Haṭhayoga’s Tantric Buddhist source text

By Graham Burns

The idea that yoga is Indian is, of course, taken for granted by almost everyone. And those who have studied the yoga traditions in more depth know of the close connections between the tradition known as haṭhayoga and Śaivism, particularly Tantric Śaivism. So the idea that the earliest known text to teach some of the important principles and practices of haṭhayoga originated not in Tantric Śaivism, but rather in a Buddhist milieu, is guaranteed to disturb the pre-conceptions of many students and scholars of yoga.

Yet, that is exactly the argument put forward by James Mallinson in an extremely well attended, and live streamed, first session of the Sanskrit Reading Room of 2017/18. Dr. Mallinson’s radical conclusion comes after years of study of a text called the Amṛtasiddhi, an influential text, many of whose verses have found their way more or less verbatim into later texts on haṭhayoga. Although modern manuscripts of the Amṛtasiddhi do indeed suggest that it was composed in a Śaivite milieu, the discovery in the China Nationalities Library in Beijing of a manuscript of the text, dated, according to its colophon, to 1160 CE, cast doubt on this ‘traditional’ attribution to the Śaivite traditions. This dating, if accurate, puts the Amṛtasiddhi earlier than any other known text to teach techniques of haṭhayoga, even if the term haṭhayoga does not itself appear in it.

Although the Amṛtasiddhi was composed in India by Tantric Buddhists, the text was later transliterated and translated into Tibetan, and it is from the bilingual manuscript that the team were able to deduce that it had, most likely, been written by Buddhists. The manuscript discovered in Beijing contained the text of the Amṛtasiddhi not just in Sanskrit, but also transliterated into one Tibetan script and translated into another. Dr. Mallinson’s reading of the manuscript with Tantric Buddhism scholar Dr. Péter-Dániel Szántό from Oxford University revealed that the Amṛtasiddhi uses numerous examples of technical terms of art from Tantric Buddhism, suggesting strongly that its origins lay in that tradition, albeit with what Dr. Mallinson described as ‘pretty eccentric’ Tantric Buddhists, who used the text to pit themselves against certain better known Tantric Buddhist traditions.

amrta 2

Our reading focussed on two parts of the Amṛtasiddhi: Chapter 7, the Bindudhāraṇaviveka, and Chapter 11, the Mahāmudrāviveka. The first ten Chapters of the text explain the workings of the ‘yogic body’, and Chapter 7 introduces the idea of a single ‘seed’ (bījam ekam), a little later in the Chapter called bindu, as the fundamental essence of the whole of existence. This seed is identified with Sadāśiva, one of the highest forms of Śiva in Śaivite Tantra, as well as with the moon, and with other exotic substances such as mada, ‘the fluid that flows from elephants’ temples when they are in rut’ (Mallinson and Singleton 2017:492 n.57). Chapter 7 makes clear that control of the bindu is brought about by control of the breath, which, in turn, leads to the very yogic goal of control of the mind, and Chapter 7.7 introduces terms from alchemy to refer to this control: if bindu is mūrcchita (stunned) it removes disease, if baddha (bound) it can make one a ‘sky-rover’ (khecara), perhaps here suggesting that the practitioner can actually fly. (In passing, many years ago I encountered a Tibetan Buddhist monk in Ladakh who recounted his ability to fly as if it were the most matter of fact accomplishment – maybe he had succeeded in ‘binding’ his bindu.)

The reference to Sadāśiva in Chapter 7.2, and other Śaivite references in the Amṛtasiddhi, led to a debate about the text’s Buddhist origins, for it would clearly have been unusual for a Buddhist text to give Śiva such prominence. Dr. Szántό, who was present at the reading, agreed that this was unusual, and suggested that the text was most likely composed for a Śaivite Tantric audience, but that, nevertheless, the extensive use of Tantric Buddhist terminology suggested that that milieu had been the text’s original starting point. As an example, Chapter 7.4 refers to ānandā viramāntāḥ (‘the [four] blisses ending in [the bliss of] cessation’), a term with no technical meaning in a non-Buddhist context, but with a clear meaning in Tantric Buddhism as the fourth and final level of ‘bliss’ in Tantric sexual ritual.

