Articles by alphabetic order
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
 Ā Ī Ñ Ś Ū Ö Ō
1 2 3 4 5 6 7 8 9 0


The Early Days of the Great Perfection II

From Tibetan Buddhist Encyclopedia
Revision as of 17:34, 30 June 2021 by VTao (talk | contribs) (Created page with "thumb SAM VAN SCHAIK 1. The origins of the Great Perfection The origins of the Great Perfection {{Wiki|movement}}, so important in the l...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
480 n.jpg





SAM VAN SCHAIK


1. The origins of the Great Perfection


The origins of the Great Perfection movement, so important in the later Tibetan tradition, have proved very difficult for modern scholarship to establish. The genuinely early texts available to scholars are like the few remaining pieces of what was once a large puzzle. Inquiries into the early history of the Great Perfection have, of necessity, been rather like plac- ing these pieces into an arrangement that merely suggests the larger whole.

Because of the scarcity of pieces, a certain amount of guesswork has had to be employed in their arrangement. Here I hope to add some more of the puzzle’s lost pieces, rearrange the existing pieces somewhat, and pro- duce an impression of the original whole. This arrangement will inevitably be refined or thoroughly reshuffled in the future, as further pieces are found.1 The earliest Great Perfection texts to be translated and made available in Tibet were those now known as the mind series (sems sde). Later devel- opments in the Great Perfection brought far more complex doctrines and practices, but the early mind series texts stayed close to one central theme: the immediate presence of the enlightened mind, and the consequent use- lessness of any practice that is aimed at creating, cultivating or uncover- ing the enlightened state. David Germano has coined the useful phrase “pristine Great Perfection” to refer to this kind of discourse. The largest and most well known of these texts is the Kun byed rgyal po’i mdo, in which one finds a rejection of the elaborate imagery and practices asso- ciated with the tantras.


1 I would like to thank Jacob Dalton, Matthew Kapstein and Tsuguhito Takeuchi for sharing their unpublished work with me, which contributed significantly to the develop- ment of this paper. Harunaga Isaacson provided some very useful references to the Indic tantric literature, and Jacob Dalton pointed out many significant Tibetan sources. As early as the ninth century, there was a recognizable form of the Great Perfection with much in common with the mind series. This is proved by certain Tibetan manuscripts found in the Dunhuang library cave, walled up in the early eleventh century. In the collection of these manuscripts at the British Library, Samten Karmay discovered two texts which looked very much like mind series literature: ITJ 594 and 647.2 The first of these texts has the title sBas pa’i rgum chung; it includes a preamble in which the author is identified as Buddhagupta, and the category of the text as Atiyoga. The second text, Rig pa’i khu byug, employs terminology famil- iar in the early Great Perfection.3 These two finds seemed to confirm that the mind series, as we know it now, is a fair representation of the kind of thing that was being called rdzogs chen in the ninth and tenth centuries. Karmay, however, suspected that the Great Perfection might have been more intimately linked in this early period with meditation techniques focussed on deities and their ma∞∂alas known as the development stage (bskyed rim) and sexual practices known as the perfection stage (rdzogs rim), in other words, the whole milieu of texts and practices known to the tradition as Mahayoga. Karmay looked at a text from the bsTan ‘gyur called Man ngag lta ba’i phreng ba, attributed to Padmasambhava, which discusses the practices of the deity yoga from the standpoint of the Great Perfection.4 David Germano has also argued that the early Great Perfec- tion derived in part from Mahayoga, and in part from a strand of thought represented by the early mind series texts. As evidence for the Mahayoga influence he cited passages in the one of the fundamental scriptures of the Mahayoga, the Guhyagarbha tantra, which employ the term rdzogs chen in relation to the idea of the immediate presence of enlightenment.5


2 The Tibetan Dunhuang manuscripts in the British Library’s Stein collection have the prefix IOL Tib J (an identification which has changed somewhat over the years, and largely been ignored by scholars who usually prefer to refer to the manuscripts with the prefix “Stein Tibetan”). The corresponding manuscripts in the Bibliothèque nationale de France’s Pelliot collection have the prefix Pelliot tibétain. Here, I have abbreviated the former to ITJ and the latter to PT. 3 Karmay 1988, chapter 2. ITJ 647 text appears as the first in the list of eighteen as the Rig pa’i khu byug. In the Kun byed rgyal po this text has the title rDo rje tshig drug. As Kar- may has shown, the six lines of the root text also appear scattered throughout other scriptures from the corpus of Atiyoga. 4 Karmay 1988, chapter 6. 5 Germano 1998, pp. 212-218. Thus both Karmay and Germano have suggested that the Great Perfec- tion developed through the intermingling of the literature of two traditions.

The first of these is the pristine ritual-free discourse of the authors of the earliest mind series texts, siddha-style yogic practitioners. The second is the elaboration of the Great Perfection as the culmination of ritual practice by commentators on the Mahayoga tantras. In fact, as I will argue below, although we do find these two kinds of literature in the early days of the Great Perfection, this does not in fact entail the existence of two separate traditions. In the following pages I trace the evolution of the rdzogs chen term itself, and the parallel evolution of “Atiyoga”, the scriptural category and so-called vehicle (theg pa) which came to be synonymous with the Great Perfection. Most of the sources for this discussion are texts from the library cave at Dunhuang, which date from before the cave was sealed at the beginning of the eleventh century. Though available to scholars for the past century, most have never been studied before. I also examine the work of two Tibetan writers involved in the creation of the Great Perfec- tion in Tibet, the ninth-century tantric exegete gNyan dPal dbyangs and the tenth-century redactor of teachings gNubs chen Sangs rgyas ye shes.

Through these enquiries I hope to shed some light into the obscure regions in which the Great Perfection scriptures were created.6 2. What is the Great Perfection? (i) rDzogs chen as a ritual moment As far as we are aware, the earliest appearance of the term rdzogs chen being used in a similar way to the Great Perfection literature, is in the Guh- yagarbha tantra. The term rdzogs chen seems to be used in the tantra in asso- ciation with a specific ritual moment, the state of being at the climax of the sexual yoga of the perfection stage immediately following consecration


6 In the following pages, I use the phraseGreat Perfection” to refer to the tradition, and the Tibetan rdzogs chen to refer to the term itself. Furthermore, I refer to the early texts of the Great Perfection as “the mind series” (a translation of sems sde). Although this usage is anachronistic in that sems sde itself does not appear in any pre-eleventh century text (in some early texts the classification is sems phyogs, but this term has also not been found in any pre-eleventh century text), I have employed “the mind series” as a useful label for this group of texts and the early form of Great Perfection discourse that they contain. with the drop of semen or bodhicitta.7 In this context, the word rdzogs chen could certainly be interpreted to have the semantic content of a great (chen) culmination of the perfection (rdzogs) stage. This usage occurs in chapter thirteen of the tantra, spoken by the Tathagata from the state of sexual union, and in chapter fourteen, which is a further poetic discourse on that state. Furthermore, chapter nineteen, which deals with the com- mitments (samaya) associated with the perfection stage yoga, also uses the term rdzogs chen. The use of the term in the sixth chapter is more general, speaking of the yogin who realizes the great perfection; yet on the evi- dence of the other occurrences of the term, this realization would be under- stood to come about through the practice of the perfection stage.

Given that rdzogs chen is closely associated in the Guhyagarbha tantra with the ritual moment of the culmination of perfection stage yoga, the question of what it signifies remains. In general, the significance seems to differ little from later Great Perfection traditions: all qualities (yon tan) and enlightened activities (’phrin las) — that is, the aims of the Bud- dhist practitioner — are complete (rdzogs) from the start (ye nas). That is to say, in another phrase that is used in the tantra far more often, every- thing is spontaneously present (lhun gyis grub).8 Furthermore, there is an emphasis on the transcendence of concepts in a state beyond the reach of thought (bsam gyis mi khyab). In spite of the association of rdzogs chen with these ideas, so familiar from the later Great Perfection texts, the phrase itself occurs only four times in the tantra, and is certainly not the defining term for this complex of ideas that it later became.

Certain texts preserved in the Dunhuang collections confirm that the term rdzogs chen was actually used in practice in the context of the ritual moment of consecration.9 For example, in PT 321, a sadhana based around a Heruka ma∞∂ala, following self-consecration and the offering of the bodhicitta to the ma∞∂ala of deities, the text mentions the ma∞∂ala of the secret great perfection (rdzogs pa chen po gsang ba’i dkyil ‘khor), which is associated with the purity of all phenomena.10


7 Dalton 2004. 8 For translations of the passages referred to here, see Germano 1994, pp. 214-215. 9 Dalton 2004. 10 PT 321, f.16r. Another manuscript, ITJ 437, a treatise on the development and per- fection stages incorporating material from the Guhyasamaja tantra and the Vajram®ta

Another piece of evidence for the association between the term rdzogs chen and the Guhyagarbha in this period is provided by an untitled poem from the Dunhuang manuscripts, PT 322B.11 This text, which has not been noticed before, takes rdzogs chen as its theme while remaining within the frame of reference of the Guhyagarbha and Mayajala tantras.12 On reading it, one feels that the term rdzogs chen has begun to represent the complex of ideas surrounding it, as it does in the later tradition. Never- theless, the author the setting of these ideas is clearly the universe of the Mayajala tantras, as the following verses demonstrate: The teaching of the primordial, spontaneously present great perfection, This sublime experiential domain of supreme insight Is bestowed as a precept upon those with intelligence; I pay homage to the definitive counsel spoken thus.

Without centre or periphery, neither one nor many, The ma∞∂ala that transcends thought and cannot be expressed, Illuminates the mind of intrinsic awareness, wisdom and knowledge; I pay homage to the great Vajrasattva. From the illusory three worlds [like] the limitless sky, Many millions of emanations are present everywhere, Surrounded by the net of insight in the expanse of sameness, I pay homage to you, the magical net (Mayajala). The ten directions and the four times secretly have the nature of the great perfection, Which itself is the suchness of the definitive essence, Primordial and spontaneously present, cause and effect inseparable, I pay homage to the supreme secret nucleus (Guhyagarbha).13


tantra, refers to the bodhicitta substance as “the great perfection, the great self, the heart nec- tar” (f.14r: rdzogs cen [sic] bdag nyid chen po thugs kyi bcud). 11 The hand seen in this manuscript is identical to several others in the Dunhuang col- lection. A group of four manuscripts in the same hand are a syncretic explication of Chan and Mahayoga meditation practices (see van Schaik and Dalton 2004). Thus the scribe of PT 322B was certainly making use of Chan texts as well, although this is not apparent in PT 322B itself. 12 The term Mayajala tantra can be used to specifically denote those tantras with Maya- jala in their title, such as the Vajrasattva mayajala tantra, or a more general group including the Guhyagarbha. The texts under examination here use the term in the latter sense. 13 PT 322B, f.1r: ye nas lhun grub rdzogs pa chen po’i chos// shes rab mchog gi spyod yul dam pa de// blo ldan rnams la man ngag sbyin pa’i phyir// gros kyi nges don ‘di skad phyag ‘tshal bstod// mtha’ dang dbus myed gcig dang du mar bral// bsam ‘das brjod myed

