^* Abbreviations: AMg = Ardhamāgadhī (Prakrit); BUS = Buddhist Hybrid Sanskrit; BR = Boehtlingk and Roth, Sanskrit-Wörterbuch (the ‘St. Petersburg Lexicon‘); Childers = R. C. Childers, A Dictionary of the Pali Language (4th impression; London, 1909); Divy = Divyāvadāna (ed. Cowell and Neil; Cambridge, Eng., 1886); Geiger = Wilhelm Geiger, Pali Literatur und Sprache (Strassburg, 1916); Hultzsch = E. Hultzsch, Inscriptions of Asoka (Oxford, 1925); IF — Indogermanische Forschungen; JAOS = Journal of the American Oriental Society; ; MIndic = Middle Indic; Mmk = Ārya-Mañjuśri-mūla-kalpa ( Sâstrî, 3 vols., Trivandrum Sanskrit Series 70, 76, 84; Trivandrum, 1920-5); Mv = Mahāvastu (ed. E. Senart, 3 vols.; Paris, 1882-97); Pischel = Richard Pischel, Grammatik der Prakrit Sprachen (Strassburg, 1900); Pkt. — Prakrit; Prāt — Le des Sarvāstivādins (ed. Finot; Journal Asiatique (Paris) 11e Série 2, Juil.-Déc. 1913, 473-543); PTSD = The Pali Text Society's Pali-English Dictionary (ed. Rhys Davids and Stede; Chipstead, Surrey, 1925); (ed. C. Bendali; St. Petersburg, 1897); Skt. = Sanskrit; (ed. Kern and Nanjio; St. Petersburg, 1912).—Pali forms cited without references can be found, with references, in PTSD. A few other abbreviations will be explained in footnotes at their first occurrence.

¹ Speyer, Vedische und Sanskrit-Syntax §21. The very complicated and difficult question of the Skt. causatives has been extensively discussed in recent years, notably by Paul Thieme, Das Plusquamperfektum im Veda 17 ff. (Göttingen, 1929), dealing primarily with the Veda but also largely with Skt.; Renou, Grammaire Sanscrite 471 ff. (Paris, 1930); Batakrishna Ghosh, Les formations nominales et verbales en p du Sanskrit 67 ff. (Paris, 1933). Without disparagement of any of these scholars, it must be said that much is still far from clear. This is partly due to the fact that the -aya- suffix was also used in more than one non-causative type of formation in different periods of Old Indic. But it is also partly inherent in the very nature and common uses of the causative itself. Let me stress one point which seems to me to have been somewhat neglected. In Classical Sanskrit, at any rate, the causative may mean ‘cause (the primary action) to be performed’. In some cases where the primary verb is transitive, this may result in a meaning not easily, if at all, distinguishable from that of the primary verb. ‘He causes (say, a piece of wood) to be cut’ (chedayati) may, of course, mean ‘he has it cut (by someone else)'; these cases are usually clear from the context and in them the causative clearly has a meaning which the simplex could not have. But chedayati may also mean ‘he causes it to become cut, gets it cut’ when the cutting is done by himself (the subject of chedayati). This means virtually the same as ‘he cuts it’. I feel sure that this accounts for some of the cases of ‘causative used in the sense of the simplex’, particularly of such forms as more commonly have distinctively Even when the simplex has a thematic present, the causative is often distinguished from it by a stronger grade of the root, as well as by the -aya- affix: bhāvayati, caus. to bhavati ‘becomes’ (root bhū).² causative meaning. So kārayati (§2) usually means ‘makes (some one else) do (something)’ or ‘gets (something) done (by someone else)', but occasionally also ‘gets (something) done, sees that it is done (by oneself)', which virtually = ‘does it'. And the same holds for Skt. chedayati; but in several of the passages in which BR attribute to it the meaning ‘cut off’ (notably ŚŚS 17.1.8; GGS 4.2.12, so read for BR 4.2.9) it has its usual (distinctively causative) meaning; the definition should be ‘abschneiden lassen’, not ‘abschneiden’.

² A summary statement of details as to form of the root in Whitney, Grammar §1042.

³ Some such formations may not have been causative originally. What is important here is that in Classical Skt. it is often impossible, or at least very difficult, to distinguish them, formally or historically, from genuine causatives in -aya-. Whether by the semantic process suggested in fn. 1 above, or by some other, it seems to me that we must speak of kārayati ‘makes, performs, does’ as ‘the same word’ as kārayati ‘causes to do or to be done’. And if so, any -aya- form which seems to be used in the sense of the simplex may likewise be a causative with altered or specialized meaning.

⁴ Its origin is still obscure, despite numerous attempts at explanation, mostly summarized by Ghosh 69 ff (see fn. 1).

⁵ Renou 469 (see f n. 1) says : ‘il arrive que le type en -āpaya- paraisse lui aussi dépourvu de valeur causative.’ But he cites no examples. If they exist, they must surely be very rare. Ghosh, to be sure, argues for complete parallelism in use between -aya- and -paya- formations; he even goes so far (87) as to call pradāpayati in KS 13.7 (189.1, 2) and TB 1.7.1.1 a ‘simple intensive', whereas it is a most obvious causative; Ghosh misunderstands the passages; unmistakable is KS 189.2 so ‘smā ime pradāpayati ‘he (Vāyu) causes these two (Heaven and Earth) to give to him'. For the rest, such speciousness as Ghosh's argument boasts is dependent on his arbitrarily limited definition of what he calls ‘true causative’ meaning, so as to exclude e.g. TS 2.2.8.4 ‘smai prâ dāpayati ‘verily he makes (men) bestow upon him’ (Keith). This, for Ghosh (86), has no ‘force de causal effectif’, because, forsooth, neither the subject nor the object of the primary verb is expressed. (Yet he apparently approves Keith's translation.) Can Ghosh seriously mean that prá dadāti could be substituted for prá dāpayati in that TS passage without changing the meaning? That is the only question which interests us here. Whether we call its meaning ‘true causative’, or ‘transitive’, or something else, pradāpayati does not mean what the simplex means. Forms in -aya- without -p- certainly do at times have a meaning scarcely distinguishable from the simplex. The fact that the -āpaya- forms were more unmistakably causative in force is certainly involved in the enormous spread of that affix in Middle Indic.

