6

Śāktism and haṭhayoga¹

James Mallinson

Introduction

In this paper I shall examine one of the ways that Śāktism endured after its heyday, the ‘Śaiva Age’ magisterially documented by Alexis Sanderson in a recent book-length article.² The thirteenth century, the end of Śaivism’s period of dominance, coincided with the rise of yoga, specifically yoga which used the techniques known as haṭha, to a position of dominance among the soteriological methods employed in India.³ It is haṭhayoga’s relationship with Śaivism and its Śākta manifestations that is the subject of this paper.

Scholarship on haṭhayoga, my own included, unanimously declares it to be a reformation of tantric yoga introduced by the gurus of the Nāth saṃpradāya, in particular their supposed founder, Gorakṣa. In much secondary literature the phrases ‘Nāth yoga’ and ‘haṭhayoga’ are used interchangeably. When other traditions are seen to employ the practices or terminology of haṭhayoga, they are said to be borrowing from the Nāths.⁴ Gorakṣa, who probably flourished in the twelfth century, and Matsyendra, who according to tradition was Gorakṣa’s guru but is likely to have lived three centuries before him, were exponents of the Śākta cult of the Paścimāmnāya or western stream of Kaula Śaivism.⁵ To this day the Nāth Yogīs are renowned as tāntrika adepts and their monasteries are often situated near important goddess temples, such as Jwalamukhi in the Kangra district of Himachal Pradesh or Devi Pattan near the India–Nepal border. The Paścimāmnāya tradition is no longer extant and Nāth liturgy is now more in keeping with the tradition of the foremost surviving Śākta cult, that of the Dakṣiṇāmnāya or southern stream of Kaula Śaivism, with the goddess Bāla- or Tripurā-sundarī being the focus of their worship.⁶

The Nāths’ Śākta credentials are pukka.⁷ The ascription to them of both the invention of haṭhayoga and the composition of the corpus of Sanskrit texts which teach its practice stands on shakier ground: it derives from the claims of the Nāths themselves and from the fact that of the small number of texts on haṭhayoga that have been edited, some are ascribed to Nāth gurus or mention them as the revealers of their doctrines. But the corpus of Sanskrit works on haṭhayoga – which is our only source for haṭhayoga’s formative period – has, perhaps surprisingly considering the widespread popularity of yoga today, been the subject of very little critical study. Much of my research of the last few years has concentrated on identifying the texts that constitute this corpus and using them to examine how haṭhayoga developed and who practised it. In the first part of this paper I shall summarise what constituted early haṭhayoga. I shall then show how, in contemporaneous taxonomies of yoga, Śākta techniques were grouped separately from haṭhayoga, under the name laya. Next I shall show how in the Haṭhapradīpikā, the text which became haṭhayoga’s locus classicus, the Śākta techniques of layayoga were included under the rubric of haṭha and how at the same time the purpose of haṭhayoga was realigned to be more in keeping with that of laya. I shall then identify the practitioners of early haṭhayoga – who were not Śāktas – and show how they have continued to be its torchbearers. In the final part of the paper I shall attempt to locate these developments in their wider context.

Early haṭhayoga

The earliest references to haṭhayoga are scattered mentions in Buddhist canonical works and their exegesis dating from the eighth century onwards, in which it is the soteriological method of last resort.⁸ In its earliest definition, in Puṇḍarīka’s eleventh-century Vimalaprabhā commentary on the Kālacakratantra, haṭhayoga⁹ is said to bring about the ‘unchanging moment’ (akṣarakṣaṇa) ‘through the practice of nāda by forcefully making the breath enter the central channel and through restraining the bindu of the bodhicitta in the vajra of the lotus of wisdom’.¹⁰ While the means employed are not specified, the ends, in particular restraining bindu, semen, and making the breath enter the central channel, are similar to those mentioned in the earliest descriptions of the practices of haṭhayoga, to which I now turn.

In seeking to establish a corpus of early works on the practices of haṭhayoga we are greatly assisted by the Haṭhapradīpikā, which can be dated to approximately 1450.¹¹ Its composer, Svātmārāma, used verses from at least twenty texts¹² to compile his hugely influential work, whose stated aim is to be a light on haṭha ‘in the darkness of a multitude of doctrines’.¹³

Only four of the texts drawn upon by Svātmārāma mention haṭha or haṭhayoga by name¹⁴ and of those only one, the Dattātreyayogaśāstra, a Vaiṣṇava work, gives a detailed exposition of its techniques. Its haṭhayoga consists of a yoga of the eight limbs also taught in, for example, Patañjali’s Yogasūtra but here ascribed to Yājñavalkya and others, or an alternative set of practices employed by Kapila and other siddhas, practices which are thus the distinguishing feature of early haṭhayoga.

These practices comprise ten physical techniques which in later works all came to be classed as mudrās. Some of these techniques are relatively simple attitudes to be assumed in meditation, ascetic practice or breath control, forerunners of which are taught or mentioned in earlier works;¹⁵the others are unique to haṭhayoga. Their primary aim, which they effect either pneumatically, by making the breath enter the central channel and rise upwards, or mechanically, is to stop the lunar bindu, or semen, which is equated with amṛta, the nectar of immortality, from dripping down from its store in the head and being consumed in the solar fire at the base of the central channel. The techniques of early haṭhayoga are thus direct methods of addressing the ancient (and still prevalent) Indic concern with the preservation of semen, which results from its being considered to be the vital principle and its loss being thought to lead to weakness and death. The haṭhayogin might practise viparītakaraṇī, inverting himself in order to use gravity to keep his bindu in his head. Or he can, applying khecarīmudrā, insert his tongue into the opening behind his uvula thereby sealing binduin the cranial cavity. Or he might, should he ejaculate, create a vacuum in his abdomen and resorb his bindu by means of vajrolimudrā.

Eight of the works used to compile the Haṭhapradīpikā teach one or more of the haṭha techniques ascribed to Kapila in the Dattātreyayogaśāstra. These constitute the corpus of Sanskrit texts on what I call ‘early’ haṭhayoga in contrast to the more catholic ‘classical’ haṭhayoga of the Haṭhapradīpikā and subsequent works.¹⁶ The texts of the corpus are, in approximate chronological order,¹⁷ the following:

Amaraughaśāsana

Dattātreyayogaśāstra

Gorakṣaśataka

Vivekamārtaṇḍa

Yogabīja

Khecarīvidyā

Amaraughaprabodha

Śivasaṃhitā

I shall now briefly summarise the techniques of haṭhayoga as taught in each of these works (other than the Dattātreyayogaśāstra, which has already been mentioned).

Amṛtasiddhi

The sectarian origins of the Amṛtasiddhi are unclear; it is explicitly Śaiva but contains no specifically Śākta teachings.¹⁸ The haṭha techniques taught in the Amṛtasiddhi (which are not named haṭha) are used to make the breath enter the central channel and raise it upwards in order to reverse the usual direction of flow in the channels of the body and stop bindu from moving downwards.¹⁹ The Amṛtasiddhi is the first text to teach the pervasive haṭhayogic physiology in which the moon is situated at the top of the central channel from where it rains down amṛta which, if preventative yogic techniques are not employed, is then burnt up in the sun situated at the bottom of the central channel.²⁰ It also introduces the widespread haṭhayogic notion of the identification of the breath, semen and mind: by stopping any one of these three, all three are stopped.²¹

The principle aim of the haṭha techniques of the Amṛtasiddhi is thus the same as those of the Dattātreyayogaśāstra: to stop or reverse the movement of bindu. Neither text associates the practices of haṭha with Kuṇḍalinī or the cakras.²²

Gorakṣaśataka ²³

Probably contemporaneous with the Dattātreyayogaśāstra, i.e. composed in the thirteenth century, are the first texts on yoga to be associated with Gorakṣa, the Gorakṣaśataka and Vivekamārtaṇḍa. Unlike the Dattātreyayogaśāstra, neither of these works calls its yoga haṭha; it is just yoga. The Gorakṣaśataka teaches that liberation is to be attained by controlling the mind through controlling the breath. One method of controlling the breath is to stimulate Kuṇḍalinī, which can be done either by using the three haṭhayogic bandhas mentioned earlier,²⁴ or through sarasvatīcālana, ‘stimulating Sarasvatī’. Sarasvatī° or śakti-cālana is not taught in the Dattātreyayogaśāstra but it is included among the mudrās of the Haṭhapradīpikā’s classical haṭhayoga. In the Gorakṣaśataka, it involves wrapping in a cloth the tongue, which is identified with Sarasvatī and said to be the goddess at one end of the central channel, and tugging on it in order to stimulate Kuṇḍalinī, who dwells at the other end.²⁵ Nowhere in the Gorakṣaśataka is the preservation of bindu or amṛta mentioned.

Vivekamārtaṇḍa

The framework of the text of the Vivekamārtaṇḍa is a description of the well-known six cakras of yogic physiology, within which are passages on a variety of yogic subjects including an enumeration of a sixfold yoga and a section on Kuṇḍalinī in which awakening her by means of ‘fire yoga’ (vahniyogena) and raising her, together with the mind and the breath, to the brahmadvāra, is said to be the way for yogins to become liberated.²⁶ Immediately after this passage on Kuṇḍalinī comes a description of five haṭhayogic mudrās,²⁷ none of which is said to bring about the raising of Kuṇḍalinī. They have a range of physical benefits, including the raising of the apāna breath, but the aim most emphasised in their descriptions, particularly in that of khecarīmudrā, is the preservation of bindu. Later in the text the haṭhayogic mudrā called viparītakaraṇī, ‘the inverter’, is taught and this too is said to be a method for stopping the downward flow of nectar; Kuṇḍalinī is not mentioned.²⁸

Yogabīja

The Yogabīja²⁹ is a dialogue between Śiva and the goddess. It makes no mention of any Nāth gurus. Its yoga is similar to that of the Gorakṣaśataka in that it uses the three haṭhayogic bandhas and śakticālanī mudrā. The awakening of Kuṇḍalinī is the purpose of many of the yoga practices taught in the text;³⁰ preserving bindu or amṛta is mentioned in passing twice.³¹

Khecarīvidyā

The Khecarīvidyā is a composite text in which teachings on the practice of the haṭhayogic khecarīmudrā have been inserted into part of a Kaula work that includes a coded description of a mantra (the Khecarīvidyā of the text’s title), and a eulogy of madirā, alcohol, which is reworked into praise of Khecarī. The practice of khecarīmudrā enables the yogin to access various stores of amṛta in the body and also to raise Kuṇḍalinī via the six cakras to the great store of amṛta in the head, with which she floods the body on her journey back down to her home at the ādhāra, the ‘base’ located at the perineum. The Vivekamārtaṇḍa is mentioned near the beginning of the Khecarīvidyā,³² but despite that text’s clear assertions that the purpose of khecarīmudrā is the sealing of bindu or amṛta in the head, that aim is not mentioned in the Khecarīvidyā (and it would be hard to reconcile with the flooding of the body with amṛta). In many respects the Khecarīvidyā’s khecarīmudrā has more in common with the tongue-pulling śakticālanī mudrā of the Gorakṣaśataka than with the cavity-sealing khecarīmudrā taught in the Dattātreyayogaśāstra and Vivekamārtaṇḍa.

Amaraughaprabodha

The Amaraughaprabodha³³ is ascribed to Gorakṣa in its manuscript colophons and mentions four gurus associated with the Nāth order in its opening verse.³⁴ It teaches a haṭhayoga (named as such) which is very similar to the bindudhāraṇa yoga taught in the Amṛtasiddhi, but it adds the awakening of Kuṇḍalinī to the benefits of the Amṛtasiddhi’s mahāmudrā and it mentions a dhyāna of śakti.³⁵

Śivasaṃhitā

The last of the texts which constitute the corpus of works on early haṭhayoga is the Śivasaṃhitā.³⁶ There is nothing in the Śivasaṃhitā to associate it with the Nāths. On the contrary, various features of the text show it to be a product of the tradition of Śaiva non-dualism mixed with the expurgated dakṣiṇāmnāya Kaula cult of Tripurasundarī known as Śrīvidyā.³⁷ This cult became the most widespread and enduring of the Śākta traditions.