(ii) rDzogs chen as a framework for all tantric practice

The earliest known commentary on the Guhyagarbha tantra is the Rin po che spar khab by the Indian author Vilasavajra, was probably written in the 770s.14 The commentary does not give any special precedence to the term rdzogs chen, and does not employ it in any specific technical sense.15 By contrast, the only other known Indic commentary on the whole tantra, the rGya cher ‘grel ba of Suryaprabhasasiµha, places far more weight on the term rdzogs chen, and uses it far more frequently.16 The status of this commentary is rather uncertain because of the author's obscurity, but we will have reason to return to Suryaprabhasasiµha later, since he may in fact have been closely involved in the development of the Great Perfection at the beginning of the ninth century. The Man ngag lta ba’i phreng ba is a treatise on chapter thirteen of the Guhyagarbha attributed to Padmasambhava. If the attribution is cor- rect, then the text would probably date from before or during Padmasamb- hava’s sojourn in Tibet in the 770s.17 While the text touches obliquely on the actual practices, the commentary primarily develops the ideas of spon- taneous accomplishment and primordial purity as the experiential climax of the practices. The author articulates the status of rdzogs chen as the


rdo rje dkyil ‘khor ni// rang rig ye shes mkhyen pa’i thugs la gsal// rdo rje sems dpa’ che la phyag ‘tshal bstod// nam ka’i mthas klags snang srid sgyu ma las// sprul pa’ bye ba mang po ci yang gda’// shes rab dra bas mnyam nyid dbyings su ‘khril// sgyu ‘phrul dra ba khyod la phyag tshal bstod// phyogs bcu dus bzhi rdzogs chen rang bzhin gsang// de nyid snying po nges pa’i de kho na// ye nas lhun grub rgyu ‘bras dbyer myed pa’i// gsang ba’i snying po mchog la phyag ‘tshal bstod// 14 This text (P.4718) is known in the canons merely as the gSang ba’i snying po’i ‘grel; the name Rin po che’i dpar khab appears in the colophon, and the text is known to the rNying ma tradition merely as the sPar khab. See Davidson 1981 pp. 6-7, where the nam- ing and dating of Vilasavajra (Tib. sGegs pa’i rdo rje) are discussed. Vilasavajra is often referred to in modern scholarship as Lilavajra. 15 See for instance P.4718: 272-5-

16 The text is the dPal gsang ba’i snying po’i rgya cher bshad pa’i ‘grel pa, P.4719 (vol.83). The differences in the use of rdzogs chen between this text and the sPar khab become obvious if one compares the treatment of the Guhyagarbha's thirteenth chapter in the two commentaries. 17 The text is found in the bsTan ‘gyur (P.4726) under the title Man ngag gi rgyal po lta ba’i ‘phreng ba. As shown by Karmay, the text dates at least to the time of Rong zom (fl. 12th c.) who wrote a commentary upon it (Karmay 1988, pp. 137-138). The transcription in Karmay 1988 is taken from this, rather than the canonical version.

culmination of the three ways (tshul) of inner yogic practice: the ways of development (bskyed), perfection (rdzogs), and great perfection (rdzogs chen).18 These three ways are subdivisions of the vehicle of inner yoga (rnal ‘byor nang ba), and not considered to be vehicles in their own right. Thus rdzogs chen is rooted in the practices prescribed by the tantra: the visualization of deities and the experience of bliss through sexual union.

It primarily functions as an interpretive framework for these experiences: The way of the great perfection is to realize that all phenomena of saµsara and nirva∞a are inseparable and have always had the nature of the ma∞∂ala of body, speech and mind, and then to meditate on that [[[realization]]].19 One feels on reading these texts that there is some unresolved tension — why practice deity yoga at all if the deity and ma∞∂ala are already spontaneously present? The existence of an early Tibetan treatise that addresses precisely these questions is evidence that the Tibetan recipients of these teachings felt the need to resolve this very tension. The title of the treatise is rDo rje sems dpa’ zhus lan (Questions and Answers on Vaj- rasattva) and it was written by an early figure in the Tibetan Mahayoga lineages called gNyan dPal dbyangs. We may provisionally date the writ- ing of the Zhus lan to the early ninth century.20


18 This use of rdzogs chen as the culmination of the development and perfection stages is also found in the exegtical tradition on the Guhyasamaja tantra, through Buddhajñana- pada’s commentary on this tantra (P.2716, vol.65, 10-1-2): If one does not realize this unchanging realization [of the perfection stage], One is not practising what is called “Mahayoga” Once one has fully understood it, This is the great perfection, the embodiment of wisdom. ‘di ni rtogs par mi ‘gyur ‘di don ma rtogs na /rnal ‘byor chen po zhes bya de la mi bya'o /de ltar rab tu shes par byas nas su /rdzogs pa chen po ye shes spyi yi gzugs/ However, as the term rdzogs chen is not found in the Guhyasamaja, its appearance here may represent a contamination from contemporary traditions of Guhyagarbha exe- gesis. 19 Man ngag lta ba’i ‘phreng ba, p. 7: rdzogs pa chen po’i tshul ni/ ‘jig rten dang ‘jig rten las ‘das pa’i chos thams cad dbyer med par sku gsung thugs kyi dkyil ‘khor gyi rang bzhin ye nas yin par rtogs nas sgom pa ste/ 20 Versions of the Zhus lan exists in the bsTan ‘gyur and in three different Dunhuang manuscripts. Three manuscript items of the text exist in the Stein and Pelliot collections of Dunhuang manuscripts. PT 837 and ITJ 470 are almost identical, and the latter appears to be a copy of the former. The third, PT 819, which is not complete, differs from the other two, and is generally closer to the version found in the bsTan ‘gyur (P.5082, Narthang rgyud ‘grel vol. ru, ff.121a-127a). These canonical verions contain mchan ‘grel notations through- Using a question-and-answer format, the Zhus lan deals explicitly with problematic issues in the application of a Great Perfection-style view to the practices of Mahayoga. So, for example, the following Great Perfection- style statement is made: This mind itself which is without basis or root Is, like the sky, not purified by cleansing.

Because enlightenment is free from production, Enlightenment does not come from cause and effect.21 And the question that very understandably follows is, how then do the blessings of the deity come about? The answer comes as a simile: When dirty water becomes clear, No effort is required for the reflections of the sun and moon to appear. Similarly, if one transforms one's own mind through yoga, No accomplishment is required for the conquerors' blessings to arise.22 The agenda here looks very similar to that of the Man ngag lta ba’i phreng ba, but expanded so that the Great Perfection approach is applied to all yogic practice, not just to the culmination of the perfection stage. The Zhus lan is also a step onwards from the Man ngag lta ba’i phreng ba in that it deals with the conflicts and problems arising from the impo- sition of this framework. The author is keen to get the message across that the practice of deity yoga is emphatically not to be abandoned, but any


out, which are not found in PT 819 and differ completely from the notation in PT 837 and ITJ 470. Orthographic features of the manuscripts indicate that PT 837 and ITJ 470 date from the post-imperial period (mid-ninth to late tenth centuries), while PT 819 may be some- what earlier. The later date of PT 837 and ITJ 470 is also suggested by the fact that the differences this version shows from the version in PT 819 are usually extra syllables in the verses, which often clarify or elaborate the meaning of the lines, but render them hyper- metrical. Thus they may be explanatory notes which have been incorporated into the main text at some point. In any case, the existence of these two variant editions of the text by the tenth century suggests that the text had been in circulation for some time at this point.

In the colophon of the Zhus lan found in PT 819 and ITJ 470, dPal dbyangs states that he wrote this explication of Mahayoga for the sNa nam and lDong clans. Though interesting, this statement is not immediately helpful in dating the work. The Zhus lan was held in high regard by gNubs Sangs rgyas ye shes, who cites it several times in his bSam gtan mig sgron: pp. 30.3, 201.6, 219.3, 255.6, and 277.3. 21 ITJ 470, section 7: /gzhi rtsa myed pa'i sems nyid ni/ /bkrus pas myi dag nam ka bzhin/ /skyed dang bral ba'i byang chub la/ /byang chub rgyu ‘bras yongs kyis myed/ 22 ITJ 470, section 12: /ci ltar chu la rnyog pa dangs par ‘gyur na/ /gnyI zla'i gzugs brnyan ‘byung ba brtsal myi dgos pa bzhin/ /rang sems rnal ‘byor dag par sgyur pa na/ /rgyal concept of the practices as a cause for enlightenment, or of the deities as separate from one’s own primordially enlightenment mind, is to be aban- doned.

The Zhus lan constantly refers to nonduality, freedom from effort, and the primordial and spontaneous presence of the enlightened mind, using terminology that is central to the mind series of the Great Perfection, such as rig pa, byang chub sems and lhun gyis grub. The term rdzogs chen itself appears just once, in an answer to a question — if there is no cause and effect, how does the yogin obtain his accomplishments? The answer is couched in another simile: When, as in the example of a king appointing a minister, The accomplishments are granted from above, this is the outer way. When the kingdom is ruled having been offered by the people, This is the way of the unsurpassable, self-arisen great perfection (rdzogs chen).23

This political metaphor is fascinating in itself, but the verse is most sig- nificant for the implication it makes that the audience for this text should consider itself to be following the way of rdzogs chen while practising Mahayoga. Here, as in the Man ngag lta ba’i ‘phreng ba, rdzogs chen is a “way” (tshul), a word suggesting a mode of practice rather than a sep- arate set of practices. It is certainly not considered a vehicle (see section vi below). According to this verse, deity yoga without the concepts of cause and effect and without the idea that one gains accomplishment from else- where is the way of rdzogs chen. Thus rdzogs chen is certainly not some- thing intrinsically separate from deity yoga.

The Zhus lan is not an isolated case, for we have another text from the Dunhuang collection that, though it does not use the term rdzogs chen, applies this Great Perfection-style framework to the practices of the tantras. The manuscript (ITJ 454) is without a title or a colophon, but what we have of it is a reasonably long section — ten scroll panels — which deals with various topics arising out of deity yoga. These include a discussion of different kinds of yoga, of the result of yogic practice, of the true nature of meditation on the deity, of mind and the phenomena of saµsara and


ba'i byin brlabs ‘byung ba bsgrub mi dgos/ 23 ITJ 470, section 9: /dper na rgyal pos blon por bskos pa ltar na/ /grub pa gong nas byin ba phyi'i tshul lo/ /'bangs kyis rgyal ba'i srid phul nas dbang bsgyur ltar/ /rang ‘byung nirva∞a, of pure perception, and the oneness of conventional and ulti- mate truth. The opening lines of the manuscript show clearly the influence of Great Perfection-style discourse on this treatise: It does not matter whether all of the phenomena of mind and mental appear- ances, affliction and enlightenment, are understood or not. At this very moment, without accomplishing it through a path or fabricating it with anti- dotes, one should remain in the spontaneous presence of the body, speech, and mind of primordial buddhahood.24

This is the perspective from which the practice of deity yoga is addressed. Thus it is said that one’s deity is simply the awareness (rig pa) of one’s own enlightened mind (byang chub sems). As for the Buddha, the concept is explained in the following way: One's own mind is primordial purity and buddhahood, and to comprehend that this mind is primordially purity and buddhahood is to be accomplished as a buddha, to see the face of the buddha, to hold the buddha in one's hands.25

Such statements occur again and again, while the practicalities of medi- tation itself are barely addressed. Yet the sphere of reference here is cer- tainly the tantras, and not just the Guhyagarbha. Seven tantras are cited, all of which are found in the later lists of eighteen Mahayoga tantras (in some cases, this is the earliest reference to these tantras yet seen).26


rdzogs chen bla na med pa'i tshul/ 24 ITJ 454, panel 1: /sems dang sems snang ba'i chos thams cad dam/ /kun nas nyons mongs pa dang/ rnam par byang ba'i chos thams cad rtogs kyang rung ma rtogs kyang rung/ ‘phral la lam gyis ma bsgrub gnyen po ma bcos te/ ye nas sangs rgyas pa sku gsung thugs lhun kyis [sic] grub par gnas pa la bya/ 25 ITJ 454, panel 4: yang na rang gyi sems ye nas rnam par dag cing sangs rgyas pa yin dang/ sems ye nas rnam par dag cing sangs rgyas pa yin pa'i don rtogs pa ni sangs rgyas su grub pa ‘am/ sangs rgyas kyi zhal mthong ba ‘am/ sangs rgyas lag tu ‘ongs zin pa yin/ 26 The cited tantras are: 1. Sangs rgyas thams cad dang mnyam par sbyor ba’i rgyud (Skt. Sarvabuddha- samayoga tantra): P.8, P.9, Tb.402-404.

2. dBang chen bsdus pa’i rgyud: Tb.595. 3. dPal mchog dang po’i rgyud: Tb.412. 4. gSang ba ‘dus pa’i rgyud (Skt. Guhyasamaja tantra): P.81, Tb.409. 5. Ri bo’i [rtsegs pa’i] rgyud: Tb.411. 6. gCig las phrin pa’i rgyud [=gCig las ‘phros pa’i rgyud]: cf. P.2032. 7. U pa ya pa sha (Skt. Upayapasa tantra): P.458, Tb.416.

