For an Indian Philology of Margins: The Case of Kashmirian Sanskrit Manuscripts

Isabelle Ratié For an Indian Philology of Margins The Case of Kashmirian Sanskrit Manuscripts In : L’espace du sens : Approches de la philologie indienne. The Space of Meaning : Approaches to Indian Philology. Ed. by Silvia D’Intino and Sheldon Pollock. With the coll. of Michaël Meyer. Publications de l’Institut de civilisation indienne 84. Paris : Collège de France and Diffusion De Boccard, 2018, p. 305-354.  For an Indian Philology of Margins The Case of Kashmirian Sanskrit Manuscripts * Isabelle Ratié Université Sorbonne Nouvelle The Virtually Unexplored Field of Indian Manuscript Margins As far as the European manuscript tradition is concerned, margins are no longer considered a marginal field of research: the success of this pun in recent scholarly literature is certainly a token of the fact that mar- gins have come to occupy a central position in medieval and Renaissance studies. 1 Several influential works on this topic 2 have opened up new *. Heartfelt thanks are due to the organizers of the conference Issues in Indian Philology: Lyne Bansat-Boudon, Silvia D’Intino and Jean-Noël Robert, as well as to the institutions that kindly granted me access to the manuscripts described below or to photographs thereof, particularly the Library of SOAS (London), the Library of the Śrī Ranbir Institute (Jammu) and the National Mission for Manuscripts (Delhi). Many thanks are also due to Chetan Pandey, who has been working tirelessly for several years to photograph as many Kashmirian manuscripts as possible and to make them available to all scholars; one of the most important annotated manuscripts mentioned below (J11) would not have come to my attention without his efforts. Finally, I am very grateful to Vincent Eltschinger for reading an earlier version of this paper and making insightful remarks. In the transliteration of marginalia I have taken the liberty of adding punctuation and standardizing the spelling of Sanskrit. 1. See for example Keefer & Bremmer Jr. (eds. 2007: 3): “The early 1990s saw a bur- geoning interest in contexts and physical layout of medieval manuscripts […]. Medieval scholarship began to look at marginalia in a new way, re-considering the implications hitherto taken for granted in the modern terms ‘main’ and ‘marginal.’ […]. Marginal infor- mation […] is now finally awarded a central position with unexpected results.” Cf. Smith (2012: 29): “For the past two decades the margin as a physical and conceptual space has occupied a central place in medievalists’ field of vision, and the notions of ‘the marginal’ and ‘marginality’ have become essential components of our intellectual and methodolog- ical kits.” See also Hooks (2012: 636): “Marginalia have become central to Renaissance studies, showing how readers understood and used the texts they encountered.” 2. They include Camille (1992), Tribble (1993), Slights (2001), Keefer & Bremmer Jr. (eds. 2007), Smith (2012), Hooks (2012), Croizy-Naquet, Harf-Lancner & Szkilnik (eds. 306 for an indian philology of margins perspectives on manuscripts by showing how much marginal annotations and drawings can tell us about textual transmission, but also medieval reading and teaching habits; and they have initiated important debates on issues such as the degree of correspondence that may exist between the marginal contents in manuscripts and their social marginality. 3 By way of contrast, so far there seems to have been very little scholarly curiosity as regards the margins of manuscripts produced on the Indian subcontinent. As far as I know, to date only one monographic study — a thus far unpub- lished PhD dissertation — 4 has been devoted to marginal annotations in South Asian manuscripts. This is all the more surprising since in medieval India, marginal annotations were long considered an essential element of manuscripts. This article does not claim to give a thorough presenta- tion of the Kashmirian Sanskrit manuscript tradition in general; 5 it is merely aimed at showing, through the example of Kashmirian Sanskrit manuscripts, that classical Indology has much to gain by paying attention to marginal annotations, because the latter, which often quote at length texts that are no longer extant, can help us retrieve significant portions of lost works; because they thus give us an opportunity to reconsider our perspective on the history of Indian scholarly traditions and the reasons why certain texts were copied in preference to others; and because they may afford us some rare glimpses into the practical aspects of intellectual life in medieval India. What do the Margins of Kashmirian Sanskrit Manuscripts Con- tain? Many Kashmirian Sanskrit manuscripts bear annotations, whether they are made of birch-bark or paper. 6 As can be seen from fig. 1 (Ap- 2015). 3. Thus according to M. Camille, the grotesque marginal characters drawn in medieval margins are to be understood in opposition to the main block of the manuscript page, which embodies the institutions at the ideological centre of medieval society, even though Camille acknowledges that the marginal images studied by him are not truly subversive inasmuch as they rather “work to reinstate the very models they oppose” (Camille 1992: 30; cf. for instance Smith 2012: 32). 4. Formigatti (2011). 5. On the latter subject, very useful works include Bühler (1877), Stein (1894), Grierson (1916), Kaul Deambi (1982), Slaje (1993), Witzel (1994) and Hanneder (2017). 6. The use of paper, later than that of birch bark, is said to have become common around Akbar’s time (Bühler 1877: 29, see Witzel 1994: 7 for a more nuanced estimate),  isabelle ratié 307 pendix III), such marginalia may include the following: 1) Folio numbers, most often accompanied by the abbreviated mention of the work’s title. In our example, “pra bhi 198” (which stands for Īśvarapratyabhijñāvimarśinī, 7 hereafter ĪPV, folio 198) can be read vertically in the bottom left corner (surrounded by a marginal annotation written obliquely). 2) Corrections to the main text, which are placed either in the margins or in an interlinear space. In fig. 1, the words °sāmarthyat pārameś- varāc chakti° (lines 6-7) have been corrected into °sāmarthyarūpāt pārameśvarāc chakti° by inserting the missing °rūpā° with a caret in the interlinear space. 8 3) Short glosses, which are most often interlinear. Thus in fig. 1, one can read kramikakāryabalāt kramikaivānumīyate between lines 2 and 3; anumānapramāṇagrāhyaiveyam ity arthaḥ between lines 3 and 4; and vedyaviśrāntāyāḥ between lines 4 and 5. 4) Lengthier annotations, usually in the margins (but they may extend to interlinear spaces if the margins are full). In fig. 1, there are three of them: one on the top margin that extends to the left margin; 9 a and Bühler explains (ibid.: 30) that by the time he visited Kashmir in the second part of the 19th century, “the method of preparing” birch-bark for the production of manuscripts had been “lost”: “It is at present impossible for the Kaśmīrians to produce new birch- bark manuscripts […]. As matters now stand, there are no birch-bark manuscripts much younger than two hundred years.” Although birch-bark manuscripts known to me often bear less annotations than paper ones, heavily annotated birchbark manuscripts are not uncommon (see e.g. fig. 2 below, or Slaje 2005: 43, which describes an “old birch-bark codex” of a śāradā-written “Yogavāsiṣṭha” that “abounds throughout in glosses written interlinear or in the margins by different hands”). 7. Witzel (1994: 11) remarks that “it would be useful to make a collection of such abbreviations as they considerably help in establishing the identity of fragments, and in any case, speed up the process of identification.” It should be noted, however, that abbreviations of a given title can fluctuate from one manuscript to the other, and some- times even in the same manuscript (thus in J2, shown here in fig. 2, Abhinavagupta’s Īśvarapratyabhijñā[sūtra]vimarśinī is abbreviated on some folios as pra bhi sū and on others as pra sū vi). 8. Note that later the last a in °sāmarthyat was made into an ā and the interlinear correction °rūpā° was changed into °rūpa° (probably because a corrector first saw that something was amiss with °sāmarthyat and only realized after making the tentative cor- rection °sāmarthyāt that the correct text was already given above the line). 9. etad uktaṃ bhavati: laukikajñānavyavahāramūlajñātr̥jñānaprakāśane sāmarthyaṃ jñāna- śaktir devasya, tatraiva saṃyojanādisvatantrātmā* vimarśaśaktir eva vastutaḥ kriyā, sā tu vedyac- 308 for an indian philology of margins shorter one on the left margin; 10 and again, a lengthy one spreading from the left to the bottom margin. 11 Although nowadays marginalia tend to be overlooked by many of those who attempt to edit Sanskrit works from Kashmirian manuscripts, the pioneering editors of the famous Kashmir Series of Texts and Studies (hereafter KSTS) did not fail to see the interest of annotations, and if they rarely made any explicit mention of the value that they ascribed to them, 12 they faithfully reproduced in footnotes all the marginalia that they could find in the manuscripts at their disposal. Thus in the example given above, the first and third interlinear glosses are found in footnotes of the edition of Abhinavagupta’s ĪPV that appeared in 1918-1921 in the chāyācchuraṇam acchatvād avacchedakam abhyupayantī kramavatī bhāti vastuto’kramatve’pi. kālānavacchedāt tāvad akramatvaṃ sthitam eveti, kim atrocyate, unmiṣate**’pi vedyagrāme māyā- pramātr̥padam apy adhyāsīno bhagavān yāvad bāhyāntararūpaṃ vedyajātaṃ prakāśasyātmavi- śrāntir itīdam ity asya vicchinnetyādinayena viśramayati, tāvat tanniṣṭho’vabhāsanakramo’pi grasta evety aham iti yā saṃyojanādicitritasamastabhāvaprakāśopahāraparyantadaśāviśeṣa- vimarśalakṣaṇā kriyā, sā niṣkrameti caturviṃśatisāhasryām. [*°svatantrātmā should probably be corrected into svatantratātmā. **unmiṣate should be corrected into unmiṣite]. 10. nīlādibhedapradarśanasāmarthyarūpā kālaśaktir iti yena śaktiviśeṣaṇābhāsabhedaḥ pra- darśyate sa śaktiviśeṣaḥ kāla ity arthaḥ. 11. madhyāyāṃ — iha māyāśaktyā vicchinnāvabhāsiny eva saṃvedyabhāge grāhyagrāhaka- bhāvaprakāśanena saṃsārasthitir nirvartyate. tatra saṃvedyabhāge vakṣyamāṇanyāyena kāla- śaktyavacchedāt kriyā kālāvacchinnā sakramatvenopapādayiṣyamāṇarūpā syāt. yā punaḥ pra- bhoḥ prakāśātmano’haṃvimarśākhyasvabhāvabhūtā sā kālāvaruddhasaṃvedyakulakavalīkara- ṇena kālagrāsād akramaiva. 12. See, however, the Preface by Mukunda Rām Śāstrī to Abhinavagupta’s Parātrīśikā- vivaraṇa (hereafter PTV): 2, which mentions a śāradā manuscript, the “marginal notes” of which “solve many a mystery for the scholar.” Of course, Western scholars too have emphasized at times the importance of Kashmirian marginalia; see e.g. Stein’s Preface to his edition of the Rājataraṅgiṇī (hereafter RT): ix: “Through the whole of Ratnakaṇṭha’s manuscript […], we find very numerous glosses, various readings and corrections marked by different hands which add, in no small degree, to the critical value of the codex”; cf. Stein (1900: 47): “The great critical value of the codex archetypus […] is due not only to the fact of its having been written by a scholar of Rājānaka Ratnakaṇṭha’s stamp, but also to the abundance of important glosses, various readings and corrections which later hands have recorded in it.” See also Wezler & Motegi (1998), in which lengthy marginal annotations were edited along with the main text (e.g. Ibid.: xxiv: “the discovery and edition of this scholion on the Yuktidīpikā […] forms one of the major improvements we think we have achieved”), and Muroya (2010) and Ratié (2017a), which both highlight the philological and historical value of Kashmirian marginalia. As far as non-Kashmirian manuscripts are concerned, note for instance that Rāhula Sāṅkr̥tyāyana, in his edition of Manorathanandin’s Pramāṇavārttikavr̥tti (hereafter PVV), chose to give in footnotes all the marginal annotations written down by Vibhūticandra.  isabelle ratié 309 KSTS; 13 and the second and third marginal annotations also happen to be printed in footnotes of the same edition. 14 It should be kept in mind, however, that the KSTS editors did not use the annotated manuscript now kept in the SOAS Library and shown in fig. 1; rather, they consulted what must have been its antigraph 15 and contained the same marginalia, which were copied from it in the SOAS manuscript along with the main text. This is one of many examples showing that marginal annotations were very often copied from one manuscript (and sometimes from several manuscripts) 16 to the next, 17 so that in all likelihood, in the case of certain texts at least, the corpus of annotations, often written by several hands in 13. Compare the gloss between lines 2 and 3 with ĪPV, vol. II: 6, n. 30 (the edition has kramikā kāryabalāt instead of kramikakāryabalāt), and the gloss between lines 4 and 5, with ibid., n. 31. 14. The second (quoted above, p. 308, n. 10) and the third (quoted above, p. 308, n. 11) are respectively found in ĪPV, vol. II: 7, n. 32 and 34 (in the latter case, the text begins with ayaṃ bhāvaḥ instead of madhyāyām). 15. Manuscript Gha in the KSTS edition (Preface, p. i) is in all probability D2, now preserved at the National Archives of India (Delhi). On the relationship between D2 and SOAS, see Ratié (forthcoming c). 16. This could explain why some annotated manuscripts bear twice or three times the same annotations with a few occasional variants (see e.g. Ratié forthcoming c on the Vivr̥ti on ĪPK 2.1.5): these annotations may have been copied from several annotated exemplars, some of which happened to share identical marginalia at slightly different places in the margins. For example, the annotation in fig. 6c is found in SOAS on the top margin of folio 202a, but also, with some variants, on the bottom margin of folio 203a; and it is likely that both annotations were copied from D2, which also bears the two quotations of the same fragment (with the same variants) in the margins of folios 203a and 204b. 17. As is often the case in Western manuscript traditions as well (see Clemens & Graham 2007: 39). For another example in the Kashmirian tradition see e.g. Vasudeva (2004: xiii), noting about the Mālinīvijayottaratantra (hereafter MVT) that “G and P share marginal notes, copied with varying degrees of fidelity from a common set of annotations.” See also Formigatti 2011: 79-80. This, of course, makes any attempt to date the annotations particularly risky. According to Wezler & Motegi (1998: xxiv), the marginal annotations found in Yuktidīpikā manuscripts may date back at least to the 14th century because they include technical Buddhist terms, the meaning of which must have been lost after Buddhism disappeared from Kashmir; but this assumption seems somewhat speculative given that the memory of Buddhist philosophical theses endured far later than the 14th century in Kashmir (even though misunderstandings certainly occurred more and more frequently — see e.g. Ratié 2017a: 167-168). 