Historical proof by textual criticism?
by Rob Mayer
Cathy and I are very gratified by the positive reception of our work in critical editing, even though I am certain I do not deserve it, and remain acutely aware of the many failings in my work. We are above all gratified because a lot more hangs on the quality of our editions than mere vanity. For if our latest edition of an early rNying ma tantra stands good, then so too does our proposed historical proof for a greater antiquity than previously demonstrable of the Noble Noose of Methods, the Mahāyoga tantra so closely associated with Padmasambhava. But if our edition proves faulty, then so too will our proof that this tantra predates the Dunhuang text by at least two copyings, and therefore might indeed date back to the late Imperial period. For these reasons, since real history depends on it, we made every effort to apply the same standards of professionalism and rigour that are commonally required in the editing of Western texts, as did such excellent scholars before us as Helmut Eimer, Paul Harrison and Jonathan Silk in their own pioneering editions from the Kanjur.
And perhaps that is why, despite my personal ineptitude, van der Kuijp not long ago singled out our work as ‘truly superb’ in the field of Tibetan critical editing, and others like Wedemeyer have concurred. Per Kvaerne has described our edition of the Noble Noose of Methods as a ‘remarkable scholarly achievement’, and ‘a massive and highly important contribution to Tibetan studies’, that ‘cannot receive anything but the highest praise’,  while for the same work Helmut Eimer warmly congratulated us on ‘ein beeindruckendes Werk der Bearbeitung’ (‘an impressive work of editing’). Hopefully, these positive reviews from such great scholars will prove to be good omens that our historical proof by textual criticism will stand the test of time!
More recently, there has also been a review from Giacomella Orofino, in what might well prove to have been the final issue of the finacially beleagured JIATS, who also makes positive remarks about our work: she says, for example, that we ‘have undertaken a very complex and important work that has the great merit of being a pioneering study’, and she describes the monograph as, ‘a dense, important work that offers valuable sources of information on the development of the Rnying ma pa tradition’. But Orofino’s review is at the same time vitiated by a great deal of confusion, including at several key and even elementary points, and especially when attempting the more technical aspects of our dense and lengthy monograph. Even in seeking to praise us, she makes surprisingly inaccurate statements of both fact and interpretation, and her two or three points of criticism are no less muddle-headed. Regarding the latter, we fully hope and expect that in due course, some truly learned scholar might propose real improvements to our undoubtedly imperfect work, and although this might prove a somewhat chastening experience, it should also be enlightening. However, Orofino’s present criticisms are as difficult to take seriously as her praises, because she is so manifestly confused in both instances.
Yet the beauty of the scholarly life lies not in the exchange of bare assertions and opinions, but in the submission of all protagonists to the gentle light of truth, and in the final analysis, that can only be established through evidence and reasoning. It is in such a spirit that we offer the list of comments numbered below, that correct Orofino’s misunderstandings. And moreover, writing this response gives us a further opportunity succinctly to encapsulate several key points about what is, unavoidably, a highly complex work. Normally, such a response would appear within the JIATS itself, and their editorial board did indeed invite me to make such a response. However, since this wonderful journal is now facing severe financial problems, and might never appear again, or at the very least will take many months or even years to produce its next issue, I decided to present my response here, very much more swiftly, rather than await an uncertain appearence in JIATS.
The book in question is: Cathy Cantwell and Robert Mayer, 2012: A Noble Noose of Methods, The Lotus Garland Synopsis: A Mahāyoga Tantra and its Commentary, Österreichische Akademie der Wissenschaften, Vienna. The review of it in question is by Giacomella Orofino, in JIATS Issue 7, August 2013.
 pages 454-455: Complimenting our courage in attempting such an arduous and complex task, Orofino portrays us at the outset of her review with an opening quote about the great difficulties of “reconstructing…..the original text of a literary work”, which implies that we actually attempted to reconstruct the original Thabs zhags Root Tantra text. Alas, not so! We pursue, of course, only the very much more modest task of reconstructing its archetype, that is, the closest ancestor of all extant versions, not the original Tibetan version. And we do not even attempt a critical edition of the Commentary, although, for some unknown reason, Orofino clearly seems to think that we do (page 460).
 A major part of the introduction to our book concerns Padmasambhava, and Orofino likewise seems pleased with this work. Yet, unfortunately, she somehow manages to misunderstand, almost to the point of reversing, the main thrust of our argumentation. On page 456, she writes:
‘[[[Cantwell]] and Mayer] have concluded that the Padma rGyal po of the .. (Thabs zhags Commentary] cannot be identified with the eighth-century historical figure of Padmasambhava; he might rather correspond to the mythologized tantric master of the later period.’