(These comprise numbers 3, 4, 12, 9, 7, 5, and 16 of the list of eighteen tantras set out in Thus in this text we see the broadening of the Great Perfection approach to tantric practice, away from reliance on one tantric tradition (the Maya- jala) towards a more general framework for all tantric practice in Maha- yoga. (iii) rDzogs chen free from yogic practice At some point the term rdzogs chen achieved an independence from the practices of Mahayoga. This is clear when we look at the mind series texts, most of which contain no references to the practices of development and perfection, or merely negative references stating that the Great Per- fection transcends such practices. One of the Dunhuang texts, ITJ 647, is a commentary on one of these mind series texts, the Rig pa’i khu byug.

We need only look at two aspects of the text, which has already been discussed in detail by Karmay. Firstly, it does employ the term rdzogs chen — once again as “way” (tshul) — in the concluding part of the text, where the subject matter is “the way of the Great Perfection view and practice of yoga.”27 Secondly, and here we begin to see the shift away from the texts of the previous two sections, there is an assertion of independence from the development and perfection stages: Because of the spontaneous accomplishment of ineffable bodhicitta, the ma∞∂ala is established spontaneously and abides without artifice, with no need for the activities of development and perfection.28

In some places this text is still concerned with interpreting tantric prac- tice — it offers an interpretation of the ritual practices of killing (sgrol ba) and sexual union (sbyor ba) in Great Perfection style framework, this is, in terms of non-conceptualization. Thus even while it grapples with tantric terminology, the text asserts a pristine form of non-practice. Yet a few scribbled notes in another hand on the last folio of the manuscript (which were not mentioned by Karmay in his study) suggest that this non-prac- tice may be more rhetoric than reality. The notes, which are very faded


Martin 1987, pp. 178-182. 27 ITJ 647, f.5r: rnal ‘byor gyi lta spyod rdzogs pa chen po’i tshul/ 28 ITJ 647, f.5v: brjod pa dang bral ba’i byang chub sems lhun gyis grub pas/ bskyed rdzogs bya mi dgos par dkyil ‘khor gdod [for gdon] pa lhun gyis gnas pa la ma bcos par and entirely blank in places, give the name of the text (Rig pa’i khu byug), which is called a “statement” (lung), and follow with the statement “if you wish to hear these sweet words…” There follows the hundred-sylla- ble mantra, written out in a form not quite the same as that which is known the later Tibetan traditions. That the recitation of the mantra is required for those who wish to receive this statement is implied if not stated out- right. The passage also contains an exhortation to “strive, day and night.”29 The Dunhuang manuscripts may yet have more to tell us about the actual practices which coalesced around this Great Perfection style rhetoric.

A brief sadhana attributed to Mañjusrimitra (ITJ 331/1) involving merely the visualization of the syllable huµ bears some resemblance to later developments in Great Perfection meditation in the Seminal Heart (snying thig) traditions.30 In any case, whether the dismissal of tantric practices in this manuscript was taken as rhetorical or literal by its scribes and readers, there is no doubt that the whole discussion of practice is couched in the language of rejection. This is the point at which it becomes possible to consider rdzogs chen as an independent approach, not only as the culmination of the per- fection stage, or a framework for all yogic practice.

(iv) rDzogs chen as a textual category

In all of the previous examples, rdzogs chen is a “way” (tshul) of prac- ticing yoga. Even in the case of the commentary on the Rig pa’i khu byug, there is no indication that the author considers there to be such a thing as a Great Perfection text as such. The author does not classify the root text with the appellation tantra, sutra, nor even the less prestigious “statement” (lung) or “precept” (man ngag). Thus there is a vagueness about the status of this text, and we do not know whether it was considered to be scripture or not.

This picture is changed by another Dunhuang text, which has previously gone unnoticed, that refers to “the sutras of the Great Perfection”. The gnas pa’o/ 29 ITJ 647, f.5v: //nang nub brtson par bya’o/ snyan thos padod na// (the hundred- syllable mantra follows). 30 This text is briefly discussed in Eastman 1983, pp. 52-53, where Eastman raises the manuscript is split between the Stein collection in London and the Pel- liot collection in Paris (the first two pages are PT 353 and the remaining nine are ITJ 507). The bulk of the text is a description of two different kinds of ma∞∂ala, inhabited primarily by various kinds of female deity.

The goddess Ekaja†i performs a central role in both ma∞∂alas, which is interesting given the significance of this deity to the later Great Perfection traditions.

The introduction to the text states that the teaching found within derives “from the tantras of the secret mantra and from the sutras of the Great Perfection”.31 This does suggest that for the author of this text, rdzogs chen indicated a category of texts with the status of scripture. It is certainly not unknown, even in the later tradition, for Great Perfection texts to be termed sutra: while most Great Perfection scriptures of the later tradition were called tantras, several mind series texts, including the Kun byed rgyal po, are still identified by their titles as sutras.32 None of these texts are mentioned in gNubs chen’s bSam gtan mig sgron (STMG), as far as I can tell, so they may represent a strand of the Great Perfection that was unknown to gNubs chen, or that developed after STMG was written.

The main part of PT 353 / ITJ 507 is a fairly straightforward ma∞∂ala visualization, and the presence of Great Perfection-style elements is lim- ited to the following lines in a brief introduction to the practice:33 All phenomena are contained in the dharmakaya. The dharmakaya is itself contained in the jñanakaya. Due to the compassion of the jñanakaya, all


question of whether the text might be considered an Atiyoga sadhana. 31 PT 353, f.1r: zhi cing rgya cher ni gsang ba’i sngags tan tra rdzogs pa chen po’i mdo las zhib tu ‘chad de/ 32 The rNying ma rgyud ‘bum contains ten Great Perfection texts apart from the Kun byed rgyal po which are entitled sutras (Tb.7-16). These are not necessarily the texts referred to by PT 353, of course. 33 This section of the text seems to be structured along the standard Mahayoga division into three concentrations: the suchness concentration (de bzhin nyid kyi ting nge ‘dzin), (ii) the all-illuminating concentration (kun tu snang gi ting nge ‘dzin), and (iii) the causal concentration (rgyu’i ting nge ‘dzin). The first generally involves a dissolution of all appear- ances in emptiness, the second the emergance of a moon disc embodying compassion for all sentient beings, and the third the appearance of a seed syllable on the disc, representing the deity. In this text, compassion is said to pervade all beings the the three realms, which is perhaps no more than an ordinary Mahayana aspiration. Nevertheless, it finds an echo in the later Great Perfection snying thig texts, in which the primordial basis (gzhi) is defined according to its essence (ngo bo), nature (rang bzhin) and compassion (thugs rje). The last sentient beings are pervaded by compassion, so that it arises in each of their own minds. The three realms are pervaded by the dharmakaya, so that they are the same. When one has been blessed by this great principle, one acts free from causation.34 (v) rdzogs chen as a vehicle

The next stage in the development of the term rdzogs chen is its trans- formation into a fully fledged vehicle (theg pa), at which point it looks much like it does throughout the later tradition. The earliest evidence of the Great Perfection being considered not just a scriptural category but a vehicle is in the bSam gtan mig sgron (STMG) of gNubs chen Sangs rgyas ye shes, probably written in the first quarter of the tenth century.35 gNubs chen refers to “the vehicle of the activity-free Great Perfection”.36 In the STMG, rdzogs chen has become synonymous with Atiyoga, and is no longer applicable in the context of Mahayoga. While gNubs chen’s approach to the practices of Mahayoga has absorbed much of the com- mentarial tradition discussed above, such as the Zhus lan, he never applies the term rdzogs chen in that context. In an echo of the Rig pa’i khu byug commentary, gNubs chen states that the Great Perfection is free from the development and perfection stages.37 In short, we first see the term rdzogs chen appearing sporadically in the Guhyagarbha tantra, which is thought to have been circulating in India by the mid-eighth century. The term is used to indicate the high level of realization that results from the sexual yoga of the perfection stage. This is supported by ritual texts from Dunhuang that use the term in the context of these practices. While the earliest dateable commentary makes little use of the term, by the late eighth century it was being given a much more detailed treatment by Guhyagarbha exegetes such as Padmasambhava.


item, compassion, is frequently characterised as “all pervasive” (kun khyab). 34 Pelliot tibétain 353, f.2r: chos thams cad ni chos kyi sku ‘dus/ /chos kyi sku yang ye shes kyi skur ‘dus/ /ye shes kyi sku’i thugs rjes/ /sems cam mtha’ dag la thugs rje des khyab par rang rang gyi sems la shar te/ /chos kyi sku des khams gsum khyab ste mnyam pa nyid du/ dgongs pa chen pos byin kyis brlabs nas/ /rgyu dang bral bar mdzad do/. 35 The dates of gNubs chen, and of the composition of the bSam gtab mig sgron, are discussed in Vitali 1996, pp. 546-547. 36 STMG p. 312: rdzogs pa chen po bya ba dang bral ba'i theg pa. In these works, rdzogs chen was the culmination of the development and perfection stage, the realization that all phenomena have always been the pure state of the deity and the ma∞∂ala.

Later, in the early ninth century, Tibetans began to write interpretive works on tantric practice. In the rDo rje sems dpa’ zhus lan, gNyan dPal dbyangs applied the ideas of Padmasambhava’s work to the practice of tantra in general, and in this context continued the use of the term rdzogs chen. As we will see below, at this same time texts were being written that looked very much like the early mind series texts. At some point, per- haps simultaneously with the writing of the tantric treatises, yet perhaps not until the late ninth century, these texts picked up the term rdzogs chen and applied it to their approach, which was very similar to that of the Man ngag lta ba’i phreng ba and the Zhus lan except in that it made no concession to the actual practice of development and perfection stages.

By the beginning of the tenth century, there was a body of texts associ- ated with the term rdzogs chen, as we see from STMG, and our gnomic reference from PT 353 to “the sutras of the Great Perfection”. And around the same time as the writing of PT 353, or perhaps even earlier, we see gNubs chen writing about rdzogs chen as a vehicle. It seems quite possible that the existence of a recognised scriptural corpus of Great Perfection lit- erature and the conferring of vehicle status upon the Great Perfection occurred at around the same time; I return to the question of gNubs chen’s involvement in this process below.

3. The development of Atiyoga (i) Atiyoga as a stage in practising yoga The earliest stages in the development of the usage of the term Atiyoga are to be seen in certain Indic tantras. The earliest of these is probably the Sarvabuddhasamayoga tantra, one of the earliest of the so-called yogini tantras, probably dating from the eighth century.38 In a verse from


37 STMG p. 360. 38 The existence of several texts comprising and related to this tantra was first noted in Davidson 1981, p. 58, p. 75, n.43. A great step forward in understanding the relation- ship between the texts is made in Weinberger 2003, pp. 260-268. My citations are from the bKa’ ‘gyur version, which is classed a “subsequent tantra” (phyi ma’i rgyud). The exis- the second chapter of the tantra, which exists in a Sanskrit citation, as well as in Tibetan, Atiyoga is compared to Anuyoga: Through Anuyoga the bliss of all yogas is practised, And through Atiyoga the true nature is fully experienced.39

Here, there appears to be some association of Anuyoga with the bliss of the sexual practices, and Atiyoga with a realization of the nature of real- ity. The context of this verse is a discussion of stages in ritual practice, which begins with Yoga, and then proceeds through Anuyoga and Ati- yoga. Thus there is a basis here for the way the term Atiyoga is used in the later tradition.40 The K®Ò∞ayamari tantra, which is thought to be later than the Sarvabud- dhasamayoga, discusses a fourfold set, Yoga, Anuyoga, Atiyoga and Mahayoga, as the four stages of yogic practice. Although the tantra goes on to define Atiyoga as the completion of all ma∞∂alas, it appears only as the penultimate stage, below Mahayoga.41 This may in fact be the same system intended in the Sarvabuddhasamayoga where, in the passage from the tantra’s second chapter discussed above, there is an implication that there may be a stage above Atiyoga, which is explicitly defined as


tence of the root tantra is difficult to determine, but see Weinberger’s analysis. 39 P.8 (vol.1) 184-4-7: rjes su sbyor bas mchod byed cing/ rnal ‘byor kun gyi bde ba dag/ bdag nyid kun tu myang byed na/ shin tu sbyor bas ‘grub par ‘gyur. The Sanskrit text of this verse is to be found in in the ninth chapter of Aryadeva’s Caryamelapakapradipa, kindly brought to my attention by Harunaga Isaacson: pujyate ‘nuyogena sarvayogasukhani tu/ samasvadayamanas tu atiyogena siddhyati//. 40 There are other occurances in the tantra of the Tibetan term (shin tu rnal ‘byor ba) which translates Atiyoga in the above passage, but without the Sanskrit we must bear in mind the possiblity that another term may have been translated in these instances. A very similar verse to the one quoted above appears in the tantra’s third chapter (185-1-5). Also in the third chapter is a verse that mentions Atiyoga alongside Mahayoga, in association with the supreme samaya (dam tshig mchog).