310 for an indian philology of margins a single manuscript, 18 kept growing for centuries. 19 The pandits of the Srinagar Oriental Research Department who edited the KSTS collection found printed footnotes to be an entirely natural locus for manuscript marginalia, as appears from the fact that in his Preface to Somānanda’s Śivadr̥ṣṭi (hereafter ŚD), Madhusudhan Kaul Śāstrī described manuscript “a” 20 as being “with footnotes on the margin”: for the Kashmirian editor, marginal annotations were nothing but footnotes. This attitude was so pervasive that the KSTS editors did not even take the trouble of specifying the source of their footnotes, which has led some modern scholars to wrongly assume that the editors were to be credited with composing them. 21 Studying these marginalia is of course extremely helpful when one endeavours to edit and translate the text that they are meant to explain, because glosses may shed much light on the meaning of passages the interpretation of which is not easy, and because some of the marginalia carefully record different readings. 22 Besides, in Kashmirian manuscripts as well as in Western manuscript traditions, many corruptions result from marginal or interlinear glosses that end up creeping in the main text, and studying them instead of dismissing them as alien to the text may enable us to spot many a scribal interpolation. Yet I would like to argue here that the interest of marginal annotations goes way beyond their ability to illuminate the main text copied in the manuscript, and that they have much to tell us about other texts as well. 18. See the famous example of the main manuscript used by Stein for his edition of the RT, the marginal annotations of which can be traced back to at least four different hands according to Stein; see RT, Preface: x-xii and Stein (1900: 47-48). See also the descriptions of manuscripts annotated by several hands in Formigatti 2011: 61-70. 19. Cf. Muroya (2010: 220) on marginal annotations in Nyāyamañjarī (hereafter NM) manuscripts: “The glosses […] were probably not composed by a single author at a specific point in time. It is more likely that the glosses were added, enlarged or modified by several persons during the transmission of the text of the NM.” 20. Nemec (2011: 81) assumes that it is currently preserved in Srinagar; in fact, however, it is most probably kept in Delhi, at the National Archives of India: see Ratié (forthcoming b: n. 14). 21. See e.g. Nemec (2011: 99, n. 2; 101, n. 11 and 13; 102, n. 20; etc.) 22. See e.g. RT, Preface: ix-xi; see also e.g. the KSTS edition of the MVT: i, and Vasudeva (2004: xiv).  isabelle ratié 311 Retrieving Lost Texts in the Margins of Kashmirian Manuscripts: the Case of Utpaladeva’s Vivr̥ti on the Īśvarapratyabhijñā Trea- tise Thus marginal annotations in Kashmirian Sanskrit manuscripts are most often quotations from other works that happen to bear on the issue at hand in the main text. 23 This means that certain texts that are poorly transmitted or have not come down to us at all might be at least in part salvaged by studying these annotations — a fact that had already been noticed, albeit to a limited extent, by the pandits of the Srinagar Oriental Research Department, who produced a partial edition of Utpaladeva’s shorter commentary on his own Pratyabhijñā treatise 24 on the basis of three manuscripts, two of which in fact contained a commentary by Abhinavagupta but had annotated margins that happened to “con- tain the Vr̥tti.” 25 Besides, Yasutaka Muroya has recently managed to fill several important lacunae in the only extant commentary on Jayanta’s NM, Cakradhara’s Nyāyamañjarīgranthibhaṅga (hereafter NMG), which had been edited on the basis of a single, incomplete manuscript, by studying marginalia in NM manuscripts. 26 There are, however, other texts that may be retrieved by examining the margins of Kashmirian manuscripts. To come back to the example used above, fig. 1 is a page from the ĪPV, a text that belongs to the corpus of commentaries on the Īśvara- pratyabhijñā treatise. This latter work, composed in Kashmir by the Śaiva nondualist Utpaladeva in the 10th century, argues on the basis of a se- ries of logical, epistemological and phenomenological analyses that we must recognize ourselves as Śiva defined as a single, omnipotent and infi- 23. See e.g. the remark in Vasudeva (2004: xix) to the effect that the marginal notes found in manuscripts G and P of the MVT “adduce explanatory material drawn from Abhinavagupta’s Tantrāloka and other exegetical sources.” Cf. Formigatti (2011: 85) on the Kashmirian śāradā manuscript (Be) of Kālidāsa’s Raghuvaṃśa, where “the scholia consist of quotations of well-known commentaries.” Formigatti (ibid.) even takes this feature as one of the characteristics of Kashmirian annotated manuscripts. 24. This edition of Utpaladeva’s Vr̥tti, published along with the Siddhitrayī, is now outdated thanks to the complete edition in Torella (2002). 25. Siddhitrayī, Preface: 3. 26. See Muroya (2010). As regards manuscripts outside Kashmir, note that the marginal annotations by Vibhūticandra in the codex unicus of Manorathanandin’s PVV are partic- ularly useful as they often enable to identify fragments of earlier commentaries that are otherwise only known today in Tibetan translation. 312 for an indian philology of margins nite consciousness out of which nothing exists. The treatise — arguably one of the most refined in the whole history of Indian philosophy — is in fact made of several layers: Utpaladeva first wrote some stanzas — the Īśvarapratyabhijñā-kārikās, hereafter ĪPK — along with a short prose commentary thereon, the (Īśvarapratyabhijñā-)Vr̥tti, after which 27 he com- posed a second, much more detailed explanation of his own work, the (Īśvarapratyabhijñā-)Vivr̥ti or Ṭīkā. Today, the kārikās and Vr̥tti are available both in manuscript form and in Raffaele Torella’s excellent edition and translation; 28 unfortunately, no complete manuscript of the Vivr̥ti or Ṭīkā has been found to date, which is all the more regrettable since the latter text — which explained at length the kārikās and Vr̥tti and even addressed a number of issues that do not explicitly appear in them— 29 certainly constituted the most crucial part of the treatise. R. Torella has edited and translated a lengthy fragment of it from a codex unicus (D) now preserved at the National Archives of India (hereafter NAI) in Delhi, 30 but this unique manuscript is terribly incomplete: whereas the treatise is summed up in 190 stanzas, the Delhi manuscript only covers the Vivr̥ti’s explanation of 13 of them. Two other crucial commentaries have come down to us, by the great Kashmirian polymath Abhinavagupta (c. 975-1025), namely the ĪPV, which offers a synthetic explanation of Utpaladeva’s kārikās, and the very lengthy Īśvarapratyabhijñāvivr̥tivimarśinī (hereafter ĪPVV), which primarily comments on Utpaladeva’s almost entirely lost Vivr̥ti. Now, if we have another look at fig. 1, we can see that the annotation on the top margin of this page of an ĪPV manuscript happens to be a quo- tation from Abhinavagupta’s other commentary on the same treatise, that is, the ĪPVV. 31 Understandably — since both texts are by the same author and comment on the same text —, manuscripts of Abhinavagupta’s ĪPV contain many quotations from his ĪPVV, but they also cite a whole range of other texts, some of which belong to the same Śaiva philosophical tra- dition (such as Somānanda’s ŚD; Utpaladeva’s commentary thereon; the latter’s Vr̥tti on the Īśvarapratyabhijñā treatise; or Kṣemarāja’s Pratyabhijñā- hr̥daya) while others were composed by representatives of rival currents 27. See Torella (2002: xli). 28. See Torella (2002). 29. This is obvious from Abhinavagupta’s commentary thereon. 30. See Torella (2007a-d). 31. The text of the annotation given above, p. 307, n. 9, is in fact (with a few omissions and variants) ĪPVV, vol. III: 3-4.  isabelle ratié 313 of thought, either Brahmanical (this is the case of Bhartr̥hari’s Vākya- padīya and Jayanta’s Nyāyamañjarī) or Buddhist (for instance Dharmakīrti’s Pramāṇavārttika[svavr̥tti] and Sambandhaparīkṣā). Most importantly, to the delight of historians of Indian philosophy, some of these marginalia also contain fragments of Utpaladeva’s lost Vivr̥ti or Ṭīkā; thus in fig. 1, the marginal annotation at the bottom 32 happens to be the beginning of the Vivr̥ti on ĪPK 2.1.2. 33 In the course of the last decade, many fragments of Utpaladeva’s magnum opus were thus retrieved from the margins of commentaries on the Pratyabhijñā treatise: short passages (each rang- ing from a sentence or a few sentences to the equivalent of a couple of printed pages) were found in the margins of one Vr̥tti manuscript studied by Yohei Kawajiri 34 and, to a greater extent, in the margins of nine 35 ĪPV manuscripts that I have been studying 36 (for pictures of such ĪPV manuscripts see here figs. 1, 3a-c, 6c and 8) and in those of an annotated Vr̥tti manuscript preserved in Srinagar and used by R. Torella for his edi- tion of Utpaladeva’s Vr̥tti. 37 Most importantly, I recently had the good fortune of spotting three chapters of the Vivr̥ti, which constitute by far the lengthiest Vivr̥ti fragment known to date (including the codex unicus studied by R. Torella), in the margins of two ĪPVV manuscripts 38 (see figs. 32. See above, p. 308, n. 11. 33. See Ratié (forthcoming c); the words tad āha māyāśakter iti, found in other sources after syāt, are omitted in SOAS. 34. See Kawajiri (2016a and 2016b) for the Sanskrit text of the fragments in question; only a diplomatic edition is offered in the first, although a few emendations are attempted in the second. 35. To the eight manuscripts mentioned in the Appendix of Ratié (2016b) should now be added manuscript S15. 36. See Ratié (2016a, 2016b, 2017a, forthcoming a and forthcoming c) for an edition, translation and study of some of these fragments. 37. Manuscript S19 was described in Torella (2002: l-li), under siglum N. According to Torella (2002: li), it is “profusely annotated by a different hand with single glosses or long passages (mostly drawn from the ĪPV) between the lines and often entirely covering the margins.” Indeed, many of the quotations are from Abhinavagupta’s ĪPV; yet as shown in Ratié forthcoming c, it also bears Vivr̥ti fragments. Although this Vr̥tti manuscript and the one studied by Yohei Kawajiri are certainly related, the latter is not a mere transcript of the former and was probably copied from another antigraph (see Ratié forthcoming c). 38. That is, J11, described in Ratié (2017a), and S12. Since S12 is written in śāradā, it is tempting to assume that J11 is a nāgarī transcript of it; it is very likely, however, that in fact J11 was not directly copied from S12 (see Ratié forthcoming c), which means that other such annotated manuscripts bearing entire chapters of Utpaladeva’s lost Vivr̥ti might have survived. 314 for an indian philology of margins 4, 5, 6a, 6b and 7), 39 and their edition and translation is underway. 40 A Jigsaw Puzzle: Examples from Utpaladeva’s Vivr̥ti, Chapter 2.1 One of the greatest difficulties encountered by those who look for fragments of lost texts in the margins of Kashmirian manuscripts is the identification of the works quoted. Although the annotations found there sometimes explicitly mention the sources of their quotations, 41 they often offer titles that differ from the one under which these texts are best known (this is the case for instance in fig. 1, which explicitly refers to both Abhinavagupta’s ĪPVV and Utpaladeva’s Vivr̥ti, but does so by respectively calling them the Caturviṃśatisāhasrī 42 and the Madhyamā 43 ); and most of the time, they remain entirely silent as to the origin of the quotes. The advent of electronic texts has greatly simplified the task of identi- fying the marginal citations, even though many texts are quoted with nu- merous variants (which makes automatic searches somewhat unreliable), not to mention that in their eagerness to rescue lost texts from obliv- ion, researchers are sometimes too hasty in offering identifications. 44 39. For an edition and translation of the marginal annotations shown in these pictures see Ratié (2017a); note that at the time I only had J11 at my disposal. 40. See Ratié (forthcoming c), devoted to Chapter 2.1; R. Torella is currently working on Chapter 2.2, and I am preparing an edition and translation of Chapter 2.3. 41. A fact that seems to be overlooked in the description of annotated ĪPV manuscripts in Formigatti (2011: 84) (“a large number of annotations are quotations taken from other works by Abhinavagupta or from well-known philosophical and theological texts […], but without mentioning the source” — italics by Formigatti). Besides the examples mentioned here, see Ratié (2017a: 168, n. 17); see also e.g. Muroya (2010: 219) on marginal annotations occasionally quoting Cakradhara’s NMG as the Ṭīkā. 42. See above, p. 307, n. 9; i.e., it is the work whose length corresponds to “twenty-four thousand” ślokas (Utpaladeva’s Vivr̥ti is sometimes called the Aṣṭasāhasrī, the work whose length corresponds to “eight thousand” ślokas, see Torella 2014: 116, n. 2). 43. See above, p. 308, n. 11. Cf. e.g. Bhāskarī, vol. I: 3, which explains that Utpaladeva has composed not only “the Vr̥tti, known as the Short Pratyabhijñā” (laghupratyabhi- jñākhyāṃ vr̥ttim) but also “the Vivr̥ti, known as the Medium Pratyabhijñā” (madhyapraty- abhijñākhyāṃ vivr̥tim), probably as opposed to Abhinavagupta’s ĪPVV, which, according to Bhāskarakaṇṭha, is “the very detailed commentary called the Great Pratyabhijñā” (br̥hat- pratyabhijñākhyā bahuvistarā ṭīkā). 44. See e.g. Kawajiri (2016b: 24), which quotes the following marginal annotation as a “fragment” that “seems to be” from the Vivr̥ti on ĪPK 2.1.4: tau cābhāsānāṃ bhāvābhāvau na vāhyahetukr̥tāv iti vistāryopapāditam iti sa eva saṃvitsvabhāva ātmā svaprasaṃkalpādāv ābhāsavaicitryanirmāṇe prabhuḥ prabhaviṣṇur iti svaṃ saṃviditaḥ [tata eva ?ai??????]. This is in fact a quotation from Abhinavagupta’s ĪPV, vol. II, p. 12, ll. 2-7.  