VI. DECLENSION. NOUNS. CONSONANT STEMS. 225 A. jarám¹; from sa-jósas- 'united', N. du. sa-jóşau beside pl. sa-jóşās. There is a further group of transitions to the a- declension starting not from contracted forms, but from the N. sing. n. understood as m.². From ávas- 'favour' is thus formed I. ávena; from krándas- 'battle-cry', D. krándaya 3 (AV.), and the compound śúci-kranda-m 'crying aloud'; from śárdhas- n. 'troop' the m. forms śárdha-m, śárdhena, śárdhāya, śárdha-sya, śárdhān4; from hédas- n., once (1. 94¹²) héla-s N. m., hence A. héda-m (AV.), L. héle; also the compound forms N. f. du. a-dveşê 'not ill-disposed', beside dvéşas- 'hate'; dur-óka-m (VII. 43) ‘un- wonted' beside ókas- 'abode'; vi-dradhé (IV. 3223) 'unclothed' (?) beside drádhasī (TS.) 'garments'; prthu-jráya-m (IV. 44¹) 'far-extending' beside jráyas 'expanse'; puru-péśasu "multiform' beside pésas form' 5. The 2. The secoud class, comprising transitious to the as declension, consists of the two groups of transfers from the radical -ā stems and the derivative -a sterns. former group embraces forms of compounds made from kṣā- 'abode', pra-já- ‘offspring', -da- 'giving', -dha- 'bestowing', besides má-s 'moon', starting from the N. in a-s which is identical in form with that of stems in -às. The forms of this type are the following: N.pl. divá-kşas-as beside N. sing. divá-kṣā-s having an abode in heaven'; A. sing. á-prajas-am (AV. VS.) 'childless', A. pl. f. iḍa-prajas-as (TS. MS.), A. sing. su-prajás-am (AV.), N. pl. m. su-prajás-as (AV. TS.) 'having a good son' beside N. sing. m. su-prajá-s; V. dravino-das6, beside N. sing. dravino-dá-s, A. dravino-da-m 'wealth-giving'; varco-dás-au (VS.) 'granting vigour'; reto-dhás-as (VS.) impregnating' beside N. sing. m. reto-dha-s; A. vayo-dhás-am (VS.), I. vayo-dhás-à (VS.), vayo-dhás-e (VS.), V. vayo-dhas, N. pl. vayo-dhás-as (AV.) beside N. sing. vayo-dhá-s 'bestowing vigour'; D. sing. varco-dhás-e (AV.) beside N. sing. varco-dhá-s (AV.) and A. sing. f. varco-dha-m (VS.); of candrá-mas-7 'moon' all the forms occurring, except the N. sing., which is their starting-point, are transfer forms: A. candrá-masam (VS. XXIII. 59¹, I. candrá-masă (AV.), D. candrá-mase (VS.), G. candrá-masas, L. candrá-masi (AV.), V. candra- mas (AV.), N. du. candrá-masā, sūrỹā-candra-másā and sūryā-candra-másau. 3. There are further several sporadic transition forms from -as stems occurring beside the ordinary corresponding -a stem. These may sometimes have started from an ambiguous N. siug., but they seem usually to be due to metrical exigencies. Such forms are the following: dákşas-e, dáksas-as, putá-dakşas-a and putá-dakṣas-as beside very frequent forms of daksa- skill'; dosás-as ³ (AV.) once beside forms of doşá- 'evening'; (viśváyu)-poṣas-am beside the common pósa-; N. sing. sá-bharūs, A. viśvá-bharas-am beside the frequent bhára- 'supporting'; veśás-as 9 (AV.) beside veśá- 'neighbour'; sépas as A. sing. in AV. (XIV. 238) for the sépam of the RV., from sépa- m. tail', N. sing. sépa-s; N. sing. m. su-sévās occurs in the AV. as a variant for su-seva-s 'very dear' of the RV.; sahásra-šokās occurs once as N. sing. m. beside the common sóka- 'flame'; A. sing. tuvi-svanás-am and N. pl. tuvi-svanás-as, beside svaná- 'sound'; A. pl. f. gharmá-svaras-as beside svará- 'roaring'; also the D. dhruvás-e (VII. 70¹) for dhruvaya to which it is preferred owing to the metre and the influence of infinitives in -áse, and similarly urdhás-e (v.645) parallel to vydhaya (VIII. 836). 4. There are besides a few quite abnormal transition forms. The occurrence of the very frequent N. pl. m. of vi- 'bird' with a singular verb (1. 1418) and once as an A. pl. (1. 104¹) may have produced the impression of a n. collective váyas- and thus led to the n. pl. váyāmsi (AV.), helped perhaps by the existence of the very frequent n. stem váyas- 'food'. The isolated form N. sing. m. sv-áñcās is probably lengthened for sv-anca-s (like su-śévās for su-śévas) which started from the A. sing. m. sv-áñc-am. The G. du. ródas-os occurs once (IX. 225), evidently on account of the metre instead of the ordinary ródasī-os 10. ILANMAN 552 thinks that medha- 'wis- dom' in N. medha, I. medháyā, N. pl. medhás, I. medhabhis started from medham as con- tracted A. of medhas found iu su-medhás-. This seems doubtful to me. 2 In VI. 666 rókas might be taken as N. sing. m. (as GRASSMANN takes it) beside rokás, m. 3 Cp. also the D. carắyai (VII. 77¹) beside caráse (1. 924, v. 474), and tárāya (11. 13¹²) beside tárase (III. 18³). 7 From más- (A. más-am, etc.), where the s belongs to the stem. It is formed probably with -as from mā- 'measure' (ma-as), being | thus in origin a contracted -as stem, in which, however, the N. más giving rise to the transition forms was understood as the lengthened form of *mas. ā-jaras-ám (B.) 6 Cp. BRUGMANN, Grundriss 2, p. 398. Indo-arische Philologic. I. 4. 8 In uşaso dosásaś ca (AV. XVI. 46) ob- viously due to parallelism with uşásas. 9 Clearly owing to the metre instead of vešásas. 4 Cp. LANMAN 353 and 554, 8. 10 Similarly aks-ós occurs in AV. V. 11⁰ (but contrary to the metre) for akşi-ós. On 5 There is also an -as stem extended with a in the D. ā-jarasắya (x. 854³) 'till the transition forms of the as declension old age', based on the adv. compound cp. LANMAN 546-558. 15