P.8 (vol.1) 184-5-7: de yi dam tshig de byung ba/ sdug pa’i dam tshig bdag rjes ‘brang/ shin tu sbyor dang sbyor chen po/ thams cad bdag nyid dam tshig mchog/.

In the sixth chapter a practitioner of Atiyoga (shin tu rnal ‘byor pa) is mentioned; the passage, which is difficult to interpret, seems to deal with sexual yogic practices.

P.8 (vol.1) 188-5-8: gzhan yang shin tu rnal ‘byor pa/ ma lus pad mo rnam sbyang phyir/ skal bzang dam pa mi ‘gyur pa/ pad me sgyu ma ‘byung bar gyur/ 41 K®Ò∞ayamari tantra 17.8, p. 123: bhavayed yogam anuyogaµ dvitiyakam/ atiyogam t®tiyam tu mahayogam caturthakam//. Tibetan in P.103 (vol.4) 16-4-1ff: dang por sgom pa rnal ‘byor te/ gnyis pa rjes kyi rnal Mahayoga in one commentary.42 In any case, both tantras use the names of the various levels of yoga to signify different ways of engaging in the tantra’s practices. On the whole the way the Sarvabuddhasamayoga treats Atiyoga looks more consistent with the later usage of Atiyoga in Tibet, and may be one of the direct sources for this usage.

The Guhyagarbha tantra does not use the term Atiyoga, but the early commentary on the tantra mentioned above, the sPar khab of Vilasavajra, mentions Atiyoga as the highest of all approaches to tantric practice: Atiyoga, in which it is taught that even the two inner [[[yogas]]: in this case, Yoga and Mahayoga] and abiding in the natural fruition of all objects are merely the obscurations of the various stages of attachment to imputa- tions.43

In all of these texts, the different yogas stand for stages in, or approaches to, tantric practice. The significance of Atiyoga is not quite clear but there seems to be a consistent theme, beginning with the Sarvabuddhasamayoga in the eighth century and picked up by the sPar khab, of Atiyoga as a reali- zation that cuts to the true meaning of reality. If we turn again to the Dunhuang manuscripts to look for early appearances of Atiyoga in Tibet, we find that these references are in fact rather late. The term appears in one of the interlinear notes to the main text of the rDo rje sems dpa’ zhus lan. Yet these notes appear only in a version of the text that probably dates to the late tenth century. The note in ques- tion is appended to a point in the main text where the following answer is given to the question of how one should perform the propitiation and accomplishment (bsnyen bsgrub) of the deity: When in ultimate propitiation and accomplishment no subject or object is per- ceived, Because there are no difficulties or effort this is the supreme propitiation and accomplishment.44


‘byor yin/ gsum pa shin tu rnal ‘byor te/ bzhi pa rnal ‘byor chen po'o/. 42 Srisarvabuddhasamayoga-∂aki∞ijalasaµvaratantrarthika (303b.6-304a.3) by rGya sbyin ldong po. Thanks to Jacob Dalton for pointing out this reference. 43 The passage is in chapter thirteen of the commentary, Peking bsTan ‘gyur, rgyud ‘grel Bu 186b: nang pa gnyis po dang/ dngos po thams cad kyi rang bzhin ’bras bur gnas kyang/ brtags pa la zhen pa'i rim pa sna tshogs kyi bsgrib pa tsam du ston pa'i a ti yo ga'o/. 44 ITJ 470, section 13: /bsnyen pa don dam par bya ba dang byed pa myi dmyigs na/

The note indicates this being without difficulties or efforts to be “an explanation of the view of Atiyoga.” (a ti yo ga’i lta ba’i bzhed). There is too little here to guess whether Atiyoga indicates anything more than a view that can be applied to the practice of deity yoga.


(ii) Atiyoga as a distinct approach


Several Dunhuang texts, of which only one has been examined, are pro- totypes, or variations of the nine vehicle system that later became nor- mative for the rNying ma traditions. One of these, PT 849, is already well known to contain a list of nine vehicles.45 However, Atiyoga occurs there only as a subdivision of the vehicle of Yoga (in the sequence Yoga, Mahayoga, Anuyoga, Atiyoga) and is not itself a vehicle. This manu- script has been dated to the late tenth or even early eleventh century, and yet there is little development here beyond the earlier uses of Atiyoga in India. Strangely, a Dunhuang text that does in fact describe the set of nine vehicles just as they are found the rNying ma tradition has previously been overlooked. In this text (ITJ 644) none of the nine are referred to as vehicles, meaning that as in the previous manuscript, Atiyoga is present, but not as a vehicle. The nine are discussed in terms of their deity line- ages (lha rgyud), which implies that each is a different way of practising deity yoga. The deity lineage of Atiyoga, which is given a definition almost identical to Anuyoga, is said to be identical to the view of Atiyoga. The practice of Atiyoga (and again this is shared with Anuyoga) is defined as spontaneous presence. Despite the conspicuous absence here of the wordvehicle,” the presence of the set of nine in a form identical to the later and definitive nine vehicle set is striking.46


/tshegs dang ‘bad pa myed pas bsnyen pa'i mchog go/. 45 This text is transcribed, translated into French and discussed in Hackin 1924. It is also discussed in Karmay 1988 pp. 148-149 (where it is incorrectly identified as PT 489). The dating of the text to the tenth century is discussed in Takeuchi unpub. 46 ITJ 644 cannot be dated with such certainty, but paleographic features and its orig- inal location in a bundle of texts mostly from this period, indicate a post-imperial date. The relevant lines on Anuyoga and Atiyoga are as follows (ITJ 644, f.1v): /a nu yo ga’i lta ba dang lha rgyud ni gcig ste/ khyad par myed do/ grub tshad ni/ yong nas lhun kyis grub par ‘dod do/ a ti yo ga’i lta ba dang lha rgyud gcig ste khyad par myed/ grub tshad ni/ lhun

For our purposes the most interesting Dunhuang text is another over- looked gem, called Teaching the views and activities of the seven great general statements (sPyi’i lung chen po bdun so so’i lta spyod pa bstan pa:

PT 656). As the title suggests, this text sets out seven approaches: the nine vehicles of the later tradition without the Pratyekabuddha and Upaya yoga stages. Each stage is a system of “general statements” (spyi’i lung) rather than a vehicle. Each approach is defined in turn according to its view, medi- tation, activities, and pledges, before we come to the seventh, Atiyoga: The view of the Atiyoga secret mantra is to view [everything] as the ma∞∂ala that is ornamented with the inexhaustible body, speech and mind. Everything that manifests — all sentient and non-sentient beings from the hell beings to those who possess the omniscient wisdom of all the buddhas and bhaga- vans, and the physical objects of the outer world which is their container — and their manifestation as forms and colours, is the body. Their being uncre- ated is the speech. Their nirvanic nonduality is the mind…. The samaya is just this one: openness…. Why is it called openness? Because of spontaneous accomplishment, we do not say: “If you maintain this, you will be accomplished, and if you do not maintain it, you will not be accom- plished.”47

In this treatment, Atiyoga constitutes a distinct approach, based on the ideas of nonduality and spontaneous accomplishment, yet still predicated on the terminology of Mahayoga, its ma∞∂alas and samayas.48 This is also seen in a passage which comes later in the text, where liberation and union (sbyor sgrol), two terms associated in the Mahayoga tantras with killing and sexual intercourse, are given an Atiyoga interpretation:


gyis grub par lta’o. 47 PT 656, l.33ff: gsang sngags a ti yo ga ni/ /sku gsang[sic] thugs myi zad rgyan gi ‘khor lor lta ba ste// de yang sangs rgyas bcom ldan ‘das thams cad khyen pa'i ye shes can dang/ sems can dmyal ba yan cad sems yod pa dang/ sems myed pa snod gyi ‘jig rten phyi'i yul bems por snang ba thams cad/ la/ kha dog dang sbyibs su snang ba ni/ sku// de nyid ma skyes pa ni gsung// //gnyi ga mye ngan las ‘das pa ni thugs so/ […] //dam tshig ni gcig ste phyal lo// […] ci'i phyir phyal zhes na/ lhun gyis grub pa'i phyir na bsrungs pas/ grub la ma srungs pas myi grubs ces bya ba myed pas// 48 The discussion of the Atiyoga interpretation of samaya is also found in a, presum- ably Indic, commentary to the Hevajra tantra by *Viravajra (dPa’ bo rdo rje): rGyud thams cad kyi gleng gzhi dang gsang chen dpal kun tu kha sbyor zhes bya ba’i rgyud kyi rgyal po’i rgya cher bshad pa rin chen phreng ba zhes bya (P.2329). This interpretation is in terms

of freedom from concepts, and from “contaminants” (zag). See 277-2-6ff. The persistence of the term atiyoga in later developments of Indic tantra is a potentially interesting topic

In the Atiyoga mantra, union is the union of space and wisdom. My con- sciousness is wisdom and all of the objects that are cut through by that [[[wisdom]]] are space. Liberation is being liberated by great sameness…. Therefore there is not even so much as the name sentient being. In reality, it does not matter whether one practises or not.49 Here we find Atiyoga defined, in distinction to other approaches includ- ing Mahayoga and Anuyoga, in terms of radical non-conceptualization, nonduality, and non-activity. While it is still referred to as “Atiyoga secret mantra”, firmly placing it in the Vajrayana fold, it is defined so that it has its own apparently self-sufficient view, meditation, activities, and vows. In the latter case, the independence from the vows of the previous approaches is made quite explicit. Thus we see here a fully realized definition of Atiyoga as a distinct approach, if not a vehicle per se.

Whether the definition of Atiyoga as distinct from Mahayoga meant that they were incommensurable in practice is difficult to determine, but another Dunhuang text, PT 44, suggests that in some circles they were con- sidered quite compatible. The text contains a difficult but very interesting passage on the application of the Atiyoga approach to a Mahayoga prac- tice, in this case, Vajrakilaya.50 The passage, in which Padmasambhava teaches his followers certain precepts after they have exhibited signs of the accomplishment of Vajrakilaya, is as follows: He taught the secret bodhicitta that is included within Atiyoga, and the sad- hanas of Vajrakilaya in accordance with the Mahayoga texts. After he showed that meditation on Vajrakilaya is the dharmadhatu, they meditated on the nonduality of objects and minds within the unborn bodhicitta. […] According to the Byi to da ma la (in Sanskrit) and the Rig pa mchog gi rgyud (in Tibetan), during the sadhana one meditates upon the secret bodhicitta.51 In the terminology of the ritual this is called “the vajra youth.” It teaches this as the authentic statement, Atiyoga. And it teaches that the state after


which I have not had the opportunity to investigate here. 49 PT 656, l.56ff: sngags a ti yo ga'i sbyor ba ni/ dbyings dang ye shes su sbyor ba'o/ bdag kyis rnam par shes pa ni/ ye shes des bcad pa'i yul thams cad na dbyings so// bsgrol ba ni mnyam pa chen pos bsgrol ba'o// […] sems can zhes bya ba'i mying yang myed pa la bya// dngos su na spyod kyang rung/ ma spyad kyang rung ngo// 50 This text has been translated in Bischoff & Hartman 1971, not very satisfactorily.