isabelle ratié 315 In any case electronic texts cannot solve the problem that we face when lengthy marginalia do not seem to belong to any known text: in the case of Utpaladeva’s lost Vivr̥ti, when, in a manuscript belonging to the Praty- abhijñā corpus, we spot a marginal annotation that does not belong to any known Sanskrit philosophical work, we are still left with the necessity of proving that such comments are not a late gloss by some learned owner of the manuscript, but actually belong to Utpaladeva’s magnum opus. One important task that may help determine that some marginal annotations indeed belong to this lost text consists in comparing the margins of the manuscript in question with the Vivr̥ti fragment edited by R. Torella on the basis of the codex unicus preserved in Delhi: if these margins can be shown to contain fragments from already known Vivr̥ti passages, in all probability they also contain other fragments hitherto unknown of the same work. Thus in fig. 3c, the text quoted in the bottom margin and spreading vertically on the right margin is in fact a quotation of a Vivr̥ti passage already known thanks to R. Torella’s work. 45 A further element that may enable us to determine that an annotation is a genuine Vivr̥ti fragment is a systematic comparison with Abhinavagupta’s com- mentary on Utpaladeva’s lost text, since we are fortunate enough to have a commentary on the latter: if the words or chains of words quoted in the ĪPVV while Abhinavagupta is explaining them are invariably found in the correct order in the annotation, the probability that we are indeed dealing with a Vivr̥ti fragment verges on absolute certainty. Unfortunately, editing texts or fragments of texts from marginal an- notations is often akin to trying to find (and fit together) the pieces of a jigsaw puzzle — and requires a similar amount of patience. This happens when dealing with rather brief fragments, each of which does not exceed two or three printed pages, as is the case in the ĪPV manuscripts of my knowledge, 46 but the analogy is painfully accurate even when we are lucky enough to have at our disposal entire chapters of a lost work. Thus in the case of the Vivr̥ti’s Chapter 2.1, we now know two ĪPVV manuscripts bearing the text in question in their margins from beginning to end, but they suffer from a number of minor lacunae. Many, however, can 45. The text, from nanv arthasya (beginning of the first line of the annotation in the bottom margin) to tatpūrvakāloparaktānubhūtabhāvāvamarśanāvasānā (end of the line written vertically in the right margin), is found in Torella (2007b: 545). See Ratié (2016a) for similar examples in this manuscript and others. 46. See Ratié (2016a, 2016b, 2017a and forthcoming a). 316 for an indian philology of margins be filled with the help of much shorter quotations of this text found in some ĪPV manuscripts. Thus in the Vivr̥ti’s explanation of ĪPK 2.1.5, the margins of ĪPVV manuscripts J11 and S12 both indicate with dots that some akṣaras could not be read by the scribe (see the beginning of the annotations shown in figs. 6a and 6b). 47 The lacuna, however, can be filled with the help of short Vivr̥ti quotations found in the margins of two ĪPV manuscripts (D2 and SOAS, see fig. 6c for SOAS). 48 Short quotations scattered in the margins of ĪPV or Vr̥tti manuscripts can thus be immensely helpful even though we have at our disposal two ĪPVV manuscripts giving the full text of several Vivr̥ti chapters; yet gath- ering and taking into account these shattered pieces of text can be some- what complex. To take but one example, the Sanskrit text of the Vivr̥ti on ĪPK 2.1.3 is given below in Appendix I, and for the sake of clarity, I have inserted in it bold numbers in brackets; for each of these numbered passages, the table in Appendix II indicates which marginal sources were used. As can be seen from this table, in order to edit this brief excerpt we have to take into account a number of sources that greatly varies from one sentence to the next — and sometimes within the same sentence. With the exception of the continuous marginal quotations found in J11 and S12, 49 the Vivr̥ti excerpts are often very short and scattered among many other quotations, which can make them particularly difficult to spot. 50 Besides, a number of richly annotated manuscripts belonging to the Pratyabhijñā corpus do not bear any significant quotations from the Vivr̥ti, so that efforts to study all marginalia in the known manuscripts of this corpus can seem infuriatingly futile or time-consuming. Yet they are worth the trouble: in 2007, thanks to R. Torella’s study of the Vivr̥ti in the fragmentary manuscript preserved in Delhi, we knew the Vivr̥ti on 47. teṣām eva ca pratyekaṃ janmapariṇāmādīnāṃ svātmany api bhāgāvabhāsa ity eṣa kā..... ekasyāpi kriyāśritaḥ. The words underlined and in bold are those quoted in ĪPVV, vol. III, p. 16. The first part of this sentence is also found at the end of the fragment quoted in the ĪPV edition, vol. II: 18, n. 77. 48. The missing compound, kālakrama, appears on the second line of the annotation. 49. And in S15 (an ĪPV manuscript), but only from ĪPK 2.1.1 to 2.1.3; the quotation breaks around the middle of the commentary on ĪPK 2.1.3, and from then only bears very few brief Vivr̥ti quotations. 50. Formigatti (2011), which contains a study of the marginalia in several ĪPV manuscripts, including one (labelled Ś7 in the thesis) that has now been shown to contain Vivr̥ti fragments (see Ratié forthcoming a), makes no mention of fragments of Utpaladeva’s lost work.  isabelle ratié 317 the Pratyabhijñā treatise’s Chapter 1.4, as well as on the end of Chapter 1.3 and the beginning of Chapter 1.5; apart from this important fragment, only a few scattered lines had been spotted. 51 A decade later, we have access to three thus far entirely unknown chapters, plus a number of shorter fragments, the most substantial of which were retrieved from the margins of ĪPV manuscripts — and it seems safe to assume that more such discoveries will be made in the coming years. Manuscript Margins and the Marginalization of Texts in Me- dieval Kashmir: the Quasi Obliteration of Utpaladeva’s Vivr̥ti There is something quite paradoxical about these recent discoveries. The quotations given in Kashmirian manuscripts obviously have as their primary function to help the reader understand the main text copied in the centre of the manuscript pages, and in fact many of them happen to be excerpts from commentaries on that main text; 52 the case of Utpaladeva’s Vivr̥ti, however, appears to involve a bizarre reversal of this usual situation: Utpaladeva’s main work seems to have become more and more marginal — in every sense of the word — while Abhinavagupta’s commentaries on it were becoming more and more central, so that today, apart from the fragmentary codex unicus preserved in Delhi, all that is known of Utpaladeva’s most important philosophical work is to be found… in the margins of its own commentaries. In this respect it is worth noting that quotations of the ĪPVV found in the margins of ĪPV manuscripts have often undergone a subtle but telling metamorphosis: it is not rare to find ĪPVV quotations in which all references to the fact that Abhinavagupta’s ĪPVV is a commentary on Utpaladeva’s Vivr̥ti have been obliterated. Thus in fig. 1, the top mar- gin gives an ĪPVV quotation 53 from which the sentence uktaṃ caitat kriyā ca paravāgvr̥ttyādiketyādipradeśeṣu 54 (“and this has been said [by Utpaladeva] in places [of his Vivr̥ti] such as [the passage beginning with] 51. For a fragment spotted in an ĪPV footnote see in particular Stern (1991: 158) and n. 55 (cf. Ratié 2016b: 386). 52. See for example Muroya (2010) on the NMG, and Formigatti (2011: 70-74) on the close relationship between commentaries and marginal annotations in North Indian manuscripts. 53. See the transliteration above, p. 307, n. 9. 54. ĪPVV, vol. III: 3, ll. 15-17. 318 for an indian philology of margins paravāgvr̥ttyādikā…”) 55 has disappeared. Similarly, the words iti sūtrārthaḥ, vr̥ttivivr̥tī spaṣṭe (“this was the meaning of [Utpaladeva’s] verse; both [his] Vr̥tti and Vivr̥ti [thereon] are [self-]evident”) 56 are no longer to be found after vastuto ’kramatve, and the quotation of the words kālagrāsāt from the Vivr̥ti that immediately follows in the ĪPVV 57 has also disappeared. In other words, in this margin of an ĪPV manuscript, Abhinavagupta’s commentary on Utpaladeva’s Vivr̥ti has been altered so as to appear to be a commentary on Abhinavagupta’s ĪPV, and all traces of the Vivr̥ti as the mūla-text have been removed! 58 Several factors might be invoked to explain this curious marginaliza- tion of Utpaladeva’s Vivr̥ti and the temptation, perceptible in the alter- ations undergone by some ĪPVV marginal quotations, to let Utpaladeva’s most important work sink into oblivion. Abhinavagupta’s detailed and elegant commentaries might have been deemed clearer than Utpaladeva’s highly inventive but terse and elliptical work; besides, with the excep- tion of Utpaladeva’s devotional hymns, the latter’s works are confined to the genre of philosophical treatises and commentaries, whereas Abhinavagupta has authored important works on various topics (the aesthetics of theatre, the aesthetics of poetry, the interpretation of Śaiva scriptures, the practice and meaning of Śaiva rituals…), so that his fame may have surpassed early on Utpaladeva’s — and this may have greatly influenced the choices that were made as to which works should be copied preferably to others within the Pratyabhijñā corpus. And once the mūla-text had become a rarity — which probably occurred before the 55. This seems to be a quotation from Utpaladeva’s Vivr̥ti on ĪPK 1.1.5 (not available to date except for Abhinavagupta’s quotations of it): see ĪPVV, vol. I: 105: ity āśaṅkyāha paravāgvr̥ttyādiketi. 56. Ibid.: ll. 18-19. 57. Ibid., l. 19. 58. This phenomenon is far from being limited to the SOAS manuscript (or D2, from which it was copied), and other attempts of this kind can be found for instance in S3, where a missing folio (18b-19a) has been replaced by another one written in a different hand. On the verso of this folio are several annotations occupying the whole page, and the bottom annotation is a quotation from the ĪPVV, but most of the references to the Vivr̥ti have disappeared and a number of glosses have been inserted. To give just one example, around the middle of line 5, the text quotes two sentences from ĪPVV, vol. I, p. 200 (… tarhi kārakavibhaktīnām api sa evārthaḥ, tad āha sarva eveti. nanu sa eva sambandho bhaviṣyaty eka ity āśaṅkyāha naikakālikam iti), but in doing so it omits tad āha sarva eveti (“this is why [Utpaladeva] says ‘sarva eva’ [in the Vivr̥ti]”) and ity āśaṅkyāha naikakālikam iti (“Having anticipated this [objection, Utpaladeva] says ‘naikakālikam’ [in the Vivr̥ti]”).  isabelle ratié 319 end of the 17th century —, 59 making the pratīkas disappear in the ĪPVV might have seemed a sensible thing to do, since by that time the ĪPVV was probably seen mainly as a useful tool for the study of Abhinavagupta’s ĪPV, which, judging from the great number of manuscripts preserved, had certainly come to be considered the most important work in the Pratyabhijñā corpus. 60 Whatever the reasons for these choices may have been, however, such decisions have profoundly affected our perception of the roles respec- tively played by Utpaladeva and Abhinavagupta in building the Praty- abhijñā system. Thus before the publication of R. Torella’s pioneering studies on Utpaladeva’s Vivr̥ti, historians of Indian philosophy had simply assumed that Abhinavagupta’s commentaries were the most innovative works in this tradition, 61 and they tended to systematically ascribe to Abhinavagupta himself the ideas that they could find in the latter’s works, but not in Utpaladeva’s stanzas or Vr̥tti; as more and more Vivr̥ti fragments are coming to light, however, we are now able to compare all of these texts, and it is becoming ever more obvious that Abhinavagupta was a brilliant commentator, but also a very faithful one, and that Utpaladeva’s Vivr̥ti was the truly innovative work. An interesting example of such a distortion in our vision of the history of the Pratyabhijñā system is the issue of the famous “doctrine of reflec- tions” (pratibimbavāda): 62 in many of his works, Abhinavagupta claims that the phenomena constituting the universe as we perceive it are similar to reflections in a mirror, because the variegated reflections that we may see in a mirror are perfectly real inasmuch as they really occur on the surface of the mirror, yet they are nothing over and above the surface of the mirror that makes them manifest by assuming their forms. In the same way, argues Abhinavagupta, all phenomena, which are perfectly real, are not to be discarded as mere illusions (contrary to what the represen- tatives of Advaitavedānta contend), yet they are nothing over and above the unique, all-encompassing and all-powerful universal consciousness that makes them manifest by assuming their forms. So far this analogy was ascribed to Abhinavagupta, or at least most scholars credited him 59. See Ratié (2017a: 167). 60. Cf. Torella (2002: xlvii), noting that among the commentaries of the Pratyabhijñā corpus, “the ĪPV […] seems to be felt as the standard one.” 61. See Torella (1988: 140) and Torella (2002: xliii). 62. On this doctrine see Lawrence (2005) and Ratié (2017b). 320 for an indian philology of margins with being the first to make a full metaphysical use of it. 63 Utpaladeva himself, however, mentions in his commentary on another Śaiva work (Somānanda’s ŚD) that he has explained at length in his detailed com- mentary (Ṭīkā = Vivr̥ti) on the Pratyabhijñā treatise how phenomena are similar to reflections, 64 and at the beginning of one of the recently edited Vivr̥ti fragments, he explicitly mentions the doctrine of reflections usually attributed to his commentator while summarizing the main theses estab- lished in the first part of his Vivr̥ti. 65 The edition and translation of three hitherto unknown Vivr̥ti chapters will undoubtedly bring to light many more similar points, and the ongoing edition and translation of Chapters 2.1 to 2.3 will certainly afford us an opportunity to thoroughly reevaluate the assumptions that still prevail in secondary literature regarding the history of the Pratyabhijñā system: studying the margins of manuscripts is not merely about attempting to save marginal works from oblivion, it may also help us understand how certain works came to be marginal- ized despite their highly innovative character, and how this process of marginalization has altered our understanding of their history. On the Purpose(s) of Annotations in Kashmirian Sanskrit Manu- scripts Little is known of the exact purpose of the annotations found in Kashmirian manuscripts. What is certain is that at least some of these manuscripts were explicitly copied for pedagogical purposes; thus the 15th -century master Śitikaṇṭha explains at the end of a NM annotated manuscript that he copied it so as to teach his pupils. 66 But of course this was not the sole purpose of manuscript copying; and what of the annotations in particular? If they too had a pedagogical purpose, was it 63. See for instance Rastogi (1984: 28) and Lawrence (2005). 64. Śivadr̥ṣṭivr̥tti: 14, 16 (see Ratié 2017b: 227, n. 64 and 232, n. 93); note that there is also one brief allusion to this doctrine in Utpaladeva’s short commentary on the Pratyabhijñā treatise (Vr̥tti: 60). For the testimonies of Abhinavagupta and Jayaratha regarding the fact that Utpaladeva himself had expounded this theory, see Ratié (2017b: 226-228, n. 62 and nn. 66-67). 