The Śākta orientation of the Śivasaṃhitā is made plain at the end of its teachings on the benefits of increasing numbers of repetitions of the Śrīvidyā mantrarāja: through thirty lakh repetitions the practitioner becomes equal to Brahmā and Viṣṇu; through sixty lakh he attains Rudra-hood; through eighty lakh, the śaktitattva. Finally, through one crore repetitions, he is absorbed into the absolute.³⁸

The Śivasaṃhitā mentions haṭhayoga by name but does not define it, nor does it make it clear which of the many practices it teaches come under its rubric. In its fourth paṭala it teaches all ten haṭhayogic mudrās taught in earlier works on haṭhayoga and, further confirming its roots in the traditions of Śrīvidyā, crowns them with its own, a haṭhayogic variety of the hand-gesture yonimudrā which occupies a central place in the rituals taught in the Vāmakeśvarīmata, one of the foundational texts of Śrīvidyā.³⁹ The Śivasaṃhitā contains detailed teachings on Kuṇḍalinī, who is said to be the ‘great goddess’ (paradevatā)⁴⁰ and to ‘take the form of the creation of the universe’ (jagatsaṃsṛṣṭirūpā).⁴¹ The purpose of its haṭhayogic mudrās is to awaken Kuṇḍalinī and make her pierce a variety of lotuses and knots as she rises upwards.⁴² Despite incorporating and reworking many of the Amṛtasiddhi’s verses on mahāmudrā, mahābandha and mahāvedha, and several verses from the Dattātreyayogaśāstra, nowhere does the Śivasaṃhitā mention bindudhāraṇa as the aim of any practice of yoga. Even vajrolimudrā, the practice of urethral suction which in the Dattātreyayogaśāstra and the Haṭhapradīpikā is for the preservation of the yogin’s bindu,⁴³ is made purely Śākta in the Śivasaṃhitā,⁴⁴ where its purpose becomes the absorption of one’s partner’s bindu or rajas, in order to combine them within one’s own body.⁴⁵

So, to summarise early haṭhayoga: in its earliest formulations, which are found in texts that are neither Śākta nor associated with the gurus of the Nāth saṃpradāya, haṭha’s distinguishing feature is a variety of physical techniques which are used to keep bindu or amṛta, i.e. semen, in the head. In contrast, contemporaneous Nāth works emphasise the purpose of their yoga, which they do not call haṭha, as being the raising of Kuṇḍalinī. To this end, the Gorakṣaśataka and Yogabīja prescribe śakticālanī mudrā, the one physical yoga technique exclusive to early Nāth yoga. Meanwhile another early Nāth work, the Vivekamārtaṇḍa, co-opts the mudrās of haṭhayoga but they are still said to work only on bindu, not Kuṇḍalinī. In later Nāth and other Śākta works of the canon, the co-option of the haṭha techniques (along with their name) is more developed, so that in the Śivasaṃhitā the purpose of the haṭhayogic mudrās has become the raising of Kuṇḍalinī and bindudhāraṇa is not mentioned.

Other methods of yoga

I shall now turn to a typology of yoga which became commonplace and which, like the description of haṭha practices, is found for the first time in the Dattātreyayogaśāstra. Dattātreya teaches three methods of yoga: mantra, laya and haṭha. All three lead to rājayoga, i.e. samādhi.⁴⁶

In some texts these different yogas are hierarchised and said to be suitable for corresponding degrees of aspirant.⁴⁷ In contrast to the attitude implied in the Buddhist texts referred to at the beginning of this paper, which place mantrayoga at the centre of their systems of practice and mention haṭhayoga as the yoga of last resort, in haṭhayogic works mantrayoga is for the lowest level of aspirant,⁴⁸ middling aspirants are suited to layayoga ⁴⁹ and haṭhayoga is for those of the highest calibre.⁵⁰

Mantrayoga

Mantrayoga corresponds to the central practice of the tantras of the Śaiva Mantramārga, namely the repetition of mantras in order to obtain siddhis. Of the Sanskrit works on early haṭhayoga, only the two texts most obviously produced in Śākta milieux, namely the Khecarīvidyā and Śivasaṃhitā, teach mantras for obtaining siddhis.⁵¹ In other works mantra practice is omitted altogether⁵² or reduced to either the ajapā gāyatrī (the involuntary repetition of ha on the outbreath and sa on the inbreath)⁵³ or the repetition of oṃ as a purificatory technique.⁵⁴

Layayoga

Layayoga means ‘yoga through dissolution’. It is achieved by a wide variety of methods, the best known of which today is the raising of the serpent goddess Kuṇḍalinī upwards from the base of the spine to union with Śiva in the head via a sequence of, usually, six cakras. The cakras are associated with progressively more subtle elements and Kuṇḍalinī’s upward journey represents a reversal of creation, a laya or ‘dissolution’.⁵⁵

The Dattātreyayogaśāstra gives a mythological explanation of the origin of layayoga which I shall now translate in full:⁵⁶

[Dattātreya said]

‘Layayoga happens as a result of the dissolution of the mind by means of saṃketas, “esoteric techniques”. Ādinātha has taught eighty million saṃketas.’

Sāṃkṛti said:

‘Please tell me, what form does Lord Ādinātha take? Who is he?’

Dattātreya said:

‘The names of Mahādeva, the great god, are Ādinātha, Bhairava and Lord of the Śabaras. While that mighty god was sporting playfully with Pārvatī in the company of the leaders of his troop in [various places such as] Mount Śrīkaṇṭha, Śrīparvata, the top of a mountain in the region of the Banana Forest, [and] the mountain at Citrakūṭa covered with beautiful trees, he, Śaṅkara, out of compassion secretly told an esoteric technique to each of them in those places. I, however, cannot teach all of them in detail. I shall gladly proclaim some of them, [such as this one] which consists of a simple practice and is easy: While staying still [or] moving, sleeping [or] eating, day and night one should meditate on emptiness. This is one saṃketa taught by Śiva. Another is said to be simply staring at the tip of the nose. And meditation on the rear part of the head conquers death. The next saṃketa is said to be simply staring between the eyebrows. And that which is [staring] at the flat part of the forehead between the brows is said to be excellent. [Another] excellent dissolution is [staring] at the big toes of the left and right feet. Lying supine on the ground like a corpse is also said to be an excellent [dissolution]. If one practises in a place free from people while relaxed, one will achieve success. Śaṅkara has thus taught many saṃketas. That dissolution of the mind which occurs by means of [these] and several other esoteric techniques is layayoga. Next hear about haṭhayoga.’

Other works from the same period also mention saṃketas taught by Śiva.⁵⁷ Jñāndev’s commentary on the Bhagavadgītā, the Bhāvārthadīpikā (popularly known as the Jñāneśvarī), concludes its lengthy teachings on Kuṇḍalinī by declaring:⁵⁸

The swallowing of the body by the body: this is a saṃketa of Śiva, although here it is Lord Viṣṇu who explains it.

The Yogabīja, whose yoga is very similar to that of the Jñāneśvarī, concludes its description of the raising of Kuṇḍalinī in a similar fashion, saying:⁵⁹

This alone is Śiva’s saṃketa; it is characterised by [also] being the saṃketa of the siddhas.

These saṃketas thus denote a variety of practices said to have been taught by Śiva, some of which have Śākta overtones, in particular the raising of Kuṇḍalinī, who in the Jñāneśvarī passage cited above is said to be ‘mother of the world’ and ‘the highest goddess’.⁶⁰ Other roughly contemporaneous texts also give teachings on laya (without calling its techniques saṃketas). These are all overtly Śaiva and some also originate in milieux influenced by Śaivism’s more Śākta manifestions. Foremost is the practice of nādānusandhāna (which is also often called simply nāda): concentration on the internal sounds which arise in the course of yogic practice, sometimes as sequences of progressively more subtle sounds that correspond to the stages of Kuṇḍalinī’s ascent. The Śivasaṃhitā says that nāda is the best form of laya.⁶¹ Sequences of the sounds of nāda are taught in a variety of Śaiva works, including Śākta Tantras such as the Brahmayāmala, Kubjikāmata and Matsyendrasaṃhitā.⁶²

The second khaṇḍa of the Amanaska, which is likely to predate the Dattātreyayogaśāstra by a century or two and shares twenty-two verses with the Kulārṇavatantra,⁶³ also mentions laya among the results of its yoga.⁶⁴ Its yoga’s principal method is śāmbhavīmudrā, in which the yogin is to gaze outwards unblinkingly, while focussing internally. This technique is said first to have been given by Śiva to Umā, the primal Śakti. It is also taught in the Candrāvalokana of Matsyendra.⁶⁵ In the Amanaska’s first khaṇḍa, which probably postdates the Haṭhapradīpikā by about a century, the rewards of gradually increasing periods of laya are taught. In the last, which results from remaining in laya for twenty-four years, the yogin becomes absorbed in the śaktitattva.⁶⁶

In the Śārṅgadharapaddhati, a lengthy compendium of verses on a wide range of subjects compiled near Jaipur in 1363, laya is said to have been mastered by Kṛṣṇadvaipāyana and others, and to involve a series of meditations working upwards through nine cakras and finally bringing about the union of the body’s three śaktis.⁶⁷ The Amaraughaprabodha defines laya as meditation on Śiva in the form of a liṅga at Kāmarūpa’s location in the body, and as flowing amṛta.⁶⁸ The Khecarīvidyā does not define laya but does teach that after five months of visualisation of Kuṇḍalinī’s ascent and her flooding the body with amṛta, the yogin achieves laya in the five elements⁶⁹ and it associates laya with unmanī, the ‘supramental state’.⁷⁰

To summarise: the saṃketas or ‘secret techniques’ of layayoga comprise a variety of meditations on, and visualisations of, places and energies in the body, in contrast with the physical practices of haṭhayoga. They are taught by Śiva, often in works of Śākta orientation.

The yoga taught in the early texts associated with Nāth gurus, with its emphasis on the raising of Kuṇḍalinī, corresponds more to the techniques of laya than to those of haṭha. An early and disjointed attempt at combining the two can be found in the Vivekamārtaṇḍa; the somewhat later Śivasaṃhitā presents a more coherent synthesis; the most influential fusion is that found in the Haṭhapradīpikā, to which we now turn.

The classical haṭhayoga of the Haṭhapradīpikā

Soon after its compilation by Svātmārāma in the fifteenth century, the Haṭhapradīpikā became the most influential work on haṭhayoga, in part, no doubt, because it was the first text explicitly to make haṭha its central concern. Texts that taught the hierarchy of mantra, laya and haṭhayogas⁷¹were known to Svātmārāma, but he reduced their typology to a distinction between haṭhayoga and rājayoga, excluding mantra and laya yoga. The practices taught in the Haṭhapradīpikā are all haṭha and they lead to rājayoga, i.e. samādhi.

Mantra practice is notable by its complete absence in the Haṭhapradīpikā.⁷² In contrast, many of the techniques of layayoga are incorporated wholesale⁷³ and are thus for the first time taught under the rubric of haṭhayoga.⁷⁴ It is also in the Haṭhapradīpikā that various other practices which were to become emblematic of haṭhayoga are for the first time taught as part of haṭha. These include non-seated āsanas, complex kumbhakas (methods of breathing), and the ṣaṭ karmāṇi, six techniques for cleaning the body.

We see in the Haṭhapradīpikā the culmination of the process begun in the Vivekamārtaṇḍa, in which the Śākta laya technique of the visualisation of the rising of Kuṇḍalinī is overlaid onto the physical techniques of haṭhayoga, techniques originally used for the purpose of bindudhāraṇa, the preservation of semen.⁷⁵ The mudrās taught in the Dattātreyayogaśāstra and Vivekamārtaṇḍa as methods of bindudhāraṇa are now said to be for raising Kuṇḍalinī.⁷⁶

The Haṭhapradīpikā’s blanket agglomeration of the various techniques of haṭhayoga and layayoga creates a somewhat incoherent whole. The shoehorning of Kuṇḍalinī into the subtle physiology first found in the Amṛtasiddhi, in which the moon is in the head and the sun already occupies Kuṇḍalinī’s seat at the ādhāra at the base of the spine, is awkward.⁷⁷ The Haṭhapradīpikā teaches khecarīmudrā twice:⁷⁸ the first keeps bindu in the head and the second floods the body with amṛta.

The reformulation of the haṭhayogic mudrās as means of raising Kuṇḍalinī found in the Śivasaṃhitā is more coherent. Unlike Svātmārāma, the compiler of the Śivasaṃhitā completely removed bindudhāraṇa from the aims of the haṭhayogic mudrās. The coherence of the Śivasaṃhitā is partly because its compiler was happy to flaunt his sectarian affiliation, giving him a freer hand than Svātmārāma had. As noted above, the Śivasaṃhitā is a product of the Śākta Śrīvidyā tradition. As well as that cult’s mantras it teaches detailed visualisations of the cakras through which its techniques make Kuṇḍalinī rise. Each cakra is to be visualised as having a specific group of syllables in its spokes, together with a siddha and a yoginī, and the yoginīs correspond to those taught elsewhere in the texts of Śrīvidyā.⁷⁹ In contrast, the Haṭhapradīpikā, despite stating that the purpose of the haṭhayogic mudrās is the raising of Kuṇḍalinī, makes no mention of cakras. Svātmārāma, in addition to casting light on haṭha, had two unspoken further aims when compiling the Haṭhapradīpikā. He sought to lay claim to haṭhayoga for the siddha tradition while continuing a process that had started at least two centuries earlier, in which Śaiva yoga was being severed from its sectarian roots. The beginnings of this process can be seen in the teachings on yoga found in the Matsyendrasaṃhitā, in the introduction to his edition of which Csaba Kiss has made the observation that the text’s cult is indicative of:

A phase in the history of yoga when yogic teachings start to become detached (perhaps not for the first time) from the mainstream religion, in this case tantric Śaivism, by eliminating sectarian boundaries through the concealment of sectarian marks such as easily decodable deity names, mantras and iconography and start to prepare for a formative period of a pan-Indian yoga, which can again become an alternative for the official/conservative religion.⁸⁰

Thus Svātmārāma, while seeking to stake a claim on haṭhayoga for the siddha tradition, also sought to avoid alienating any of that tradition’s disparate elements. The list of synonyms of samādhi which he gives at the beginning of the Haṭhapradīpikā’s fourth upadeśa is a roll call of the various goals of different siddha traditions’ methods of yoga,⁸¹ but nowhere does he mention sect-specific features such as metaphysics, mantras, or sets of cakras.⁸²

Svātmārāma’s appropriation of haṭhayoga for the siddha tradition is evinced by his naming as masters of haṭhavidyā approximately thirty mahāsiddhas, including Matsyendra, Gorakṣa, Virūpākṣa, Cauraṅgī and Carpaṭi, whose names also occur in lists of Nāth gurus.⁸³ The texts that he used to compile the Haṭhapradīpikā, however, come from a much broader range of traditions. Thus Svātmārāma borrows approximately twenty verses from the Dattātreyayogaśāstra, yet he makes no mention of Dattātreya anywhere in the text. The Dattātreyayogaśāstra is the product of a Vaiṣṇava tradition which coalesced into the yogi suborders of the Daśanāmī Saṃnyāsīs, in particular the Giris and the Purīs.⁸⁴ Dattātreya is the tutelary deity of the Jūnā Akhāṛā, which today is the largest of the Saṃnyāsī akhāṛās and whose members are predominantly Giris and Purīs, and there is a long history of rivalry, or at least differentiation, between the Nāths, with Gorakṣa as their first guru, and the Saṃnyāsīs, with Dattātreya as their iṣṭadevatā.⁸⁵ To this day the Nāth Yogīs of northern India, despite displaying an otherwise broad inclusivity, will have little to do with Dattātreya.