Although the Amṛtasiddhi does not itself teach Tantric sexual rituals, this association may also find reflection in the somewhat mysterious idea of ‘male’ and ‘female’ bindu, introduced in Chapter 7.8. While male bindu is easily identified as semen, the identity of the female seed, rajas, is more uncertain, though it was noted by one reader that the ancient Greek medical tradition sometimes referred to both male and female ‘seeds’. Although rajas is described as red, possibly suggesting menstrual fluid, it is also said to unite with the male bindu externally in the act of procreation, while uniting the two internally is said to be the ‘highest dharma and the best yoga’ (7.13: paro dharma… yogaḥ paro mataḥ). The male bindu is said to be ‘lunar’ and the female ‘solar’, making this reference to their inner union a possible precursor of the later popular notion that the term haṭhayoga itself refers to some form of ‘uniting’ of ‘solar’ and ‘lunar’ energies in the body.

Chapter 7 closes by re-emphasising the relationship of bindu, breath and mind, while Chapter 11, by contrast, discusses one of the most important methods of achieving the control of bindu. This is the practice of mahāmudrā, a yogic practice which finds its way into later texts of haṭhayoga, and one which is still reasonably well known and widely practised in contemporary yoga. Mahāmudrā (the ‘great seal’) is another well-known Tantric Buddhist term, even if there meaning something somewhat different to the practice described in the Amṛtasiddhi, and Dr. Szántό noted that this section of the Amṛtasiddhi had appeared in an even older Tibetan text. Conversely, as Dr. Mallinson pointed out, mahāmudrā was an unknown term in Śaivite circles before this text.

The practice of mahāmudrā is explicitly said to hold bindu, in turn leading to control of body, speech and mind (11.7: kāyavākcittasādhanam, again an unusual triad in Śaivite texts but a common one in Tantric Buddhism). The method of practice, as described in the text, is perhaps better shown in illustration, as in these images (the first from a manuscript of the 18th century Jogpradīpakā and the second of the famous 20th century yogi Tirumalai Krishnamacarya).

amrta 1

However, for those of us who practise mahāmudrā regularly, the text’s instruction to ‘raise the buttocks onto a seat’ (11.4: āsane kaṭim āropya) while the heel is pressed into the perineum (11.3: yoniṃ saṃpīḍya vāmena pādamūlena) caused some confusion.

amrta 3

Ultimately, the very practice of mahāmudrā, according to the Amṛtasiddhi, stops death (11.10: anayā bādhyate mṛtyor…), and its description also brought the first, fascinating, session of the Sanskrit Reading Room for 2017/18 to a close.

Graham Burns is a final-year doctoral researcher at SOAS.


Forthcoming (2018). “The AmṛtasiddhiHaṭhayoga’s Tantric Buddhist Source Text.” In Śaivism and the Tantric Traditions: A Festschrift for Alexis Sanderson, ed. Dominic Goodall, Shaman Hatley & Harunaga Isaacson. Leiden: Brill.ṛtasiddhi_Haṭhayogas_Tantric_Buddhist_Source_Text

Mallinson, James and Singleton, Mark 2017 Roots of Yoga London: Penguin Random House

Image credits:

Image 1 (Amṛtasiddhi manuscript): China Nationalities Library of the Cultural Palace of Nationalities MS 005125 (21).

Image 2: British Library Add. 24099 Jogpradīpakā āsana No. 19

Image 3: Source unknown


Winter Term Schedule 2017

Wednesday October 11, 3-5 pm

Dr James Mallinson (SOAS)


Room T102, 22 Russell Sq, SOAS

(With a tea reception to follow, 5-6 pm)

Wednesday October 25, 3-5 pm

Dr Péter Szántó (University of Oxford)

Padminī of Mahāsukhavajra and commentary to Caṇḍamahāroṣaṇatantra

Room T101, 22 Russell Sq, SOAS

Wednesday November 8, 3-5 pm

Dr Camillo Formigatti (John Clay Sanskrit Lib)