65. See Ratié (2017a: 172). 66. See the colophon of manuscript P in Graheli (2015: 46): śiṣyān adhyāpayituṃ śiti- kaṇṭhasvāminā guruṇā gautamamatatattvavidā [tva]ritaṃ śrīnyāyamañjarī likhitā. “This Nyāya- mañjarī was swiftly copied by the master Śitikaṇṭha, expert in the truth of Gautama’s doctrine, in order to teach his pupils.” On this Śitikaṇṭha, see below.  isabelle ratié 321 the only one? And how exactly were they used to teach? In this respect Camillo Formigatti notices the following: In the case of esoteric doctrines like the non-dualistic Śaiva phi- losophy of Abhinavagupta, an oral transmission of the teachings was predominant, with a direct relationship between the guru and the disciples, in an unbroken chain of transmission of the secret knowledge. This is to a certain extent reflected in the character, content and visual organization of the annotations occurring in the manuscripts of the ĪPV… 67 C. Formigatti is thus inclined to think that the annotations might have been written by students taking notes during a class. This, however, does not seem to be entirely consistent with the fact that many annotations were obviously copied either from manuscripts of the quoted texts (since, as pointed out by Formigatti, they were probably too long to be committed to memory) 68 or from the margins of one or more antigraph(s). Formigatti concedes this point but suggests that there might have been a difference in this respect between the long scholia written in the margins (presumably copied from earlier manuscripts) and shorter glosses: Above all, it is clear that the long scholia have been copied from manuscripts of the works quoted. However, the short glosses may have a different origin. The esoteric character of the doctrines expounded in works like the ĪPK and the ĪPV presupposed a direct relationship between the teacher and the disciples. As pointed out by K. Pandey, Abhinavagupta used to explain its texts to his own disciples, who diligently wrote down his explanations on their own manuscripts. A similar situation may have been at the origin of some of the short annotations. 69 This hypothesis explicitly rests on the idea that the Pratyabhijñā cor- pus was meant to impart an “esoteric” knowledge and thus required an essentially oral form of teaching, as opposed to other texts (belonging to poetry and poetics) that, according to C. Formigatti, might have been studied by reading directly the texts (since there was nothing secret about them), which would explain why in the latter case, the annotations may take the form of a single running commentary instead of a series of quo- tations and sometimes have a clearly defined system of references to 67. Formigatti (2011: 84). 68. Ibid.: 85. 69. Formigatti (2011: 80). 322 for an indian philology of margins the main text. 70 In order to justify this interpretation of the distinction between the two kinds of annotations and the thesis that the study of the “esoteric” Pratyabhijñā corpus required an oral teaching, Formigatti in- vokes the authority of K.C. Pandey 71 to assert that Abhinavagupta himself was portrayed by one of is own disciples, Madhurāja, as teaching students who kept “writing whatever [the master Abhinavagupta] said” (uktam uktam likhadbhiḥ). 72 One might suspect, however, that this is a somewhat romantic view of the Pratyabhijñā tradition, first because the Pratyabhijñā corpus explic- itly presents itself as not being esoteric at all: rather, Utpaladeva insists that he is presenting an “easy” (sughaṭa) 73 soteriological path by trans- lating the metaphysical doctrines contained in the “secret scriptures” (rahasyāgama) into the philosophical terminology common to all the great religious movements of his time and accessible to all. 74 Besides, the con- tent of short glosses is usually far from being esoteric: as pointed out by C. Formigatti himself, 75 most of them are very basic explanations that simply offer a synonym for a particular word in order to clarify its mean- ing or provide the analysis of a compound — and this could certainly be taken as an important clue that these annotations indeed had a pedagog- ical purpose; but arguing on the mere grounds of the alleged esoteric character of these texts that their glosses had to have an oral source, or that they were written down by students rather than by their masters, might be going one step too far. 76 Finally, there seems to be no evidence 70. Formigatti (2011: 85): “Both the Kāvyādarśa and the Raghuvaṃśa were very popular texts, well-known and read also outside the Indian subcontinent. In the course of the centuries, they had reached the status of ‘classical works’ […]. Taking into account the content of the annotations and the auxiliary signs added in the main text, one should not be very far from reality by asserting that the annotations in this kind of manuscripts have been written as an aid for students who used the manuscripts for their study.” 71. See Pandey (1963: 20-22). 72. Gurunāthaparāmarśa (hereafter GNP) 4. The text was first edited in Raghavan (1980: 1-16). 73. ĪPK 4.16. 74. See Torella (2002: xxx) and Ratié (2017c). 75. Formigatti (2011: 84). 76. In the RT for instance (admittedly not an “esoteric” work — but probably no less so than the Pratyabhijñā treatise), according to Stein, short glosses of this kind (which “refer exclusively to points of grammar and construction”) were written down by Ratnakaṇṭha himself; Stein takes them to be “Ratnakaṇṭha’s own memoranda jotted down in the course of copying” (RT, Preface: xii, n. 3).  isabelle ratié 323 that the so-called “testimony” of the South Indian author Madhurāja regarding Abhinavagupta has any historical merit: as has been pointed out by Alexis Sanderson, it could very well be that Madhurāja only ever studied Abhinavagupta’s works, and the so-called “pen picture” 77 offered in the GNP might be a sheer product of Madhurāja’s imagination. 78 Nonetheless, C. Formigatti’s suggestion that marginal annotations constituted a “mnemonic support” for teachers during their classes is certainly very interesting and cannot be dismisssed out of hand, although — as acknowledged by Formigatti himself — neither can the hypothesis (apparently embraced for instance by Marc Aurel Stein) 79 that they were rather notes taken “by an erudite studying the text on his own.” 80 It is also possible that they were written down by the master but intended for students entrusted with the manuscripts, or that, as suspected by C. Formigatti, they were at times written by the students themselves 81 — and in fact, all of these suppositions need not be mutually exclusive. Thus the distinction between personal and pedagogical notes, just as that between single running commentary and series of glosses and/or quotations, may well have been less clear-cut than we tend to assume, all the more since in some cases at least, annotating manuscripts may have been a veritable team work. 82 Besides, according to Albrecht Wezler and 77. Pandey (1963: 20). 78. Sanderson (2007: 381, n. 486): “It has been claimed that Madhurāja was a pupil of Abhinavagupta (Pandey 1963: 20-22) […]. But not one of the verses […] states that he was anything more than a devotee of Abhinavagupta’s works and an adept in the Parākrama, the system of meditative worship taught in the Parātrīśikā.” 79. Stein (1900, vol. I: 47) expresses the opinion that Rāmakaṇṭḥa copied “the majority of his manuscripts chiefly for his own use.” 80. Formigatti (2011: 85). 81. Cf. the suggestion in Muroya (2010: 235-236) regarding an annotation making a reference to Śitikaṇṭha’s work in the margin of a NM manuscript (the annotation is quoted below, p. 330, n. 106): “it may even be possible to speculate that the […] gloss referring to the Bālabodhinīnyāsa was composed by one of Śitikaṇṭha’s disciples who was very involved in studying the NM and left his annotations and glosses on a manuscript of this work […] for his own future reference or for the purpose of stimulating others to expand their study of the NM.” 82. M.A. Stein thus hypothesizes that the annotator referred to as A2 in his RT edition, whom he later identified as Bhaṭṭa Haraka, a contemporary of Ratnakaṇṭha (Stein 1900, vol. I: 47), was working by the latter’s side (RT, Preface: xii): “While Ratnakaṇṭha was copying the text of O, A2 appears to have revised what the former had written, and to have added from the original the marginal notes and various readings which Ratnakaṇṭha omitted to copy.” 324 for an indian philology of margins Shujun Motegi, to whom we owe a remarkable edition of the Yuktidīpikā (hereafter YD), at some point the numerous marginal annotations found in YD manuscripts were extracted from the latter’s margins and gathered in an attempt to produce a single running commentary out of them. 83 Admittedly, the narrative 84 offered by the two editors as regards this attempt to transform marginalia into a single text is not without serious flaws, 85 and one can only hope that the history and exact nature of these 83. See the Introduction to Wezler & Motegi (1998), which explains that manuscripts K (written in śāradā script, preserved in Srinagar) and D (written in nāgarī, preserved in Delhi) are replete with marginal annotations, and that the content of these notes also appears in the form of a continuous text copied in a few independent śāradā-written folios preserved inside manuscript K (see Wezler & Motegi 1998: xiv-xv for a detailed description of K, and xvi-xvii for that of D). According to the editors (ibid.: xxiv, cf. 10, n. 2), there is no doubt that the apparently continuous text is nothing but an incomplete attempt to turn a collection of heterogeneous marginalia into a single commentary. 84. According to Wezler and Motegi, K must have been moved from Srinagar to the NAI in Delhi in 1948, and D is a very late copy that must have been made in Delhi during K’s stay there (hence the nāgarī script, common in Delhi), after which K was returned to Srinagar (where it is currently preserved) and then only, in Srinagar, did someone attempt to turn the marginal annotations found in D into what looked like a single running commentary (the one now found in K). See Wezler & Motegi (1998: xvi): D “is evidently a very modern transcript of K, made only after Independence, when, most probably for reasons of safety in the wake of the Kashmir imbroglio, a number of manuscripts were brought down from Srinagar to the National Archives in Delhi to be returned only later to their rightful owner(s).” Cf. Ibid.: xxiv: “Significantly, this attempt seems to have been undertaken by the Delhi copyist, i.e. the scribe who made D by copying K, and that [sic] what he did was transcribed into śāradā most probably when K was returned to Srinagar.” No explanation is given as to why the editors believe that the attempt was made by the scribe who made D, whereas as far as I understand, the latter does not contain any continuous version of the marginalia; I assume that they have in mind the fact that the marginal annotations in D are numbered, contrary to those in K, but maybe they meant that the scribe who copied D also produced an independent manuscript putting together all the annotations that disappeared after being transcribed into śāradā? 85. Bronkhorst (2003) criticizes certain aspects of the YD edition’s stemma codicum but does not focus on the history of the annotated manuscripts as it is presented in the Introduction to the edition. It should be noted, however, that manuscript K certainly never travelled to Delhi, because the manuscripts that were taken from Srinagar and deposited at the NAI in 1948 were never returned to their owners and are still in Delhi to date: the entire collection of the “Manuscripts belonging to the Archaeology and Research Department, Jammu and Kashmir Government, Srinagar” (which, according to the “List of Gilgit Manuscripts and Sanskrit Manuscripts” of the NAI, consists of 212 items that “were temporarily transferred to the National Archives of India for safe custody and preservation on Oct. 19, 1948”) was still available at the NAI when I last visited them in 2012. Besides, manuscript D could hardly have been copied in Delhi: the transcript  isabelle ratié 325 very substantial notes will soon be examined afresh: are they really, as the editors of the YD assumed from the start, independent short marginal glosses that were artificially gathered at a very late stage? Or might some of these annotations at least belong to a lost single commentary, the fragments of which have been overlooked so far? Only a new examination of the manuscripts and the marginalia’s relationship with the text on which they comment (a study far beyond the scope of this article) could help answer this question. In any case it is quite possible that some of the texts that we regard today as running commentaries were at first annotations that were later gathered to form a single text — thus Nagin Shah, the editor of Cakradhara’s commentary on the NM, states about the latter: This commentary is important because it is the only extant com- mentary on the famous Nyāyamañjarī of Jayanta Bhaṭṭa […]. It is not a commentary in the usual sense of the term explaining each and every term. It is not even a running commentary. It is of the nature of an annotation. 86 To sum up, in Kashmir the annotations may have had several uses, pedagogical and personal, and their content and status were sometimes al- tered to suit new needs: this plasticity is obvious in the attempt described above to transform Abhinavagupta’s commentary on Utpaladeva’s Vivr̥ti into a commentary on another text by Abhinavagupta; if Wezler and Motegi are right, in the case of the YD at least, independent scholia may also have been transformed into a single commentary; and it is tempt- ing to suspect with Shah that Cakradhara’s commentary on the NM was originally little more than a set of marginal notes. was probably executed either in Jammu or in Srinagar prior to 1948, since it belongs to the collection of Kashmirian manuscripts that were deposited at the NAI at that time. Finally, one fails to see how someone in Srinagar may have used D to produce an artificial running commentary if D was made and remained in Delhi (unless the editors of the YD assumed that the scribe who copied D had also produced an independent version of the gathered marginalia in nāgarī that was somehow brought to Srinagar, contrary to D, and that disappeared after being transcribed into śāradā; but even this far-fetched hypothesis seems hardly tenable given that one does not see how K might have travelled to Delhi and back to Srinagar). 86. Shah (1972: 4). 