Munis and siddhas

Some of the Haṭhapradīpikā’s teachings on the more gymnastic āsanas are taken from the Vasiṣṭhasaṃhitā.⁸⁶ As an ancient ṛṣi not closely associated with an ascetic order, there were no sectarian reasons to sideline Vasiṣṭha and he does get a brief mention in the Haṭhapradīpikā when it is said that the āsanas taught therein are those that were accepted by munis such as Vasiṣṭha and yogins such as Matsyendra.⁸⁷

This distinction between munis or ṛṣis and yogins or siddhas is found in a wide range of texts.⁸⁸ In a somewhat confused analysis based in the main on the Dattātreyayogaśāstra and the Vivekamārtaṇḍa, the Śārṅgadharapaddhati says that haṭhayoga is of two sorts, one practised by Gorakṣa and others, another by Mārkaṇḍeya and others.⁸⁹ The latter refers to the yoga taught to Mārkaṇḍeya by Dattātreya in the Mārkaṇḍeyapurāṇa. Elsewhere in the Śārṅgadharapaddhati rājayoga is said to be of two varieties.⁹⁰ The first, which was mastered by Dattātreya and other mahātmans, corresponds to the bindudhāraṇa haṭhayoga of the Dattātreyayogaśāstra; the second, whose practitioners are not identified, uses the breath to raise Kuṇḍalinī upwards through five cakras. The roughly contemporaneous Maithili Varṇaratnākara gives a list of siddhas and a list of munis.⁹¹Among the former are a number of Nāth gurus, including Gorakṣa; among the latter are Mārkaṇḍeya and Kapila.

What this boils down to is a distinction between layayoga-practising siddhas such as Gorakṣa and haṭhayoga-practising munis such as Kapila or Dattātreya, which manifests among today’s ascetics as a distinction between the Śākta Nāths and the relatively more orthodox Daśanāmīs and Rāmānandīs. This distinction should not be applied too rigorously, however, because there are many anomalies, in particular lexical ones. Thus Kapila, though most commonly said to be a muni, is described as a siddha in texts as diverse as the Bhagavadgītā and the Dattātreyayogaśāstraitself.⁹²

The muni tradition of haṭhayoga

Kapila’s ambiguous status is important for our understanding of the muni tradition which developed the techniques of early haṭhayoga. The first formulation of the practices distinguishing it from other methods of yoga is taught in the Dattātreyayogaśāstra where it is said to be the doctrine of the school of Kapila and other siddhas; the practices are taught as an alternative to the way of the kavi, the eightfold yoga practised by Yājñavalkya.⁹³ Bronkhorst has shown how Kapila was associated with the practice of ‘non-Vedic’ asceticism⁹⁴ and the Dattātreyayogaśāstra appears to confirm this when it contrasts Kapila’s yoga with that of the more orthodox Vedic ṛṣi Yājñavalkya.

The practices of ascetics beyond the Vedic pale are only mentioned in passing in early textual sources. In fact, very little of the actual practices of ascetics is taught in texts, Sanskrit or otherwise. This is not surprising: asceticism and scholarship are uncommon bedfellows.⁹⁵ The only ascetic practice that is treated in any depth in Sanskrit texts is yoga, and this is because it is one of the few of their practices which is not restricted to ascetics alone. The Śivasaṃhitā, for example, is explicitly aimed at householders.⁹⁶ The techniques of haṭhayoga are not taught in Sanskrit texts until the eleventh century or thereabouts. It is my contention that they did not appear ex nihilo, but that they developed from techniques practised by ascetics from at least the time of the Buddha. The Buddha himself is said to have tried both pressing his tongue to the back of his mouth, in a manner similar to that of the haṭhayogic khecarīmudrā,⁹⁷ and ukkuṭikappadhāna, a squatting posture which is perhaps a forerunner of the vīrāsana mentioned as part of ascetic practice in the Mahābhārata⁹⁸ and may be related to haṭhayogic techniques such as mahāmudrā, mahābandha, mahāvedha, mūlabandha⁹⁹ and vajrāsana¹⁰⁰ in which pressure is put on the perineum with the heel, in order to force upwards the breath or Kuṇḍalinī.¹⁰¹ Elsewhere in the Pali Canon these same practices are associated with tāpasas and Ājīvikas,¹⁰² who, together with other austerities, are also said to practise the ‘bat-penance’ (vagguli-vata), which is generally assumed to mean suspending oneself upside-down from a tree, thus inverting oneself in a fashion not dissimilar to the haṭhayogic viparītakaraṇī mudrā. In early Sanskrit sources too we find mentions of ascetics practising austerities suspended upside-down.¹⁰³ All these practices are undertaken by ascetics who are celibate and it seems likely that some of their austerities were linked with the preservation of semen in the manner of similar techniques taught in the texts of early haṭhayoga as methods of bindudhāraṇa, and that these ancient and previously obscure ascetic practices came to light as the mudrās of early haṭhayoga.

Within these older sources we find no mention of forerunners of vajrolimudrā,¹⁰⁴ the quintessential and crudest method of bindudhāraṇa, in which semen is resorbed through the urethra. Not all ascetics, whether those of mythology or scriptural prescription, were celibate¹⁰⁵ and it may be that some of their historical counterparts used the technique of vajrolimudrā in order to be able to have sex and preserve their bindu, to have their cake and eat it, as it were. It is this understanding of vajroli that has remained predominant in India, rather than the more infamous aim attributed to it in Śākta haṭhayogic texts such as the Śivasaṃhitā, that of absorbing the commingled sexual fluids of both the male and female partners.¹⁰⁶ Although our earliest clear-cut reference to vajroli is in the Dattātreyayogaśāstra,¹⁰⁷ a recent edition of a section of the Brahmayāmala which teaches the asidhāravrata, the ‘knife’s edge penance’,¹⁰⁸ suggests a milieu in which vajroli might have been used. As taught in the Brahmayāmala, the asidhāravrata was a form of coitus reservatus quite distinct from Śākta sexual practices which employed sexual fluids in ritual. Its emphasis on continence would have been nicely complemented by vajrolimudrā, which could have been employed should the ascetic have fallen off the knife-edge.

Among the sources cited thus far as evidence for the ancient muni tradition of bindu-oriented haṭhayoga, other than the Amṛtasiddhi there are no Śaiva works. Indeed, in Śaiva works that predate the Amṛtasiddhi we find no mention of bindudhāraṇa, the key principle of early haṭhayoga. But the Śaiva associations of some of early haṭhayoga’s practices are clear, even as taught in the Vaiṣṇava Dattātreyayogaśāstra.¹⁰⁹ Most obviously, there are the names of mudrās such as jālandharabandha, uḍḍīyānabandha, mahāmudrā, mahāvedha, vajroli, sahajoli, amaroli and khecarīmudrā, all of which are redolent of Śaivism. Mahāmudrā is said in the Dattātreyayogaśāstra to have been taught by Bhairava.¹¹⁰ And in vajrolimudrā there is perhaps a link between haṭhayoga and the practices of the earliest Śaiva ascetics. The asidhāravrata mentioned above was probably practised by Pāśupatas and Atimārga ascetics, who were separate from the Mantramārga traditions in which Śāktism developed.¹¹¹

Corroborating the link between the practices of haṭhayoga and those of earlier tapas-practising ascetics is the fact that the ascetic traditions whom our textual sources show to have been most closely associated with the practice of haṭhayoga since its first textual codification, traditions which are today best represented by the Daśanāmī Saṃnyāsīs and the Rāmānandīs, are those most closely associated with the practice of tapas. To this day their ascetic practices include many of the austerities mentioned together with the forerunners of haṭhayogic techniques in our earliest sources, and some of the āsanas taught in later manuals of haṭhayoga correspond exactly to ancient austerities such as standing on one leg or holding one or both arms in the air for long periods.¹¹² The origins of many of the physical practices of haṭhayoga are thus to be found in tapas, in particular the bodily austerities undertaken by the ascetics most often associated with the practice of yoga in early sources, rather than in the more rarefied formulations of mental yoga taught in, say, the Yogasūtra and its commentaries or Buddhist works.¹¹³ Indeed, formulations of yoga as yoga prior to the appearance of the haṭhayogic corpus show it to be a mental, meditative practice, its physicality limited to sitting in a suitable position and regulating the breath.

It was onto the bindudhāraṇa-oriented haṭhayoga of this ascetic tradition that the Kuṇḍalinī-oriented layayoga of the siddha tradition was grafted. The name haṭha suggests the difficult asceticism with which its early techniques had been associated, but this did not sit well with all of the practitioners of the newly Śākta haṭhayoga, either laya-practising siddhas or householders unsuited to practising tapas.¹¹⁴ In the Yogabīja we see an attempt to obfuscate the primary meaning of haṭha by resorting to an esoteric nirukti which is often repeated in later works and secondary literature: ha means ‘sun’, ṭha means ‘moon’ and their yoga or union is haṭhayoga.¹¹⁵ The rehabilitation of the word haṭha in the face of yoga’s universalisation was ultimately unsuccessful, however, and the use of the word in the context of yoga fell out of favour in all but scholastic typologies of the subject and, in recent years, in secondary and popular literature on yoga. In vernacular literature haṭh continued to refer to extreme ascetic practices.¹¹⁶

Conclusion

I shall conclude with some speculative remarks on the broader context of the Śākta co-option of haṭhayoga.

First, my findings raise the question of why these physical yoga techniques started to be codified approximately 1,000 years ago, particularly if, as I contend, they were already ancient. This seems to me to be connected with the severing of yoga from its sectarian moorings mentioned earlier, which served to make yoga available to all. Much of the practice of haṭhayoga as taught in its texts appears to be for ascetics. Several of its techniques are distasteful or difficult and take a lot of time and effort to master: the yogin is often instructed to carry out his practice in an isolated hut. But at the same time most of the texts of haṭhayoga are explicitly anti-sectarian and universalist,¹¹⁷ and state or imply that anyone can practise yoga.¹¹⁸ Vajrolimudrā in particular – if only as an ideal¹¹⁹ – would have assisted in opening up the yoga practice of ascetics to all. The Śivasaṃhitā says that through its practice even householders can attain liberation, without observing the restrictions taught in the texts of yoga.¹²⁰

Written in simple Sanskrit and free from the abstruse metaphysics of the Yogasūtra and its exegesis, or the esoterica of Śaiva yoga manuals, the texts of haṭhayoga are the first works on yoga that are accessible to all. This made its aims, liberation and siddhis, also accessible to all, without the need for priestly intermediaries, ritual paraphernalia or sectarian initiations. We see a similar democratisation of religion in the bhakti cults that started to develop during the same period and this may have been a corollary of the demise of Śaivism, at least as a grand, state religion.

The democratisation of yoga was responsible for the production of its texts. Ascetics had learnt haṭhayoga through oral teachings for centuries, but when its teachings were opened up to householders texts were produced, perhaps as the result of patronage by these new practitioners, who would have enlisted pandits to codify the teachings of ascetic gurus. At first these yoga manuals were written in Sanskrit, which would have limited their audience, but vernacular texts containing teachings on yoga soon appeared. Jñāndev’s late-thirteenth-century Jñāneśvarī contains exquisite teachings in Marathi on the ascent of Kuṇḍalinī, and the Tamil Tirumantiram, which probably dates to a similar period,¹²¹ has extensive teachings on haṭhayogic techniques. Vernacular yoga manuals first appear in the north in the sixteenth century, with the Persian Baḥr al-Ḥayāt.¹²² This is followed in the seventeenth century by the Braj Bhāṣā Jñānasamudra and Sarvāṅgayogapradīpikā of the Dādūpanthī scholar Sundar Dās, and a variety of Persian translations of Sanskrit works, such as the Khulāṣah al-Khulāṣah of Devī Dās Kāyastha.¹²³ A noteworthy eighteenth-century vernacular manual of yoga is the Braj Bhāṣā Jogpradīpakā of the Rāmānandī Jayatarāma.¹²⁴

The universalism and antisectarianism taught in the texts of haṭhayoga, which facilitated the coming together of the Śākta and ascetic yogas, was mirrored by the ascetic milieu of the time, which saw a common ascetic identity develop, drawing on a wide range of sources, from Śākta to Sufi, with the result that, to this day, the yogins of the various North Indian ascetic orders are very similar in appearance and lifestyle. However, by the seventeenth century, out of this relatively homogeneous ascetic soup, various ascetic sects did solidify. The co-option of haṭhayoga by the fledgling Nāths may have been part of this process of sect-formation, symbolising their abandonment of more licentious Kaula practices in favour of a celibate ascetic lifestyle. Thus the final verse of the Gorakṣaśataka is the following declaration:¹²⁵

We drink the dripping liquid called bindu, ‘the drop’, not wine; we eat the rejection of the objects of the five senses, not meat; we do not embrace a sweetheart [but] the Suṣumnā nāḍī, her body curved like kuśa grass; if we have intercourse + … + it takes place in a mind dissolved in the void, not in a vagina.