The translation here is my own attempt. A discussion of the text, and an improved trans- lation of its first half appears in Kapstein 2000. 51 This is a reference to Rig pa mchog gi rgyud (skt. *Vidyottama tantra), P.402. See liberation is the aim of the ritual. And one should meditate on the lack of difference between bodhicitta and great bliss. According to the precepts on this, there are said to be four [kinds of] great statement. Although they can be distinguished according to their particular features, all of the correct state- ments are said to be Atiyoga.52

The manuscript is about as late as a Dunhuang text can be: Tsuguhito Takeuchi has recently dated the writing of the text to between 978 and circa 1010, when the cave was sealed.53 Despite the difficulty of the passage, and there are several elements of the translation above that are uncertain, we can see that Atiyoga here refers to a teaching on the nature of mind, identifying it with wisdom, and referring to this as bodhicitta.54 Therefore this Atiyoga closely resembles the Atiyoga of the mind series texts. In this manuscript Mahayoga and Atiyoga seem to be highly com- patible, with Atiyoga meditation being applicable to the practice of Maha- yoga deities like Vajrakilaya. Furthermore, Atiyoga is intimately linked with the key Mahayoga rite of “liberation” (sgrol ba). Though the term for the sexual practices, “union” (sbyor ba), is not mentioned here, the reference to “great bliss” clearly implies a sexual yogic practice. As in the Spyi’i lung chen po bdun passage quoted above, Atiyoga is intimately conjoined with these key Mahayoga ritual concepts. While that was a reading of ideas of sameness and nonconceptuality onto the term libera- tion and union, here in PT 44, there is at least an implication of the way in which practitioners of Mahayoga combine their ritual practice with the nondual and nonconceptual space of Atiyoga.


Bischoff & Hartman 1971, pp. 12-13. 52 PT 44, ff.13-19: gsang ba'i byang cub kyis sems a ti yo gar gtogs pa dang // phur bu'i bsgrub thabs ma ha yo ga'i gzhung bzhin bstan pa phur ba'i bsgom pa chos kyis dbyings su gsal nas // byang cub kyis sems bskyed par ma byas kyis bar du // […] rgya gar skad du byi to da ma la'o // bod skad du rig pa mchog kyi rgyud to / bsgrub thabs su byang cub kyis sems gsang bar bsgom mo // las kyi skul tshig du rdo rje gzhon nu zhes bya'o // khungs lung a ti yo gar bstan pa dang // bsgral nas gnas kyis cho ga'i gza gtad gang yin bar bstan pa dang // byang chub kyis sems bde ba chen po las myi dams par bsgom mo / de ltar man ngag kyis don du lung chen po bzhir bstan no // khyad bar gyis so sor phye nas kyang // yang dag pa'i lung thams cad kyang a ti yo gar bstan to // rdzogs sto // || // 53 Takeuchi unpub. 54 Although in PT 44 the Atiyoga teachings are given by Padmasambhava to his dis- ciples, there is no attempt to credit him as their originator, as there often is in the later treas- This text shows that in the late tenth century, though there is a distinc- tion between the texts of Atiyoga, called “authentic statements” (khungs lung)55 and the texts of Mahayoga, there was certainly a compatibility, even an interdependence, between the practices of Mahayoga and the meditation on the true nature of mind in Atiyoga.


(iii) Atiyoga as a textual category


There is just one Dunhuang text that identifies itself as belonging to the category (phyogs) of Atiyoga; it is the sBas pa’i rgum chung (ITJ 594), one of the Great Perfection texts studied by Karmay that looks like an early mind series text, although it is not found in any of the surviving col- lections. The words that place the text itself within the category of Atiyoga occur in an introduction to the text written on the page before the text proper begins. In similar language to PT 44, the introduction also states that the text is derived from “the statements on bodhicitta” (byang chub sems kyi lung). The introduction is probably a later supplement added by the writer of the manuscript, and the orthography of the manuscript itself suggests, again, a tenth-century date. The text itself is ascribed to a Sangs rgyas sbas pa. The identity of this figure, and his relationship to the author known as Buddhaguhya or Bud- dhagupta in the bsTan ‘gyur is uncertain, and the question is complicated by the fact that the names Buddhagupta and Buddhaguhya (as well as their Tibetan equivalents Sangs rgyas sbas pa and Sangs rgyas gsang ba) seem to have been used interchangeably.56 The existence of a figure known both as Buddhaguhya and Buddhagupta as the author of the a number of Yoga tantra treatises is quite certain. And there is early evidence for the existence of a Buddhagupta who was the author of works on Mahayoga tantras. STMG mentions a Buddhagupta in connection with both Maha- yoga and Atiyoga.57 A previously unnoticed Dunhuang mentions the name


ure (gter ma) tradition. 55 The distinction between four kinds of statement (lung) is intriguing, but difficult to link to divisions of Atiyoga preserved in the later tradition. 56 This question of identification is discussed in Karmay 1988, pp. 61-63. See David- son 2002, pp. 153-159, and Hodge 2003, pp. 22-24, for discussions of the Buddhaguhya/ Buddhagupta who wrote the Yoga tantra commentaries.

Buddhagupta in connection with the a supreme essential teaching of the Vajrayana, alongside the names Srimañju (Mañjusrimitra?) and Huµkara, who are both strongly associated with the Mahayoga tradition.58 The later rNying ma histories identify a single Buddhaguhya as the author of com- mentaries on Yoga tantra, Mahayoga and Atiyoga. According to traditional histories, he was a student of Buddhajñanapada and Vilasavajra, both of whom were familiar with the term rdzogs chen, as we saw earlier.59 Thus there are good reasons to believe that the author of the sBas pa’i rgum chung really was the Mahayoga exegrete Buddhagupta (who may be identical with the Yoga tantra exegete of the same name). But whoever the author of the text really was, ITJ 594 is evidence that by the tenth cen- tury the term Atiyoga had begun to signify the textual classification of a type of discourse very similar to those texts which survived in the later Tibetan tradition as the mind series.60


(iv) Atiyoga as a vehicle


We have already seen that the first dateable use of the term rdzogs chen as a vehicle (theg pa) is in STMG, and this is also the case with Atiyoga. In fact, these two terms, the development of which, as we have seen, followed parallel yet distinct trajectories, have become entirely syn- onymous in STMG.61 It was Atiyoga that came to be most closely asso- ciated with vehicle status in the rNying ma tradition, in the structure of the nine vehicles. The nine-vehicle structure was known to gNubs chen, the author of STMG, since he discusses it in his Mun pa’i go cha, a com- mentary on the dGongs pa ‘dus pa’i mdo.62 gNubs chen, who was closely


57 STMG pp. 204, 223, 344, 414. 58 ITJ 1774: slob pon ni ‘bu ta kub ta dang/ shi ri man ‘ju dang/ hung ka ra 59 See Gu bkra’i chos byung p. 100-101 and Dudjom 1991, vol.1, pp. 465-466. 60 Another case of the use of the term Atiyoga appears in PT 699, a commentary on a Chan text ascribed to Bodhidharma. The commentary ends with a discussion of the char- acteristics of a master of the sutras, a master of the tantras, and a master of Atiyoga. A trans- lation of this passage, and a study of the commentary, which brings a Mahayoga structure to the Chan text, appear in van Schaik and Dalton 2004. The root text exists in two other Dunhuang manuscripts, ITJ 689/1 and ITJ 1774. Interestingly, in ITJ 1774 the Chan root text is followed by a fragment (which I have mentioned briefly already) about a teaching of the secret mantra which includes the name of Buddhagupta. 61 See for instance p. 376. involved in the promulgation of this text in Tibet, may be one of the first writers to have described the nine-vehicle system in detail.

Although as we saw above there are certain Dunhuang texts that appear to be early instances of the nine-vehicle system, none of these actually refers to Atiyoga as a vehicle. The closest they come to such a defini- tion is to include Atiyoga as one in a series of “statements” (lung).

Therefore, gNubs chen remains the earliest known exponent of the nine vehicles, as vehicles per se. This places the first appearance of Atiyoga as a vehicle within gNubs chen’s lifetime, in the late ninth and early tenth centuries.63 We should note here that there is a text attributed to sKa ba dPal brt- segs in the bsTan ‘gyur, entitled The Instructions / Teachings on the Stages of the View (lTa ba rim pa’i man ngag or lTa ba rim pa’i bshad pa).64

This text refers to Mahayoga, Anuyoga and Atiyoga as vehicles, and describes the general approach of Atiyoga in some detail. Though unlikely, it is possible that Mahayoga, Anuyoga and Atiyoga were designated vehi- cles in a late eighth century text, but not again until the early tenth cen- tury. However, I am inclined to doubt the attribution of this text to dPal brtsegs and therefore its status as an eighth-century text. gNubs chen quotes dPal brtsegs several times in STMG, but these passages, including one that mentions Atiyoga, are not from the lTa ba rim pa’i man ngag.65 One would not expect gNubs chen to overlook a source from a well-known author which offered a venerable tradition for his assertion that Atiyoga was a vehicle. Furthermore, as Karmay has noticed, Bu ston, though including this text in the bsTan ‘gyur along with the other texts attributed to dPal brtsegs, was inclined to doubt its authenticity.66 To sum up, the term Atiyoga first appears in certain Indic tantras, per- haps the earliest of which is the Sarvabuddhasamayoga. Atiyoga appears in association with Mahayoga and Anuyoga; all three are ways of prac- tising the deity visualization and sexual yoga described in the tantra.


62 See Dalton 2002. 63 On the dating of gNubs chen’s works see Vitali 1996. 64 P.4728 and 5843. 65 STMG 132, 151, 153, 406. The citation in the Atiyoga section actually refers to a sKol mdo lo tsa ba dPal brtsegs, who may be a different figure entirely (STMG p. 406).

On the Tibetan scene there is little sign of Atiyoga until the beginning of the tenth century. Perhaps in the intervening years it was rendered redun- dant by the popularity of the term rdzogs chen, used to refer to the highest level of yogic practice in the Mayajala tantras. The earliest appearances of Atiyoga in the Dunhuang texts may well be later than the works of gNubs chen (early tenth-century), in which Atiyoga is granted the full status of a vehicle. Nevertheless, the various uses of Atiyoga within the Dunhuang texts discussed above suggest that the place which it holds in STMG (as a vehicle and textual category) remained only one of many meanings of the term throughout the tenth century.

The question of whether Atiyoga should be considered a vehicle in itself continued to be an issue within the Tibetan tradition, and as late as the thirteenth century its status as such was disputed by Sa skya Pa∞∂ita (1182- 1251) in his sDom gsum rab dbye. By this time, Atiyoga was well estab- lished as a scriptural category in rNying ma systems, yet Sa pa∞ seems to have considered this an inauthentic innovation.