326 for an indian philology of margins A Learned Tradition: on Annotated Manuscripts Copied by Fa- mous Kashmirian Authors Despite these many uncertainties (which make further studies on Kashmirian marginal annotations all the more desirable), one thing at least is beyond doubt: in Kashmir, most śāradā manuscripts (and their an- notations) were not produced by professional scribes with no knowledge of their content, but by learned men. 87 And not only did these pandits understand what they copied: they had a keen interest in studying and teaching this content, and they were often authors themselves. This is obvious from the transmission of Abhinavagupta’s ĪPV. Thus S3, an annotated ĪPV manuscript (see figs. 3a, 3b and 3c), is not dated but must have been copied in the 17th century; 88 in the colophon, the scribe identifies himself as Rājānaka Śaṅkarakaṇṭha, and this character was in all likelihood the author of a Bhairavastotra 89 and father of Rājānaka Ratna- kaṇṭha, 90 a renowned Śaiva author who presents himself in his works as the son of Rājānaka Śaṅkarakaṇṭha. As for Ratnakaṇṭha, who must have 87. This was already noted in Bühler (1877: 32) (“a great many of them have been written by Pandits, not by professional writers”). See also e.g. Slaje (1993: 20) and Witzel (1994: 13). 88. The scribe’s name is that of the father of a well-known 17th century Kashmir- ian author (see below), and the recent, hand-written Sanskrit note in nāgarī script attached to the beginning of the manuscript specifies that the person who copied it, namely Śaṅkarakaṇṭha, “lived in Kashmir during the reign of Shah Jahan” (śāhajahāna- rājye kāśmīreṣu jīvitavān āsīt), and adds that “there is no other manuscript older than this manuscript” (prater asyā [sic] prācīnā nānyā kācana pratir vidyate), by which it is probably meant that the library contains no older manuscript of the ĪPV. 89. See Sanderson (2007: 397, n. 555). 90. I thank Alexis Sanderson for first drawing my attention to the fact that a Rājānaka Śaṅkarakaṇṭha known to him “as the author of a Bhairavastotra transmitted in some manuscripts of the Kalādīkṣāpaddhati was the father of the well-known scholar Rājānaka Ratnakaṇṭha, who wrote his commentary on the Stutikusumāñjali of Jagaddhara accord- ing to its initial and final verses (which also tell us that Śaṅkarakaṇṭha was his father) from Śaka 1602 [AD 1680] to Vikrama 1738 [AD 1681] during the reign of Aurangzeb (avaraṅgamahīpāle kr̥tsnāṃ śāsati medinīm). So the date proposed on the slip may well be right.” (Personal written communication, 2008.) The “slip” referred to here is the hand-written note (see above, n. 88) according to which Śaṅkarakaṇṭha lived during Shah Jahan’s reign.  isabelle ratié 327 been active around 1680, 91 he has authored stotras 92 as well as astronom- ical treatises and commentaries on poetic and aesthetic works, 93 and he was famous not only for composing works of his own, but also for his impressive abilities as a manuscript copyist; one of the most remarkable manuscripts in his abundant production is the richly annotated “codex archetypus” of Kalhaṇa’s RT. Marc Aurel Stein remarks in this connection that as he is writing his translation of the RT, the memory of Ratnakaṇṭha “still lives in Kaśmīr Paṇḍit tradition as that of a great scholar and very fast writer” capable of copying manuscripts at a particularly impressive rate; Stein further mentions “the considerable number of manuscripts written by him which are still extant in Kaśmīrian libraries both in and outside the valley,” as well as Ratnakaṇṭha’s very distinctive handwrit- ing and the difficulty of deciphering it. 94 Now, this Śaiva figure — who happened to be the son of the man who copied the oldest known anno- tated ĪPV manuscript — was also in all probability the Ratnakaṇṭha whom Bhāskarakaṇṭha, the author of the only hitherto published commentary (Ṭīkā) on Abhinavagupta’s ĪPV, 95 mentions as his master and describes as an “expert in all fields of scholarship” (samastavidyānipuṇa). 96 The history of the transmission of annotated ĪPV manuscripts is thus inextricably linked with that of the study and exegesis of the Pratyabhijñā corpus in medieval Kashmir — and there are other such examples. Thus according to the colophon at the end of S17 — a śāradā manuscript of Utpaladeva’s Īśvarapratyabhijñāvr̥tti now preserved in Srinagar —, 97 the 91. On Ratnakaṇṭha, his dates and works, and the name of his father, see Hanneder, Jager & Sanderson (2013: 12-20). Note that in the case of the RT’s “codex archetypus” as well as in that of the oldest known annotated ĪPV manuscript, it is the name of the scribe that enables us to provide a date for the manuscript (see Stein 1900, vol. I: 46 for the RT manuscript). 92. See Hanneder, Jager & Sanderson (2013). 93. These include commentaries on Ratnākara’s Haravijaya, Mammaṭa’s Kāvyaprakāśa, Vāsudeva’s Yudhiṣṭhiravijaya and Jagaddhara’s Stutikusumāñjali. 94. See Stein (1900, vol. I: 47); cf. Witzel (1994: 13). For facsimiles showing samples of Ratnakaṇṭha’s handwriting, see RT: first (unnumbered) page and Hanneder, Jager & Sanderson (2013: 44). 95. It is best known as the Bhāskarī. 96. Cittānubodhaśāstra 298: 359. That Ratnakaṇṭha was Bhāskarakaṇṭha’s master is pointed out in Sanderson (2007: 422) and Hanneder, Jager & Sanderson (2013: 15). 97. For a detailed description, see Torella (2002: lii-liii). S17 appears to have been transcribed in J11 before the beginning of the ĪPVV. 328 for an indian philology of margins manuscript was made by a certain Sundara Bhaṭṭāraka. 98 Interestingly, this figure is very probably the Bhaṭṭāraka Sundara who authored a hith- erto unpublished commentary (Kaumudī) on the ĪPK, which, as far as I know, is preserved in a single (annotated) manuscript, S18. 99 According to its colophon, Bhaṭṭāraka Sundara was the son of a man named Śaṅkara — or rather, Śrīśaṅkarācāryavarya — who was, just as himself, a worshipper of Tripurasundarī (Śrīvidyā) — and again, the study of the Pratyabhijñā tradition and the transmission of its manuscripts appears to have been a family tradition, since one of the annotated ĪPV manuscripts preserved in Srinagar was obviously copied by Bhaṭṭāraka Sundara’s father. 100 The colophon of Bhaṭṭāraka Sundara’s Kaumudī explicitly states that the work was “composed” (racita) by this author, and one could surmise that the manuscript was nonetheless copied in its entirety (pūrṇīkr̥ta) by someone else, were it not for the fact that manuscripts S17 and S18 were evidently handwritten by the same person (and, as far as I can judge from the scans at my disposal, on the same paper), so that S18 is likely to be an autograph. If it is the case, given that the colophon of the Pratyabhijñākaumudī was 98. According to Torella (2002: liii), the colophon of S17 reads śrīmatsundarabhaṭṭārakeṇa pratyabhijñāvr̥ttiḥ sampūrṇīkr̥teti śubham astu devādīnāṃ mānuṣyakītānāṃ. In fact both S17 and J11 read mānuṣyakīṭakānām (although one would rather expect mānuṣyakīṭāntānām or mānuṣyakīṭakāntānām). Torella (2002: xliv, n. 76) understands this as meaning that Sundara Bhaṭṭāraka “would have composed a commentary to take the place of the final part of the Vr̥tti (missing in all the Kashmirian manuscripts).” It seems to me, however, that this is not what is meant here, since in fact Sundara Bhaṭṭāraka’s commentary as preserved in S18 (to which R. Torella had not had access when he wrote his book) does not just fill the final gap: it comments on the ĪPK from the beginning of the treatise (see ĪPK 1.1.1 on folio 1b, ĪPK 1.1.2 on folio 11a-b, ĪPK 1.1.3 on folio 15a, etc.). Besides, the sentence in question in S17 occurs in the colophon of a manuscript that does not contain Sundara Bhaṭṭāraka’s commentary, even in part: the Vr̥tti in S17 breaks at the usual place and, as noticed in Torella (2002: liii), “the remaining kārikās [are] without commentary.” It therefore seems much more likely that the colophon in S17 simply states that “this Praty- abhijñāvr̥tti [manuscript] was completed by the master Sundara Bhaṭṭāraka — prosperity to [all] from gods to men and worms.” 99. I thank Alexis Sanderson for drawing my attention to this manuscript. 100. In the colophon of S18 (on which see below, p. 329, n. 101), Bhaṭṭāraka Sundara describes himself as a sākṣācchaṅkarāvatāraśrīmacchrīvidyāpārijātaśrīmacchrīśaṅkarācārya- varyaputramaṇi, “the gem that is the son of the venerable master Śrīśaṅkarācāryavarya, who was a wish-granting tree? of the venerable Śrīvidyā (śrīmacchrīvidyāpārijāta) and an avatar of Śaṅkara in person.” At the end of the ĪPV in S5 (in fact a composite manuscript made of several works), on folio 203a, the end of the colophon states that the work was copied by a Śrīśaṅkarācāryavarya (śrīśaṅkarācāryavaryavicitritā vijñeyeśvarapratyabhijñety om) who describes himself as °śrīmacchrīvidyāpārijātanūtanasvāmī.  isabelle ratié 329 completed in the year 74 (most probably of the laukika era) on the 7th lunar day of the dark fortnight, 101 it could have been composed at the end of the 17th or 18th century. 102 That those who copied śāradā manuscripts in medieval Kashmir were also authors in their own right was by no means a peculiarity of the Pratyabhijñā tradition. To give just one additional example that does not belong to the nondualistic Śaiva milieu, a heavily annotated NM birch-bark 101. The long colophon spreads on folios 80a and b (arab numbers). Its beginning runs as follows from what I can read: oṃ varṣe vedatapodhane’tha? nabhasi pakṣe ca śukletare saptamyāṃ khalu puṣṭisaubhagayute chattre ca lagne sthire. śrīvidyāpadapadmasevanalasad- buddhyātitīkṣṇāgrayā | śrībhaṭṭārakasundareṇa racitā pūrṇīkr̥tā kaumudī || 1 || The śāradā numerals 7 and 4 are given above the chronogram “veda[4]-ascetics[7]” (vedatapodhana), which reads backwards according to the rule aṅkānāṃ vāmato gatiḥ (on which see e.g. Sircar 1965: 230). The colophon goes on with five more verses (also numbered) and some prose, and its very end says again: saṃ 74 śrāvaṇe kr̥ṣṇapakṣe saptamyāṃ revatyāṃ saubhāgyaviṣkambhe chattrayoge ca vr̥ścike lagne ca samāptiṃ gatā pratyabhijñākaumudī. Note that the chronogram is the result of a correction; I cannot read the original words crossed out and replaced by tapodhane, and I suspect that the latter word might be by a second hand, but I have no certainty in this regard. The end of the colophon might also have been written (or altered) in part by a second hand (possibly the one who added some simplistic glosses, such as kr̥ṣṇe for śukletare, or who wrote again in the interlinear space the words samāptiṃ gatā pratyabhijñākaumudī as if they were unclear to this reader). 102. A possible match would be the 2nd of September 1798 according to Michio Yano’s pañcānga program (http://www.cc.kyoto-su.ac.jp/~yanom/pancanga/ index.html), which I used after adding 46 to the laukika year to obtain the last dig- its of the corresponding śaka year (= [17]20). But it could also be the 28th (the 18th in the Julian calendar) of August 1698 (same source). The first hypothesis would carry more weight if Bhaṭṭāraka Sundara’s father could be identified with the Svāmiśrīmacchaṅkara who copied S2 (an ĪPV manuscript) in the year 49 (of the laukika era), in the bright fortnight of the month of Mārga[śīrṣa] (svāmiśrīmacchaṅkaraiḥ likhyate’sau saddr̥ḍhārthā pratyabhijñā kr̥pāloḥ saṃ 49 mārga śu ti pralikhanaprārambhaḥ); the end of the manuscript gives again the name of the copyist, Svāmiśrīmacchaṅkara, and the date: vikramāditya saṃ 1830, which corresponds to 1773/1774 A.D. (and the laukika year [48]49). Unfortunately, manuscripts S2 and S5 (the latter was most probably written by Bhaṭṭāraka Sundara’s father: see above p. 328, n. 100) were obviously written by different hands. Besides, according to Kannupillai (1922: 399), the nakṣatra mentioned in S18, namely revatī, does not seem to correspond to the dates mentioned above (or to any other plausible year 74 of the laukika era or even of the śaka and vikrama eras). Unfortunately I am way too ignorant in astro- nomical matters to explain this (or to check the other astronomical data given in the colophon); I also have doubts as to possible alterations by a second hand in the colophon of S18 (see above, n. 101). Hopefully one of the learned readers of this book can come up with a more satisfactory solution than the suggestions above as regards the date of the Pratyabhijñākaumudī. 330 for an indian philology of margins manuscript dated 1472 CE 103 was copied by a master Śitikaṇṭha (śitikaṇṭha- svāmilikhita), son of Arjunasvāmin, and as noticed by Alessandro Graheli, a certain Śitikaṇṭha also copied a manuscript of Vāsudeva’s Nyāyasārapada- pañcikā (a commentary on the Nyāyasāra, another Naiyāyika Kashmirian work) now preserved in Vienna. 104 Now, as shown by Y. Muroya, 105 this Śitikaṇṭha, whose authority is invoked in a marginal annotation to the NM, 106 and who presents himself as an expert in Nyāya, 107 also hap- pened to be the author of a commentary (Nyāsa) on a grammatical work, Jagaddhara’s Kātantrabālabodhinī, and he might also have composed a Nyāyasiddhāntamañjarīdīpikā. 108 The fact that most of these manuscripts were copied by scholars has obvious advantages for modern editors: they are often copied with much care, 109 and, of course, enriched with erudite annotations. On the other hand, these learned scribes, who often worked from several manuscripts and even commentaries 110 (which renders the application of the stem- matic method particularly problematic in many cases), 111 usually felt entitled to edit the text 112 whenever they deemed it necessary (and did so with mixed fortunes); 113 their marginal glosses, however accurate, sometimes led to unwarranted corrections in the text; 114 and their hand- writing, devoid of the calligraphic qualities of professional scribes, was 103. For a thorough description see Graheli (2015: 45-49), under siglum P. 104. See Slaje (1990: 87). 105. Muroya (2010: 234-236). 106. Muroya (2010: 234): ity asya prapañcas tu śitikaṇṭḥācāryaviracite bālabodhinīnyāse draṣṭavyaḥ (“The details of this [discussion], however, should be looked up in the Bāla- bodhinīnyāsa composed by the master Śitikaṇṭha”). 107. See above, p. 320, n. 66. 108. On this hypothesis, see Graheli (2015: 48, n. 14). Although it has also been surmised that the Śitikaṇṭha who authored the Bālabodhinīnyāsa and the one who composed the nondualistic Mahānayaprakāśa were one and the same person (Grierson 1929: 73-74, Witzel 1994: 27), as pointed out in Sanderson (2007: 300-301), this is highly improbable. 109. See e.g. Stein (1900, vol. I: viii): “the writer of this codex was not a mere copyist, but a scholar of considerable attainments who, we have every reason to believe, has copied his original manuscript with great care and accuracy.” 110. See Torella (2002: xlii). 111. Hanneder (2017: 62 ff). (See also Hanneder 1998: 46-48.) 112. See e.g. Jager (2010: 293). 113. On “learned mistakes” in Kashmirian manuscripts, see e.g. Witzel (1994: 3-4), Torella (2002: xlvi). 114. See the example in Hanneder (2017: 63-64) about the Maṅkhakośa.  isabelle ratié 331 often misread by later copyists. 