Deconstructing a critical edition: The Anāthapiṇḍadāvadāna from the Avadānaśataka

The Weston Library, Bodleian Library, University of Oxford

Wednesday November 15, 3-5 pm

Dr Måns Broo (Abo Akademie University)

Hari Bhaktivilāsa

Room 404, 30 Russell Sq, SOAS

Wednesday November 29, 3-5 pm

Graduate Sanskrit Reading Room

Maddalena Italia (SOAS)

Translation immoral? Contamination, hybridity, and vociferous silences in

early twentieth-century translations of Sanskrit erotic poetry

Rohini Bakshi (Independent Scholar)

Vedic Chanting Method and Demonstration

Graham Burns (SOAS)

Formulation, Substitution and Veneration: three controversial Upanisadic terms.

Room T102, 22 Russell Sq, SOAS

Wednesday December 6, 3-5 pm

Dr Lidia Wojtczak (SOAS)

Pimps and beggars. Poets through the lens of Sanskrit satirical verse.

Room 404, 30 Russell Sq, SOAS

Wednesday December 13, 3-5pm

Dr Marie-Hélène Gorisse (Ghent University)

Ratnakīrti’s Īśvarasādhanadūṣaṇa

Room T101, 22 Russell Sq, SOAS

For more details, subscribe to or our Facebook group: Sanskrit Reading Room. Sessions will be streamed live or available as an audio recording.

For any enquiries contact Karen O’Brien-Kop (Karen_O’ or Avni Chag (

To book a space in any of these sessions please email Ruth Westoby (


Winter term poster.jpg

Maṇḍana’s Hermeneutic Strategy

By Karen O’Brien-Kop

Maṇḍana Miśra, the circa 7th-century philosopher, is known both as a Mīmāṃsā and as a Vedānta thinker. In this session Dr Theodore Proferes, Senior Lecturer in Ancient Indian Religions at SOAS, focused on the opening of one of Maṇḍana’s later works, the Brahma Siddhi, which is devoted to an exposition of the ideas of Advaita Vedānta.

Mīmāṃsā and Vedānta are two paired schools of philosophy in the classical tradition.  Mīmāṃsā is a school of scriptural exegesis focused on the early Vedic texts and emphasizes the study of ritual action (karmakāṇḍa), while Vedānta focuses on the later Upaniṣadic texts and on the study of knowledge (jñāna-kāṇḍa). Dr Proferes firstly outlined the Mīmāṃsā position that formed the backdrop to Maṇḍana’s thinking – in order to highlight how Maṇḍana’s work played a key role in conjoining the two kāṇḍas, ritual action and knowledge. Although focussed on scripture, Mīmāṃsā was associated exclusively with ritual and not with texts; the texts were only relevant in that they contained injunctions on how to act. Niyoga was the term used to connote injunctive force, and a niyogavādin was one who believed that the primary function of texts was to enjoin. Although the action that is being enjoined in the Upaniṣads seems different, in that we are being enjoined to ‘know thyself’, the self is subsidiary in this instruction, and it is still the injunction to perform action that makes the text meaningful. Thus the revelation that occurs in this context is the revelation of the action, and not of the nature of the self.

Dr Proferes then read from the opening of the Brahma Siddhi, which contains a plan for the whole work. In this section, Maṇḍana sets out a series of arguments that are systematically treated in the rest of the Brahma Siddhi. This first chapter is dedicated to refuting the first argument – Maṇḍana announces the intention to correct wrong claims made by other scholars regarding the epistemological authority (pramāṇa) of the Upaniṣads, and he sets out to explain how Brahman and readings of the Upaniṣads are consistent. The self-contained character of the first chapter has led some scholars, such as Vetter, to assert that this section may originally have been an independent text (Vetter 1969). Dr Proferes also made the case that this opening section is distinct from the rest of Brahma Siddhi in that it sets itself out as a hermeneutic teaching. From the very first word, Maṇḍana is deeply engaged in a hermeneutic enterprise and provides a method for how to read the Upaniṣads. In the opening maṅgala of chapter one (the Brahmakāṇḍa), Maṇḍana Miśra lists key concepts of Prajāpati:

ānandamekamamṛtamajam vijñānamakṣaram asarvam sarvamabhayam namasyāmaḥ prajāpatim

We make obeisance to Prajāpati [who is] bliss, the singular, undying, unborn, cognition, the imperishable syllable, all, not all, fearless.

