^{page 92 note 1} Some shorter passages are discussed sufficiently fully to obviate translation. The translation is intended throughout merely as a literal gloss on the Sanskrit.

^{page 92 note 2} The variety of readings are attributed by him to ‘scribal errors, additions and interpolations in different manuscripts, but these can be easily detected and eliminated by critically checking and collating the text’. It is thus astonishing that he accords (op. cit., pp. 30 ff.) high praise to Subba Rao's, D. retrograde step (J. Or. Inst. Baroda, II, 2, 1952Google Scholar, and appendix 6 in the second edition of the Baroda Nātyaśāstra, vol. I, 1956) in forcing a translation from the text of the Maṇḍapavidhānādhyāya as printed in the Baroda edition, and accrediting this text to the author.

^{page 92 note 3} I refer here to the text of the vulgate compilation from which stem the manuscripts collected by M. R. Kavi. For evidence of an earlier recension see below, p. 98.

^{page 93 note 1} Otherwise the presence of ca is inexplicable. To delete ca and retain -angābhyām would destroy the parallelism of the clauses. In this sentence pūrṇavṛttivṛttyangānām, used of rūpakas other than nāṭaka and prakaraṇa (since they are said to be derived from these two), will doubtless refer to nāṭikā, toṭaka, etc. Here vṛttinyūna is used to mean apūrṇavṛttivṛttyanga.

^{page 93 note 2} See below, p. 109 f.

^{page 93 note 3} For the probable solution to the problem posed by the designation triśṛngāra, see below, p. 107.

^{page 94 note 1} See below, p. 115 f.

^{page 94 note 2} Respectively kaiśikī, sātvatī, ārabhaṭī, and vaidarbha, pāñcāla, gauḍa.

^{page 94 note 3} cf. (p. 451,11.18 f.) Udbhaṭa's ceṣṭātmikā nyāyavṛttir anyāyavṛttir vāgrūpā discussed below.

^{page 95 note 1} The details of the development must remain vague; a possible train of events might be the following:

(1) A set of nyāyas or acting methods, termed bhārata, sātvata, kaiśika (pāñcāla, ārabhaṭa, vārsagaṇa, etc. ?), whose nature can be dimly perceived from the Nāṭyaśāstra definitions of nyāyas and vṛttis.

(2) From the miscellany of terms attested in the Nāṭyaśāstra (ch. viii), the emergence of four standard abhinayas and the application to these, on the basis of fanciful etymology, of the nyāya terms bhārata, etc.

(3) The incorporation of the philosophical concept vāgvyāpāra, its three divisions receiving designations (pāñcāla, etc.) parallel with or modelled on those of the abhinayas and classed as anyāyavṛttis in contradistinction to the nyāyavṛttis (the three abhinayas other than bhāratī which now belongs with the vāgvyāpāras).

(4) The notion of ‘qualities’ developed and inflicted on the nyãyavṛttis through the influence of the Nāṭyaśāstra myth of the invention of the vṛttis, and on the anyāyavṛttis aided by the tendency to interpret their designations as geographical terms. The tendency for vaidarbha to assume all qualities is presumably based as much on the influence of bhāratī (which as the generic anyāyavṛtti must receive all qualities) as on critical assessment of the literature of the south.

(5) Invention of the śṛngārādi rasas, inevitably superimposed on both sets of vṛtti.

^{page 95 note 2} I am indebted to Professor J. Brough for the suggested interpretation of this passage used in my translation and for other valuable guidance in the presentation of this article.

^{page 98 note 1} Previous discussions (e.g. K. M. Varma, Seven words in Bharata) start from the assumption that the Rasādhyāya must be reckoned among the oldest material in the Nāṭyaśāstra. Rasa, however, is the culmination, scarcely the starting-point of Indian aesthetic theory. Against Varma, if any inference can be drawn from Nś. 6.8 ff., it is that, in substantial agreement with Abhinava, kārikā is a versified sūtra (v. 11 sūtrataḥ…arthapradarśinī) and nirukta is versified bhāṣya (v. 13 sthāpito'rtho bhaved yatra…arthasūcakam, dhātvarthavacanena). On this terminology, what I have termed kārikās would be ānuvaṃśyas; but it is far from certain that the terminology holds good for the (Gupta ?) period of the assumed vulgate compilation.

^{page 99 note 1} In the definition of the samavakāra, triśṛngāra is not a reference to rasa. See below, p. 107. As a general term denoting poetic value, the word is common. See below, p. 104.

^{page 99 note 2} Information on the manuscript readings is drawn from M. R. Kavi's apparatus. I assume that unidentified readings are those of manuscript M, since the preface to vol. I indicates that this manuscript is closest to Abhinava's pratīkas at this point.

^{page 99 note 3} I refer throughout to the four pādas of a śloka and the four divisions of an āryā (dividing at the caesuras) as a, b, c, d.

^{page 100 note 1} 11. 37 f. The translators (Dillon, Fowler, and Raghavan, Philadelphia, 1960) contrive to translate as though the text had °ceṣṭitam and construe yat as a relative pronoun, although the omission here of the correlative tat in d (vijñeyaṃ nātakaṃ nāma) suggests that Sāgaranandin was citing a version altered to suit the commentatorial explanation of yat as yasmāt.

^{page 101 note 1} The more obvious emendation śreṣṭhibrāhmarṇa- being avoided since śreṣṭhin was already accounted for in v. 50 (Kāvyamālā, v. 101).

^{page 102 note 1} vyutpādya: ‘to be used as instructive material’ (=vineya p. 410, 11. 17 FF.); upāyavyutpatti = upāyopadeśa (p. 442, 1. 2). Cf. vyutpatti and vyutpādana ‘edification’ (p. 412, 1. 7 and 1. 13).

^{page 105 note 1} For this earlier reading in v. 12, see above, p. 101.

^{page 105 note 2} The words following angāni (lakṣaṇany alaṃkārāk sarve guṇāḥ) may result from reading salakṣaṇam in an early commentary as part of the text and understanding it as sālaṃkāraṃ saguṇam, a better attempt than Abhinava's ankapraveśakayor lakṣaṇayuktam which reflects the obsoleteness of lakṣaṇa in the sense ‘alaṃkara’. The appearance of lakṣaṇa in Nlrk. and Abhinava might be held to result, like sandhi and anga, from glosses on vastuśarīra, but lakṣaṇa would he an odd choice and its appearance in a prescription which continued kevalam…syāt strongly suggests that it stems from a commentator's clarification ‘having the same definition except that…’.

^{page 106 note 1} We may compare the situation in the commentaries on v. 47 (Km., v. 98) where Nlrk.'s source had a reference to anga, understood as sandhyanga, and Abhinava has a reference to anka, understood as ankapraveśakādi. If, as I assume, both stem from a gloss anka or anga on the word kāvyaśarīram of the text, it is probable that the early commentator intendedanka which alone would be comprehensible without qualification. As shown above, it is probable that Abhinava's source commentary had conflated a commentary resembling that known to the Nlrk. (which will have contained the gloss in the form anga) with another which contained the gloss anka.

^{page 108 note 1} Read sā ca instead of nonsensical na ca.

^{page 108 note 2} See śuddhipattrikā.

^{page 113 note 1} ca is here presumably an error for a daṇḍa.

^{page 114 note 1} Read utsṛṣṭatvād veti instead of utsṛṣtatvānye tadā.