115 The Nāgarī “Revolution” and the Fate of Traditional Annota- tions At some point around the middle of the 19th century, 116 however, this tradition underwent a spectacular change: more and more Kashmirian Sanskrit manuscripts were copied in a variety of nāgarī script rather than śāradā. 117 According to Georg Bühler, this shift from one script to another had a social component that was to have a crucial impact on textual trans- mission: his famous 1877 report explains that “all nāgarī manuscripts are written by professional scribes” who have no interest in understanding the text and by whom these documents are “mostly prepared for the market.” 118 As a result, the advantages and inconvenients pertaining to the old way of transmitting texts were inverted: many nāgarī manuscripts pro- duced in Kashmir and the neighboring Jammu from older śāradā exemplars are admirable works of calligraphy — and this aesthetic concern extends to the layout of marginalia, as can be seen e.g. from the often fanciful arrangement of marginal annotations in J11 (see fig. 7) — yet they tend 115. See Stein (1900, vol. I: 47, n. 6) about the famous case of Ratnakaṇṭha’s RT manuscript: “His handwriting as it appears there, with its very cursive and peculiar characters, presents unusual difficulties even to the practised reader of śāradā writing. To these difficulties must be ascribed, at least partly, the numerous clerical corrruptions and blunders which we meet, to a greater or smaller extent, in all modern transcripts of the codex archetypus.” 116. According to Bühler (1877: 33), “The devanāgarī manuscripts written in Kaśmīr are all very modern. I was told that these characters had come into more general use during the last thirty years only, since the annexation of Kaśmīr to the Jamū dominions.” It is likely, however, that the process started somewhat earlier (see below, p. 333, n. 127), even though it did not become “general” before the time indicated by Bühler. 117. Although the vast majority of “transcripts” from that time onwards were made from śāradā to nāgarī, copies were sometimes made from the latter to the former. See e.g. Slaje (2005: 46) for the curious case of a śāradā manuscript copied in 1934/35-38/39 AD, which seems to be “a śāradā transcript” of a Yogavāsiṣṭha edition printed in nāgarī, and which, according to W. Slaje, gives “the impression of a ‘collector’s edition.’” On possible motives for the practice (attested beyond Kashmir in many areas of Indic culture) of producing manuscripts from printed editions, see Francis (2017: 341-348). For an example of transcription from nāgarī to śāradā largely predating the phenomenon of massive transliteration described here, see Witzel (1985). 118. Bühler (1877: 33). 332 for an indian philology of margins to abound in crude mistakes. 119 Many of the latter seem to stem from a relative lack of familiarity with the śāradā script, 120 so that, as suggested by Bühler, in such cases “the best plan for restoring corrupt passages is to try to find the śāradā ligature which most closely resembles the corrupt devanāgarī group.” 121 The marginalia copied along with the main text in nāgarī transcripts are no exception to this rule: the scribe of J11 for 119. Bühler (1877: 33) already noted that since these nāgarī manuscripts are made by professional scribes, they “are, for this reason, even if they have been afterwards corrected by Pandits, less trustworthy than even śāradā paper manuscripts.” See also Stein (1894: ix, n. 2): “Many of the Jammu manuscripts, written by these Kashmirian copyists, are excellent specimens of modern Indian calligraphy and from this point of view truly fit for a Royal Library. Unfortunately, they hide only too frequently under this attractive exterior a very defective text, the result of careless revision.” Cf. Stein (1900, vol. I: 45), on “all the defects inherent to devanāgarī transcripts made in Kaśmīr from śāradā Manuscripts.” See also e.g. Graheli (2015: 45) (about a nāgarī manuscript that “betrays a close connection to the śāradā script, and thus to Kashmir”): “The calligraphic and consistent ductus suggests that the copyist was a professional scribe, rather than a scholarly one, as indicated also by the presence of copious Sanskrit errors.” Cf. Hanneder (2017: 52): “In Kashmirian literature we see that devanāgarī versions were often not prepared by Pandits but by non-expert scribes and that they abound in all kinds of mistakes. Here to edit a text without the help of śāradā manuscripts is not advisable.” 120. See e.g. Stein (1900, vol. I: 53), about the scribe of a nāgarī transcript who was “an exceptionally careless and ignorant copyist, incapable even of always reading correctly the śāradā writing of his original;” Stein also mentions “numberless faults of transcription in L, and the occasional occurrence in it of śāradā characters which the scribe has slavishly reproduced instead of transcribing them.” Note that although most modern scholars studying nāgarī manuscripts copied from śāradā examplars thus point out how careless the transcription is, some seem to think that the scribes were nonetheless familiar with śāradā, and that the not infrequent cases (such as the one mentioned by Stein) where śāradā conjuncts are reproduced instead of being transcribed betray this familiarity rather than puzzlement over a character that could not be read. See e.g. Vasudeva (2004: xxii), about a nāgarī transcript that “shows abundant errors of metathesis, transposition, euphonic combination etc., many deriving from a careless transcription of the śāradā original”: “occasionally the scribe has reverted to the presumably more familiar śāradā kta-ligature and medieval e-mātrā.” Cf. Graheli (2015: 38), about a nāgarī manuscript that bears some “unequivocal indications of a close derivation from a śāradā antigraph,” among which the fact that “some śāradā akṣaras were inadvertently and intermittently written by the scribe”; “the copyist was most likely a professional scribe, familiar with both nāgarī and śāradā.” See also Witzel (1994: 14), according to which “copying from śāradā into nāgarī… automatically increased the chance for writing mistakes, even if the scribe knew both scripts perfectly well. Unconsciously, though, certain characters looking alike in both scripts but expressing different sounds frequently are confused. In addition a number of ligatures of śāradā are badly copied or even taken over from the śāradā original into the nāgarī manuscript.” 121. Bühler (1877: 33). For a list of typical mistakes, see Witzel (1994: 42).  isabelle ratié 333 instance (see figs. 4 and 7) seems to have had much trouble distinguishing semi-homograph śāradā conjuncts. 122 According to Bühler, nāgarī manuscripts, because they were meant to be sold, are “also not unfrequently ‘cooked,’ i.e. the lacunae and defects in the original are filled in according to the fancy of the Pandit who corrects them.” 123 As already pointed out by Michael Witzel, however, this assess- ment deserves to be nuanced, since many nāgarī manuscripts faithfully indicate all lacunae or illegible passages by using dots 124 (see e.g. the marginal annotations in figs. 4 and 6b) and, as noted by Stein, sometimes they simply reproduce śāradā conjuncts that could not be identified, 125 whereas there are known instances of “cooked” śāradā manuscripts. 126 The precise circumstances of this shift from śāradā to nāgarī remain to be determined; 127 in any case the tradition of annotating manuscripts, 122. Besides very frequent confusions between e.g. cc and śc, see e.g. fig. 7: in the bottom margin, after °dena sa eva kālakramo bhavati (beginning of the first line), this fragment of Utpaladeva’s Vivr̥ti on ĪPK 2.1.5 reads bhāvavikāraṣaṃkenānena instead of the correct bhāvavikāraṣaṭkenānena, because the scribe has mistaken the semi-homograph śāradā conjunct ṭk for ṅk (= ṃk). On semi-homograph akṣaras in śāradā, see Slaje (1993: 43-45). 123. Bühler (1877: 33). The author reports his personal experience in this respect (ibid.): “In no part of India have I […] been told of the practice of restoring or ‘cooking’ Sanskrit books with so much simplicity as in Kaśmīr. I was asked by my friends if the new copies to be made for me were to be made complete or not; and one Pandit confessed to me with contrition, after I had convinced him of the badness of the system, that formerly he himself had restored a large portion of the Viṣṇudharmottara.” 124. Witzel (1994: 15): “This feature, mentioned so prominently by Bühler, is perhaps not as common as he presents it. There probably existed a small ‘cottage industry’ in his time as there already had been a demand by the Mahārāja for some 20 years who got many Kashmirian manuscripts copied for his Raghunāth Temple Library at Jammu. Bühler’s visit and his acquisitions for the Bombay government of course, further intensified this process. Cooking does not seem to be the case in the Vedic manuscripts, as hardly anybody (except Bühler) had an interest to buy them. Even the copies of the Paippalāda Saṃhitā that were made for the Mahārāja clearly show the lacunae of the original, as does the Paippalāda Saṃhitā manuscript itself.” For other such examples, see Graheli (2015: 37 and 44). 125. See above, p. 332, n. 120. 126. Witzel (1994: 15) cites the example of a colophon forgery; note, however, that many śāradā manuscripts also indicate lacunae with dots (or more commonly short dashes) — see the remark by Witzel quoted above, n. 124, and Slaje (1993: 21); cf. for instance Graheli (2015: 53), and the marginal annotations in figs. 5 and 6a. 127. Bühler links it with the annexation of Kashmir by the neighbouring kingdom of Jammu (see above, p. 331, n. 116) and insists that even in his time the pandits of Kashmir seemed to have very little acquaintance with the nāgarī script (Bühler 1877: 33). It should be noted, however, that the Kashmirian professional scribes were famed for their 334 for an indian philology of margins which had been so lively in Kashmir for centuries, seems to have as it were frozen around this time: the marginal annotations found in śāradā anti- graphs were copied along with the main text in most nāgarī transcripts, but probably because the scribes of these transcripts were professional scribes rather than pandits, they usually refrained from adding new an- notations of their own, and although the nāgarī manuscripts were often corrected by pandits, very few seem to have added lengthy annotations to these transcripts. 128 Besides, in most śāradā manuscripts very little effort is made to link the annotations with the main text, the readers being expected to understand their relationship from the content of the marginalia, 129 and as a result, in Kashmirian nāgarī manuscripts, annota- calligraphic nāgarī, and this tradition, which must have predated the Jammu annexation (see e.g. Witzel 1994: 19 and Hanneder 1998: 44 for late 18th -century examples), was not brought to Kashmir proper from Jammu, since the scribes employed in Jammu were rather brought from the valley of Kashmir. See Stein (1894: viii-ix): “The scribes employed at Jammu in copying manuscripts for the Temple Library were almost exclusively drawn from Kashmir, which until quite recently could boast of a plentiful supply of ‘lekhārīs’ writing a devanāgarī hand of peculiar elegance.” See also Cox (2010) for the example of a 12th -century scribe presenting himself as a kāśmīrabhaṭṭa and drawing up the so- called Nīlgunda plates in nāgarī (with a śāradā invocation) in the South Indian kingdom of the Western Cālukyas. In any case the Jammu Raghunath Mandir Library’s eagerness to constitute a large collection of Sanskrit works accessible in nāgarī not only by having its own staff copy śāradā exemplars but also by purchasing transcripts made in the valley (Stein 1894: iv ff.) must have been a very important factor (although by no means the only one) in the massive production of nāgarī manuscripts in the second half of the 19th century. Note also that Kashmirian Sanskrit manuscripts certainly aroused some interest beyond the Jammu Library (and Western scholars such as Bühler and later Stein); thus as mentioned in Tantrasāra (Preface: i), some of the nāgarī transcripts used for the KSTS editions came from “the collection of books bearing on the Śaiva philosophy in general which one Rīvatī Raman of Southern India got copied during his visit to Kashmir for acquisition of rare manuscripts. On his death here in Kashmir these books fell into the hands of his servants,” and Mukund Rām Śāstrī bought an important part of the collection from one of these servants (ibid.; see also PTV, Preface: 1-2). 128. For an example of nāgarī transcript with both corrections and alternative readings offered in the margins, see Hanneder (1998: 43). 129. Although mere corrections are often indicated with a caret (kākapāda, see e.g. fig. 1, after °marthya° at the beginning of line 7, and before rūpā in the interlinear space), in the śāradā manuscripts of my knowledge, glosses and marginal annotations are seldom linked with a sign to a specific passage in the main text. Cf. Formigatti (2011: 84), about ĪPV manuscripts: “there is no apparent overall strategy to link unambiguously the annotations to the text commented. Almost every form of reference sign is absent, numbers referring to the line in the main text are lacking and with very few exceptions, no pratīka is used. Only the intuitive strategy of putting the short interlinear glosses directly above the  isabelle ratié 335 tions are often misplaced. 130 Thus as can be seen in fig. 8, S7 — a nāgarī annotated ĪPV manuscript — gives a Vivr̥ti fragment explaining ĪPK 1.3.5 on the bottom margin of the lower page, 131 but in fact the main text copied on that page explains ĪPK 1.4.1, 132 whereas the verse explained by the Vivr̥ti fragment quoted here actually appears in the main text four pages before this point! 133 Conclusion: on Margins and the Indologists’ “Embarrassment of Riches” In his recent article on textual criticism in Indology and European philology, M. Witzel has emphasized how philologists working on Indian texts are faced with a situation strikingly different from that of Classical studies in terms of manuscript numbers: The Indian situation is thus radically different from the one in Classical or Biblical studies, where only a limited number of pre- Renaissance manuscripts have come down to us. In South Asia, an estimated 30 million manuscripts exist in public and in frequently neglected private collections. Each village will have a Brahmin’s family collection, of varying sizes, and there are some 600,000 vil- lages… This situation creates not only problems of access but also an embarrassment of riches that no single scholar can access, pro- cess, and use even for one particular edition. 