In the secret mantra of the new schools, The superiority of the great yoga67 Is that there is no class of tantras superior to it. There is also no object of meditation Higher than the great yoga. The wisdom that arises from it Is unelaborated and inexpressible; Therefore it is not a stage in the vehicles. If one understands this tradition properly, Then the view of Atiyoga too Is wisdom and not a vehicle; To make the inexpressible an object of expression Is not the intention of scholars — thus it is said.68


66 Karmay 1988, p. 149. 67 Sa pa∞, perhaps to make the analogy between Mahayoga (rnal ‘byor chen po) and *Anuttarayoga (rnal ‘byor bla na med pa) clearer, refers to the latter as rnal ‘byor chen po here. Neverthless, it is clear that the higher tantras of the new schools are signified. 68 sDom gsum rab dbye, Sa skya bka’ bum Vol.V, 312-1-2: /des na gsang sngags gsar ma la/ /rnal ‘byor chen po'i lhag na ni/ /de bas lhag pa'i rgyud sde med/ /bsgom pa'i dmigs pa nyid kyang ni/ /rnal ‘byor chen po'i gong na med/ /de las skyes pa'i ye shes ni/ /spros pa med cing brjod bral bas/ /theg pa rim pa mi bzhed do/ /lugs ‘di legs par shes gyur na/ /a ti yo ga'i lta ba yang/ /ye shes yin gyi theg pa min/ /brjod bral brjod byar byas pa ni/

In the opinion of Sa pa∞, Atiyoga is a wisdom arising out of the prac- tice of the development and perfections stages, and not a vehicle, though it may be designated as a view. The last lines look like a reference to Atiyoga being made the special subject of treatises, like the statements and precepts we have been examining in the Dunhuang manuscripts. Sa pa∞ accepted the usage of Atiyoga in tantras such as the Sarvabuddhasama- yoga that were considered to be genuine by the Sa skya, and rejected the later developments of texts devoted to Atiyoga alone, and the development of Atiyoga into a vehicle for these texts. Sa pa∞ probably considered these to be Tibetan innovations, and on the strength of the evidence reviewed above, he may well have been right. 4. gNyan dPal dbyangs, gNubs chen, and the Great Perfection scriptures (i) dPal dbyangs and the six lamps Now that we have seen how the terms rdzogs chen and Atiyoga first developed in Tibet, I would like to offer a theory on the way the Great Perfection texts themselves came into being. At the centre of this theory is the obscure figure of gNyan dPal dbyangs, whom we have already come across as the author of the rDo rje sems dpa’ zhus lan. I will take dPal dbyangs as emblematic of the original creators of the earliest Great Perfection texts. This approach has its limitations of course, for we are only dealing with a single figure who may be exceptional rather than typical, but as we will see, the conclusions drawn from the case of dPal dbyangs are at least in agreement with the sources reviewed above. dPal dbyangs is significant because he is a named figure from the early ninth century, writing texts that look like the early mind series texts, which, because they make no reference to any Great Perfection scriptures or vehicle of Atiyoga, seem to come from the same period and milieu as the early mind series texts. There are six texts attributed to the dPal dbyangs in the bsTan ‘gyur, known to the tradition as The Six Lamps (sgron ma drug). They are the following: 1. Thugs kyi sgron ma (P.5918) 2. rNal ‘byor spyod pa’i lugs nges pa’i don la ji bzhin sgom thabs kyi sgron ma or Man ngag rgum chung (P.5922)


3. mTha’i mun sel sgron ma or lTa ba rgum chung (P.5920) 4. lTa ba yang dag sgron ma (P.5919) 5. Thabs dang shes rab sgron ma (P.5921) 6. lTa ba rin po che sgron ma (P.5923)69 The importance of these texts has been recognised by Guiseppe Tucci and Samten Karmay, yet the implications of their existence have not been fully worked out.70 The first of the texts, Thugs kyi sgron ma, is by far the longest, and differs from the other five in being written as a treatise on Mahayoga (and the Mayajala tantras in particular) and its relationship to other modes of practice. Yet there is little discussion of actual practices except to undercut their basis in conceptual structures and on the whole the emphasis is on nonduality and the spontaneous presence of wisdom.71 This text is therefore similar in approach to the rDo rje sems dpa’ zhus lan in its application of the doctrines of nonduality and spontaneous pres- ence to the practice of development and perfection stage yoga.

The other five texts, all under two folios in length, are discourses in the style of the early mind series texts.72 The term dzogs chen itself does not appear in them at all, though we know that dPal dbyangs was famil- iar with the term, which he used in the Zhus lan. The doctrine of the texts is referred to in other words, such as “the yoga of one’s own mind” (rang sems rnal ‘byor).73 If dPal dbyang’s works were written at a time when Great Perfection texts with scriptural status were in circulation, one would expect some reference to them somewhere, yet no such reference is made. The only reference to a textual tradition is to an unspecified group of statements (lung) and precepts (man ngag) in the third verse of the lTa ba yang dag sgron ma:

Non-conceptualization, the nature of ultimate truth, The great way of awareness — if one focuses on it with the mind, THE EARLY DAYS OF THE GREAT PERFECTION 191 /mkhas pa'i dgongs pa min zhes bya/. 69 Nos.1, 2 and 4 also appear in the bKa’ ma shin tu rgyas pa, vol.86, pp. 283-340. 70 Tucci 1958, pp. 141-150; Karmay 1988, pp. 59-69. 71 For example, the discussion of prayer (smon) in 232-4-4 72 The statement in Karmay 1988, p. 66 that all except nos.2 and 3 are on the doctrines of the Guhyagarbha is incorrect. One will achieve resolution through awareness with the statements and pre- cepts, And become convinced of the correct nature of phenomena.74

dPal dbyangs then goes on to give an introduction into a Great Perfec- tion-style view. He places this text in the tradition of statements and pre- cepts, a tradition of authored texts rather than canonical scripture. These six lamps might then serve as a paradigm for the way the early Great Perfection scriptures came into being. This of course depends on several other factors, including the dates of their composition, their relationship to other figures and texts in the early Great Perfection tradition, and their history in the centuries after their composition. So, to start with, we need to look a little more closely at gNyan dPal dbyangs himself.

Samten Karmay has covered most of the scarce sources available on this figure.75 In the later rNying ma tradition he is barely remembered, figuring only as one of the students of gNyags Jñanakumara, himself said to have been a follower of Padmasambhava.76 Earlier sources are barely more informative, although they do place dPal dbyangs in Great Perfec- tion lineages. In Nyang ral Nyi ma ‘od zer’s history of Buddhism, gNyan dPal dbyangs occurs in certain enumerations of teachers (a tsa rya) along with Vairocana and gNyags Jñanakumara. The Blue Annals is slightly more informative: regarding the “mind class” (sems phyogs), the text indicates that the source of the teaching is a Buddhagupta (sangs rgyas gsang ba) whose disciple was Vimalamitra. It goes on to tell how gNyags Jñanakumara received teachings from Vimalamitra, Vairocana and g.Yu sgra snying po, and that one of the students of gNyags Jñanakumara was a sNyan dPal dbyangs.77 Additionally, the Blue Annals also mentions the transmission of the six lamps in the twelfth century.78 Several lines from the sBas pa’i rgum chung, the Dunhuang text attrib- uted to Buddhagupta, are found in two of six lamps (nos. 2 and 3 which


73 lTa ba rin po che sgron ma, l.53. 74 lTa ba yang dag sgron ma ll.9-16: rnam pa mi rtog don dam rang bzhin gyi /rig pa'i tshul chen sems kyis dmigs ba na /lung dang man ngag rig pas thag gcod te/ chos kyi rang bzhin yang dag yid ches bya/. 75 Karmay 1988, pp. 66-69. Karmay shows that gNyan dPal dbyangs is clearly not the same as the sBa’ dPal dbyangs who was the second abbot of bSam yas. 76 See Dudjom 1991, vol.1, pp. 601-605.

77 Blue Annals p. 170. have alternative titles also ending with rgum chung) as well as in the Zhus lan. Thus dPal dbyangs appears to have been incorporating into his own writings lines from a work he considered a statement or precept by an earlier figure in his teaching lineage, Buddhagupta.79 However, the most direct influence on dPal dbyangs may in fact have been a hitherto unsuspected Indian tantric exegete. Something near to the name of this figure appears in one of the six lamps themselves. In the opening verses to the lTa ba yang dag sgron ma, in which dPal dbyangs sets out his reasons for writing the text, he pays homage to his teacher: Due to my teacher dPal ldan thugs kyi dkyil ‘khor nyi ma ‘od, Blazing rays of light came forth in my mind's darkness. A fraction of this true nature of the intrinsic profound, unborn mind, Transcending songs and words, transcending the characteristic of the sky, manifested.80

Although the word ston has meanings other than “teacher,” the use of the honorific dpal ldan (skt. sri) strongly suggests that this is indeed a person. The best candidate for the identity of the teacher dPal dbyangs mentions here is a Nyi ma’i ‘od kyi seng ge (*Suryaprabhasasiµha) who appears as a minor figure in some histories. A long commentary on the Guhyagarbha tantra mentioned earlier is ascribed to this Suryaprabha- sasiµha. The commentary is held to have been translated into Tibetan by Vairocana, though this is problematic.81 Suryaprabhasasiµha’s students are said to have included gNyags Jñanakumara and Jñanakumara’s own student Sog po dPal gyi ye shes.82 This ties in rather well with dPal dbyang’s status as a student of Jñanakumara. If one of Jñanakumara’s students was able to receive teachings from Suryaprabhasasiµha then


78 Blue Annals, p. 659. 79 In theory, the incorporation could have been the other way around, as the Dunhuang manuscript may be no earlier than the end of the tenth century. But because the early evi- dence for the existence of a Buddhagupta involved in Mahayoga that we reviewed earlier is quite strong, it seems likely that this figure was an influence on the work of dPal dbyangs. 80 lTa ba yang dag sgron ma ll.1-8: dpal ldan thugs kyi dkyil ‘khor nyi ma'i ‘od ston gis/ shin tu ‘bar ba'i zer tsam bdag blo mun bar byung/ dbyangs dang tshig ‘das namkha'i mtshan nyid las ‘das pa'i/ rang sems skyed med zab mo'i don ‘di phyogs tsam snang/. 81 Dudjom Rinpoche takes this event as the starting point of the rNying ma lineages in Khams (Dudjom 1991, vol.1, p. 687). However, the colophons of the canonical editions do not give Vairochana as the translator, but a figure called Padma ru tshe. A discussion of this commentary and a summary of the first chapter appears in Martin 1987.

dPal dbyangs, another student, could also have had the opportunity to do so. The attribution of the long commentary on the Guhyagarbha to Surya- prabhasasiµha needs to be further analysed, but it is interesting to com- pare this text to those we have been examining. The long commentary is far closer to the approach of dPal dbyangs than to the only other early commentary on the Guhyagarbha, the sPar khab. The whole of the text is shot through with the concepts of spontaneous accomplishment, non- duality, and bodhicitta as the nature of all phenomena. Like dPal dbyangs in the Zhus lan, Suryaprabhasasiµha is concerned with questions arising out of the application of these approaches to the practices of Mahayoga.83 It is a pity that the accounts of this teacher are from much later histo- ries, yet the fact that the works of dPal dbyangs and Suryaprabhasasiµha employ a similar approach to the Mahayoga material goes some way to support the traditional account that Suryaprabhasasiµha was living and writing towards the end of the eighth century through to the early ninth. And while the verse cited above does not provide a clear identification, we can at least accept the possibility that Suryaprabhasasiµha was the teacher of dPal dbyangs.84 The picture that I would like to suggest here, based on dPal dbyangs and the attribution of the sBas pa’i sgum chung to Buddhagupta, is that within certain lineages of tantric exegesis, especially those focussing on the Guhyagarbha, it became customary by the late eighth century to write texts, presented as statements (lung) or precepts (man ngag), which elab- orated the ideas of nonduality, spontaneous presence, primordial purity and so on, removed from direct reference to the actual yogic practices derived from the tantra. Another figure in the mind series lineages, Mañ- jusrimitra, seems to confirm this picture, as he is known to have written


82 Blue Annals, pp. 108, 158. 83 For example, Suryaprabhasasiµha’s discussion of how the siddhi appear shows a con- cern with reconciling a position of nonduality and spontaneous presence with ordinary accomplishments, similar to section 12 of the Zhus lan (see dPal gsang ba’i snying po’i rgya cher bshad pa’i ‘grel pa, P.4719 (vol.83), 27-4-3ff.) 84 If Suryaprabhasasiµha was the same person as the legendary Great Perfection teacher Sri Siµha, who is associated with the early mind series literature, this would follow the pat- tern we have seen in the work of Buddhagupta and dPal dbyangs, of authors writing both Mahayoga tantric commentaries and brief, “pristine” Great Perfection texts. There is as the most important Mahayoga commentary on the Mañjusrinamasaµgiti tantra.85

Thus the early or proto- Great Perfection texts were written by the same people who were producing more conventional direct exegesis of the tantras. There were clearly two different kinds of texts being written, yet the tradition of placing these two kinds of discourse into two different textual categories, Mahayoga and Atiyoga, had yet to be developed. The model for this dual mode of literary production may have been the Guhya- garbha itself, wherein, as mentioned earlier, chaper thirteen employs the term rdzogs chen in the context of the realization arising out of the bliss of sexual union, whereas chapter fourteen employs the term in a more gen- eral yogic song on the state of ultimate realization. (ii) gNubs chen and the lamp for the eyes of contemplation