⁶ Edgerton, Lang. 13.111 f.

⁷ Ghosh 67 is therefore wrong in saying that ‘in the Middle Indic dialects all [my emphasis] causatives contain the element -p-.’ To be sure, Pali chindayati is very rare. I know only the gerund chindayituāna Mahāvamsa 9.17; and in this passage the meaning seems to be substantially identical with the simplex, ‘having cut’ (cf. fn. 1 above). A certain Pali case of a new stem in -e- with causative meaning based on a MIndic present is tīreti ‘finishes, executes, accomplishes’, based on *tīrati (unrecorded, but the type is standard, Geiger §175.2; Skt. tīryate) ‘is got over, is penetrated’, originally passive to tarati ‘gets over, penetrates’. Note that tīreti cannot possibly be derived formally from tarati. — Another case is Pali laggeti ‘make:3 stick, fastens’, to Pali laggati (by the side of lagati) ‘sticks’.

⁸ No explanation of the spread of the -āpaya- causative is offered in the standard textbooks. What I believe is the correct explanation was suggested by Manu Leumann, IF 57.224 (for a similar theory, published much earlier, see Tedesco, JAOS 43.389). It started with pairs of participles ending in -ita-, non-causative, and -āpita-, causative, from certain old roots in root-final Skt. -ā (heavy bases) : particularly sthā and its compounds, also certain compounds of dhā, dā, and mā ‘measure’. For example, Skt. utthita- (to ud-sthā, pres. , ‘arise') means ‘arisen'; while utthāpita- (to utthāpayati ‘makes arise, raises') means ‘raised’ or the like. By analogy with such pairs, other participles in -ita-, of whatever origin, came to form causative participles in -āpita-. Such participles in -ita- were regular with presents in -ayati (MIndic -eti), especially causatives and denominatives. But also, from early times, we find participles in -ita- associated with thematic presents in -ati e.g. RV , AV patitá- to palati, likhitá- to likhati. So, on the model of utthāpita- to utthita-, there were created such forms as BUS pradīpāpita- ‘caused to be lighted, ordered lighted’ to pradīpita- ‘lighted’ (ppp. of pradīpayati ‘lights’, caus. to pradīpyate ‘is alight') ; ‘caused to be punished’ to ‘punished’ (from denominative ‘punishes') ; and likhāpita- ‘caused to be written’ to likhita- ‘written’. From such participles it was a short and inevitable step to new causative presents like pradī-· pāpayati, , likhāpayati, and a host of similar MIndic forms, with full inflections.

⁹ §182: ‘Die doppelkausative Bedeutung ... ist vielfach abgeschwächt [my emphasis], tritt aber doch mitunter deutlich hervor.‘

¹⁰ I am here following, at least in essence, a suggestion made by Manu Leumann, IF 57.223, top; one of his illustrations is Asokan lekhāpeti, blend of likhāpeti (fn. 8 above) and lekheti (lekhayati, regular Skt. caus. to likhati). — The occurrence together of chindāpayati first, and then chedāpayati (resuming it), in the above Mv passage may be significant. But such juxtapositions need not invariably have been involved even in the creation of such blend forms, and certainly not in their literary use, once they were created. Pali has chedāpeti, not usually in such contexts, and sometimes clearly as a causative to chiudati, e.g. Sutta Nipāta comm. 3.3 (ed. Smith, vol. 1, PTS. 81 ; London, 1916) (bhikkhū ...) chindanti pi chedāpenti pi, ‘the monks were cutting trees and having them cut’.

¹¹ Actually the form is Senart's emendation; I think it probably right. But the precise meaning is not wholly certain; perhaps it means rather ‘cause (the brahmans) to be fed’, ‘have them fed’, i.e. causative to bhojeti (bhojayati), not to bhuñjati. It is part of a king's instructions to his wife; in line 12 below she carries them out, and it is said that she ‘feeds’ (bhojeti) the brahmans.

¹² Most recently by Manu Leumann, IF 57.223, who is so mystified by the forms that he comments: ‘Man hat durchaus den Eindruck, dass zur Zeit dieser Inschriften (um 250 v. Chr.) der Typus der II. Kausative [i.e. the MIndic causatives in -āpaya-] noch nicht stabilisiert war.‘

¹³ Ratnachandra, Ardha-Magadhi Dictionary, 3.147 (1930) ; Sheth, (Prakrit-Hindi Dictionary). In the following I am obliged to confess that I have had to rely largely on these two secondary sources. The only text I have used is Jacobi's Ausgewāhlte Erzählungen in (Leipzig, 1886). I shall refer by ‘Meyer’ to John Jacob Meyer's translation of this work, Hindu Tales, London, 1909. Except this, the dictionaries cite for these forms partly manuscripts, partly printed works not known to exist in this country, and only in a case or two books reported to be in American libraries other than Yale. I have limited my textual study to Jacobi; the very little more which I could have done by interlibrary loan would still have left the work far from complete.

¹⁴ Ratnachandra 3.147, cited from the ; according to Emeneau's Union List of Printed Indic Texts and Translations in American Libraries (New Haven, 1935), it seems that this work is not available in this country. The form is cited between dāvei and dāvae, hence I suspect may possibly be a misprint for dāvimi (= dāvemi).