We nowhere conclude that the Padma rGyal po of the Thabs zhags Commentary cannot be identified with the eighth-century figure of Padmasambhava. Since we are aware from the ethnographic record that gurus of Padmasambhava’s type are often mythologized within their own lifetimes, we cannot accept without further analysis that Padmasambhava too was not mythologized in the 8th century. Hence we seek to complicate the facile assumption of an 8th century historical Padmasambhava and a later mythologised one, by presenting evidence that Padmasambhava is already highly mythologised in such a very early text as the Thabs zhags Commentary, in precisely the same terms as later employed by Nyang ral. Moreover, the Dunhuang version mythologises him yet further in its unique marginal annotations.
 In the following sentence, also on page 456, she compounds the confusion by misreading our analysis of Padmasambhava’s putative authorial relations to the Man ngag lta ba’i phreng ba and the Commentary on “A Noble Noose of Methods”:
‘This [[[non-identity]] of the Padma rGyal po of the Thabs zhags Commentary with an 8th century historical Padmasambhava) can also be confirmed by the fact that it is not possible to observe remarkable similarities between the Commentary on “A Noble Noose of Methods” and the Man ngag lta ba’i phreng ba, the famous early work attributed to the historical Padmasambhava.’
We have in fact shown something entirely different, because we question the popular modern presumption, deriving originally from Ken Eastman’s tentative comments which were magnified uncritically by subsequent scholars, that the Thabs zhags Commentary is attributable to Padmasambhava. We are certainly not saying that some later Padmasambhava composed the Thabs zhags Commentary; rather, we are saying that the text itself does not by any means make clear that its references to Padmasambhava should be taken as authorial attributions at all. On the contrary, Padmasambhava seems to be presented not as the merely human author of the Thabs zhags Commentary, but more likely as the transcendent enlightened inspiration for the Root Tantra itself. In her above statement, Orofino seems to miss this, our key point, implying that we believe the Padma rGyal po who is eulogised at the end of the Thabs zhags Commentary must be its human composer. Moreover, we do not even assert (or deny) that the Man ngag lta ba’i phreng ba is by the historical Padmasambhava. We simply observe that if there is any written work which might, according to the criteria of the modern academic historian, possibly be by the historical Padmasambhava, then the Man ngag lta ba’i phreng ba seems to be the best candidate so far. But there is nothing in the Thabs zhags Commentary which might suggest it to be by the same author.
 On page 457, Orofino seems to misconstrue the relevant findings of Zimmerman and Skilling about the Bathang Kanjur. These are not, as Orofino says, that the Bathang collection as a whole might even possibly predate the 14th century Tshalpa Kanjur. Their relevant findings are, by contrast, that on close examination, some of Bathang’s individual constituent texts preserve very old readings indeed, even (Zimmerman on the Tathāgatagarbha Sūtra) predating Ralpachen’s 9th century reforms. That is why we make the point that Zimmerman and Skilling’s finding is consistent with our stemma, in that the Bathang Thabs zhags also shows readings of great antiquity, predating the Dunhuang text.
 On page 458, Orofino writes as though we never intended to make a full-on stemma codicum, even though this is quite clearly exactly what we do intend to do:
“One should notice, though, that rather than constructing a stemma codicum in a strict Lachmanian (sic) sense, the two scholars [[[Cantwell]] and Mayer] have outlined a complex diagram of the possible transmissions of the Noble Noose of Methods, displaying the relationship between them. In this diagram the length of the lines of descent has no significance from a historical point of view.”
Since it is said that Lachmann never made a stemma codicum himself (Greetham 1994:323), and since her statement was incomprehensible to us, we asked Orofino for clarification. She confirmed that she meant we had not attempted or intended to make a stemma codicum in the classic tradition of Lachmann’s main modern interpreters and intellectual descendants, such as M. L. West and Paul Maas, and that she should have said, “in the classical Lachmanian sense”, rather than in “a strict Lachmanian sense.” On the contrary, let us assure our readers that: (i) we have in fact made a stemma codicum; (ii) it is in the style of Lachmann’s main modern disciples, Paul Maas, and Martin L. West, our emeritus colleague at Oxford; (iii) it does have historical significance, allowing us to recover archetypal readings (but yes, the length of diagram lines is of course immaterial); (iv) it is the most salient, prominent outcome of our text critical efforts in this monograph.