The very first word of the text, ānanda, or ‘bliss’, is then singled out as a key term for discussion. The claims about ānanda go thus: if Brahman is of the nature of ‘bliss’, then the activity of one seeking liberation, a mumukṣu, would be the result of desire for bliss. Such an action would not lead to the goal, because vision of the self is said to belong only to the person who is tranquil and controlled, states that are incompatible with desire. Thus ānanda has to be defined negatively rather than positively. And so according to this line of reasoning, ānanda is not a synonym for sukha (happiness) but rather denotes the absence of duḥkha (suffering). Ānanda means ‘aduḥkha’ or non-suffering. This key strategy of negation informs the hermeneutic of the rest of the text.

In the 4th-8th centuries, Dr Proferes continued, there was a large and diverse group of thinkers who were shifting the terms of philosophical and religious debate – figures such as Kumārila, Prabhākara, Śaṅkara, and Bhartṛhari. At this point in time there was a battle going on between different groups for a unified reading of the Veda, as both karma and jñāna (action and knowledge). So Maṇḍana sets forth a hermeneutic principle, an approach to interpretation of meaning, which is consistent with the Advaita position of non-duality. Through the strategic use of niṣedha (prohibition in Mīmāṃsā), he draws on the apoha theory of the Buddhists (semantics based on the principle of negation or exclusion) and possibly also on the Prapañca-vilayavādins (who espoused an early doctrine about the nature of reality and how to read the Upanisads). The Brahma Siddhi is thus not concerned with ontology but with hermeneutics.

The last question that Dr Proferes explored was how to understand the relationship between Maṇḍana’s earlier Mīmāṃsā output  (the Mīmāṃsānukramāṇikā, the Bhāvanāviveka, and the Vidhiviveka) and this later Advaita work. Maṇḍana was a follower of Kumārila and also an adherent of the sphoṭa theory of language proposed by Bhartṛhari, in which the linguistic entity is eternal. The two philosophical strands of Maṇḍana’s earlier Mīmāṃsā and later Vedānta work can, in fact, be seen as complementary. The method of reading that Maṇḍana advocates unites all methodologies and interpretation theories of the Upaniṣads. In the end, direct realization of Brahman leads to liberation, not knowledge alone, and only the act of meditation, or upāsanā, produces this direct realization.


Proferes, T. 2017. ‘Sanskrit Reading Room’, talk held at SOAS, University of London, on June 14.

Vetter, T. 1969. (trans.) Brahma-Siddhi. Bohlau


(Karen is a PhD Candidate in the Study of Religions and Philosophies at SOAS, and a co-organiser of the Sanskrit Reading Room.)

Mahābhārata: The Critical Edition and Beyond; Inclusions, Omissions, Translations.

By Rohini Bakshi

At a recent Sanskrit Reading Room presented by Professor James Hegarty, Reader in Indian Religions at Cardiff University, the astounding variety and variation among different editions of the Mahābhārata were brought to life. We came to realise that words and verses notwithstanding, important episodes may be presented differently, or even omitted from some editions altogether. In narrating the same episode, verse order is sometimes altered, which has real impact on narrative continuity and the literary and philosophical impact of a given passage.

Prof. Hegarty gave us some background on the creation of the critical edition of the Mahābhārata. Given the bewildering variety and variation in the different recensions, in the late 19th century a movement began to create a ‘critical edition’ of the epic. This task was undertaken at the Bhandarkar Oriental Research Institute, Pune, in 1917 and was completed nearly 50 years later in 1966. Teams of Indian and Western Indologists, supported by traditionally trained śāstrīs and highly qualified university students undertook this gargantuan task.