134 Upon reading this description, those who have been working in the commented work or the longer marginal scholia near the passage to be explained is employed in all manuscripts. For this reason, it is often difficult to determine to which word or passage in the main text a marginal annotation refers, and only its content may help find it out.” Note, however, that in the two ĪPVV manuscripts bearing Vivr̥ti fragments of which I am aware to date, an effort has been made to rubricate the words of the mūla-text quoted by Abhinavagupta in his commentary and the corresponding words in the Vivr̥ti fragment quoted in the margin (see figs. 4 and 5). 130. See e.g. Muroya (2010: 229), about misplaced annotations in a nāgarī transcript. 131. For an edition and translation of this fragment (beginning with asato vā), see Ratié (2016b). 132. ĪPK 1.4.2 also appears on that page at the very bottom of the main text (all verses are rubricated). 133. Admittedly, as pointed out in Formigatti (2011: 84), annotations may also be mis- placed in śāradā manuscripts; it is possible (although unlikely) that in the case of S7, the śāradā antigraph already suffered from this systematic shift between the notes and the main text. 134. Witzel (2014: 42). 336 for an indian philology of margins field of classical studies may picture up India as the promised land of all philologists. Alas, it should be kept in mind that the embarrassment of riches conjured up by M. Witzel only concerns a very limited number of works, 135 the success of which happened to be sufficient to ensure their transmission in impressive numbers of surviving manuscripts; but the main concern of indologists is not, in truth, to avoid drowning in an ocean of manuscripts. Thus in the field of Indian philosophy (an example among many others), a great deal of major Sanskrit treatises are lost, and their existence is only known from a few testimonies or scant fragments in later works, or from Tibetan and/or Chinese translations; besides, among the Indian philosophical works that are known in their original language, many had to be edited on the basis of an often incomplete codex uni- cus. That countless South Asian works have thus virtually vanished is surely much worse for our understanding of the intellectual, artistic and religious history of India than the plethora of manuscripts with which students of e.g. the Bhagavadgītā may have to deal, and one of our only chances to rescue a few of these works from oblivion lies in the study of manuscript margins, where, as shown above, quotations that sometimes extend to several chapters can be found. Studying these marginalia may also enable us to understand the pro- cess of marginalization through which some works came to be neglected, despite their inventivity and the intense exegetic or critical reaction that they may have initially triggered: understanding the shifts of interest that have sometimes occurred in a given tradition from one part of a corpus to another (for instance, as pointed out above, from Utpaladeva’s Īśvarapratyabhijñā-Vivr̥ti to Abhinavagupta’s ĪPV) is crucial, if only because our perspective on the historical importance of authors and works is to a great extent determined by such shifts, and the study of margins can help us correct many a distorted point of view in this respect. Because in Kashmir, for centuries manuscripts were copied and anno- tated by pandits who studied and taught their content, and even wrote entire commentaries on them, marginalia also constitute a virtually un- explored 136 source for the history of learning and teaching habits in 135. Witzel (2014: 44), under the heading “Abundance of Indian manuscripts,” mentions as examples “the works of Śaṅkara, or the Kāśikā.” 136. That is, with the few exceptions mentioned above. Studies focusing on the different hands in which the marginalia of a given South Asian manuscript were written and their chronology (as in Stein’s edition of the RT or the still unpublished Formigatti 2011) are  isabelle ratié 337 medieval India, as well as for the understanding of the exegetical his- tory of a given current of thought: as shown above, the history of the commentaries on the Pratyabhijñā treatise cannot be dissociated from that of the transmission of the annotated manuscripts pertaining to this tradition. And marginalia are by no means confined to the Pratyabhijñā corpus: in Kashmirian manuscripts, they are found along with many kinds of Sanskrit texts, which include philosophical treatises 137 but extend to many other genres, for example those of poetry and poetics. 138 In short, studying marginal annotations is not a luxury to be afforded only when we have eventually dealt with the main texts transmitted in the millions of manuscripts mentioned by M. Witzel: marginalia are often no less important than these texts, and given that margins are usually the first to incur damage when manuscripts decay, 139 their examination should not be postponed for too long — one can only hope that the margins of South Asian manuscripts can soon be given the scholarly attention they deserve. scarce to say the least. A reason for this might be that, as pointed out in Slaje (1993: 17), we still lack a detailed study of the evolution of the śāradā script from the 14th century onwards that could help us determine more easily when a specific set of annotations was added. 137. Besides the works mentioned above — particularly Muroya (2010), devoted to marginalia in NM manuscripts, and Wezler & Motegi (1998), on very lengthy marginal annotations to the YD — see e.g. Srinivasan (1967), which mentions numerous marginal annotations in Kashmirian manuscripts of Vācaspatimiśra’s Tattvakaumudī, or Slaje (2005), which briefly describes several annotated Mokṣopāya manuscripts. 138. See Formigatti (2011) on marginal annotations in manuscripts of the Raghuvaṃśa and Kāvyādarśa. 139. See e.g. fig. 3a here; note that attempts to repair manuscripts or at least prevent further damage by pasting strips of paper often render marginalia illegible (see e.g. the examples mentioned in Graheli 2015: 46). 338 for an indian philology of margins Appendix I Utpaladeva’s Īśvarapratyabhijñā-Vivr̥ti on ĪPK 2.1.3 1 [1] kriyāsakramatvavicāraprasaṅgād īśvarapratyabhijñopayogād api ca vakṣya- māṇāt kālasvarūpaṃ nirūpayitum 2 āha: [ĪPK 2.1.3] kālaḥ sūryādisaṃcāras tattatpuṣpādijanma vā | śītoṣṇe vātha tallakṣyaḥ krama eva sa tattvataḥ || [2] padārthasvarūpaviśeṣaparicchedahetuvyavahārāya 3 kālapadārtho’bhyupa- gato’vacchedakaś 4 ca prasiddha eva bhavatīti bahutarapratipādyapratipādaka- prasiddho nālikādivyapadeśyasūryādisaṃcāraviśeṣo 5 nirdiṣṭaḥ prasiddha- kriyopalakṣaṇārthaḥ, tad āha sā sā prasiddheti. 6 [3] viśiṣṭāpi 7 kramanirvartya- tayā 8 prasiddhatvāt tasyāḥ saiva kālaḥ. 9, 10 tena godohādīnām api grahaṇam. sūryādīti puṣpādigrahaṇād anyagrahaparigrahaḥ. 11 [4] puṣpādijanma sūryādi- saṃcārāvyutpannānāṃ hālikādīnāṃ prasiddham, 12 tena te sūryādisaṃcāraṃ vasantādisaṃjñam 13 apy upalakṣayanti. atrāpy 14 ādigrahaṇāc chālyādijanma 1. For a translation and study, see Ratié (forthcoming c). Appendix II provides a list of the manuscripts in the margins of which the passages preceded here by a bold number in brackets are preserved. The words underlined and in bold are quoted by Abhinavagupta in ĪPVV, vol. III: 4-9. 2. nirūpayitum S15, J11 : nirūpayatum S12. 3. padārthasvarūpaviśeṣaparicchedahetuvyavahārāya S12, J11 : padārthasvarūpaviśeṣa- paricchedahetur vyavahārāya Kawajiri (2016b) : padārthasvarūpaviśeṣaṃ paricchedahetur vyavahārāya S15. 4. ’vacchedakaś S15, Kawajiri (2016b) : ’vabodakaś S12, J11. 5. nālikādivyapadeśyasūryādisaṃcāraviśeṣo S12, S15 : nālikādivyapadeśasūryādisaṃcāra- viśeṣo J11 : [ā?si?krā]divyapadeśyaḥ sūryādisaṃcāraviśeṣo Kawajiri (2016b). 6. sā sā prasiddheti S15, Kawajiri (2016b) : sā prasiddheti S12, J11. 7. viśiṣṭāpi conj. : viśiṣṭāsya J11 : viśiṣṭasya S12 : viśiṣṭakriyā S15. See ĪPVV, vol. III, p. 5. 8. kramanirvartyatayā S12, S15 : kramanivartyatayā J11. 9. kālaḥ S12, J11 : kātraḥ S15. 10. The words cinmaya eveti were marked by the KSTS editors as a pratīka, but they probably belong to Abhinavagupta’s commentary (the main text in J11 omits the final iti, so does manuscript kha according to ĪPVV, vol. III, p. 6, n. 2). 11. puṣpādigrahaṇād anyagrahaparigrahaḥ S12, J11, S15 : puṣpādijanma ādigrahaṇād anya- grahaparigrahaḥ S19 : ādigrahaṇād anyagrahaparigrahaḥ Kawajiri (2016b). 12. sūryādisaṃcārāvyutpannānāṃ hālikādīnāṃ prasiddham S12, J11, S15, D2, SOAS, S19, n. 38 ĪPV : sūryādisaṃcārāvyutpannānāṃ hālikādīnāṃ puṣpādijanma Kawajiri (2016b). 13. vasantādisaṃjñam J11, S15 : vasantādikasaṃjñam D2, SOAS, n. 38 ĪPV. 14. atrāpy S12, J11, S15 : atr° D2, SOAS, n. 38 ĪPV.  isabelle ratié 339 gr̥hītaṃ 15 pakṣiviśeṣāgamanāder 16 [5] hālikāprasiddhasyopalakṣaṇaṃ caitat. tat sarvaṃ sā seti vīpsayā vivr̥tam. [6] na kevalaṃ pratītyavacchedāyaivopa- yogī 17 kālo yāvat siddharūpaḥ 18 sahakārikāraṇatādyupayogy apy uṣṇādiḥ 19 prasiddhaḥ, 20 [7] tad āha śītādīti. etad anyadarśanasiddhāntopadarśana- mātraṃ [8] nātra kvacid abhiniveśaḥ svecchāmātreṇāvacchedakānām upa- kalpitatvāt prasiddheś cānavasthitatvāt. [9] śītādīnāṃ kāryajananopayoge ca kāraṇataiva na kālatvam, kālatvaṃ punar avacchedakatvād eva, [10] etac cānavasthitam 21 ity uktam. [11] tad evam etat sarvopalakṣyamāṇaṃ yat kiñcit tad eva sthirarūpaṃ padārthapratītyavacchedopayogi kālo’stu. sa ca pratītiparyālocanayā krama evopapannaḥ, [12] tad āha tadupalakṣiteti. [13] tathā hi tasya vasantāder 22 ādityādyudayakramarūpasyaivādityādy- upādhibhūtasyānubhavaḥ. [14] tenaiva 23 sūryādisaṃcārādinā bhāvānām 24 abhūtānāṃ bhāvopalakṣaṇāt krama evopalakṣitaḥ, yenopādhinā 25 kramiko- ’rthaḥ puṣpādir 26 devadattādir api vāvabhāsyate, tad āha bhinnābhāsa- māneti. 27 [15] vasantādiśabdo bhūtabhaviṣyadādiśabdaś ca na sākṣāt kālātmā- nam āha yathā kriyāśabdaḥ kriyām, api tu yugapadvyavasthitasiddhārtha- 15. ādigrahaṇāc chālyādijanma gr̥hītaṃ S12, J11, S15, Kawajiri 2016b : ādigrahaṇāc chālyādi- janma kārṣikāṇāṃ prasiddhaṃ gr̥hītaṃ D2, SOAS, n. 38 ĪPV. 16. pakṣiviśeṣāgamanāder S12, J11 : pakṣiviśeṣāgamanādi S15 : tato’py anabhijñānāṃ tat tad iti vīpsayā pakṣiviśeṣāgamanāder grahaṇaṃ D2, SOAS, n. 38 ĪPV. 17. pratītyavacchedāyaivopayogī D2, Kawajiri (2016b) : pratītyāvacchedāyaivopayogī SOAS : pratītyavacchedāyaivopabhogī S12, J11, S15. 18. siddharūpaḥ S12, J11 p.c., S15, Kawajiri (2016b), SOAS : siddho rūpaḥ J11 a.c. : siddha- rūpoḥ D2. 19. °upayogy apy uṣṇādiḥ S12, J11, S15, Kawajiri (2016b) : °upayogī coṣṇatādiḥ D2, SOAS (with an attempt to delete the ḥ). 20. prasiddhaḥ S12, J11, S15, Kawajiri (2016b) : prasiddha iti D2, SOAS. 21. etac cānavasthitam S12, J11, S15 : etac cāvacchedakatvam anavasthitam D2, S3, SOAS. 22. tathā hi tasya vasantāder S12, J11, S15, S3 : vasantāder hi kālasya D2, SOAS, ĪPV n. 45. 23. tenaiva conj. (cf. ĪPVV, vol. III, p. 8: tenaiveti) : tena vā S12, S3 : yataḥ D2, SOAS, n. 45 ĪPV. 24. bhāvānām D2, S3, SOAS, n. 45 in ĪPV : bhāvanām S12. 25. yenopādhinā S12, S3 : tenopādhinā D2, SOAS, n. 45 ĪPV. 26. puṣpādir S3, D2, SOAS, n. 45 ĪPV : puṣpādi S12. 27. tad āha bhinnābhāsamāneti S12 : tad āha bhinnābhāsa S3 (there is a word underneath but no longer legible, the margin seems to have been cut) : ity arthaḥ D2, SOAS, n. 45 ĪPV. 340 for an indian philology of margins pratipādanadvāreṇa, āsīd bhavati bhavitetyādayas tu śabdāḥ kālānubhavocita- pratikāriṇaḥ. tad evaṃ janmasattāvipariṇāmādikramātmaiva 28 sa vyavatiṣṭha- mānaḥ kālalakṣaṇo’rthas tathaiva vicāryaḥ, tad āha tasyaivopayogād iti. krama eva hi tāvad 29 bhedāvacchedāyopayujyamāne’vasīyata ity arthaḥ, anena tattva- ta 30 iti vivr̥tam. 28. janmasattāvipariṇāmādikramātmaiva conj. : janmasattādipariṇāmādikramātmaiva S12. 29. tāvad corr. : tāva S12. 30. tattvata conj. : tattva S12.  isabelle ratié 341 Appendix II Marginal Sources for the Vivr̥ti on ĪPK 2.1.3 [1] S12; J11; S15. [2] S12; J11; S15; Kawajiri (2016b). [3] S12; J11; S15; S19 and Kawajiri (2016b) only have a very brief fragment.1 [4] S12; J11; S15; D2; SOAS; ĪPV, vol. II, p. 9, n. 38; S19 and Kawajiri (2016b) only have very brief fragments.2 [5] S12; J11; S15. [6] S12; J11; S15; D2; SOAS; Kawajiri (2016b). [7] S12; J11; S15; Kawajiri (2016b). [8] S12; J11; S15.3 [9] S12; J11; S15; S3; D2; SOAS; Kawajiri (2016b). [10] S12; J11; S15; S3 (followed by iti ṭīkā); D2; SOAS. [11] S12; J11; S15; S3; D2; SOAS. [12] S12; J11; S15. [13] S12; J11; S15; S3: D2; SOAS; ĪPV, vol. II, p. 10, n. 45. [14] S12; S3; D2; SOAS; ĪPV, vol. II, p. 10, n. 45. [15] S12. 1. puṣpādigrahaṇād anyagrahaparigrahaḥ S12, J11, S15 : puṣpādijanma ādigrahaṇād anya- grahaparigrahaḥ S19 : ādigrahaṇād anyagrahaparigrahaḥ Kawajiri (2016b). 2. 1) sūryādisañcārāvyutpannānāṃ hālikādīnāṃ prasiddham S12, J11, S15, D2, SOAS, S19, n. 38 ĪPV : sūryādisañcārāvyutpannānāṃ hālikādīnāṃ puṣpādijanma Kawajiri (2016b). 2) ādigrahaṇāc chālyādijanma gr̥hītaṃ S12, J11, S15, Kawajiri 2016b : ādigrahaṇāc chālyādijanma kārṣikāṇāṃ prasiddhaṃ gr̥hītaṃ D2, SOAS, n. 38 ĪPV. Kawajiri (2016b) takes chālyādijanma and gr̥hītaṃ as forming a compound, and believes that sūryādisañcārāvyutpannāṃ hālikādīnāṃ puṣpādijanma and ādigrahaṇāc chālyādijanma gr̥hītaṃ constitute a single fragment. 3. Kawajiri (2016b): 23 does not notice that there is a lacuna between 7 and 9, as can be seen from Abhinavagupta’s commentary (ĪPVV, vol. III: 7: ity āha prasiddheś ceti…, etc.). 342 for an indian philology of margins Appendix III A Few Annotated Manuscripts Figure 1 – Ms. SOAS, Īśvarapratyabhijñāvimarśinī, London, SOAS Library (photograph: Library of SOAS)  isabelle ratié 343 Figure 2 – Ms. J2, Īśvarapratyabhijñāvimarśinī, Jammu, Śrī Ranbir Institute (photograph: I. Ratié) 344 for an indian philology of margins Figure 3a – Ms. S3, Īśvarapratyabhijñāvimarśinī manuscript, Srinagar, Ori- ental Research Library (photograph: National Mission for Manuscripts) Figure 3b – Idem Figure 3c – Idem  isabelle ratié 345 Figure 4 – Ms. J11, Īśvarapratyabhijñā-vr̥tti and -vivr̥tivimarśinī, Jammu: Śrī Ranbir Institute, Raghunath Mandir Library (photograph: C. Pandey) 346 for an indian philology of margins Figure 5 – Ms. S12, Īśvarapratyabhijñāvivr̥tivimarśinī, Srinagar: Oriental Re- search Library (photograph: National Mission for Manuscripts)  isabelle ratié 347 Figure 6a – Ms. S12 (detail of a margin) Figure 6b – Ms. J11 (detail of a margin) Figure 6c – Ms. SOAS (detail of a margin) 348 for an indian philology of margins Figure 7 – Ms. J11 (detail of a margin) Figure 8 – Ms. S7, Īśvarapratyabhijñāvimarśinī, Srinagar, Oriental Research Library  isabelle ratié 349 References Manuscript Sources D. Īśvarapratyabhijñāvivr̥ti. Delhi: National Archives of India (“Manuscripts be- longing to the Archeology and Research Department, Jammu and Kashmir Government, Srinagar,” in List of Gilgit Manuscripts and Sanskrit Mss), no. 30, vol. IX [paper, śāradā script]. D2. Īśvarapratyabhijñāvimarśinī. Delhi: National Archives of India (“Manuscripts belonging to the Archeology and Research Department, Jammu and Kashmir Government, Srinagar,” in List of Gilgit Manuscripts and Sanskrit Mss), no. 5, vol. II [paper, śāradā script]. J2. Īśvarapratyabhijñāvimarśinī. Jammu: Śrī Ranbir Institute, Raghunath Mandir, no. 2 in Stein (1894) [śāradā script, birch bark; the folios are not bound in the correct order, as noticed in Stein (1894): 220: “pattrāṇi vaiparītyena nibaddhāni bandhakena”]. J11. Īśvarapratyabhijñāvivr̥tivimarśinī (“Pratyabhijñāvr̥ttī īśvarapratyabhijñāvivr̥ti”). Jammu: Śrī Ranbir Institute, Raghunath Mandir Library, no. 5077 [paper, nāgarī script]. S2. Īśvarapratyabhijñāvimarśinī. Srinagar: Oriental Research Library, no. 1035 [pa- per, śāradā script]. S3. Īśvarapratyabhijñāvimarśinī (“Īśvarapratyabhijñākaumudi”). Srinagar: Oriental Research Library, no. 838 [paper, śāradā script]. S5. Īśvarapratyabhijñāvimarśinī (“Sāṃkhyatattvakaumudī by Vācaspati Miśra”; the ĪPV is copied after the latter). Srinagar: Oriental Research Library, no. 1212 [paper, śāradā script]. S7. Īśvarapratyabhijñāvimarśinī. Srinagar: Oriental Research Library, no. 2250 [pa- per, nāgarī script]. S12. Īśvarapratyabhijñāvivr̥tivimarśinī (“Īśvarapratyabhijñā(vimarśinī)”). Srinagar: Oriental Research Library, no. 2403 [paper, śāradā script]. S15. Īśvarapratyabhijñāvimarśinī. Srinagar: Oriental Research Library, no. 787 [paper, śāradā script]. S17. Īśvarapratyabhijñāvr̥tti. Srinagar: Oriental Research Library, no. 860 [paper, śāradā script]. S18. Īśvarapratyabhijñākaumudī. Srinagar: Oriental Research Library, no. 1089 [paper, śāradā script]. S19. Īśvarapratyabhijñāvr̥tti. Srinagar: Oriental Research Library, no. 824 (wrongly numbered 823 on the Manus data sheet of the National Mission for Manuscripts) [paper, śāradā script]. SOAS. Īśvarapratyabhijñāvimarśinī. London: School of Oriental and African Studies Library, no. 207 in R.C. Dogra, A Handlist of the Manuscripts in South Asian Languages in the Library [of SOAS], London: SOAS, 1978/ no. 44255 [paper, śāradā script]. 