The Śākta appropriation of haṭhayoga – and yoga more broadly conceived – was entirely successful, to the extent that even the celibate bindu-yoga traditions were made more Śākta through incorporating Kuṇḍalinī into their practice. She features widely in the so-called ‘Yoga Upaniṣads’ compiled in the seventeenth and eighteenth centuries; several of the mudrās taught in the Jogpradīpakā are said to raise Kuṇḍalinī; and a Daśanāmī Saṃnyāsī whom I met at Gangotri in 2006 told me that his mastery of vajrolimudrā enabled him to resorb his semen should Kuṇḍalinī make him involuntarily ejaculate as she passed through his svādhiṣṭhāna cakra.

Haṭhayoga’s Śākta veneer obscured the origins of its practices. In secondary literature it is widely assumed that the mudrās of haṭhayoga arose in Śākta milieux. As I have shown, it is likely that they in fact developed amongst non-Vedic ascetic traditions. To add to the textual evidence I have cited, there are two pieces of external, negative evidence that suggest that the bindudhāraṇa techniques unique to early haṭhayoga were not part of Śākta practice. First, khecarīmudrā, vajrolimudrā and viparītakaraṇī are not found in Tibetan sources, implying that they were not part of the tantric traditions, in particular that of the siddhas, including ‘Nāths’ such as Matsyendra and Gorakṣa, which travelled to Tibet during the centuries prior to the fixing of the Tibetan Buddhist canon. Second, I have long puzzled over how, if khecarīmudrā developed in Śākta milieux, it is never compared with sexual intercourse in our textual sources, when there are such obvious parallels between the tongue entering the void above the palate and the penis entering a vagina; I can only surmise that khecarīmudrā developed outside of a milieu in which sexual rituals were practised, such as the ancient non-Vedic ascetic tradition.

The Śākta appropriation of haṭhayoga and its associated adoption by the Nāths meant that yoga and yogis of all stripes came to be associated by some with the more malevolent practices of Śāktism,¹²⁶ an association that persists to this day despite the oldest and most persistent haṭhayogatradition being that of the generally benevolent munis, represented today by the Daśanāmī Saṃnyāsīs and Rāmānandīs. It is the muni tradition that has been responsible for all the new texts and exegesis of haṭhayoga since the time of the Haṭhapradīpikā. Meanwhile the Śākta Nāth Yogīs have produced no further manuals of haṭhayoga and ceased to practise it.¹²⁷ Those Nāths whose sādhanā is anything other than an ascetic lifestyle and archetypal North Indian bhakti remain true to their roots and perform Śākta tantric ritual.

Notes

1  I would like to thank the organisers of the Śākta Traditions conference held in Oxford in September 2011 for inviting me to present the first draft of this paper and the participants at the conference for their helpful comments. I would also like to thank Jason Birch, Patton Burchett and Véronique Bouillier for their detailed comments on subsequent drafts.

2  Sanderson 2009.

3  It is at this time that yoga is first included among the darśanas, in the c. twelfth-century Sarvasiddhāntasaṃgraha, which was composed by an anonymous Advaitavedāntin (Halbfass 1988: 352). Over the subsequent centuries the orthodox (in particular Vedāntin) interest in the haṭhatechniques of yoga continued to grow, culminating in the composition in the seventeenth to eighteenth centuries of a group of ‘Yoga Upaniṣads’ which consist for the most part of passages from earlier haṭha texts (Bouy 1994).

4  Examples of this are legion. To give just three, from different contexts: Bouy (1994: 6) who describes all the texts used to compose the ‘Yoga Upaniṣads’ as Nāth; Vaudeville (1957: viii–xii) who says that Kabīr borrowed various yogic elements from the tradition of the Nāths; and van der Veer (1989: 89) who says that the Rāmānandīs have been ‘deeply influenced’ by the Nāths.

5  For an overview of the history of the Nāth saṃpradāya see Mallinson 2011c.

6  See e.g. Bouillier 1997: 147; Vilāsnāth 2010.

7  In the late medieval period ascetic Nāths claiming allegiance to Gorakhnāth subscribed to nirguṇī and, subsequently, Śaiva tenets (see Mallinson 2011c). Gorakhnāth is portrayed in later legends as a celibate Śaiva ascetic who reformed the licentious ways of his Kaula tantric guru Matsyendranāth (Mallinson 2007: 186 n129). He is also said to have overcome the goddess in magical contests (Bouillier 2008: 44). With Gorakhnāth coming to be seen by most Nāths as the head of the Nāth saṃpradāya, the nirguṇī Śaivism associated with him came to overshadow the order’s Śākta roots. Nevertheless, as noted above Nāth liturgy retains Śākta features and in some hagiographies Gorakhnāth is said to be a worshipper of the goddess (e.g. in the current story of the origins of the Gorakh Ḍibbī temple at Jwalamukhi).

8  On the Buddhist texts which mention haṭhayoga, see Birch 2011.

9  In this definition haṭhayoga is a gloss of the mūla’s ‘haṭha’. This is the first of many instances of practices referred to in primary sources simply as haṭha being said to constitute haṭhayoga in later systematisations, exegesis and secondary literature. This has resulted in the Haṭhapradīpikāoften being referred to in secondary literature as the Haṭhayogapradīpikā when only a tiny fraction of its manuscripts call it thus. We see in this the scholastic tendency towards categorisation, to identify different types of yoga and to pit them against one another. But it should be noted that the situation is more nuanced in the texts themselves: even those which do categorise different varieties do not see them as altogether different types of yoga but as contrasting (yet sometimes complementary) methods of achieving yoga. Thus compounds such as haṭhayoga and layayogashould be understood as tatpuruṣas with an instrumental case relationship: ‘yoga by means of haṭha’ etc. The one exception to this is rājayoga, which, as the aim of all methods of yoga, should be understood either as a karmadhāraya or as a tatpuruṣa of genitive relationship: ‘the royal yoga’. A small number of texts do identify their yoga as rājayoga but this is to emphasise the superiority of their method above all others rather than give a definition of rājayoga (see e.g. Amanaska 2.3–4; Vasudeva 2011).

10  This earliest definition of haṭhayoga is repeated verbatim in other Buddhist exegetical works (see Birch 2011). The verse from the Kālacakratantra on which Puṇḍarīka is commentating (4.119cd) says that if the siddhi desired by mantra-practitioners does not arise as a result of purification, yogic withdrawal and so forth, then they should accomplish it by forcefully (haṭhena) restraining bindu in the vajra in the lotus:

saṃśuddhipratyāhārādibhir vai yadi bhavati na sā mantriṇām iṣṭasiddhir nādābhyāsād dhaṭhenābjagakuliśamaṇau sādhayed bindurodhāt |

Puṇḍarīka glosses haṭhena with haṭhayogena and defines it thus (Vimalaprabhā vol. 2, p. 212):

idānīṃ haṭhayoga ucyate | iha yadā pratyāhārādibhir bimbe dṛṣṭe saty akṣarakṣaṇaṃ notpadyate ayantritaprāṇatayā tadā nādābhyāsād dhaṭhena prāṇaṃ madhyamāyāṃ vāhayitvā prajñābjagatakuliśamaṇau bodhicittabindunirodhād akṣarakṣaṇaṃ sādhayen niḥspandeneti haṭhayogaḥ ||

11  Bouy 1994: 82–85.

12  For the details of the texts used and verses borrowed, see my forthcoming monograph Yoga and Yogis: the Texts, Techniques and Practitioners of Traditional Haṭhayoga.

13  Haṭhapradīpikā 1.3:

bhrāntyā bahumatadhvānte rājayogam ajānatām |

haṭhapradīpikāṃ dhatte svātmārāmaḥ kṛpākaraḥ ||

14  Dattātreyayogaśāstra 9, 26; Yogabīja 143, 149; Amaraughaprabodha 3, 4, 5, 7, 10, 24, 29, 73 and Śivasaṃhitā 5.12, 5.23, 5.222, 5.254.

15  For precursors of jālandharabandha, see e.g. Sarvajñānottara Yogapāda 11ab as edited by Vasudeva (2004: 398 n.77): āsanaṃ ruciraṃ baddhvā ūrdhvakāyam adhaḥśiraḥ which teaches that in all the four seated āsanas that have just been named the head is to be held down; cf. Vaikhānasasmārtasūtra 8.8 in which having the face turned downwards is included in a list of ascetic practices. See Mallinson 2007: 17–24 on forerunners of the haṭhayogic khecarīmudrā. On precursors of mūlabandha, see ibid., p. 15. These techniques are often prescribed, without being named, as adjuncts to other mudrās and āsanas in haṭhayogic texts. See e.g. Amṛtasiddhi 11.4, Dattātreyayogaśāstra 133, Jñāneśvarī 6.207–208, Matsyendrasaṃhitā 3.28–30, Vivekamārtaṇḍa 8, Khecarīvidyā 3.25.

16  A measure of the validity of using the Haṭhapradīpikā to establish this corpus in this way – as well as of the Haṭhapradīpikā’s inclusivity – is that we know of no text which predates it and teaches early haṭha techniques but was not used in its compilation. (As noted above, some of the haṭha practices, in particular the three bandhas, are taught in earlier works, but as one among a variety of attitudes to be assumed in meditational, ascetic or yogic postures: they are not treated individually, nor do they take the names by which they are known in haṭhayogic works.)

17  In terms of chronology, the corpus splits neatly into two halves. The Amṛtasiddhi, Dattātreyayogaśāstra and Vivekamārtaṇḍa are named or cited in fourteenth-century works (Bu ston Rin chen grub’s 1322 catalogue of canonical works in the case of the Amṛtasiddhi (see Schaeffer 2002), the 1363 Śārṅgadharapaddhati in the case of the Dattātreyayogaśāstra (over thirty verses cited) and the Khecarīvidyā (1.14) in the case of the Vivekamārtaṇḍa). Verses from the central core of the Gorakṣaśataka are found in the Yogabīja, which also borrows from the Dattātreyayogaśāstra. This, together with the coherence of the Gorakṣaśataka’s teachings, suggests that the Gorakṣaśataka was the source. The four texts that constitute the second half of the corpus are all to some extent derivative of those in the first half, indicating their later date. Further details of their dependence on the earlier works are given in the notes to the descriptions of the texts below.

18  Schaeffer, on the strength of the Amṛtasiddhi teaching haṭhayoga and the text cycle which it is part of being traced back to Virūpākṣa/Virūpākṣanātha, identifies the text as Nāth (2002: 518–519). In the text itself, however, there is nothing to suggest that it was composed in a Nāth, or even Paścimāmnāya, milieu. As well as there having been a siddha called Virūpākṣa/Virūpā, the name Virūpākṣa has been used from at least the twelfth century to refer to the form of Śiva that presides over the Vijayanagar region (Verghese 1995: 18) and the Amṛtasiddhi could also be a product of that Śaiva tradition, in particular the Kālamukha cult that flourished there prior to the Saṅgama kings’ patronage of Vaiṣṇavism.

19  Amṛtasiddhi 11–13.

20  In the Amṛtasiddhi (as in later haṭhayogic texts) the archaic triad of sun, moon and fire is reduced to the pair sun and moon, the sun and fire being identified as one (Amṛtasiddhi 5.2).

21  See Amṛtasiddhi viveka 7.

22  The Dattātreyayogaśāstra makes no mention of the usual six cakras of yogic physiology (although at verse 131 the sahasrakamala, which is sometimes added to that schema as a seventh cakra, is said to be the source of amṛta) but does refer to Kuṇḍalinī in passing, saying that in the paricaya stage of yoga she and the breath are made to move by fire (107). The Amṛtasiddhi mentions neither Kuṇḍalinī nor cakras. (At 7.8–13 it does say, however, that a solar rajas, the feminine equivalent of bindu, is found in all beings ‘wrapped in the goddess element’ (devītattvasamāvṛtaḥ), and that its union with the lunar bindu is yoga; see footnote 100 for the text of this passage. Cf. Vivekamārtaṇḍa 54–57; Yogabīja 89.)

23  On the confusion between the Gorakṣaśataka and the Vivekamārtaṇḍa caused by the latter also coming to be known, among other names, as the Gorakṣaśataka, see Mallinson 2011b: 262–263.

24  The text says that the three bandhas are used to restrain the breath in order to stimulate Kuṇḍalinī, who is to be stimulated in order to control the breath, suggesting that the Kuṇḍalinī paradigm has been imposed onto a yoga that worked on breath alone.

25  Later redactions of the Gorakṣaśataka’s verses, such as that in the Haṭhapradīpikā, through textual corruption and, we must assume, either a rupture in the transmission of the practice or a dissociation of the textual and practical traditions, say that the cloth is to be wrapped around the waist (see Mallinson 2011b).

26  Vivekamārtaṇḍa 31–39. In the Vivekamārtaṇḍa Kuṇḍalinī is not said to interact with the cakras in any way.

27  Vivekamārtaṇḍa 40–63 teaches mūlabandha, uḍḍīyānabandha, jālandharabandha, khecarīmudrā and mahāmudrā.

28  Vivekamārtaṇḍa 113–115.

29  The Yogabīja includes a number of verses from the Dattātreyayogaśāstra and Gorakṣaśataka and is likely to postdate them (Yogabīja 116–117, 119c-121b, 122c-123b = Dattātreyayogaśāstra 144–145, 141c-143b, 138; Yogabīja 103, 105c-106b, 109ab, 110c-115d, 121c-122b = Gorakṣaśataka 35c-36b, 38c-39b, 45ab, 47a-52b, 61c-62b). The following verses are found in both the Yogabīja and Haṭhapradīpikā but in no other texts, so it is likely that the Yogabīja is their source: Yogabīja 92, 94, 118c-119b, 123c-124b, 125 (= Haṭhapradīpikā 3.109, 3.112, 3.54, 2.46, 3.111).

30  Yogabīja 92a-98b, 108–112, 123c-132d.

31  Yogabīja 123, 140.

32  Khecarīvidyā 1.14.

33  The Amaraughaprabodha shares verses with the Amṛtasiddhi, Amanaska and Śivasaṃhitā (Amaraughaprabodha 20, 29, 32cd, 37ab, 38, 39cd, 40ab, 45 = Amṛtasiddhi 16.1c-2b, 11.3, 11.9cd, 11.3cd, 14.6, 13.5cd, 13.7cd, 19.2 (many other verses in the Amaraughaprabodha are derivative of verses in the Amṛtasiddhi, particularly those describing the four classes of aspirant – compare Amaraughaprabodha 18–24 with Amṛtasiddhi vivekas 15–18); Amaraughaprabodha 8 = Amanaska 2.32; Amaraughaprabodha 3, 45 = Śivasaṃhitā 5.12, 3.31 (Amṛtasiddhi 19.2 is likely to be the original source of the latter)). It also includes a quotation attributed to a Śrīsaṃpuṭa. It is thus to some extent a compilation, but it contains nineteen verses found in the Haṭhapradīpikā and not elsewhere (Amaraughaprabodha 9, 30–31, 32ab, 34 = Haṭhapradīpikā 4.14, 3.10–11b, 3.12ab, 3.19; Amaraughaprabodha 35ab, 35cd, 36, 40c-41d, 42ab, 42cd, 43ab, 43c-44b = Haṭhapradīpikā 3.22ab, 3.23cd, 3.24, 3.26c-27d, 3.29ab, 3.30ab, 3.29cd, 3.30c-f; Amaraughaprabodha 44cd, 46–53 = Haṭhapradīpikā 1.61ab, 4.70–77), so in the absence of an alternative source for those verses it seems likely that the Amaraughaprabodha was compiled before the Haṭhapradīpikā.

34  Amaraughaprabodha 1:

oṃ namo ‘stv ādināthāya mīnanāthāya vai namaḥ |

namaś caurāṅgināthāya siddhabuddhāya dhīmate ||

35  Amaraughaprabodha 29–33, 55.

36  The Śivasaṃhitā shares sixteen verses with the Amṛtasiddhi (Śivasaṃhitā 2.1b, 2.1cd, 2.2, 2.3, 2.4ab = Amṛtasiddhi 1.15b, 1.16ab, 1.17ab+1.16cd, 1.17c-1.18b, 1.19ab; Śivasaṃhitā 2.6c-9d, 2.11–12, 3.31, 4.27ab, 4.27dc, 4.28ab, 4.34cb, 5.13, 5.17c-20b = Amṛtasiddhi 3.1–4, 4.3–4, 19.2, 11.3cd, 11.4bc, 11.5ab, 11.7cd, 15.1, 16.1–3; Śivasaṃhitā 4.31, 4.38, 5.14–15 ≈ Amṛtasiddhi 11.6, 12.6, 15.3; several other verses in the Śivasaṃhitā’s descriptions of mahāmudrā, mahābandha and mahāvedha are derivative of verses in the Amṛtasiddhi); four verses with the Dattātreyayogaśāstra (Śivasaṃhitā 3.102–105 = Dattātreyayogaśāstra 35–38; the Śivasaṃhitā also paraphrases passages from the Dattātreyayogaśāstra: compare Śivasaṃhitā 5.71, 3.42–48, 3.60–61, 3.62–63, 3.72–75, 4.88ab etc. with Dattātreyayogaśāstra 22c-23b, 72c-82b, 98c-100b, 89c-90b, 112–121, 157cd); one verse with the Śāradātilaka (Śivasaṃhitā 2.16 = Śāradātilaka 25.32); and one with the Vivekamārtaṇḍa (Śivasaṃhitā 3.81 = Vivekamārtaṇḍa 119). Ten of the fifteen verses which the Śivasaṃhitā shares with the Haṭhapradīpikā are not to be found in other texts (Śivasaṃhitā 3.5a-6b, 3.43, 3.108–109, 4.21, 4.23ab, 4.47ab, 4.88ab, 5.47, 5.222a-d, 5.254 = Haṭhapradīpikā 4.22a-23b, 2.14, 1.18–19, 3.2, 3.6ab, 3.24cd, 3.87ab, 1.43, 2.76, 1.11).

37  See the invocations of the goddess Tripurabhairavī at 5.84 and 5.240, and teachings on the associated tripartite Śrīvidyā mantrarāja/mūlavidyā. The mantrarāja consists of vāgbhava + kāmarūpa + śakti, whose condensed forms are aiṃ, klīṃ,and sauḥ; it is taught at 5.80–85 and 5.234–235. Its three components also combine to make the mūlavidyā in the Vāmakeśvarīmata (1.93–101). The yoginīs said in the Śivasaṃhitā to be situated at the cakras correspond to those in Bhāskararāya’s Saubhāgyabhāskara commentary on Lalitāsahasranāmastotra 148–158. The assertions that the yogi can make himself irresistible to women (5.104), that he can become a second Kāma (3.93), and instructions to meditate on Kāma (4.3), have parallels with the ‘love magic’ taught in the Vāmakeśvarīmata and the associated earlier Nityā tantras, a Śākta textual corpus from which the cult of Tripurā developed and which was independent of the Vidyāpīṭha literature (Sanderson 2009: 47; cf. Golovkova 2010: 9, 41–53). The Śivasaṃhitā’s vedantic teachings (see in particular the first paṭala) confirm the connection with the cult of the southern Śaṅkarācāryas of Kanchi and Shringeri.

38  Śivasaṃhitā 5.250a-251b:

triṃśallakṣais tathā japtair brahmaviṣṇusamo bhavet |

rudratvaṃ ṣaṣṭibhir lakṣaiḥ śaktitattvam aśītibhiḥ ||

koṭyaikayā mahāyogī līyate parame pade |

39  Vamakeśvarīmata 1.164 and 2.13, and 3.27 in Jayaratha’s Vāmakeśvarīmatavivaraṇa ad loc.

40  Śivasaṃhitā 2.23b.

41  Śivasaṃhitā 2.24a.

42  Śivasaṃhitā 4.23–24.

43  The Dattātreyayogaśāstra’s teachings on vajroli (vv. 150c-159b) give only bindudhāraṇa as its aim, not the absorption of commingled sexual fluids. The Haṭhapradīpikā’s teachings on vajroli (vv. 3.82–89) describe it as a technique of bindudhāraṇa for the male yogin (a verse found in some witnesses after 3.89, but not in the edited text, enjoins the absorption of both rajas and bindu). At the end of the section teaching amaroli and sahajoli, two practices supplementary to vajroli, it is said (vv.~3.95–98) that the true yoginī absorbs male bindu mixed with her own rajas, thereby achieving total success (sarvasiddhim), knowing the past and present and becoming a khecarī.

44  Śivasaṃhitā 4.78–87.

45  Kuṇḍalinī is not mentioned in the context of vajroli even in the Śivasaṃhitā, but in the c. 17th-century Haṭharatnāvalī the practice is explicitly said to awaken her (2.82).

46  Dattātreyayogaśāstra 10a-11b.

47  The different degrees of aspirant are also taught in the Amṛtasiddhi (vivekas 15–18) but the Amṛtasiddhi does not give a corresponding typology of yogas.

48  Dattātreyayogaśāstra 12–14.

49  Śivasaṃhitā 5.17–20.

50  Śivasaṃhitā 5.21–23.

51  The teachings on the mantra of Khecarī that were central to the earliest form of the Khecarīvidyā (before the teachings on the haṭhayogic khecarīmudrā were incorporated into it) become a relic in its later redactions. The coded definition of the mantra was obscure to commentators and the verses in which it is taught were unintelligible to scribes, so were corrupted in most, if not all, recensions of the text. See Khecarīvidyā 1.30c-38b and Mallinson 2007: 199 n.225.

52  This is most striking in the Haṭhapradīpikā, on which see below.

53  Vivekamārtaṇḍa 28–29, Gheraṇḍasaṃhitā 5.86–87; cf. Śāradatilaka 25.50–52.

54  Dattātreyayogaśāstra 87c-88d, Vivekamārtaṇḍa 70, Śivasaṃhitā 3.54, 3.71; cf. adhyāya 6 of the Yogayājñavalkya.

55  Dissolution into the elements is also mentioned at Yogasūtra 1.19: bhavapratyayo videhaprakṛtilayānām. Śaiva works predating the haṭhayogic corpus often mention laya, particularly in the context of meditations that lead the yogin up through a hierarchy of elemental tattvas, and they teach some of the techniques associated with the layayoga taught in haṭhayogic works, but they do not teach laya to be a particular variety of yoga (personal communication from Alexis Sanderson, June 2011).

56  Dattātreyayogaśāstra vv. 15–26:

layayogaś cittalayāt saṃketais tu prajāyate |

ādināthena saṃketā aṣṭakoṭiḥ prakīrtitāḥ ||15||

sāṃkṛtir uvāca

bhagavan ādināthaḥ saḥ kiṃrūpaḥ kaḥ sa ucyatām ||16||

dattātreya uvāca

mahādevasya nāmāni ādināthaś ca bhairavaḥ |

śabareśaś ca devo ’yaṃ līlayā vicaran prabhuḥ ||17||

śrīkaṇṭhaparvate gauryā saha pramathanāyakaiḥ |

mithaḥ śrīparvate caiva kadalīvanagocare ||18||

girikūṭe citrakūṭe supādapayute girau |

kṛpayaikaikasaṃketaṃ śaṃkaraḥ prāha tatra tān ||19||

tāni sarvāṇi vaktuṃ hi na śaknomi tu vistarāt |

kāni cit kathayiṣyāmi sahajābhyāsavat sukham ||20||

tiṣṭhan gacchan svapan bhuñjan dhyāyec chūnyam aharniśam |

ayam eko hi saṃketa ādināthena bhāṣitaḥ ||21||

nāsāgradṛṣṭimātreṇa aparaḥ parikīrtitaḥ |

śiraḥpāścātyabhāgasya dhyānaṃ mṛtyuṃ jayet param ||22||

bhrūmadhyadṛṣṭimātreṇa paraḥ saṃketa ucyate |

lalāṭe bhrūtale yaś ca uttamaḥ saḥ prakīrtitaḥ ||23||

savyadakṣiṇapādasya aṅguṣṭhe layam uttamam |

uttānaśavavad bhūmau śayanaṃ coktam uttamam ||24||

śithilo nirjane deśe kuryāc cet siddhim āpnuyāt |

evaṃ ca bahusaṃketān kathayām āsa śaṅkaraḥ ||25||

saṃketair bahubhiś cānyair yaś cittasya layo bhavet |

sa eva layayogaḥ syād haṭhayogaṃ tataḥ śṛṇu ||26||

B = Dattātreyayogaśāstra, ed. Brahmamitra Avasthī, Svāmī Keśavānanda Yoga Saṃsthāna 1982 • J1 = Mān Siṃh Pustak Prakāś 1936 • W1 = Wai Prajñā Pāṭhaśālā 6/4–399 • M = Mysore Government Oriental Manuscripts Library 4369 • W2 = Wai Prajñā Pāṭhaśālā 6163 • U = Yogatattvopaniṣad

15a °layāt] J1W1MNW2; °layaḥ BNU15b saṃketais] BNW1W2; saṃketas J1MN, kotiśaḥ U • tu prajāyate] parikīrtitaḥ U15d saṃketā aṣṭa°] BJ1W1W2; saṃketās sārdha° M 16a °nāthaḥ saḥ] B; °nāthasya J1W1, nāthaś ca M, nāthaś ca W2 16b kiṃrūpaḥ kaḥ sa ucyatām] B; ko yaṃ rūpaḥ sa ucyate J1, kiṃrūpaṃ kaḥ sa ucyatāṃ W1, ko ‘yaṃ puruṣa ucyate M, ko yaṃ rūpaḥ sa ucyatāṃ W2 17b °nāthaś ca bhairavaḥ] MW2; °nāthādikāny api BW1, °nāthasya bhairavaḥ J117c śabareśaś ca devo ’yaṃ] MW2; śiveśvaraś ca devo ’sau B, śivarīśaś ca devo yaṃJ1, śiveśvaraś ca devo yaṃ W1 17d vicaran] J1W1MW2; vyacarat B18a śrīkaṇṭha°] BJ1W1 ; śrīkaṇṭhaḥ MW2 18b saha pramatha°] BJ1W1W2; sahasrapramatha° M (unmetrical) • °nāyakaiḥ] M; °nāyakān BJ1W1W2 18c mithaḥ śrīparvate caiva] M; himākṣaparvate caiva BJ1W1, himālayas tu parvate W2 19bsupādapayute girau] BW1; sapādaniyatai gurau J1, sa yathā niyayoginaḥ M, sapādanilaye girau W2 19d śaṃkaraḥ prāha tatra tān] BJ1W1W2; śabarebhyo hi dattavān M 20b na śaknomi tu vistarāt] BW1; na śaknomīti vistarāt J1, nānyaś śakto hi taṃ vinā M, nan saknomi tu vistarāt W2 21bdhyāyec] W1M; dhyāyan BJ1W2, dhyāyen U • chūnyam ahar niśam] niṣkalam īśvaram U22c °pāścātya°] M; °paścāc ca BJ1W1, paścāt W2 (unmetrical)22d mṛtyuṃ] mṛtyu W2 • jayet param] BW1; jaye param J1W2, °jayo varaḥ M 23clalāṭe bhrūtale yaś] BJ1; lalāṭe bhūtale yaś W1W2, līlāvibhūtilepaś M 24c uttāna°] BJ1M; uttānaḥ W1W2 • °śavavad] BW1M; °śikha/rā\d J1, śaravad W2 24d śayanaṃ coktam uttamam] śapasanacottamottamaṃ J1 (unmetrical), śayanaṃ cottamottamam M 25a śithilo] BW1W2; śithilā J1, haṃseva M 26d haṭha°] M; karma° B, dharma° J1W1, dhaṭa° W2

57  The saṃketas of the siddhas are also mentioned frequently in the Hindi verses attributed to Gorakṣa. See e.g. the Gorakhbāṇī, sākhīs 106, 115, 217.

58  Jñāneśvarī 6.291:

piṃḍeṃ piṃḍācā grāsu | to hā nāthasaṃketicā ḍaṃsu |

pari dāunu gelā uddeśu | māhāviṣṇu || 291 ||

See Kiehnle 2005: 31. Kiehnle (ibid.) understands nātha° here to refer to the Nātha saṃpradāya of Yogīs but both internal and external evidence suggest otherwise: the poet is pointing out that the teaching derives from Śiva even though it is being told by Viṣṇu, and the verse was written several centuries before nātha was used to refer to members of an order of yogis (Mallinson 2011c).

59  Yogabīja 136ab:

sa eva nāthasaṃketaḥ siddhasaṃketalakṣaṇaḥ |

60  Jñāneśvarī 6.272, 6.281.

61  Śivasaṃhitā 5.44 and 5.47.

62  On nāda in Śaivism see Vasudeva 2004: 273–280.

63  On the Amanaska, see Birch 2012.

64  See Amanaska 2.22, 2.57.

65  Amanaska 2.10–11 (2.10 = Candrāvalokana 1, Haṭhapradīpikā 4.36):

antarlakṣyaṃ bahirdṛṣṭir nimeṣonmeṣavarjitā |

eṣā hi śāmbhavī mudrā sarvatantreṣu gopitā ||2.10||

ādiśaktir umā caiṣā matto labdhavatī purā |

adhunā janmasamskārāt tvam eko labdhavān asi ||2.11||

66  Amanaska 1.82:

caturviṃśatibhir varṣair layasthasya nirantaraṃ |

śaktitattvasya siddhiḥ syāc chaktitattvamayo bhavet |

67  Śārṅgadharapaddhati 4350–4363.

68  Amaraughaprabodha 27.

69  Khecarīvidyā 2.39.

70  Khecarīvidyā 2.113ab.

71  I.e. the Dattātreyayogaśāstra, Yogabīja, Amaraughaprabodha and Śivasaṃhitā.

72  Mantrayoga does perhaps get a lexical nod in the name by which Svātmārāma classifies the cleansing practices, the verses teaching which are among the few that I am yet to find in earlier works and which may thus be the work of Svātmārāma himself. Despite teaching seven such practices, he calls them the ṣaṭ karmāṇi, ‘six acts’, which is the name also given to a group of six magical methods of overpowering one’s enemies that can be activated by means of mantras taught in texts of the Śaiva Mantramārga (see Bühnemann 2000; the same phrase can also refer to the six duties of a brahmin).

73  For the most part the laya techniques are incorporated without alteration. An exception is the saṃketa of lying like a corpse taught in the Dattātreyayogaśāstra (24cd), which in the Haṭhapradīpikā (1.32) becomes an āsana, namely śavāsana, ‘the corpse pose’ widely taught in modern yoga.

74  Laya is often mentioned in the Haṭhapradīpikā, particularly in the context of practices that elsewhere come under the name of layayoga, and at one place (4.103) haṭha and laya are differentiated, but the subject matter of the text is, as evinced by its title, explicitly said to be haṭha, with the implication that all the practices taught therein are haṭha practices.

75  I know of one exception to the principle of the raising of Kuṇḍalinī not involving physical techniques in pre-haṭha Śaiva works: in his commentary ad Netratantra 7.30, Kṣemarāja says that clenching and unclenching the anus makes Kuṇḍalinī point upwards.

76  Haṭhapradīpikā 3.5:

tasmāt sarvaprayatnena prabodhayitum īśvarīm |

brahmadvāramukhe suptāṃ mudrābhyāsaṃ samācaret ||

77  Cf. Vivekamārtaṇḍa 56ab, where the homology of the masculine bindu, śiva and indu is felicitous, but that of the triple-gendered rajas, śakti and sūrya seems forced.

78  Haṭhapradīpikā 3.31–53, 4.43–4.55.

79  See footnote 35.

80  Kiss 2009: 97.

81  Haṭhapradīpikā 4.2–4.4:

athedānīṃ pravakṣyāmi samādhikramam uttamam |

mṛtyughnaṃ ca sukhopāyaṃ brahmānandakaraṃ param ||4.2||

rājayogaḥ samādhiś ca unmanī ca manonmanī |

amaratvaṃ layas tattvaṃ śūnyāśūnyaṃ paraṃ padam ||4.3||

amanaskaṃ tathādvaitaṃ nirālambaṃ nirañjanam |

jīvanmuktiś ca sahajā turyā cety ekavācakāḥ ||4.4||

82  The six-cakra system was yet to achieve hegemony in yogic discourse. See e.g.the Śārṅgadharapaddhati which mentions groupings of both six and nine cakras (4298–4300, 4351–4360).

83  Haṭhapradīpikā 1.5–8.

84  Textual, ethnographic and iconographic sources show that the Śaiva orientation of these suborders did not take hold until the seventeenth century as part of the formalisation of the Daśanāmī order (see my forthcoming monograph Yoga and Yogis: The Texts, Techniques and Practitioners of Traditional Haṭhayoga).

85  See for example the early eighteenth-century Bachittar Nāṭak of Guru Gobind Singh, caupāī 23–24 (translated in Kohli 2005: 126):

Then I created Datt,

who also started his own path,

His followers have long nails in their hands,

And matted hair on their heads,

They do not understand the ways of the Lord. 23

Then I created Gorakh,

Who made great kings his disciples,

His disciples wear rings in their ears,

And do not know the love of the Lord. 24

The second largest akhāṛā of the Saṃnyāsīs, the Mahānirvāṇī, has as its tutelary deity Kapila, who, as mentioned above, is credited in the Dattātreyayogaśāstra as being the originator of some of the practices of haṭhayoga.

86  The Vasiṣṭhasaṃhitā’s verses on these āsanas are also found (sometimes with changes of metre) in earlier Pāñcarātrika texts, texts which are thus the first to teach non-seated āsanas as techniques of yoga. See for example those on mayūrāsana at Vimānārcanakalpa paṭala 96; Pādmasaṃhitā yogapāda 1.21–22, Ahirbudhnyāsaṃhitā 31.36–37; Vasiṣṭhasaṃhitā yogakāṇḍa 1.76–77; Yogayājñavalkya 3.15a-3.17b and Haṭhapradīpikā 1.30.

87  Haṭhapradīpikā 1.18. Elsewhere in the Haṭhapradīpikā we find more references to these two traditions: after the description of bhadrāsana, which, like mayūrāsana, is taught in a variety of Pāñcarātrika works that predate the Haṭhapradīpikā (Vimānārcanakalpa paṭala 96; Pādmasaṃhitā yogapāda 1.16c-1.17d; Ahirbudhnyāsaṃhitā 31.41; Vasiṣṭhasaṃhitā 1.79; Yogayājñavalkya 3.12), it is said that siddhayogins call it gorakṣāsana (Haṭhapradīpikā 1.54).

88  As well as the examples given here, see also Vyāsa’s Bhāṣya on Yogasūtra 3.51, Vivekamārtaṇḍa 74 and Jñāneśvarī 6.154–156.

89  Śārṅgadharapaddhati 4372.

90  Śārṅgadharapaddhati 4364–4371.

91  Varṇaratnākara pp. 57–58.

92  Bhagavadgītā 10.26, Dattātreyayogaśāstra 131.

93  Dattātreyayogaśāstra 29, 130–131.

94  Bronkhorst 2007: 61–68. To the examples given by Bronkhorst may be added verse 20.153 of the Bṛhatkathāślokasaṃgraha which suggests the antinomian nature of the soteriological practices of Kapila: of Caṇḍasiṃha’s city Budhasvāmin writes, ‘There the vices that usually terrify those who want to be liberated from the wheel of rebirth are prescribed by Kapila and others in treatises on liberation’ (yena doṣena saṃsārāt paritrasyanti mokṣavaḥ |sa tasmin mokṣaśāstreṣu śrūyate kapilādibhiḥ ||.)

95  Studying books is scorned by the Nāths (Yogī 1924: 11) and this attitude is found among other traditional yoga-practising ascetics of today. At the 2010 Haridwar Kumbh Mela I was informed, over the din of competing loudspeaker systems blaring out pravacans, ‘lectures’, from ascetic-scholars, that within living memory it was normal for any Rāmānandī ascetic who dared to expound the śāstras to the public to be thrown off his āsan, ‘seat’, and beaten up by his peers.

96  See Śivasaṃhitā 5.258–260, the final verses of the text.

97  See Mallinson 2007: 17–19.

98  Mahābhārata 12.292.8, 13.7.13, 13.13.10, 13.13.54. Cf. Vaikhānasasmārtasūtra 8.8 and Manusmṛti 11.110. In its earliest definition, in the Tattvavaiśāradī of Vācaspati Miśra (ad Yogasūtra 2.46), vīrāsana has the yogin squatting on one foot with the other on its opposite thigh. This posture was demonstrated to me by the Tapasvī Nāth Yogī Mānav Nāth Jī at Jwalamukhi on 11 November 2011 and said to be part of his practice of tapas, not yoga.

99  See e.g. Dattātreyayogaśāstra 132–136, 143–145.

100  See e.g. Gorakṣaśataka 15. This variety of vajrāsana is variously known as svastikāsana, siddhāsana and mūlabandhāsana (see Goodall 2004: 349–350 n.730).

101  The Buddha is said to try this and a variety of other ascetic techniques in the Mahāsīhanādasutta (Majjhima Nikāya 1:78).

102  Ājīvikas are said to practise ukkuṭikappadhāna in the Naṅguṭṭhajātaka (Jātaka 1:493) and the Kassapasīhanādasutta (Dīghanikāya 1:166), tāpasas in the Setaketu and Uddālaka Jātakas (Jātaka 3:235, 4.299). Ukkuṭikappadhāna is also mentioned at Dhammapada 141 and Aṅguttara Nikāya 1.296 and 2.206.

103  See e.g.Vaikhānasasmārtasūtra 8.8 (Caland 1929: 191), which mentions those ‘who hang with their head downwards’ in a list of celibate hermits who practise a variety of austerities, and the following references in the Mahābhārata: 1.13.10–13 and 1.41.1–3, in which the ūrdhvaretas muniJaratkāru finds his ancestors performing penance by hanging over a big hole (cf. Agastya at 3.94.11–14, whose ancestors say they are performing the penance in order to get offspring); 1.26.2, in which Garuḍa sees some Vālakhilyas hanging upside-down; 3.185.4–5, in which Manu practises extreme tapas (as well as hanging upside down, he is also said to stand on one leg and hold up his arms); 12.126.18, in which the muni Tanu meditates upside down; 13.7.8–13, in which Bhīṣma describes a variety of ascetic practices to Yudhiṣṭhira, amongst which is hanging upside down. The practice of hanging upside-down was relatively common until recently but appears from my ethnographic inquiries to have died out. See e.g. the c.1830 illustration of tapkar āsan from an illustrated manuscript of Jayatarāma’s Jogpradīpakā reproduced at Bühnemann 2007: 51, and the photograph of an ‘Urdhamukhi Sadhu’ at Oman 1984: 46. Hanging upside down or performing a headstand has been a Sufi practice for at least a thousand years (Sanaullah 2010: 643).

104  McEvilley’s claim (1981: 66) that vajroli is referred to in the Bṛhadāraṇyakopaniṣad is not borne out by the text.

105  Dattātreya himself is said in the seventeenth adhyāya of the Mārkaṇḍeyapurāṇa to have indulged in wine, women and song in an unsuccessful attempt to dissuade some young sages from seeking his tutelage. The final verse of the same adhyāya (17.25) says that he practises tapas and also that he is ‘a master of yoga meditated upon by yogins seeking liberation’ (yogīśvaraś cintyamāno yogibhir muktikāṅkṣibhiḥ). The dharmaśāstras allow vānaprasthas to take wives with them to the forest but they are to remain chaste and undertake more extreme austerities than the solitary saṃnyāsīs (Kane 1974: 928–9).

106  E.g. Śaṅkaradigvijaya 9.90, where Kṛṣṇa is said to use vajroli in order to be able to sport with the gopīs and Sarvāṅgayogapradīpikā 3.14, where Śiva is said to use it while making love to Gaurī so that Kāma cannot reach him. Cf. Siddhasiddhāntapaddhati 2.13; see also footnote 114.