V S Sukthankar, general editor of the critical edition of the Mahābhārata, explained that it is an approximation of the earliest recoverable form of the text. He was careful not to claim that it was a ‘discovery’ of the original Mahābhārata. In Sukthankar’s own words “What the promoters of this scheme desire to produce and supply is this: a critical edition of the Mahābhārata in the preparation of which all important versions of the Great Epic shall have been taken into consideration, and all important manuscripts collated, estimated and turned to account.”

The general idea was to collect as many manuscript copies from around the country (literally Kashmir to Kanya Kumari) to represent as many regions as possible by using Sanskrit manuscripts in different scripts (Malayāḷam, Grantha, Nevāri, Devanāgarī, Bengalī, Śāradā to name a few) and to balance newer, well maintained manuscripts with older ones which might be partial or even illegible – the latter being considered more valuable. As it turned out, the Malayāḷam and Śāradā variants – so far removed from each other geographically – shared commonalities and gaps, allowing the editors to take informed calls about what the archetype might have been.

The editor and team for each parvan went through and tallied all the manuscripts they collated – stanza by stanza, line by line, word by word. In the final form, verses which occur on the maximum number of manuscripts are deemed to be authentic and included. Some which are included have wavy lines indicating that they are doubtful. Those verses that do not make the cut are put in footnotes at the bottom of the page – and longer passages are put into appendices at the end of the parvan. A typical page from the critical edition looks like this:


Ever since the critical edition (also known as the Poona edition) of the Mahābhārata was completed, practically all Western scholarly work has focused on it. The Chicago translation began in the 1970s with J. A. B. van Buitenen and was continued by his students after his death. It is still incomplete. Indian Economist and Sanskritist Bibek Debroy recently completed his English translation, also based on the critical edition. The Clay Sanskrit library deviated from this trend by translating the vulgate (on which the famous 17th century commentary of Nīlakaṇṭha is based.)

How does the critical edition stand in relation to the vulgate and other recensions? For instance, two episodes which are considered axiomatic by Indian audiences are not in the critical edition: The story of Gaṇeśa as scribe and, more famously, the re-robing of Draupadī by Kṛṣṇa through a divine miracle. This means these episodes did not exist in enough of the manuscripts to make the cut. What is the fate and relevance of all such episodes that have been left out of the critical edition?

Prof. Hegarty stated that while the critically reconstituted text has been applauded by ‘philologists’ it has been rubbished by those who have an interest in the ‘anthropology’ of the Mahābhārata tradition. These scholars are equally (if not more) interested in the footnotes and appendices, i.e. in that which has been left out. He drew our attention to their extraordinary variety – material from every manuscript was consulted in the process of creating the critical edition. He called them an ‘embarrassment of riches’ and was frankly surprised that there wasn’t an efflorescence of publications that takes up the ‘rejected’ rich range of literary data. Instances of critical instability, he explained, offered excellent products for study.

“The advantage of translating from the footnotes and appendices is that one has a clear and detailed account of the manuscripts from where these readings came. This allows us to do all sorts of interesting things; we can reconstitute individual manuscripts, groups of manuscripts or translate all the variants – a sort of mega-composite that goes beyond even the Vulgate and is the mirror image of the critically edited text, which contains only that which was common to all manuscripts,” said Prof. Hegarty.

Considering the Ādi Parvan alone, some 235 manuscripts were consulted, which included 32 manuscripts in the Bengalī script, 31 in Grantha, 28 in Teḷugu, 26 in Malayālam, 5 in Nevāri, 3 in Śāradā, 1 Maithilī and 1 Devanāgari. Of these 235 some 60 were finally used. In the process of constructing the archetype, it matters not only what is used, but also what is left out. He demonstrated for instance, that a reconstituted text including all the footnoted material which had been left out of the critical edition made this section of the Ādi Parvan decidedly more Vaiṣṇava in orientation:

I will tell in full the great creation of

Sagacious Vyāsa, who is known to all.                                                (23)

All honour to him, of limitless might!