350 for an indian philology of margins Editions of Sanskrit Texts Bhāskarī. Īśvarapratyabhijñāvimarśinī of Abhinavagupta. Doctrine of Divine Recognition. Ed. by K.A. Subramania Iyer and K.C. Pandey. Vol. I-II: Sanskrit Text with the Commentary Bhāskarī. Delhi: Motilal Banarsidass, 1986 [1938]. Cittānubodhaśāstra. Rājānaka Bhāskarakaṇṭḥa’s Cittānubodhaśāstram. Ed. by Sushama Pandey. Varanasi: Tara Book Agency, 1990. GNP = Gurunāthaparāmarśa (of Madhurāja). Ed. by P.N. Pushp. Kashmir Series of Texts and Studies 85. Srinagar, 1960. ĪPK = Īśvarapratyabhijñākārikā. See Torella (2002). ĪPV = Īśvarapratyabhijñāvimarśinī. Ed. by Mukund Ram Shāstrī and Madhusudan Kaul Shāstrī. 2 vols. Kashmir Series of Texts and Studies 22 & 33. Srinagar, 1918-1921. ĪPVV = Īśvarapratyabhijñāvivr̥tivimarśinī by Abhinavagupta. Ed. by Madhusudan Kaul Shāstrī. 3 vols. Kashmir Series of Texts and Studies 60, 62 & 65. Srinagar, 1938-1943. MVT = Mālinīvijayottaratantram. Ed. by Madhusudan Kaul Shāstrī. Kashmir Series of Texts and Studies 37. Bombay, 1922. NM = Nyāyamañjarī of Jayantabhaṭṭa with Ṭippaṇī. Nyāyasaurabha by the editor. Ed. by K.S. Varadhacharya. 2 vols. Mysore: Oriental Research Institute, 1969-1983. NMG = Nyāyamañjarīgranthibhaṅga. See Shah (1972). PTV = Parātrīśikāvivaraṇa. The Parā-Trimshikā with Commentary, the Latter by Abhi- nava Gupta. Ed. by Mukund Rām Shāstrī. Kashmir Series of Texts and Studies 18. Bombay, 1918. PVV = Pramāṇavārttikavr̥tti. “Dharmakīrti’s Pramāṇavārttika with Commentary by Manorathanandin”. Ed. by Rāhula Sāṅkr̥tyāyana. In: Journal of the Bhan- darkar Oriental Research Institute 24-26. 1938-1940, Appendix. RT = Rājataraṅginī. Kalhaṇa’s Rājataraṅginī or Chronicle of the Kings of Kashmir. Ed. by M.A. Stein. Vol. I: Sanskrit Text with Critical Notes. Bombay: Education Society’s Press, 1892. [Repr. Delhi: Motilal Banarsidass, 2003 [1988], as vol. III of Stein (1900).] ŚD = Śivadr̥ṣṭi of Śrīsomānandanātha with the Vr̥tti by Utpaladeva. Ed. by Madhusudan Kaul Shāstrī. Kashmir Series of Texts and Studies 54. Srinagar, 1934. Siddhitrayī. The Siddhitrayī and the Pratyabhijñākārikāvr̥tti of Rājānaka Utpaladeva. Ed. by Madhusudan Kaul Shāstrī. Kashmir Series of Texts and Studies 34. Srinagar, 1921. Tantrasāra. The Tantrasāra of Abhinavagupta. Ed. by Mukund Rām Shāstrī. Kashmir Series of Texts and Studies 17. Srinagar, 1918. Vr̥tti = Īśvarapratyabhijñāvr̥tti. See Torella (2002). YD = Yuktidīpikā. See Wezler & Motegi (1998).  isabelle ratié 351 Translations, Studies, Catalogues Bronkhorst, Johannes. 2003. “Review of Yuktidīpikā. The Most Significant Com- mentary on the Sāṃkhyakārikā… by Albrecht Wezler and Shujun Motegi”. Zeitschrift der Deutschen Morgenländischen Gesellschaft 153.1: 242-247. Bühler, Georg. 1877. Detailed Report of a Tour in Search of Sanskrit MSS Made in Kaśmīr, Rajputana and Central India. Extra Number of the Journal of the Bombay Branch of the Royal Asiatic Society. Camille, Michael. 1992. Image on the Edge. The Margins of Medieval Art, Essays in Art and Culture. London: Reaktion Books. Clemens, Raymond and Timothy Graham. 2007. Introduction to Manuscript Studies. Ithaca and London: Cornell University Press. Cox, Whitney. 2010. “Scribe and Script in the Cālukya West Deccan”. The Indian Economic and Social History Review 47.1: 1-28. Croizy-Naquet, Catherine, Laurence Harf-Lancner, and Michelle Szkilnik, eds. 2015. Les Manuscrits médiévaux témoins de lectures. Paris: Presses Sorbonne Nouvelle. Formigatti, Camillo. 2011. “Sanskrit Annotated Manuscripts from Northern India and Nepal”. PhD thesis. University of Hamburg. Supervised by H. Isaac- son. Francis, Emmanuel. 2017. “The Other Way Round. From Print to Manuscript”. In: Indic Manuscript Cultures through the Ages. Material, Textual, and Historical Investi- gations. Ed. by Vincenzo Vergiani, Daniele Cuneo, and Camillo Formigatti. Berlin: de Gruyter, p. 319-351. Graheli, Alessandro. 2015. History and Transmission of the Nyāyamañjarī. Critical edi- tion of the Section on the Sphoṭa. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Grierson, George. 1916. “On the Śāradā Alphabet”. Journal of the Royal Asiatic Society 17: 677-708. — 1929. “The Language of the Mahā-Naya-Prakāśa. An Examination of Kāshmīrī as Written in the 15th Century”. Memoirs of the Asiatic Society of Bengal 11.2: 73-129. Hanneder, Jürgen. 1998. Abhinavagupta’s Philosophy of Revelation. Mālinīśloka- vārttika 1.1-399. Gronigen Oriental Series 14. Gronigen: Egbert Forsten. — 2017. To Edit or Not to Edit. On Textual Criticism of Sanskrit Works. Pune Indologi- cal Series. Pune: Aditya Prakashan. Hanneder, Jürgen, Stanislav Jager, and Alexis Sanderson. 2013. Ratnakaṇṭhas Stotras – Sūryastutirahasya, Ratnaśataka und Śambhukr̥pāmanoharastava. Mün- chen: P. Kirchheim Verlag. Hooks, Adam G. 2012. “Marginalia”. In: The Encyclopedia of English Renaissance Literature. Ed. by Garrett A. Sullivan Jr. and Alan Stewart. Vol. 2. Oxford: Wiley-Blackwell, p. 636-639. 352 for an indian philology of margins Jager, Stanislav. 2010. “Editing Rājānaka Ratnakaṇṭha’s Sūryastutirahasya and Ratnaśataka”. Wiener Zeitschrift für die Kunde Südasiens 52-53: 285-294. Kannupillai, L.D. 1922. An Indian Ephemeris A.D. 700 to A.D. 1799. Vol. VI: A.D. 1600 to A.D. 1799. Madras: Superintendent Government Press. Kaul Deambi, B.K. 1982. Corpus of Śāradā Inscriptions of Kashmir, with Special Refer- ence to Origin and Development of Śāradā Script. Delhi: Agam Kala Prakashan. Kawajiri, Yohei. 2016a. “New Fragments of the Īśvarapratyabhijñā-vivr̥ti”. In: Utpaladeva, Philosopher of Recognition. Ed. by Raffaele Torella and Bettina Bäumer. Delhi: DK Printworld, p. 77-101. — 2016b. “New Fragments of the Īśvarapratyabhijñā-vivr̥ti (3)”. Nagoya Studies in Indian Culture and Buddhism (Saṃbhāṣā) 33: 17-46. Keefer, Sarah Larratt and Rolf H. Bremmer Jr., eds. 2007. Signs on the Edge. Space, Text and Margin in Medieval Manuscripts. Paris and Louvain: Peeters. Lawrence, David P. 2005. “Remarks on Abhinavagupta’s Use of the Analogy of Reflection”. Journal of Indian Philosophy 33: 583-599. Muroya, Yasutaka. 2010. “A Study on the Marginalia in Some Nyāyamañjarī Manuscripts. The Reconstruction of a Lost Portion of the Nyāyamañjarī- granthibhaṅga”. Wiener Zeitschrift für die Kunde Südasiens 52-53: 218-267. Nemec, John. 2011. The Ubiquitous Śiva. Somānanda’s Śivadr̥ṣṭi and His Tantric Inter- locutors. Religion in Translation. Oxford and New York: New York University Press. Pandey, Kanti Candra. 1963. Abhinavagupta. An Historical and Philosophical Study. 2nd ed. [Varanasi: 1935]. Varanasi: Chowkhamba Sanskrit Series Office. Raghavan, V. 1980. Abhinavagupta and His Works. Varanasi: Chaukhamba Orien- talia. Rastogi, Navjivan. 1984. “Some More Nyāyas as Employed by Abhinavagupta”. Annals of the Bhandarkar Oriental Research Institute 65.1: 27-42. Ratié, Isabelle. 2016a. “Some Hitherto Unknown Fragments of Utpaladeva’s Vivr̥ti (I). On the Buddhist Controversy over the Existence of Other Conscious Streams”. In: Utpaladeva, Philosopher of Recognition. Ed. by Raffaele Torella and Bettina Bäumer. Delhi: DK Printworld, p. 224-256. — 2016b. “Some Hitherto Unknown Fragments of Utpaladeva’s Vivr̥ti (III). On Memory and Error”. In: Around Abhinavagupta. Aspects of the Intellectual History of Kashmir from the Ninth to the Eleventh Century. Ed. by Eli Franco and Isabelle Ratié. Berlin: Lit Verlag, p. 375-400. — 2017a. “In Search of Utpaladeva’s Lost Vivr̥ti on the Pratyabhijñā Treatise. Report on the Latest Discoveries (with the Vivr̥ti on the End of Chapter 1.8)”. Journal of Indian Philosophy 45: 163-189. — 2017b. “An Indian Debate on Optical Reflections and its Metaphysical Im- plications. Śaiva Nondualism and the Mirror of Consciousness”. In: Indian Epistemology and Metaphysics. Ed. by Joerg Tuske. London: Bloomsbury, p. 207- 240.  isabelle ratié 353 — 2017c. “Scholasticism and Philosophy. On the Relationship between Rea- son and Revelation in India”. Théorèmes 11: 1-15. url: http://journals. openedition.org/theoremes/1132. — Forthcoming a. “Some Hitherto Unknown Fragments of Utpaladeva’s Vivr̥ti (II). Against the Existence of External Objects”. In: Śaivism and the Tantric Traditions. Volume in Honour of Alexis G.J.S. Sanderson. Ed. by Dominic Goodall et al. — Forthcoming b. “On the Ṣaḍdhātusamīkṣā, a Lost Work Attributed to Bhartr̥- hari. An Examination of Testimonies and a List of Fragments”. Journal of the American Oriental Society. — Forthcoming c. Utpaladeva on the Power of Action. A First Edition, Annotated Translation and Study of Īśvarapratyabhijñāvivr̥ti, Chapter 2.1. Sanderson, Alexis. 2007. “The Śaiva Exegesis of Kashmir”. In: Mélanges tantriques à la mémoire d’Hélène Brunner. Ed. by Dominic Goodall and André Padoux. Pondicherry: Institut Français de Pondichéry and École Française d’Extrême- Orient, p. 231-442. Shah, Nagin J. 1972. Cakradhara’s Nyāyamañjarī-Granthibhaṅga. L.D. Series 35. Ahmedabad. Sircar, D.C. 1965. Indian Epigraphy. Delhi: Motilal Banarsidass. Slaje, Walter. 1990. Katalog der Sanskrit-Handschriften der Österreichischen National- bibliothek (Sammlungen Marcus Aurel Stein und Carl Alexander von Hügel). Vienna: Verlag der Österreichischen Akademie der Wissenschaften. — 1993. Śāradā. Deskriptiv-synchrone Schriftkunde zur Bearbeitung kaschmirischer Sanskrit-Manuskripte. Reinbek: Dr. Inge Wezler Verlag für Orientalistische Fachpublikationen. — 2005. “The Mokṣopāya Project (III). Manuscripts from the Delhi and Śrīnagar Collections”. In: The Mokṣopāya, Yogavāsiṣṭha and Related Texts. Ed. by Jürgen Hanneder. Aachen: Shaker Verlag, p. 37-54. Slights, William W.E. 2001. Managing Readers. Printed Marginalia in English Renais- sance Books. Ann Arbor: University of Michigan Press. Smith, Kathryn A. 2012. “Margin”. Studies in Iconography: Medieval Art History today – Critical Terms: 29-44. Srinivasan, Srinivasa Ayya. 1967. Vācaspatimiśras Tattvakaumudī. Ein Beitrag zur Textkritik bei Kontaminierter Überlieferung. Hamburg: de Gruyter. Stein, Marc Aurel. 1894. Catalogue of the Sanskrit Manuscripts in the Raghunatha Temple Library of His Highness the Mahārāja of Jammu and Kashmir. Bombay: Nirnaya-Sagara Press. — 1900. Kalhaṇa’s Rājataraṅgiṇī, A Chronicle of the Kings of Kaśmīr, Translated, with an Introduction, Commentary, and Appendices. 2 vols. Vol. I: Introduction, Books I-VII; vol. II: Book VIII, Notes, Geographical Memoir, Index, Maps. Westminster: Archibald Constable and Company. Stern, Elliot. 1991. “Additional Fragments of Pramāṇaviniścaya I-II”. Wiener Zeitschrift für die Kunde Südasiens 35: 151-168. 354 for an indian philology of margins Torella, Raffaele. 1988. “A Fragment of Utpaladeva’s Īśvarapratyabhijñā-vivr̥ti”. East and West 38: 137-174. — 2002. Īśvarapratyabhijñākārikā of Utpaladeva with the Author’s Vr̥tti. Critical Edi- tion and Annotated Translation. [Roma: 1994.] Delhi: Motilal Banarsidass. — 2007a. “Studies on Utpaladeva’s Īśvarapratyabhijñā-vivr̥ti. Part I: Anupalabdhi and Apoha in a Śaiva Garb”. In: Expanding and Merging Horizons. Contributions to South Asian and Cross-Cultural Studies in Commemoration of Wilhelm Halbfass. Ed. by Karin Preisendanz. Wien: Verlag der Österreichischen Akademie der Wissenschaften, p. 473-490. — 2007b. “Studies on Utpaladeva’s Īśvarapratyabhijñā-vivr̥ti. Part II: What is Memory?” In: Indica et Tibetica. Festschrift für Michael Hahn zum 65. Geburtstag von Freunden und Schülern überreicht. Ed. by Konrad Klaus and Jens-Uwe Hartmann. Wiener Studien Zur Tibetologie und Buddhismuskunde 66. Wien: Universität Wien, p. 539-563. — 2007c. “Studies on Utpaladeva’s Īśvarapratyabhijñā-vivr̥ti. Part III: Can a Cog- nition Become the Object of Another Cognition?” In: Mélanges tantriques à la mémoire d’Hélène Brunner. Ed. by Dominic Goodall and André Padoux. Pondicherry: Institut Français de Pondichéry and École Française d’Extrême- Orient, p. 475-484. — 2007d. “Studies on Utpaladeva’s Īśvarapratyabhijñā-vivr̥ti. Part IV: Light of the Subject, Light of the Object”. In: Pramāṇakīrtiḥ. Papers Dedicated to Ernst Steinkellner on the Occasion of his 70th Birthday. Ed. by Birgit Kellner et al. Wiener Studien Zur Tibetologie und Buddhismuskunde 70.2. Wien: Univer- sität Wien, p. 925-940. — 2014. “Utpaladeva’s Lost Vivr̥ti on the Īśvarapratyabhijñā-kārikā”. Journal of Indian Philosophy 42: 115-126. Tribble, Evelyn B. 1993. Margins and Marginality. The Printed Page in Early Modern England. Charlottesville: University of Virginia Press. Vasudeva, Somdev. 2004. The Yoga of the Mālinīvijayottaratantra. Chapters I-4,7, II-17, Critical Edition, Translation & Notes. Pondicherry: Institut Français de Pondichéry and École Française d’Extrême-Orient. Wezler, Albrecht and Shujun Motegi, eds. 1998. Yuktidīpikā. The Most Significant Commentary on the Sāṃkhyakārikā, Critically Edited. Stuttgart: Franz Steiner Verlag. Witzel, Michael. 1985. “Die Atharvaveda-Tradition und die Paippalāda-Saṃhitā”. Zeitschrift der Deutschen Morgenländischen Gesellschaft Supplement 6: 256-271. — 1994. “Kashmiri Manuscripts and Pronunciation”. In: A Study of the Nīlamata. Aspects of Hinduism in Ancient Kashmir. Ed. by Yasuke Ikari. Kyoto: Institute for Research in Humanities, Kyoto University, p. 1-53. — 2014. “Textual Criticism in Indology and in European Philology during the 19th and the 20th Centuries”. Electronic Journal of Vedic Studies 21.3: 9-90.