 On page 458, she writes that
“Unfortunately, one should notice that the severely truncated form of the Bstan ’gyur versions sets limitations to a clear-cut examinatio.“
We do indeed say (p.47) that almost 40% of the [[bsTan \\‘gyur root text lemmata are lost. Moreover, the lemmata are lost for some of the critical readings which demonstrate a number of the indicative errors of the sub-branch descending from “c”. So we cannot know whether they might in fact have been indicative errors of the whole branch descending from “b”.
Yet Orofino’s statement is misleading if the reader might get the impression that this loss hampers our main overall picture of five separate branches from the archetype. It does not – that is clear. The main limitation is in terms of our understanding of the status of “b” – whether it only had a limited number of distinctive readings, or whether it was already riddled with many of “c”‘s errors as well.
 On page 459, Orofino confuses our own stemmatic critical text with part of its apparatus. She says:
‘The authors assert that their aim is (25) “.. accepting and even celebrating the ongoing permutations of these texts, but still finding value in stemmatic techniques as a way of recovering both their original archetypes and also significant moment in their history.” Given such premises, one might wonder whether this ambivalent approach might not cause some confusion in establishing the critical text.’
Our resolve to show alongside our own stemmatic edition full data sets of how the textual tradition has ebbed and flowed through long periods of its history, has had no bearing whatosever on the establishment of our own ‘critical text’. It has bearing only on the non-critical unedited text we choose to append as various types of apparatus. These include separate text-envelopes, presenting the many commentarial intrusions into the root text that characterise the transmitted canonical versions from the most popular bKa’ ‘gyur and rNying ma’i rgyud ‘bum editions. Our own ‘critical text’ for the root text is based on stemmatic principles, and is not in itself ambivalent. Yet, our theoretical standpoint, that it is important to recognise all the historically significant presentations of the text, is indeed built into our presentation of the wider data sets that comprise the monograph as a whole, and we make no apology for this. There is no confusion beteween this and our stemmatic reconstruction. It is important to recall that the purpose of an edition is to facilitate further scholarship by providing various data sets, and for this reason, we consider it inadequate simply to construct our own edition without also representing the versions that Tibetans have in real history actually relied upon. Yet we do not muddle these historical traditions with our own edition. Instead, what we do is highlight the readings and whole passages of text which differ in the Tshal pa bKa’ ‘gyur and in the Bhutanese rNying ma’i rgyud ‘bum versions, including separate clearly marked text envelopes with passages of text given in these widespread and historically important traditional transmissions of the text. We consider critical editing neither to be about producing monoliths, nor about producing abstract aesthetic artifacts: we consider critical editing to be about presenting a range of useful and comprehensive data sets.
 On page 459, Orofino says we claim to have reconstructed the entire archetypal text (contradicting her implication elsewhere that we were reconstructing the original; nor is it clear how we could have achieved any reconstruction at all, if, as she says elsewhere, we had not even attempted a full-on stemma codicum). Be that as it may, we nowhere make any claim to have reconstructed the entire archetypal text. We cannot quantify exactly how much has been reconstructed. We hope it is a good proportion, but certainly cannot claim everything, and were at pains never to claim this. Out of this basic confusion, Orofino consequently criticises our decision to preserve archaic orthography in the apparatus. On the contrary, to place uncertain reconstructions of archaic orthography in the main body of what is in the final analysis our own text, would be a misleading affectation—better present them in the apparatus.
She then adds some largely tangential comments about the need to study Old Tibetan.
1. As we say, we draw attention to all readings which might possibly have been in the archetype, including many cases of very minor variants where our five groups descending from the archetype are divided, so that stemmatic logic will not establish with certainty the the precise spelling or particle in the archetype. We still draw attention to readings which might have been in the archetype by italicising them in the apparatus, so we are not ignoring or belittling the evidence of the possibly earliest readings.
2. These possibly archetypal readings which are in the notes rather than the body of our edition unfortunately do not represent a distinctive form of Old Tibetan which could give us any great breakthrough in research into Old Tibetan. There are in any case not a lot of them, as browsing the apparatus of our edition will make clear (perhaps Orofino, like some non-specialists, tends to overestimate a priori the degree to which the formal chos skad of canonical texts has changed since early times; yet one must be mindful that the modern editions of the Kanjur and Tenjur remain to this day packed with translations made in those early times). One example of our possibly archetypal readings (Ch. 37) is rngul, instead of dngul; it is well-known that the head letter ra is often used instead of da prefix in archaic texts. But even this kind of example is extremely rare in our text. A more typical example is pa’i as opposed to pa yi (various instances); or gcig du as opposed to gcig tu (Ch.16) or krō dha instead of kro dha in mantras (Ch.14), or the particle, ste instead of de (Ch.12). Should we be accused of failing to apply academic research on Old Tibetan by our italicisation of these minor variants in the apparatus, rather than their inclusion in our edition?
3. We give detailed information on the codicological features and orthographical peculiarities of the Dunhuang ms. (a detailed section p.32-34), the only edition of which the copy itself is very old (in the other versions which inherit the earliest readings, it is quite likely that some archaic features were dropped).
 On page 460, Orofino opens her discussion of the Commentary text with the words
“As regards the critical edition of the Commentary…”
We made no critical edition of the commentary, and have never called our edition of it critical. We made a simple diplomatic edition based on the Dunhuang ms, the most complete version, straightforwardly collated against the Tenjur version without any choices of individual spellings on our part, and to complete the data set, collated also those commentarial intrusions found in the various root text versions. In the few places where the Dunhuang text was missing, we continued to present the Tenjur text, and very rarely when it too was corrupted, the Tshal pa Kanjur version (clearly marked off so that it could not be confused with the Dunhuang manuscript transcription). It is important to grasp at the outset that there are no recensional variations between the Dunhuang and Tenjur witnesses of the Commentary (the Tshal pa fragments are of course derivative of the Tenjur version): they only differ in transmissional errors and in the sections of them that survived. Orofino has palpably not understood this fundamental consideration. Our diplomatic edition was thus an informed and well-judged strategy, based on the consideration that there are no recensional differences between these versions, and on the overwhelming importance of making this unique commentary readily available for study in terms of its meanings and doctrines. See also the next point:
 On page 460, Orofino continues her discussion about our presentation of the Commentary text by saying that
“Collating witnesses that have a gap of more than seven hundred years between them is…not without controversial aspects from a methodological point of view.”
She means the Tenjur is seven hundred years later than the Dunhuang text. Actually, for what is primarily a diplomatic transcription like this of two versions with no recensional variations between them, it might well be described as quite uncontroversial. Be that as it may, our stemma of its lemmata has shown the Tenjur to have older indicative readings than the Dunhuang witness by at least two generations of copying, a finding that Orofino herself accepts elsewhere. It is therefore not only self-contradictory, but also naive and misleading, for her here to assert so simplistically that the Tenjur text is seven hundred years later than the Dunhuang text (cf recentiores non deteriores, the more recent copyist may have copied an older manuscript).
 On page 460, Orofino says about our Commentary edition that
” the great linguistic lack of homogeneity between the two lines of witnesses (Tenjur and Dunhuang)….makes the authors’ choice of reconstructing the overall shape of the text…problematic.”
This statement is factually mistaken. There is no “great lack of linguistic homogeneity” between the Tenjur and Dunhuang texts, and in fact, there are no recensional variations between them, merely transmissional ones. Hence these texts are abundantly suited to be collated together.
 Moreover, Orofino’s idea that it is problematic that we decided to reconstruct the overall shape of the text, is absurd. The Dunhuang text omits a few lines here and there, or gets the order of a passage of text muddled (as in one section) or omits a couple of very short chapters, perhaps by eyeskip. Yet there is little doubt that these corruptions are due to scribal lapse, not to design: they are transmissional, not recensional. Where they survive in the Tenjur, we therefore very correctly supply the reader with the missing lines or passages, and correct the order. Of course, all this is transparently marked and suitably signalled with headings in the main text (such as the opening to Chapter 31), not only hidden in the apparatus, so that it is entirely clear where the Dunhuang manuscript lacked the text, and from which source the text is continued. Not to include the omitted passages at all, for the sake of avoiding an imginary “variance between different methodological procedures”, seems in this case an absurdity. As we have already pointed out above, the Dunhuang and Tenjur texts are the same basic recension, not different ones, varying only in transmissional error; and, as we have also pointed out above, the indicative readings of the Tenjur text’s lemmata are slightly older, not 700 years younger, than the Dunhuang text.
Finally, we should mention that this remarkable text is the most important source so far known to scholarship for an understanding of early rNying ma tantrism. Our task is to present comprehensive and transparent data sets of it useful for scholars. We must in the final analysis consider what textual criticism can do for rNying ma studies, not what rNying ma studies can do for textual criticism. We aim not at abstract aesthetic perfection, but at the most complete, transparent, presentation of functional data.
Orofino herself has never attempted an edition outside of the Kanjur, nor a stemmatic textual reconstruction. One of Orofino’s two published editions is a very brief diplomatic transcription of part of the Mañjuśrīnāmasaṃgīti (Orofino 2007). The other is a parallel presentation with minimal apparatus of two different Tibetan translations of the Sekkodeśa (1994), in which she seeks, controversially, to emend the root text to conform with various of its later commentaries. Hence she writes (p. 38): ‘my aim was to present a text [of the Sekkodeśa) conforming with the readings and meanings of the various commentaries, prātikas etc. both in Sanskrit and Tibetan’.
A very major consideration in editing Kanjur texts, especially the more famous ones, is the existence of multiple Tibetan translations of the same text from Sanskrit, as also of Tibetan translations of differing Sanskrit versions. Such recensions are thus inherently open, not closed, and therefore problematic for stemmatic reconstruction, since there is no single archetype from which all versions descend. The variant versions can moreover interact further within Tibet, complicating matters through contamination. Such factors create a proliferation of variants that are recensional (made deliberately) rather than transmissional (arising by accident), and at a very deep and imponderable level.
Yet rNying ma tantric texts are in many cases quite different, and Orofino seems unable to envisage this; in fact, she does not make it entirely clear if she is even considering the crucial distinction that must be made between redactional and transmissional variants. The largely indigenous, Tibetan-produced texts of the NGB have a very much smaller proportion of translations from Sanskrit at all, let alone multiple ones of any kind. The five rNying ma texts we have investigated so far moreover each appear to originate in a single moment of origin within Tibet, whether redactorial, or, in one case, possibly translational. And in the case of these five Mahāyoga texts, we can also see there is a much lower proportion of recensional differences.
This is not a statement we make idly. We are not simply assuming these variations are transmissional rather than recensional. Over the years, we have devoted a great deal of energy to minutely analysing the nature and causes of textual variation as found in rNying ma tantras. In our unpublished work sheets from editing two major NGB texts, which we hope to publish electronically when the technology is developed, we exhaustively analyse every single variant in every single witness of the texts to ascertain as exactly as possible what the probable reason for variation was, be it eyeskip, haplography, dittography, homophony, etc. etc. Some indication of the scope of this work can be seen in our article of 2006, “Two Proposals for Critically Editing the Texts of the rNying ma’i rGyud ‘bum”, Revue d’Etudes Tibétaines, Vol.10, April 2006: 56-70.
Interestingly, Orofino seems to accept our analysis of the earliest reconstructable readings—her main criticism seems in fact to be that we should keep to it more rigidly in the main text of our edition! But she is confused in points of detail and basic principle as set out above, and also more generally by the transmissional environment of the Thabs zhags and its Commentary, where transmissional errors predominate over recensional ones, where there is no multiple translation from Sanskrit causing an open recension, where a very early witnesses from Dunhuang is available, and where the recension has proven closed. Hence it is unsurprising that at several points in her review, she palpably seems to misconstrue what we do, and inadvertently projects the more familiar problems of her previous Kanjur editing onto the rather different rNying ma works we have attempted. Above all, she seems to be constantly deceived by an a priori expectation that recensional variation must necessarily prevail at all points: yet it does not. In addition, she seems to have neglected some of the basics of textual criticism, such as the distinction between an archetype and an original, or the principle of recentiores non deteriores.
Maas himself conceded that where there was substantial contamination, stemmatics could not reconstruct the archetype, although of course it still remained an indispensible tool for identifying the separate descents of different groups of texts, and bringing out the evidence for contamination. It is equally clear that stemmatic analysis also runs into limitations where recensions are open, rather than closed; but even here, one can still often use stemmatics to establish hypearchetypes, or even to establish the archetypes for each of the various versions. Many Classicists might agree that most transmissions of Greek and Latin texts from Classical Antiquity are so complex and rich, so full of contamination and so cloaked by long periods of transmissional obscurity, that in all likelihood few if any them is in truth amenable to stemmatic reconstruction of the archetype. This situation applies even more to the great majority of Buddhist Sanskrit texts. Likewise, it seems that most (and possibly all) Kanjur texts so far edited have similar conditions, so that the Maas-style system of stemmatic reconstruction of archetypal readings cannot be unproblematically applied to them either, as Harrison, Eimer, Skilling, Silk and others have pointed out. Thus, these authors have sometimes used stemmatics only for examining variants, and figuring out the stemmatic relations among witnesses, identifying hypearchetypes, and so on, although others might try to stemmatically reconstruct multiple archetypes. Jonathan Silk (1994), for example, uses stemmatics to reconstruct the archetypes of both Tibetan translations of the Heart Sūtra, his Recensions A and B, although he notes complexities in the case of Recension A.
