¹ On this language see my articles: ‘The Prakrit underlying Buddhistic Hybrid Sanskrit’, BSOS 8.501–16; ‘Nouns of the a-declension in Buddhist Hybrid Sanskrit’, HJAS 1.65–83; ‘The Meter of the Saddharmapuda↓rīka’, KSCV 39–45; ‘The Aorist in Buddhist Hybrid Sanskrit’, JAOS 57.16–34. Reprints of most of these are still in the possession of the author, who will gladly distribute them gratis, as long as they last, to responsible scholars on request. The first-mentioned contains a statement of the whole problem in general terms. I use the term ‘protocanonical Prakrit’ of the hypothetical Prakrit dialect on which the bastard Sanskrit of the Buddhists was based. The forms used in this article are taken almost wholly from SP and LV; all from verses except those specified as prose. The following abbreviations are used:

AMg. : Ardhamāgadhī.

Ap. : Apabhramśa.

BSkt.: Buddhist (Hybrid) Sanskrit.

BSOS : Bulletin of the School of Oriental Studies, London.

Cale. : see under LV.

Foucaux: Ph. E. Foucaux, Rgya Teh'er Rol Pa (Tibetan LV., ed. and transi.), Paris, 1847–8; or, the same author's translation of the Sanskrit LV (1884, 1892).

Geiger: Wilhelm Geiger, Pali, Strassburg, 1916.

HJAS : Harvard Journal of Asiatic Studies.

JAOS: Journal of the American Oriental Society.

JM: Jaina Māhārārī.

JRAS : Journal of the Royal Asiatic Society.

KSCV: Kuppuswami Sastri Commemoration Volume, Madras, n.d. [1936].

La k.: La kāvatāra Sūtra, ed. Nanjio, Kyoto, 1923.

LV: Lalitavistara, referred to by page and line of Lefmann's edition; the Calcutta (Bibliotheca Indica) edition is occasionally referred to as Cale.

m.c. : metri causa.

Mhv. : Mahāvastu, ed. Senart.

O: collective siglum used by Kern and Nanjio in referring to the readings of a group of fragmentary mss. of the Kashgar recension of SP.

Pischel: R. Pischel, Grammatik der Prakrit-Sprachen, Strassburg, 1900.

Renou: Louis Renou, Grammaire Sanscrite, Paris, 1930.

Sen: Sukumar Sen, ‘Syntax of Buddhistic Sanskrit’, Journal of the Department of Letters, University of Calcutta, 17 (1928).

SP: Saddharmapundarīka, referred to by page and line of Kern and Nanjio's edition; that of Wogihara and Tsuchida (Tokyo, 1935–6) is occasionally referred to as WT.

Tib: Tibetan (translations of SP and LV).

v.l.: varia lectio.

WT: see under SP.

² Rather than with dissyllabic roots, as suggested by Renou 125. This suggestion is indefensible, because primarily the gerund is formed from the weak grade of all roots, so that even dissyllabic roots should and regularly do show no i (cf. e.g. bhūtvā, tīrtvā). Whenever it occurs it is secondary, and in nearly every case the itvā forms were in my opinion based on the present stem rather than the root. The form in i is, says Whitney (956, cf. 991), ‘used especially with roots having finals that are only with difficulty, if at all, combinable with t.' As a pragmatic statement of fact I think this statement still has value (of course with many qualifications), tho one would prefer to formulate it differently now. The only historically original and ‘correct’ gerunds in itvā are those from roots in i, like jitvā, and those, like sthitvā, in which the i is part of the weak grade of heavy bases, not of the suffix. These may, of course, have helped in the spread of the suffix itvā. The gerunds in (ay)itvā from aya presents are unhistorical in containing the -ay- of the present stem; see note 4.

^2a At least in the text as printed; for utsjya 267.8 (prose) the fragment publ, by La Vallée Poussin, JRAS 1911, 1074, reads parityaktvā.

³ But perfect forms are not so rare in SP and LV as Sen (54) says; his list of roots showing perfects is far from complete.

⁴ In Sanskrit, gerunds from causatives and denominatives (and aya verbs generally) keep or insert the syllable -ay- before the gerund ending itvā, as was noted above, altho their participles show the historically expected form without -ay-; thus kārita pple., but kārayitvā gerund. In our dialect also this is the usual procedure; but we find not a few forms in itvā without -ay-. Thus, cintitva, to be read SP 117.4 for cintetva of both editions (all mss. quoted as cintitv; meter requires short final); janitvā LV 234.7, and °tva LV 50.14, 422.4, etc.; choritva SP 251.5 (BSkt. chorayati); tāritva SP 149.12; pāitva SP 85.12; pūjitva SP 217.9; bhāvitva SP 92.8, 116.4; harşitva SP 25.7; compound bases, prajñapitvā SP 194.3; vivarjitva SP 99.4. It might be suggested that these preserve the historically original forms (§3). But this would be a false theory. They are purely secondary and Prakritic. This is proved by the fact that in our dialect other aya forms lose the syllable -ay-, not only futures (e.g. vitrāsişyase LV 334.7), but even presents: moce ‘I set free’ LV 219.4 (= mocaye, mocayāmi); rocanti ‘they accept, approve’ SP306.4 (= rocayanti, Pali rocenti); pratisamodate ‘greets in return’ LV68.11ff., in prose (= pratisamodayate, which is used just above in the same passage, LV 68.6; this instructively shows the exact equivalence of the two forms; cf. Pali paisamodeti); tād↓ LV 341.13 = tād↓n = tād↓yan, etc. Conversely, also, our dialect has not a few forms in aya without any trace of causative meaning, precisely equivalent to forms without -ay-; e.g. gacchayamī LV 223.1, m.c. for gacchayāmi, = gacchāmi ‘I go’. In other words, the paling out of the distinctive force of the aya formations, which clearly starts in Sanskrit, has gone so far that in our language aya and non-aya forms are freely used interchangeably. Hence no specific significance can be attached to gerunds lacking this -ay-.—It is perhaps worth noting that, altho e for ayi (as well as aya) of Sanskrit is found repeatedly, I have noted only one gerund in etvā for ayitvā from an aya stem: sthāpetv(ā) in sthāpetv-upāya SP 91.8. And this is a kind of petrified form, practically a preposition, ‘except’, =Pali thapetvā. In SP 117.4 the edition's cintetva is an emendation, see above.

⁵ According to Senart, Mhv. i p. xii, note 1, Nepalese mss. make no distinction between hna and h a. Probably, therefore, gh a- should be adopted (with Senart in the edition of Mhv.) always instead of ghna- which is often read in Lefmann's text. There are, however, other cases of n, even intervocalic, where Sanskrit phonology requires n; cf. kariyāna §18.