107  Vajroli may also be obliquely referred to at Amṛtasiddhi 7.8–13:

sa bindur dvividho jñeyaḥ pauruṣo vanitābhavaḥ |

bījaṃ ca pauruṣaṃ proktaṃ rajaś ca strīsamudbhavaṃ ||7.8||

anayor bāhyayogena sṛṣṭisaṃhārau jāyate nṛṇāṃ |

yadā tv abhyantare yogas tadā yogo hi bhaṇyate ||7.9||

kāmarūpo vased binduḥ kūṭāgāra+ṇya+ koṭare |

pūrṇagiri sadā sparśād vrajanti madhyamāpathe ||7.10||

yonimadhye mahākṣetre javābindūrasannibhaṃ |

rajo vasati jantūnāṃ devītatvasamāvṛtaḥ ||7.11||

binduś candramayo jñeyo rajaḥ sūryamayas tathā |

anayoḥ saṃgaṃaḥ sādhyaḥ kūṭāgāre ’tidurghaṭe ||7.12||

eṣa tattva paro dharma eṣa yogaḥ paro mataḥ |

eṣa muktiprado mārgaḥ eṣa guhyatamaḥ paraḥ ||7.13||

8a dvividho] em.; vividho J1 9b sṛṣṭisaṃhārau] em.; sṛṣṭisaṃhāra J1 10b kūṭāgāra°] em.; kūṭādhāra J1 10d vrajanti] em.; vājaṃti J1 13b yogaḥ] em.; yoga J1 13d paraḥ] em.; paraṃ J1

108  Hatley 2011.

109  The Dattātreyayogaśāstra names Śaiva ascetics among the possible practitioners of its yoga teachings when it mentions ‘he who says namaḥ śivāya’ (namaḥśivāyavācī) at verse 44.

110  Dattātreyayogaśāstra 132.

111  Hatley 2011.

112  These austerities include going naked, sitting in the summer sun surrounded by fire, remaining immersed in cold water at night in winter, and spending long periods standing up (sometimes on just one leg) or holding up one or both arms. For examples see the references in notes 97 and 98, as well as the following: Mārkaṇḍeyapurāṇa 31.24; Kathāsaritsāgara 6.4.11; Kūrmapurāṇa 2.27.29–30; Matysapurāṇa 35.17, 171.1 etc; Vāyupurāṇa (Revākaṇḍa) 54.50; Bhāgavatapurāṇa 7.3.2 and Liṅgapurāṇa 1.69.76.

113  Tapas is one of the five niyamas, ‘observances’, prescribed at Yogasūtra 2.32. In his Bhāṣya thereon, Vyāsa defines tapas as the endurance of opposites (dvandvasahanam); ‘opposites’ include hunger and thirst, heat and cold, and sthānāsana, ‘standing up and sitting down’, which may refer to the physical postures maintained for long periods by ascetics. Other parts of haṭha practice also correspond to the tapas of old. Prāṇāyāma, in concert with its expiatory and purificatory capabilities (on which see Vasudeva 388–389), is said in dharmaśāstric literature to be the highest form of tapas (Manu 2.83, 6.70; Viṣṇu 55.17; Vaśiṣṭha 25.5). Matsyendrasaṃhitā 4.2 says that prāṇāyāma is both yoga and tapas; Amanaska 2.42 says that the various techniques of prāṇāyāma are unpleasant and difficult (duḥkhātmakair durjayaih).

114  In some modern manifestations of yoga the physical practices of haṭhayoga are identified with tapas. See Smith 2009 and Sarbacker 2009: 174–175.

115  Yogabīja 148c-149b, which is cited at Haṭhatattvakaumudī 55.29; cf. Haṭharatnāvalī 1.22. Of the various possible referents of sun and moon, the most commonly given in this context are the piṅgalā and iḍā nāḍīs (i.e. the channels carrying prāṇa which run from the base of the central column to the right and left nostrils respectively); see e.g. Vivekamārtaṇḍa 32. Other possibilities include the prāṇa and apāna breaths (Amṛtasiddhi 6.11–13), Śakti and Śiva, as menstrual fluid and semen (Vivekamārtaṇḍa 74–76), or the tip of the tongue and the forehead (Khecarīvidyā 2.72–74). For a detailed discussion of this understanding of haṭha, see Birch 2011: 532–534.

116  The Yoga Upaniṣads largely consist of verses from haṭhayogic works (Bouy 1994) yet only the Yogatattvopaniṣad mentions haṭha (in verses taken from the Dattātreyayogaśāstra: Yogatattvopaniṣad 19, 24 = Dattātreyayogaśāstra 9c-10b, 26). The one Sanskrit text produced in a Nāth milieu after the formation of the Nāth saṃpradāya, namely the circa 1700 Siddhasiddhāntapaddhati, makes no mention of ‘haṭha’ yoga. In Hindi works, what might be understood as haṭhayoga is usually referred to simply as yog/jog. Other than in the doxographies of Sundar Dās and the Bhaktamāl of his contemporary and fellow Dādūpanthī Rāghav Dās, the word haṭh (the Hindi form of haṭha) is used only once in the context of yoga in the corpus of medieval Hindi literature, in the Prāṇ Saṃkalī of Cauraṅgināth (v. 256), in which haṭh jog is said to be the union of the sun and moon. Elsewhere in the medieval Hindi corpus, haṭh in the context of religious practice refers to extreme austerities and is usually scorned (see e.g. Gorakhbāṇī sākhī 31, Guru Granth Sāhib 905.4 and 1305.6 (in the latter haṭh is associated with hanging upside down), and the verses attributed to Carpaṭnāth found in a manuscript dated 1711 CE and translated in Singh 1937: 68–71). Jayatarāma’s 1737 Braj Bhasha reworking of the long recension of the Haṭhapradīpikā is called Jogpradīpakā and, despite the many references to haṭha in its source text, does not use the word haṭha/haṭhin its 964 verses, except when listing the Haṭhapradīpikā among its sources (v. 956). Hindi-speaking yogis in India today do occasionally use the compound haṭh-yog, perhaps influenced by Sanskrit texts. In their understanding haṭhayogaencompasses difficult and uncomfortable practices, including non-physical ones such as mauna, a vow of silence. I have heard such ascetics translate haṭhayoga into English as the conveniently homophonic ‘hard yoga’.

117  E.g. Dattātreyayogaśāstra 41a-42b:

brāhmaṇaḥ śramaṇo vāpi bauddho vāpy ārhato ’thavā |

kāpāliko vā cārvākaḥ śraddhayā sahitaḥ sudhīḥ ||

yogābhyāsarato nityaṃ sarvasiddhim avāpnuyāt |

118  Women are said to practise vajroli at Dattātreyayogaśāstra 155 and Haṭhapradīpikā 3.84, 3.95–98. Cf. Haṭharatnāvalī 2.80–117. Amṛtasiddhi 19.6 says that both the avadhūta and gṛhastha can practise yoga. The Śivasaṃhitā is aimed squarely at householders (but it is less open to women: unlike in the Dattātreyayogaśāstra and Haṭhapradīpikā women are not mentioned as practitioners in its description of vajroli).

119  Householder wrestlers in Kota, Rajasthan, for whom the refinement and preservation of bindu is an important part of their practice, speak highly of vajroli but do not practise it (personal communication from Norbert Peabody, 11 June 2010).

120  Śivasaṃhitā 4.79:

svecchayā vartamāno ’pi yogoktaniyamair vinā |

mukto bhaved gṛhastho ’pi vajrolyabhyāsayogataḥ ||

Cf. Dattātreyayogaśāstra 152:

svecchayā vartamāno ’pi yogoktaniyamair vinā |

vajroliṃ yo vijānāti sa yogī siddhibhājanaḥ ||152||

121  See Goodall 1998: xxxvii n.85 and Goodall 2004: xxix.

122  On the Baḥr al-Hayāt and related texts, see Ernst 2003 and Sakaki 2005.

123  I am grateful to Dr Kazuyo Sakaki for providing me with a comprehensive list of Persian works on yoga.

124  I do not include here the Gorakhbāṇī, the esoteric pads and sakhīs ascribed to Gorakṣa but probably written in the sixteenth century, whose obscurantist esotericism is aimed at initiates and distinguishes them from the more straightforward vernacular manuals listed here.

125  Gorakṣaśataka 101 (cf. Haṭhapradīpikā 3.46–48.):

yā bindvākhyeti dhārā vigalitamanasā tāṃ pibāmo na madyaṃ

pañcānām indriyāṇāṃ viṣayanirasanaṃ carvayāmo na māṃsam |

āliṅgāmo na kāntāṃ kuśakuṭilatanuṃ nāḍikāṃ tāṃ suṣumnāṃ

śūnye citte pralīne + pravilaya vivaśan + maithunaṃ cen na yonau ||101||

126  See Mallinson 2012.

127  I know of two exceptions to prove this rule. First, Bouillier (2008: 128) reports the failed attempts of Svāmī Śiv Nāth Jī, the mahant of an āśram in Orissa, to interest any of the Nāths’ main jamāt (itinerant monastery) in learning haṭhayoga. Second, on 11 November 2011 ‘Yogī Bābā’ Anūp Nāth Jī demonstrated a sequence of complex haṭhayogic āsanas to me at the Nāth monastery at Jwalamukhi. When I asked him where he had learnt it from, he said that it had come to him automatically when he was a boy.

128  This work is reproduced at Callewaert and de Beeck 1991: 489–510. I have used the verse numbering of the latter.

References

Primary sources

Aṅguttaranikāya, parts I and II, ed. R. Morris. London: Pali Text Society. 1885 and 1888.

Amanaska, ed. Jason Birch. Submitted as part of a D. Phil. thesis, University of Oxford. 2012.

Amaraughaprabodha, of Gorakṣanātha, ed. K. Mallik in Mallik 1954.

Amṛtasiddhi of Virūpākṣanātha. Maharaja Mansingh Pustak Prakash, Jodhpur, Acc. no. 1242.

Ahirbudhnyāsaṃhitā, ed. M.D. Ramanujacharya under the supervision of F. Otto Schrader; revised by V. Krishnamacharya. Madras: Adyar Library and Research Centre. 1966.

Kathāsaritsāgara of Somadevabhaṭṭa, ed. Pandit Durgāprasāda and Kāśīnātha Pāṇḍuraṅga Paraba. Bombay: Nirnaya Sāgara Press. 1889.

Kūrmapurāṇa, ed. A.S. Gupta. Varanasi: All-India Kashiraj Trust. 1967.

Khecarīvidyā, ed. James Mallinson. The Khecarīvidyā of Ādinātha. A critical edition and annotated translation of an early text of haṭhayoga. London: Routledge. 2007.

(Śrī) Guru Granth Sāhib, with complete index prepared by Winand M. Callewaert. 2 parts. Delhi: Motilal Banarsidass. 1996.

Gorakṣaśataka. Government Oriental Manuscripts Library, Madras, MS no. R 7874.

Gorakṣaśataka, ed. F. Nowotny: Das Gorakṣaśataka, Cologne, 1976 (Dokumente der Geistesgeschichte). [This text is referred to as the Vivekamārtaṇḍa in this article; for the reasons why see Mallinson, James. 2011b, ‘The Original Gorakṣaśataka’, pp. 257–272 in D.G. White, ed., Yoga in Practice. Princeton, NJ: Princeton University Press.]

Gorakhbāṇī, ed. P.D. Baḍathvāl. Prayāg: Hindī Sāhity Sammelan. 1960.¹²⁸

Gheraṇḍasaṃhitā, ed. and tr. J. Mallinson. New York: YogaVidya.com. 2004.

Candrāvalokana. Government Oriental Manuscripts Library, Madras, MS no. D 4345.

Jātaka, ed. V. Faussell. 4 vols. London: Trübner & Co. 1877–1887.

Jñānasamudra, see Sundar-granthāvalī.

Jogpradīpakā of Jayatarāma, ed. M.L. Gharote. Jodhpur: Rajasthan Oriental Research Institute. 1999.

(Śrī)-Jñāneśvarī of Jñānadeva, ed. G.S. Naṇadīkar. 5 vols. Mumbai: Prakāś Gopāl Naṇadīkar. 2001.

Tirumantiram of Tirumūlar. See Natarajan, B. 1989. Holy Hymns (a translation of the Tirumantiram of Tirumūlar). Madras: ITES Publications.

Dattātreyayogaśāstra, unpublished critical edition by James Mallinson.

Dīghanikāya, vol. 1, ed. T.W. Rhys Davids and J. Estlin Carpenter. London: Pali Text Society. 1890.

Dhammapada, ed. Sūriyagoḍa Sumaṅgala Thera. London: Pali Text Society. 1914.

Pādmasaṃhitā, ed. S. Padmanabhan and R.N. Sampath. Madras: Pancaratra Parisodhana Parisad. 1974.

Prāṇ Saṃkalī of Cauraṅgināth. See Gorakhbāṇī.

Bṛhatkathāślokasaṃgraha of Budhasvāmin, ed. and tr. James Mallinson (as The Emperor of the Sorcerers). 2 vols. New York: Clay Sanskrit Library/JJC Foundation. 2005.

Bhagavadgītā, ed. J.A.B. van Buitenen in The Bhagavadgītā in the Mahābhārata: Text and Translation. Chicago, IL: University of Chicago Press. 1981.

Bhāgavatapurāṇam, ed. Vasudeva Śarman. Bombay: Nirnaya Sagar. 1905.

Majjhimanikāya, vol. 1, ed. V. Trenckner. London: Pali Text Society. 1888.

Matsyapurāṇa, ed. The Pandits of the Ānandāśrama. Ānandāśrama Sanskrit Series no. 54. 1907.

Matsyendrasaṃhitā, See Kiss, Csaba. 2009. ‘Matsyendranātha’s Compendium (Matsyendrasaṃhitā). A Critical Edition and Annotated Translation of Matsyendrasaṃhitā 1–13 and 55 with analysis’. Unpublished D.Phil. thesis submitted to Oxford University.