As we have seen, the first datable instance of an unambiguous use of Atiyoga as a vehicle is in the works of gNubs chen Sangs rgyas ye shes, from the late ninth and early tenth centuries. In his bSam gtan mig sgron (STMG), gNubs chen makes a strong distinction between the vehicles of Mahayoga and Atiyoga. Dealing with them in two different chapters, he quotes from a number of sources. The sources for Mahayoga are not the sources for Atiyoga, and vice versa. The texts quoted in the Atiyoga chapter include many of those classified in the later tradition as the eight- een early translations of the mind series, as well as a number of other texts found in the rNying ma rgyud ‘bum or Bai ro rgyud ‘bum. None of the passages from these texts that are cited in STMG make an explicit


yet, unfortunately, no basis for such an identification. 85 Germano 2002. p. 247 points out that three seminal Great Perfection figures were Mañjusrinamasaµgiti exegetes: dGa’ rab rdo rje, Mañjusrimitra and Vimalamitra. In this paper Germano writes insightfully on the early relationship between Mahayoga and the Great Perfection (p. 240): The origin of the Great Perfection within Mahayoga points however to their symbiotic relationship. The former offered a built in deconstruction of the latter’s own architec- tronic doctrinal and ritual complexity, as well as a mitigating influence on its empha- sis on the visual logic of deities and the wrathful logic of subjugation. Mahayoga in turn offered the Great Perfection a backdrop for its radical rhetoric of negation and the natural inherence of Buddhahood, a safety net in which it could perform its acro- batics within empty space, semantic fields that gave its own denials substance even distinction between Atiyoga and Mahayoga. We know that many mind series texts, such as the Kun byed rgyal po and the Khyung chen gi rgyud,

do make this distinction in no uncertain terms. If such passages were available, he would have been likely to have used them. So it looks like they appeared some time after the STMG, perhaps in part to address its concerns, while the authors of the earlier mind series texts were not as concerned with distinguishing the vehicle of Atiyoga from other vehicles. Although most of the texts cited in the Atiyoga chapter of STMG can be identified with texts in later collections, a number of the cited texts can- not be found. One such text, referred to by gNubs chen as rGum chung, is in fact the Man ngag rgum chung of dPal dbyangs. Although gNubs chen cites the rGum chung four times in the Atiyoga chapter of STMG, the authorship of the text is never acknowledged.86 This is in contrast to other chapters in STMG, where gNubs chen quotes from the Zhus lan and the lTa ba yang dag sgron me, mentioning dPal dbyangs as responsible for the works.87 This may be indicative of a pattern in STMG. In fact, only two of the citations in the Atiyoga chapter are attributed, one being a teaching of Lo tsa ba dPal brtsegs, and the other a text called Klong ‘grel by Vimalamitra.88 In the other cases only the short titles are given, some familiar mind series texts like the bDe ‘byams and sPyi chings, others which are not easy to identify in later collections. The identifiable texts are among the shorter mind series texts in the canon. Few of them are styled tantra.89 It is possible that, like the Man ngag rgum chung, these


under the erasure of negation. 86 STMG pp. 382, 404-405, 440. In the first instance the title is given as sBum chung. 87 The Zhus lan is cited on pp. 30, 201, 219, 255 and 277. The lTa ba yang dag sgron ma is cited on p. 49 as a sgom lung of mKhan po dPal dbyangs. 88 The dPal brtsegs texts is cited on p. 406. The Vimalamitra text is cited on p. 9 and p. 276. This is perhaps a commentary to the mTsho klong (Byang chub kyi sems rgya mtsho klong dgu’i rgyud, Tb.69), which is cited twice in the Atiyoga chapter. There is a com- mentary on the Klong drug rgyud attributed to Vimalamitra in the bKa’ ma shin tu rgyas pa (vol.100) but as the root text is one of the Seminal Heart tantras which were not in cir- culation until the eleventh century at the earliest, this is most unlikely to be the text referred to here. As a final note, there is also a mysterious text referred to in STMG as coming from the lineage of the three acaryas (pp. 424, 434). 89 These are the Sems sgom pa’i rgyud (perhaps Sems bsgom rdo la gser zhun, cited many times in STMG), Khyung chen lding ba’i rgyud, rMad du byung ba’i rgyud, Yul kun

la ‘jug pa’i rgyud, Man ngag rna ba rgyud (though the last may be a transformation of texts had previously been transmitted as statements and precepts of par- ticular teachers. The almost total lack of authorial attribution in the Ati- yoga section of the STMG suggests that gNubs chen himself may have been complicit in the activity of transforming authored texts into scripture. In fact, names of teachers are mentioned quite frequently in the Atiyoga chapter, though mainly in the interlinear notes to the text. These were apparently written by gNubs chen, unless his references to himself in the notes are a later falsification.90 The teachers are associated with different approaches to teaching the Great Perfection, rather than with specific texts. Their names are as follows: dGa’ rab rdo rjeJam dpal bshes gnyen rGyal po ‘Da’ he na ta lo Ra tsa ha ta O rgyan ma ha ra tsa Ku ku ra dza ‘Bu ta kug ta Shri seng nga / sing ha dGe slong ma Kun dga’ ma Bi ma la mi tra Bai ro tsa na A tsa ra gSal ba rgyal Eleven of the names appear in the early mind series lineage list found in the Bairo rgyud ‘bum. The remaining name, an Acarya gSal ba rgyal can be identified with one of gNubs chen’s teachers, gSal ba’i rgyan, an early author of texts based on the dGongs pa ‘dus pa’i mdo.91 The dGongs pa ‘dus pa’i mdo, later classified as Anuyoga, is clearly considered an Atiyoga text in STMG, as it is quoted more frequently than any other text in the Atiyoga section. Thus gNubs chen must have considered gSal ba’i rgyan an Atiyoga teacher. The appearance of gSal ba rgyal is fur- ther reason to believe that the interlinear notes are the work of gNubs chen.


rNa brgyud kyi man ngag, also cited in STMG). 90 STMG, pp. 375, 419. The author of the notes refers to himself as “the little monk” (ban chung), an epithet which gNubs chen uses to refer to himself in the main text. 91 This figure is generally known either as gSal ba’i rgyan or bDe ba gsal mdzad. His texts on sadhanas and empowerments based on the dGongs pa ‘dus pa’i mdo were collected The attribution of different approaches to the Great Perfection to these figures itself suggests that they may have been responsible for some lit- erary production, and that these names themselves were originally been the authorial names attached to the Great Perfection texts that gNubs chen quotes. The appearance of Buddhagupta’s name in ITJ 597 may be an example of a continuing tradition in which the authorial names were still attached to the texts, which as “statements” (lung) were accepted to be the writings of named teachers. The appearance of these names the interlinear notes of STMG as the exponents of particular approaches to the Great Perfection may have been gNubs chen’s strategy for acknowl- edging their contributions after stripping their names from the texts they created.

These Atiyoga teachers do not appear in any kind of lineage or hier- archy in STMG, and it is quite likely that the arrangement of these names into a mind series lineage seen in other texts is a later development. By the twelfth century these names, as well as a number of others from sources now unknown, had been arranged into a lineage. This can be seen in two texts from the Bai ro rgyud ‘bum, a collection of mind series texts associated with the Zur family, who are considered the inheritors of gNubs chen’s transmissions.92 The arrangement of the early Great Perfection teachers into the lineage found in the Bai ro rgyud ‘bum is awkward, and there has always been a difficulty in dealing with it in the Great Perfection tradition.93 The author of one of these texts, the Bai ro’i ‘dra ‘bag chen mo, suggests that the order of the lineage is uncertain (go rims ma nges pa).94 Later Great Perfection Seminal Heart (snying thig) traditions com- pressed the list by taking out many of the names. Perhaps the reason for the clumsy nature of the lineage list in the Bairo rgyud ‘bum is that the names were not consecutive figures in a lineage, but authors from a span of not more than two or three generations. Where gNubs chen made a space for the names as teachers of different approaches


by gNubs chen. See Dalton 2002, p. 145, n.227. 92 See Kapstein unpub., pp. 6, 9. The earlier of the two texts is the Pa’ sgrub rnams kyi thugs bcud snying gi nyi ma (vol.1, pp. 1-172) and the later is the Bai ro ‘dra bag chen mo (vol.8, pp. 442-489), a biography of Vairocana. Kapstein argues that the former text dates from the twelfth century, while the latter is a later addition to the collection. 93 See Kapstein unpub., pp. 5-6. to the Great Perfection, the author of the Bairo rgyud ‘bum text, or the tradition that he was repeating, did the same thing by placing the names in a diachronic rather than synchronic arrangement. Despite the ensuing difficulties, the lineage arrangement served as a way of preserving the names of these Great Perfection authors, although many were, in any case, forgotten by the later tradition.

Due to the paucity of sources, it is perhaps impossible to determine whether gNubs chen was repeating a received tradition or himself attempt- ing to found a tradition. There is however, reason to believe that gNubs chen was attempting to define a tradition around the names Great Per- fection and Atiyoga. Firstly, as we have seen, gNubs chen’s works are the earliest reliably dateable instance that we see Atiyoga being defined explic- itly as a vehicle (theg pa). If gNubs chen was involved in early attempts to establish a tradition, then to give the tradition vehicle status would have been highly desireable. Furthermore, the great range of texts quoted in the Atiyoga section of STMG suggests that gNubs chen may have been drawing these texts together, to make them all part of a group under the label of Atiyoga, where they were previously disparate texts from a number of lineages. The texts themselves, while sharing numerous resemblances, are by no means a homogeneous group. gNubs chen may well have faced the same diffi- culty that confronts us when examining our Dunhuang texts: while the texts bear certain resemblances to what we know in the later tradition as the Great Perfection, they are also quite different from each other. gNubs chen’s enumeration of different approaches to the view of Atiyoga sug- gests an attempt to unify as a whole a somewhat disparate group of teach- ings. His bringing together of the texts in the Atiyoga chapter of the STMG can itself be seen as a act of imposing a coherence and unity upon the texts. Yet the differences between his chapter and the accepted texts of the later tradition, such as his inclusion of the dGongs pa ‘dus pa’i mdo and several other texts that do not form part of later canons, and the conspicuous absence of the Kun byed rgyal po’i mdo, suggest a movement still in development, one that has yet to achieve a stable canon. Thus even if gNubs chen was not the first to gather together these texts and catego- rize them as belonging to a vehicle of their own, he was certainly one of the earliest to do so.


5. Conclusions

Returning to the simile I introduced at the beginning of this paper, I will now take the various puzzle pieces that have been examined alone and try to place them into an arrangement suggesting the most convinc- ing complete picture. We must begin with the Guhyagarbha tantra, since it seems that the actual term rdzogs chen was more closely associated with this text than any other. The term itself signified the realization aris- ing out of the culmination of sexual yogic practice. The tantra and cer- tain others of a group known as the Mayajala tantras were circulating in some form by the mid-eighth century, and by the 770s, commentaries were being written on them. By the 790s, certain commentaries and trea- tises began to elaborate upon the ideas of primordial purity, spontaneous presence and nonduality which played a role in the Guhyagarbha (as they do in most Mahayoga tantras). In the transmission lineages for the Guhyagarbha and its family of tantras in the late eighth and early ninth centuries there was a trend for writing short texts expounding these ideas outside of the tantra’s ritual framework. The activity of writing these texts was not considered to be inconsistent with writing exegetical works on the tantras themselves. Therefore, these earliest forerunners of the Great Perfection texts were probably written to provide an interpretive framework for the practice of the development and perfection stage practices of the Mayajala tantras. We may compare them to latter siddha literature like the mahamudra instructions of the Naropa lineage, and the *margaphala (lam ‘bras) system from the Virupa lineage. All of these systems took as their fundamental tenet the primor- dially enlightened mind, and it seems that there was a need in all of these tantric transmission lineages for an interpretive framework into which the discourse of the tantras could be placed. By the time the mahamudra and margaphala teachings were brought to Tibet, no trace was found of the forerunners of the Great Perfection texts, which has since led to doubts over their Indic heritage. Though there is little doubt that most of the texts in the canons of Great Perfection scriptures originated in Tibet, Indic models may well have existed at one time. If the Indic Great Perfection-style texts were strongly associated with the transmission lineages for the Guhyagarbha tantra, then the fact


that the Tibetans who brought back texts from India in the eleventh cen- tury failed to find any of these texts would be due the to same reason they failed to find the Guhyagarbha itself — neither would not have survived the apparent demise of these lineages in India by the eleventh century.

