su-barhișas, su-jy'tişas and su-jyotişas, svá-rocisas ¹.- 2. náhuṣas, mánușas, V. manușas, vanúşas, vápusas. N. A. n. 1. arcímsi, jyótīmṣi, barhímși (VS. xxVIII. 21), śocímși, havímsi. -2. áyumşi, cákṣumsi, janúmsi, tápūmsi, párūmṣi (TS. VS. AV.), yájūmṣi (VS. AV.), vápumsi. 224 A. m. 1. su-jyótisas. -2. náhuşas, mánuşas, vanúsas². I. I. n. havirbhis ³. 2. n. dhánurbhis (AV.), yájurbhis (VS.Iv.1), vápur- bhis. — D. I. m. urdhvá-barhirbhyas (VS. xxxvIII. 15) 'being above the litter'. 2. n. yájurbhyas (VS. xxxvIII. II). G. 1. n. jyótisām, havíṣām (AV.). - 2. n. cákṣuṣām (AV.), janúṣām, yájuṣām (AV.), vápuṣām. - m. vanişām, viśvá-manuṣām belonging to all men'. L. I. n. havissu. ß. Stems in -as. 343. Primarily this declension consists almost entirely of neuters, which are accented on the root, as mán-as- 'mind'; but these as final members of adjective compounds may be inflected in all three genders, as N. m. f. su-mánās, n. su-mánas 'well-disposed'. There are besides a few primary masculines, which are accented on the suffix, being either substantives, as raks-ás- 'demon', or adjectives, some of which occur in the f. also (as well as n.), as ap-ás- 'active'; and one feminine, us-ás- 'dawn'4. a. Strong cases are regularly distinguished only in the N. A. pl. 1., where the vowel of the suffix is lengthened and nasalized 5 (as in the is and us stems), as ámhāmsi 'troubles'. Otherwise the stem usás- shows in the A. sing., N. A. du., N. V. pl., strong forms with lengthened vowel in the suffix, which occur beside the unlengthened forms 6, the latter being nearly three times as frequent in the RV. The long vowel is here required by the metre in 20 out of 28 occurrences, and is favoured by the metre in the rest7; so that the lengthening may be due to metrical exigencies. The strong form uşásas 8 occurs once (X. 39¹) for the weak uşásas as the G. sing. or A. pl. There is further the single strong form N. du. m. tośásā9 bestowing abundantly' 10. b. Supplementary stems ending in -as beside -an are ýbhvas- ‘skilful' (as well as ŕbhva-) beside ŕbhvan-, and síkvas- 'strong' (as well as síkva-) beside šíkvan-. c. There is here a large number of transition forms both to and from the ẵ- declension. Many pairs of stems in -a and -as are common in both forms and seem therefore to be of independent derivation. But there are also a good many such doublets of which the one is the normal stem, while the other has come into being through mistaken analogy or metrical exigency. 1. In the transitions to the a- declension several may be explained as starting from the misleading analogy of contracted forms. To this group belong the following: from angiras-, a name of Agni, L. sing. angire beside N. pl. angiras; from án-āgas- 'sinless', A. pl. án-agan beside A. sing. án-agām and N. pl. án-āgās; from ap-sarás- 'water-nymph', apsarábhyas (AV.), apsarásu (AV.), apsara-patés (AV.) beside apsarám (AV.); from usás- 'dawn', N. du. usé (VS.), uṣabhyām (VS.), beside A. sing. uşám and pl. usás; from jarás- m. 'old age', f. sing. N. jará (AV. VS.), D. jarayai (AV.), beside I Also the transition forms arcáyas, | Socáyas. 2 Also the transition form vanin. 3 Also the transition form arci-bhis. 4 The derivation of a few stems ending in -as is obscure; as upás- 'lap' (only L. sing.), and risadas- 'destroying enemies'. 5 On the origin of this form as a com- bination of -an-i and -as-i see JOHANSSON, BB. 18, 3 and cp. GGA. 1890, p. 762. The lengthened vowel never occurs in the Pada text in these forms (excepting of course the N. uşás). 7 See ARNOLD, Vedic Metre, p. 130, 11 (a); cp. LANMAN 546. 8 The Pada text has uşáṣaḥ. 9 Here the long vowel appears in the Pada text also. This is the only form made from the stem tośás- (from tus- 'drip'). 10 The form sa-psarāsas ‘enjoying in com- mon (?)', occurring once, must be the pl. of sa-psará- (BR., GRASSMANN), not of *sa- psarás- (LANMAN 546, GELDNER, VS. 3, 197), because -as is never accented in Bahuvrihi compounds; see WACKERNAGEL 21, p. 301 d, note. su-medhás- 'very wise' is not a Bahuvrihi. 1
