¹ Much of the contents of this article was presented at the meeting of the American Oriental Society in Baltimore, Easter 1939. I have again to thank Professor Edgerton for much help and advice.

² The essentials of this theory were presented at the meeting of the Linguistic Society of America in Providence, in December 1940, in a paper entitled The alleged Slavic present type *umĭretŭ ‘he will die’.

³ Renou had some doubts about this formation: in Valeur du parfait 141 (1925) he says that the ya-present alone is old, and BSL 33.25 he calls mara- ‘un présent mal assuré, soumis à condition ... ou subjectif’ (another remark p. 10); but he did not arrive at a solution.

⁴ 8.75.7 kám u ṣvid asya sénayāgnér ...

paṇím góṣu starāmahe

‘what Paṇi shall we lay low in the fight with his, Agni's, weapon?‘ Probably aorist subjunctive, although the root aorist (ástar, star) is only active and stárate 1.129.4 is likely to be present indicative.

⁵ AV ā́ śiṣāmahe 18.1.37 (mistakenly Whitney, Index, ā śiṣ° and 18.1.31) is a variant of RV ā́ śiṣāmahi 8.24.1 (pret. indic.).

⁶ Neisser: 'Der Conj. müsste durchaus kárāmahai lauten.' Renou: 'on a ... kárāmahe, alors que le subjonctif gvédique exige ici l'élargissement -ai.' In his treatise La décadence et la disparition du subjonctif (Monographies sanskrites I, 1937), Renou does not take up the type kárāmahe.

⁷ īḷāmahe 10.85.22 (late), not modal, probably is thematic indicative; so Whitney, Roots. The same is likely for dadāmahe 3.21.5 (Renou, BSL 33.17).

⁸ But aśnavāmahe MS (Bloomfield–Edgerton, Ved. Variants 1.§§26, 253).

⁹ But Macdonell classifies the same form in VS as aorist subj., §502.

¹⁰ Geldner: ‘Wer ist der Gott ..., an dessen teuern Namen wir jetzt denken? Wer gibt uns ... zurück?‘ So also Grassmann.

¹¹ I do not believe in Renou's theory of the originally indefinite character of this type. Rather the old short-vowel subjunctive, which is not sufficiently characterized in form, is already beginning to lose its original value and entering a process of transformation.

¹² Cf. the variation RV vánāmahai : SV °he Bloomfield–Edgerton 1.§124.—According to Neisser, BB 7.224, who admits no mahe-forms in the subjunctive, vanāmahe is always indicative: ‘wir wünschen’.

¹³ According to Whitney, Roots, and Macdonell §§431, 433, the stem vaná- is a present of the 6th class; as preterit indicatives do not occur, a definite decision is difficult; but a thematic aorist beside root aorist seems to be more likely than a present vaná- beside vána- and vanó-.

¹⁴ ‘Von den Göttern wollen wir dies erlangen’; Geldner: ‘das setzen wir bei den Göttern durch’; wrongly Grassmann.

¹⁵ Delbrück, Altind. Tempuslehre 81 (1876).

¹⁶ Discussion of the form in Renou, Décadence (cf. fn. 6) §7 (1937).

¹⁷ Neisser, BB 7.216, 219, 225 (1883); Bartholomae, KZ 29.276–9 (1888); Macdonell §502 (but the ending -anti is not named in §414); Renou, BSL 33.8.

¹⁸ Avestan has only -å̄nte: hačå̄ntē ‘they shall pursue’, zayå̄nte ‘they shall be borne’; the short-vowel subjunctive also has only -ante (Bartholomae, Grundriss §§340, 371).

¹⁹ Because of RV āntrá- ‘intestines’, I should not, with Renou, think of phonetic shortening of *-ānta into -anta for the bulk of the Rigveda; also his ‘finale “commune”‘ seems to me unlikely; on the other hand, I think it quite possible that this particular injunctive form in -anta was allowed a wider syntactic use than the injunctive in general has.

²⁰ Bartholomae, KZ 27.212–3, thinks of a proportion vártantai : váriante = váriai : várie, which is hardly likely.

²¹ W. Geiger §§137, 191; Dict. PTS sub marati and miyyati; Andersen, A Pāli Reader, Glossary 203–4.

²² Childers' quotation ‘maranto Dh. 86‘ is misleading; all his quotations beyond (page) 76 refer to the commentary.—Similarly the Dictionary of the Pali Text Society does not distinguish between Jātaka verse and commentary.

²³ ii, 242.4; i, 253.6; 359.23; 360.3; 361.5; and iii, 158.10 respectively.

²⁴ 704.15 (Trivandrum Sanskrit Series 84).

²⁵ A repetition of the Proto-Indic change in Skt. brávīti ‘he speaks’ = Avest. mraoiti.

²⁶ Morgenstierne, Report on a Linguistic Mission to Afghanistan 71.

²⁷ Op.cit. 85.

²⁸ One might envisage the possibility that the indicative marati developed out of the aorist modes: subj. marāti, opt. *maret, and imp. *mara. But canonical Pāli has only the optative mīyetha, miyye; mare- appears only post-canonically, together with the indicative marati. So in the period previous to the rise of marati aorist modes no longer existed.

²⁹ marituṁ : marissati on the model of tarituṁ : *tarissati; *maritvā on the model of marituṁ.

³⁰ Dict. PTS sub udrīyati regards avadīya- as a hypersanskritism for *oddīya- = *uddīya-; but as Sanskrit commonly has ava-dṝ- from the Rigveda on, but only once, Daś., uddīrṇa-(v.l. udīrṇa-), it is more probable that Pāli udrīya- itself is from *avadriya-, *odrīya-.

³¹ Wackernagel §180b : ‘ṛ scheint vor y phonetisch zu ri geworden zu sein und die spärlichen nachvedischen Fälle der Lautfolge ṛy im einfachen Wort auf Systemzwang zu beruhen.‘

³² In Pāli, the passive *tīrati is not quoted, but is presupposed by tīreti ‘to finish’; tīreti : *tīrati on the model of pūreti ‘to fill’ : pūrati ‘to be filled’.—Skt. (Whitney, Roots) ‘tīryate? E’.