In Tibet, by the beginning of the ninth century, literary production of the texts later classified as the mind series was well underway. gNyan dPal dbyangs, probably following the model of his teacher Suryaprabhasasiµha, wrote treatises on Mahayoga ritual with a particular emphasis on nonduality and spontaneous presence, as well as short treatises that took the ideas of nonduality and spontaneous presence and expounded them without reference to the ritual universe of the Mahayoga tantras. The model for this latter kind of text might have been the fourteenth chapter of the Guhyagarbha, but as more and more such texts were written, many seem to have deliberately eschewed all reference to Mahayoga ritual except to declare it unnecessary. The Dunhuang manuscripts show a variation in how much such texts chose to acknowledge the ritual universe of the Guhyagarbha. Unlike the later mind series texts such as the Kun byed rgyal po, these early texts were not presented as the words of the buddha Samantabhadra. Rather they were short, pithy instructions. In concordance with this, they were known as statements (lung) or precepts (man ngag) not as sutras or tantras. Thus there is no need to posit two streams of textual creation coming together in the early Great Perfection literature. Rather, it seems that two types of composition occurred within Mayajala commentarial traditions. The heterogeneous nature of these early texts is demonstrated by the very person who attempted to bring them together as a coherent group cat- egorized as Atiyoga, gNubs chen Sangs rgyas ye shes. This term Atiyoga, which had begun in certain tantras, perhaps most importantly the Sarva- buddhasamayoga, as a stage in the practice of yoga, appears in several Dunhuang manuscripts, probably from the tenth century, in a number of enumerations of different systems of tantric practice. Although none of these manuscripts explicitly designates Atiyoga as a vehicle, in most it has come to have the same status as Mahayoga, which would entail a scriptural content. Because all of these manuscripts appear to be from around the same period as the STMG (the early tenth century), or later, it appears that gNubs chen was one of the first to designate Atiyoga as a vehicle and to associate a group of scriptures with it. Therefore gNubs


chen may well be a central point in the coalescence of the Great Perfection as a tradition. In order that these Great Perfection texts be considered scriptures, it was necessary remove from them any association with author- ship. The names of teachers in the Atiyoga chapter are generally men- tioned only in the interlinear notes, and associated with approaches to the teaching, rather than specific texts.

The names of Great Perfection teachers and the texts cited in the Atiyoga chapter of STMG correspond closely, if not exactly, with the texts and lineage found in the early mind series collection known as the Bairo rgyud ‘bum. The lineage of Great Perfection teachers found in this collection, which probably dates to the twelfth century, appears to be a construction in which the teachers whose names were preserved by gNubs chen, and no doubt in other lost texts, were placed in a diachronic list, even though as such the lineage was recognised to be untenable, and was severely truncated by later traditions. Much remains to be established regarding these early days of the Great Perfection. But if the Dunhuang texts tell us anything, it is that we cannot posit a Great Perfection tradition existing in the ninth and tenth centuries. The texts, which in varying degrees acknowledged or moved away from their basis in the Mayajala tantras, were not the homogeneous group presented in the bSam gtan mig sgron. Thus the development of the Great Perfection as a distinct tradition begins to look like the work of certain determined individuals. Future research may discover who these were, especially in the years between gNubs chen and the patriarchs of the Zur lineage. Another important area for further research is the devel- opment of the mind series literature from the earliest texts (found in the bSam gtan mig sgron citations and the Dunhuang manuscripts) to the later sutras and tantras, of which the Kun byed rgyal po is only one of many.

Their provenance is as yet unknown, but these later texts seem to reinforce the existence of a pristine Great Perfection in total distinction from Mahayoga, and thus carry on the work of gNubs chen. The Dunhuang manuscripts which we have been looking at, mostly dating from the late tenth century, show us that whatever the efforts of gNubs chen, Atiyoga and the Great Perfection continued right through to the beginning of the eleventh century to be seen as a way of approaching Mahayoga ritual and meditative practice, rather than a distinct approach.


Furthermore, not long after the latest of these Dunhuang manuscripts, we see the appearance of new trends in Great Perfection literature with the emergence of the Seminal Heart in the eleventh century. The Semi- nal Heart literature presents a pristine rhetoric interwoven with instruc- tions on actual meditative practices that absorb and transform the prac- tices of Mahayoga and later Indic traditions.95 So it appears that the “pristine” form of Great Perfection only ever existed as one strand in Great Perfection discourse, and never dominated the field. Rather than a chronological development from the pristine to the ritualized, the pris- tine discourse always existed alongside other forms of Great Perfection literature that engaged fully with structured practice, first in the form of Mahayoga and later in the specialized tantric meditations of the Seminal Heart. References Tibetan and Sanskrit Sources P. bKa’ ‘gyur and bsTan ‘gyur, Peking edition. Tb. rNying ma rgyud ‘bum, mTshams brag edition. Gu ru bkra shis

Gu bkra’i chos byung. Krung go’i bod kyi shes rig dpe skrun khang, 1990. gNyan dPal dbyangs lTa ba yang dag sgron ma: P.5919 lTa ba rin po che sgron ma: P.5923 Thabs dang shes rab sgron ma: P.5921 Thugs kyi sgron ma: P.5918 mTha’i mun sel sgron ma: P.5920 rDo rje sems dpa’ zhus lan: P.5082 rNal ‘byor spyod pa’i lugs nges pa’i don la ji bzhin sgom thabs kyi sgron ma: P.5922 gNubs chen Sangs rgyas ye shes: mDo ‘grel mun pa’i go cha. In Bdud ’joms ’Jigs bral Ye shes Rdo rje, ed. (1982). Rnying ma bka’ma rgyas pa. Kalimpong, W.B.: Dubjang Lama. Vols. 50-51. Rnal ‘byor mig gi bsam gtan. S.W. Tashigangpa, Ladakh, 1974. THE EARLY DAYS OF THE GREAT PERFECTION 203 94 See Karmay 1988, p. 20. Padmasambhava

Man ngag lta ba’i phreng ba. P.4726 [transcription and translation in Karmay 1988, pp. 152-171] Sa skya Pa∞∂ita sDom gsum rab dbye, in Sa skya bka’ ‘bum (vol.V). The Toyo Bunko, Tokyo, 1968. Suryaprabhasasiµha dPal gsang ba’i snying po’i rgya cher bshad pa’i ‘grel pa: P.4719 Vilasavajra rGyud kyi rgyal po chen po dpal gsang ba’i snying po’i ‘grel pa (rin po che spar khab): P.4718 Anonymous works and collections Kun byed rgyal po’i mdo: Tb.1 K®Ò∞ayamari-tantra with Ratnavali Pañjika of Kumaracandra. Rinpoche, Sam- dhong and Dvivedi, Vrajvallabh, eds. Rare Buddhist Text Series-9. Central Institute of Higher Tibetan Studies, Sarnath, 1992. Guhyagarbha Tantra: Tb.417 The rGyud ‘bum of Vairocana: a Collection of Ancient Tantras and Esoteric Instructions. Leh: S.W. Tashigangpa, 1971.

Sarvabuddhasamayoga tantra: P.8 Dunhuang manuscript sources


ITJ IOL Tib J PT Pelliot tibétain ITJ 331/1: Short sadhana attributed to Mañjusrimitra. ITJ 437: Short treatise on Mahayoga. ITJ 454: Long but incomplete treatise on Mahayoga. ITJ 470, PT 819, PT 837: rDo rje sems dpa’ zhus lan. Asc. gNyan dPal dbyangs. ITJ 594: sBas pa’i rgum chung. Asc. gSangs rgyas sbas pa. ITJ 647: Rig pa’i khu byug with anonymous commentary. PT 44: The story of Padmasambhava’s introduction of the Vajrakilaya practices to Tibet. Author unknown. Dated: ca. 978–ca.1010. PT 321: Heruka sadhana. PT 322B: Prayer extolling the Great Perfection and the Mayajala tantras. 204 SAM VAN SCHAIK PT 353, ITJ 507: Thabs dang shes rab tu ldan pa’i rang bzhin gyi dkyil ‘khor mtshan nyid. Description of two ma∞∂alas. PT 644: A treatment of the nine vehicles (though without the term “vehicle”) and the vidyadhara levels. PT 656: Spyi’i lung chen po bdun so so’i lta ba dang spyod pa bstan pa gsungs pa’i don man ngag. Presentation of a seven-part doxographical system. PT 849: Miscellany containing, among other things, lists of Buddhist vehicles, tantras and kings. Dated: late 10th-early 11th c. Secondary Sources Bischoff, F.A. & Hartman, C. 1971. ‘Padmasambhava's Invention of the phur-bu Ms. Pelliot tibétain 44', Études tibétains dédiées à la mémoire de Marcelle Lalou. Librairie d'amérique et d'orient Adrien Maisonneuve, Paris. Dalton J. 2002. The Uses of the Dgongs pa ‘dus pa’i mdo in the Development of the Rnying-ma school of Tibetan Buddhism. Doctoral dissertation, Univer- sity of Michigan. — 2004. ‘The Development of Perfection: The Interiorization of Buddhist Ritual in the Eighth and Ninth Centuries’. The Journal of Indian Philosophy 32, pp. 1-30. Davidson, R.M. 1981. ‘The Litany of the Names of Mañjusri', Tantric and Taoist Studies in Honour of R.A. Stein (ed. Michael Strickman), Institt Belge des Hautes Études Chinoises, Brussels. — 2002. Indian Esoteric Buddhism: A Social History of the Tantric Movement. Columbia University Press, New York. Dudjom Rinpoche, Jigdrel Yeshe Dorje. 1991. The Nying-ma School of Tibetan Buddhism: Its Fundamentals and History (2 vols.), translated and edited by Gyurme Dorje and Matthew Kapstein. Wisdom Publications, Boston. Eastman, K.W. 1983. ‘Mahayoga Texts at Dunhuang', Bulletin of the Institute of Cultural Studies, Ryukoku University 22, pp. 42-60. Germano, D. 1994. ‘Architecture and Absence in the Secret Tantric History of the Great Perfection (rdzogs chen)’, JIABS 17.2, pp. 203-335. — 2002. ‘The Seven Descents and the Early History of rNying ma transmissions’, in The Many Canons of Tibetan Buddhism (eds Helmut Eimer and David Germano), Leiden: Brill. Hackin. 1924. Formulaire sanscrit-tibétain du Xe siécle. Librairie orientaliste Paul Geunthner, Paris. Hodge S. 2003. The Maha-Vairocana-Abhisaµbodhi Tantra With Buddhaguhya’s Commentary. Routledge Curzon, London. Karmay, S. 1988. The Great Perfection. E.J Brill, Leiden. Martin, D. 1987. ‘Illusion Web — Locating the Guhyagarbha Tantra in Buddhist Intellectual History’ in Silver on Laps: Tibetan Literary Culture and History (ed. C.I. Beckwith). Bloomington, The Tibet Society. THE EARLY DAYS OF THE GREAT PERFECTION 205 Meinert, C. 2002. ‘Chinese Chan and Tibetan rDzogs chen: Preliminary Remarks on Two Tibetan Dunhuang Manuscripts’, in Religion and Culture in Secular Tibet (ed. Henk Blezer). Brill, Leiden. Roerich, G.N. 1976. The Blue Annals. Delhi: Motilal Banarsidass. van Schaik, Sam & Jacob Dalton. 2004. “Where Chan and Tantra Meet: Bud- dhist Syncretism in Dunhuang,” in The Silk Road: Trade, Travel, War and Faith. London: British Library Press. Takeuchi, T. unpub. ‘Sociolinguistic Implications of the use of Tibetan in East Turkestan from the end of Tibetan Domination through the Tangut Period (9th-12th c.)’. Unpublished conference paper. Tucci, G. 1958. Minor Buddhist Texts II. Is.M.E.O, Rome. Vitali, R. 1996. The Kingdoms of Gu.ge Pu.hrang. Tho.ling gtsug.lag.khang lo. gcig.stong ’khor.ba’i rjes.dran.mdzad sgo’i go.sgrig tshogs.chung, Dharam- sala. Weinberger, S. 2003. The Significance of Yoga Tantra and the Compendium of Principles (Tattvasaµgraha Tantra) within Tantric Buddhism in India and Tibet. Doctoral dissertation, University of Michigan.



Source

Wikipedia:The Early Days of the Great Perfection II