260 I. ALLGEMEINES UND SPRACHE. 4. VEDIC GRAMMAR. The forms occurring are: G. m. indrā-varuṇayos, īśāndyos ‘ruling', deváyos, mitráyos, mitrá-váruṇayos, yamáyos 'twins', várunayos, vaikarṇáyos ‘descendants of Vikarṇa'. - n. pasy-ds (for *pāsyd-y-os) 'pressing stones', puran-y-bs¹ (for *purāṇá-y-os) 'ancient'. L. m. ámsayos, áśvayos, upākáyos 'closely connecteď, kárṇayos (AV.), jámbhayos (TS. IV. 1. 10³) jaws', tuvi-jatayos 'of powerful nature', dámstrayos (AV.), nadáyos 'roarers', mitráyos, muskáyos 'testicles', várunayos, váhisthayos 'drawing best', vi-vratayos 'refractory', şthuráyos 'strong', hástayos. Pl. N. V. m. Here there are two forms. In the more common form the normal ending -as coalesces with the final of the stem to -ās, e. g. devás. The less common form appears to be made by adding the normal ending -as over again², e. g. devás-as. The form in -as is about twice as frequent in the RV. as that in -asas³, the former being made from 808 stems, the latter from 403. In the original parts of the AV. -ās is 24 times as frequent as -āsas, the former occurring 1366 times, the latter only 57 times 4. Both forms frequently occur side by side, the choice of the one or the other being often no doubt determined simply by the metre; e. g. brháa vadema vidáthe suvírāḥ (II. 1¹) ‘abounding in heroes we would speak aloud in the assembly', but suviraso vidátham á vadema (II. 12¹5) 'abounding in heroes we would speak to the assembly'. Examples of the most frequent forms made with the two endings are: 1. devāsas (86), jánāsas (41), sómāsas (41), sutāsas (29), ādityásas (24) ‘Adityas', yajñíyāsas (21) ‘holy', amŕtīsas (11). -2. dévis and devas (311), sómās (42), ādityás (39), sutás (27), jánās (24), amŕtās (22), yajñíyās (10). N. A. n.5 Here, as in the N. m. there are two forms, a shorter and a longer, the former being the older and original, as well as the more fre- quent one. The older form is made not by adding the normal ending -i, but by lengthening the final -a of the stem, e. g. havya ‘oblations'. The later form ends in -âni and is doubtless due to transitions from the stems in -an7 which form the n. pl. N. A. with both -ā and -āni, e. g. nămã and námäni. The form in -a is in the RV. made from 394 stems, that in -āni from 280, the proportion of the occurrences of the former being roughly three to every two of the latter. The proportion in the AV. is almost exactly reversed, the form in -a being there made from 102 stems, that in -āni from 1588. The two forms are so common side by side that when two n. plurals occur in the same Pada, the one generally ends in -ani and the other in -ā⁹; e. g. yá te bhīmáni áyudhā (IX.61³⁰) 'thy terrible weapons'. This phenomenon ¹ With y inserted though -a is dropped. a collective; this would account for the 2 See BRUGMANN, Grundriss 2, p. 661, agreement of the singular verb with this pl. where several examples are given of endings in Greek; cp. also sárvā tá astu (RV. I. being repeated in other languages. 1628); see BRUGMANN, Grundriss 2, p. 682. 3 The form in -āsas seems to be an Indo- Iranian innovation, as there are no certain traces of it in other Indo-European languages; cp. BRUGMANN 1. c. 7 The G. áhānām is an example of the transference of another case from an -an stem. 4 In the original Mantra portions of the TS. the pl. in -ās is very numerous, but I have noted only 11 forms in -asas. In the Khilas, ſorms in -ās are three times (30) as numerous as in -āsas (10). 8 In the independent Mantra portions of the TS. the forms in - seem to outnumber those in -āni in about the same proportion as in the RV.. there are at least 20 forms of the former and 14 of the latter. In the Khilas the two forms are almost equally divided, as 10 examples of -ã and 12 of -āni occur. 5 There is no example of a V. in the RV., and only one, cittani (III. 24), in the AV. where the Mss. have cittáni. 6 This form in -a is commonly supposed to have started from a N. sing. f. in -ā as 9 Similarly, the form in - appears beside n. pl. forms in -īni, uni or even -āmsi, -īmși, -ūmşi; e. g. bhūrīņi bhadrá (1. 16610);