But we submit that at least some of the rNying ma tantra transmission is fundamentally different. After critically editing five rNying ma Mahāyoga tantras, each one in an exhaustive manner designed to reconnoitre and explore this hitherto unknown territory in as much detail as possible, we became struck by how fundamentally the recension of these five Mahāyoga texts differed from the Kanjur findings so far. Firstly, their recensions were closed, rather than open. Secondly, when compared to the Kanjur findings so far, their transmissions were neither rich nor complex. In the case of the Thabs zhags, with its Dunhuang witness, we avoid long periods of obscurity. Even when, as with the Thabs zhags, a wide variety of witnesses were available that spanned huge expanses of time and geography, their transmissions in fact seemed so narrow and so simple by comparison to the Kanjur, and their differences so much transmissional rather than recensional, that we began to suspect wholesale reconstruction of archetypal readings might even be applicable. I am reasonably convinced that Pasquali himself would have agreed, for he was a neo-Lachmannian, not an anti-Lachmannian. He did not reject stemmatic reconstruction per se, but rather pointed out that it could only really work under particular conditions. Since we believe its transmission does enjoy precisely such conditions, the Thabs zhags, extant in so many witnesses from different times and places, became our second test bed for this theory. Our first test bed, using the same methods and for the same reasons, was the edition of the Kīlaya Nirvāṇa Tantra (Cantwell and Mayer 2007), which was received with critical acclaim from Leonard van der Kuijp, amongst others (Van der Kuijp 2010: 448-449).
Cathy Cantwell and Robert Mayer, 2006. “Two Proposals for Critically Editing the Texts of the rNying ma’i rGyud ‘bum”, Revue d’Etudes Tibétaines, Vol.10, April 2006: 56-70
Cathy Cantwell and Robert Mayer, 2007. The Kīlaya Nirvāṇa Tantra and the Vajra Wrath Tantra: two texts from the Ancient Tantra Collection, Vienna, The Austrian Academy of Sciences Press.
Leonard van der Kuijp, 2010 “Faulty Transmissions: Some Notes on Tibetan Textual Criticism and the Impact of Xylography”, in Chayet, Anne (et al., eds), Edition, éditions. l’écrit au Tibet, évolution et devenir. Indus Verlag, München.
Per Kvaerne, Compte-rendu de “Cathy Cantwell, Robert Mayer: A Noble Noose of Methods, The Lotus Garland Synopsis: A Mahâyoga Tantra and its Commentary”, Revue d’Etudes Tibétaines, no. 28, Octobre 2013, pp. 127-129.
Paul Maas, Textual Criticism (Oxford: Clarendon, 1958), 49; trans. of Textkritik (3rd ed.; Leipzig: Teubner, 1957; 1st ed., 1927).
Giacomella Orofino, 1994. Sekoddeśa. A critical edition of the Tibetan Translations. Rome, IsMEO.
Giacomella Orofino, 2007. “From Archaeological Discovery to Text Analysis: the Khor chags Monastery Findings and the Mañjuśrīnāmasaṃgīti Fragment”, in Discoveries in Western Tibet and the Western Himalayas, eds Amy Heller and Giacomella Orofino. PIATS 2003: Tibetan Studies: Proceedings of the Tenth Seminar of the International Association for Tibetan Studies, Oxford, 2003. Brill, Leiden.
Jonathan Silk 1994 The Heart Sūtra in Tibetan: A Critical Edition of the Two Recensions Contained in the Kanjur. Arbeitskreis für Tibetische und Buddhistische Studien, Universität Wien.
Christian K. Wedemeyer, 2010. Review of The Kīlaya Nirvāṇa Tantra and the Vajra Wrath Tantra: two texts from the Ancient Tantra Collection, by Cathy Cantwell and Robert Mayer. Religious Studies Review, Volume 36, Issue 1, March 2010, Page: 101.
 Van der Kuijp 2010: 448, here referring especially to Cantwell and Mayer 2007, which was made using identical technical methods and theoretical approaches as Cantwell and Mayer 2013.
 Religious Studies Review Volume 36, Issue 1, March 2010, Page: 101,
 Kvaerne 2013: 128, 129
 Personal communication, 8/10/2012