Manusmrti: with the Sanskrit Commentary Manvartha-Muktavali of Kulluka Bhatta, ed. J.L. Shastri. Delhi: Motilal Banarsidass. 1983.

Mahābhārata, ed. V. Sukthankar, with the cooperation of S.K. Belvalkar, A.B. Gajendragadkar, V. Kane, R.D. Karmarkar, P.L. Vaidya, S. Winternitz, R. Zimmerman and other scholars and illustrated by Shrimant Balasaheb Pant Pratinidhi (since 1943 ed. S.K. Belvalkar). 19 vols. Poona: Bhandarkar Oriental Research Institute, 1927–1959.

Mārkaṇḍeyapurāṇa, ed. K.M. Banerjea. Calcutta: Bishop’s College Press. 1862.

Lalitāsahasranāmastotra with the commentary Saubhāgyabhāskara of Bhāskararāya, revised by Vāsudev Lakṣmaṇa Śāstrī Paṇśīkar (4th edn). Bombay: Nirṇaya Sāgara Press. 1935.

Liṅgapurāṇa. Gurumandal Series no. XV. Calcutta. 1960.

Yogatattvopaniṣad. See Śāstrī, A.M., ed. 1920. The Yoga Upaniṣads. Madras: Adyar Library..

Yogabīja, ed. Rām Lāl Śrīvāstav. Gorakhpur: Śrī Gorakhnāth Mandir. 1982.

Yogayājñavalkya, ed. K.S. Śāstrī. Trivandrum Sanskrit Series no. 134. Trivandrum. 1938.

Varṇaratnākara of Jyotirīśvarakaviśekharācārya, ed. S.K. Chatterji and B. Misra. Calcutta: Royal Asiatic Society of Bengal. 1940.

Vasiṣṭhasaṃhitā (Yogakāṇḍa). Revised edition, ed. Swami Digambarji, PitambarJha, Gyan Shankar Sahay (first edition); Swami Maheshananda, B.R. Sharma, G.S. Sahay, R.K. Bodhe (revised edition). Lonāvalā: Kaivalyadhām Śrīmanmādhav Yogamandir Samiti. 2005.

Vāyupurāṇa (Revākhaṇḍa), ed. (as Śrīskandamahāpurāṇam) Kṣemrāj Śrīkṛṣṇadās.Bombay: Veṅkateśvara Steam Press. 1910.

Vimalaprabhā of Kalkin Śrī Puṇḍarīka on Śrīlaghukālacakratantra by Śrī Mañjuśrīyaśas, chief editor Samdhong Rinpoche, ed. Vrajavallabh Dwivedi et al. Rare Buddhist Texts Series 13 (3 vols). Sarnath: Central Institute of Higher Tibetan Studies. 1994.

Vimānārcanākalpa, ed. Raghunāthacakravārtin and Setu Mādhavācārya. Madras: Venkateshwar Press. 1926.

Vivekamārtaṇḍa of Gorakṣadeva. Oriental Institute of Baroda Library. Acc. no. 4110.

Viṣṇusmṛti, ed. V. Krishnamacharya.Adyar Library Series 93. Madras: The Adyar Library and Research Centre. 1964.

Śaṅkaradigvijaya of Mādhava, ed. Mādhavānand Bhāratī. Lakhanaū: Muṃśī Navalakiśora. 1865.

Śārṅgadharapaddhati, ed. Peter Peterson. Bombay: Government Central Book Depot. 1888.

Śāradātilakam of Lakṣmaṇadeśikendra. Paṭala 25, as edited in pp. 343–354 of Bühnemann 2001.

Śivasaṃhitā, ed. and tr. J. Mallinson. New York: YogaVidya.com.

Sarvāṅgayogapradīpikā, see Sundar-granthāvalī, ed. Purohit Hari N. Śarmā. Calcutta. 1936.

Siddhasiddhāntapaddhati of Gorakṣanātha, ed. M.L. Gharote and G.K. Pai. Lonavla: Lonavla Yoga Institute. 2005.

Sundar-granthāvalī, ed. Purohit Hari N. Śarmā. Calcutta. 1936.

Haṭhapradīpikā of Svātmārāma. Rajasthan Oriental Research Institute, Jodhpur. MS no. 6756.

Haṭhapradīpikā of Svātmārāma, ed. Svāmī Digambarjī and Dr Pītambar Jhā. Lonavla: Kaivalyadhām S.M.Y.M. Samiti. 1970.

Haṭharatnāvalī, ed. M.L. Gharote,P. Devnath and V.K. Jha. Lonavla: Lonavla Yoga Institute. 2002.

Secondary literature

Birch, Jason. 2011. ‘The Meaning of Haṭha in Early Haṭhayoga’. Journal of the American Oriental Society, 131(4): 527–552 .

Birch, Jason. 2012. ‘The Amanaska’. D.Phil. thesis submitted to the University of Oxford.

Bouillier, Véronique. 1997. Ascètes et Rois. Un Monastère de Kanphata Yogis au Népal. Paris: CNRS.

Bouillier, Véronique. 2008. Itinérance et Vie Monastique: Les Ascètes Nāth Yogīs en Inde Contemporaine. Paris: Maison des sciences de l’homme.

Bouy, Christian. 1994. Les Nātha-Yogin et les Upaniṣads. Paris: Diffusion de Boccard.

Bronkhorst, Johannes. 2007. Greater Magadha: Studies in the Culture of Early India. Vol. 19 of Handbook of Oriental Studies, Section Two, India. Leiden and Boston, MA: Brill.

Bühnemann, Gudrun. 2000. “The Six Rites of Magic”, In Tantra in Practice, ed. David Gordon White, pp. 447–462. Princeton, NJ: Princeton University Press.

Bühnemann, Gudrun. 2001. The Iconography of Hindu Tantric Deities volume II. The Pantheons of the Prapañcasāra and the Śāradātilaka. Groningen: Egbert Forsten.

Bühnemann, Gudrun. 2007. Eighty-four Āsanas in Yoga. A Survey of Traditions (with Illustrations). Delhi: D.K. Printworld.

Caland, W. 1929. Vaikhānasasmārtasūtram: The Domestic Rules and Sacred Laws of the Vaikhānasa School Belonging to the Black Yajurveda. Calcutta: Asiatic Society.

Callewaert, W.M. and op de Beeck, B. 1991. Devotional Hindī Literature: A Critical Edition of the Pañc Vānī or Five Works of Dādū, Kabīr, Nāmdev, Rāidās, Hardās with the Hindī Songs of Gorakhnāth and Sundardās, and a Complete Word-Index. 2 vols. New Delhi: South Asia Institute, New Delhi Branch, Heidelberg University.

Ernst, Carl W. 2003. ‘The Islamization of Yoga in the Amrtakunda Translations’ Journal of the Royal Asiatic Society, Series 3, 13(2): 1–23.

Ernst, Carl W. 2009. “Being Careful with the Goddess: Yoginis in Persian and Arabic Texts”. In Performing Ecstasy: The Poetics and Politics of Religion in India, ed. Pallabi Chakrovorty and Scott Kugle, pp. 189–203. New Delhi: Manohar.

Golovkova, Anna. 2010. ‘The Cult of the Goddess Tripurasundarī in the Vāmakeśvarīmata’. Unpublished thesis submitted for the degree of Master of Studies at Oxford University.

Goodall, Dominic. 1998. Bhaṭṭa Rāmakaṇṭha’s Commentary on the Kiraṇatantra. Volume 1: Chapters 1–6. Critical Edition and Annotated Translation. Pondicherry: Publications de l’Institut français d’Indologie no. 86.1.

Goodall, Dominic. 2004. The Parākhyatantra. A Scripture of the Śaiva Siddhānta. A critical edition and annotated translation. Pondicherry: Publications de l’Institut français d’Indologie no. 98.

Halbfass, Wilhelm. 1988. India and Europe. Albany, NY: State University of New York Press.

Hatley, Shaman. 2011. ‘Erotic Asceticism: The Knife’s Edge Observance of Brahmayāmala xl (asidhāravratapaṭala), with a Critical Edition and Translation’. Tantric Studies, 3, forthcoming.

Kane, P.V. 1974. A History of Dharmaśāstra. Vol. 2, Part 2. Poona: Bhandarkar Oriental Research Institute.

Kiehnle, Catharina. 2005. “The Secret of the Nāths: The Ascent of Kuṇḍalinī according to Jñāneśvarī 6. 151–328”. In Bulletin des Études Indiennes 22–23, pp. 447–494.

Kiss, Csaba. 2009. ‘Matsyendranātha’s Compendium (Matsyendrasaṃhitā). A Critical Edition and Annotated Translation of Matsyendrasaṃhitā 1–13 and 55 with Analysis’. Unpublished D. Phil. thesis submitted to Oxford University.

Kohli, Surendar Singh. 2005. (The) Dasam Granth: Bachittar Nāṭak of Guru Gobind Singh. New Delhi: Munshiram Manoharlal.

McEvilley, Thomas. 1981. ‘An Archaeology of Yoga’. Anthropology and Aesthetics no. 1, pp. 44–77.

Mallik, Kalyani, ed. 1954. The Siddha Siddhānta Paddhati and Other Works of Nath Yogis, Poona: Poona Oriental Book House.

Mallinson, James. 2007. The Khecarīvidyā of Ādinātha. A Critical Edition and Annotated Translation of an Early Text of Haṭhayoga. London: Routledge.

Mallinson, James. 2011a. “Haṭha Yoga”. In Brill Encyclopedia of Hinduism, ed. K.A. Jacobsen, H. Basu, A. Malinar and V. Narayanan, pp. 770–781. Leiden: Brill.

Mallinson, James. 2011b. “The Original Gorakṣaśataka”. In Yoga in Practice, ed. D.G. White, pp. 257–272. Princeton, NJ: Princeton University Press.

Mallinson, James. 2011c. “Nāth Saṃpradāya”. In Brill Encyclopedia of Hinduism, ed. K.A. Jacobsen, H. Basu, A. Malinar and V. Narayanan, pp. 407–428. Leiden: Brill.

Mallinson, James. 2012. ‘The Yogīs’ Latest Trick’. Review article in Tantric Studies (University of Hamburg). Also pp. 165–180 in Journal of the Royal Asiatic Society, 24(1) (2014).

Natarajan, B. 1989. Holy Hymns (a translation of the Tirumantiram of Tirumūlar). Madras: ITES Publications.

Oman, John Campbell. 1984 [1903]. The Mystics, Ascetics and Saints of India. New Delhi: Cosmo Publications.

Sakaki, Kazuyo. 2005. ‘Yogico-tantric Traditions in the Ḥawd al-Ḥayāt’, Journal of the Japanese Association for South Asian Studies, 7: 135–156.

Sanaullah, M. 2010. “Elements of Yoga in Sufism”. In History of Yoga, ed. S.P. Singh, pp. 629–652. Vol. XVI, part 2 in the History of Science, Philosophy and Culture in Indian Civilization Series. New Delhi: Project of History of Indian Science, Philosophy and Culture.

Sanderson, Alexis. 1988. “Śaivism and the Tantric Traditions”. In The World’s Religions, ed. S. Sutherland, L. Houlden, P. Clarke and F. Hardy, pp. 660–704. London: Routledge.

Sanderson, Alexis. 2009. “The Śaiva Age”. In Genesis and Development of Tantrism, ed. Shingo Einoo, pp. 41–349. Tokyo: Institute of Oriental Culture, University of Tokyo.

Sarbacker, Stuart Ray. 2009. “The Numinous and Cessative in Modern Yoga”. In Yoga in the Modern World: Contemporary Perspectives, ed. Mark Singleton and Jean Byrne, pp. 161–183. London: Routledge.

Śāstrī, A.M., ed. 1920. The Yoga Upaniṣads. Madras: Adyar Library.

Schaeffer, Kurtis R. 2002. ‘The Attainment of Immortality: from Nāthas in India to Buddhists in Tibet’ Journal of Indian Philosophy, 30(6): 515–533.

Singh, Mohan. 1937. Gorakhnāth and Medieval Mysticism. Lahore: Dr Mohan Singh.

Smith, Benjamin Richard. 2009. “With Heat Even Iron Will Bend”. In Yoga in the Modern World: Contemporary Perspectives, ed. Mark Singleton and Jean Byrne, pp. 140–160. London: Routledge.

van der Veer, Peter. 1989. Gods on Earth. The Management of Religious Experience and Identity in a North Indian Pilgrimage Centre. Delhi: Oxford University Press.

Vasudeva, Somadeva. 2004. The Yoga of the Mālinīvijayottaratantra. Publications de l’Institut français d’Indologie no. 97. Pondichéry: Institut français d’Indologie.

Vasudeva, Somadeva. 2011. ‘Haṃsamiṭṭhu: “Pātañjalayoga is Nonsense”’. Journal of Indian Philosophy 39(2): 123–145.

Vaudeville, Charlotte. 1957. Kabir Granthavali (Doha). Publications de l’Institut français d’Indologie no. 12. Pondichéry: Institut français d’Indologie.

Verghese, Anila. 1995. Religious Traditions at Vijayanagara. New Delhi: Manohar.

Vilāsnāth, Yogī. 2010. Śrī Nāth Rahasya. 2 vols. Haridvār: Akhil Bhāratavarṣīya Avadhūt Bheṣ Bārah Paṃth Yogī Mahāsabhā.

Yogī, Candranāth. 1924. Yogisampradāyāviṣkṛti. Ahmedābād: Śivanāth Yogī.