By grace will I tell Nārāyaṇa’s tale.                                                (21*1-2)

No sacred waters nor cloistered grove

Can offer reward as this story can.                                               (21*3-4)

Nothing that is, was or will be equals

Nārāyaṇa; by these words is wealth achieved!                          (21*5-6)

Poets told it and, telling it, they will

Tell it again: what was the world over.                                               (24)

Kept by Brahmins; whether epitomised

Or told in full; the three worlds know it.                                            (25)

Adorned with virtue and well chosen words,

Varied in rhythm; it delights the wise.                                               (26)

Quite rightly, he points out that some recensions have far more detail in their narrative, which the reader of the critical edition misses out on. He treated us to a reconstituted text from Virāṭa Parvan, which included verses from the Malayāḷam manuscripts. These contain a large number of readings unique to themselves. We came to see that the inclusion provided a) a display of virtuosic myth-knowledge on the theme of powerful beings ‘in disguise’ and b) a richer description of Yudhiṣṭhira’s disguise as the Brahmin gambling master, Kaṅka. “None of this rich textual ‘life’ is apparent if one translates either just the critically reconstituted text or the Vulgate,” he said.

Prof. Hegarty is driven by two deeps interests – one is the incredible variety of materials left out of the critical edition, and the second is how to render the Sanskrit into elegant English. He shared two translations of the opening of 17th-century English poet Milton’s Paradise Lost, a staple of epic poetry that informed the translation styles of European scholars. One of these two translations was rather tongue-in-cheek, suggesting how an Indologist might translate to preserve accuracy of meaning at the expense of aesthetics. I have to say that in all honesty I preferred the second, but that says as much about me as it does about the translation. For those who missed the session, here are the two translations:

Of Mans First Disobedience, and the Fruit
Of that Forbidden Tree, whose mortal taste
Brought Death into the World, and all our woe,
With loss of Eden, till one greater Man
Restore us, and regain the blissful Seat, [5]
Sing Heav’nly Muse, that on the secret top
Of Oreb, or of Sinai, didst inspire
That Shepherd, who first taught the chosen Seed,
In the Beginning how the Heav’ns and Earth
Rose out of Chaos: Or if Sion Hill [10]
Delight thee more, and Siloa’s Brook that flow’d
Fast by the Oracle of God; I thence
Invoke thy aid to my adventrous Song,
That with no middle flight intends to soar
Above th’ Aonian Mount, while it pursues [15]
Things unattempted yet in Prose or Rhime.

Sing heavenly Moûsa (a minor Greek deity) of the first defiance of humankind and of the fruit of the proscribed tree, whose death-taste, brought mortality into the world and the loss of Eden (a pre-lapsarian paradise in Christian traditions) and considerable sorrow, until a more powerful man restored us and re-obtained the throne of ecstasy; [sing] of Mt. Horeb (the place of origin of the ten commandments – a set of ethico-legal requirements authoritative within Jewish and Christian traditions) or Sinai (see above on Mt. Horeb), which inspired that shepherd to teach the people of Israel (a reference to Exodus 19-20) about the creation of both heaven and earth from the void or, if Mt. Zion is more pleasing or the brook of Siloa (a rock cut pond outside Jerusalem) that passed the Ark of the Covenant [sing of these]. I thus request your help in this brave composition, which intends to fly unimpeded above Mt. Helicon pursuing things that have not been attempted either in prose or by means of rhyme.

Upon being asked about the dangers of mistranslation (intentional or otherwise) he was quite unequivocal. “Literal is not always faithful… A literal translation can misrepresent and distort.  (Yet) I am all for the ‘literal’ style of translation, (as) there are ‘dangers’ in all forms of translation. Translation is one of those things that the more you think about it the more impossible it becomes, and yet it is done every day. I experience the Mahābhārata as a dynamic, rhythmic, powerfully exciting and insightful text. I know that this is how it has been experienced by generations of Indians. I would like to create English that reflects this.”

You can read more about Prof. Hegarty and his work here:


  • Sanskrit Reading Room with Prof. James Hegarty, June 7th, 2017
  • Follow up email exchange with Prof. Hegarty
  • Prologomena to the critical edition of the Ādi Parvan, Book 1 of the Mahābhārata: