Next Article in Journal
The “Whence” of Evil and How the Demiurge Can Alleviate Our Suffering
Next Article in Special Issue
Amṛtasiddhi A Posteriori: An Exploratory Study on the Possible Impact of the Amṛtasiddhi on the Subsequent Sanskritic Vajrayāna Tradition
Previous Article in Journal
“I Get Peace:” Gender and Religious Life in a Delhi Gurdwara
Previous Article in Special Issue
Love me for the Sake of the World: “Goddess Songs” in Tantric Buddhist Maṇḍala Rituals
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:

The Visualization of the Secret: Atiśa’s Contribution to the Internalization of Tantric Sexual Practices

Department of Religious Studies, Santa Clara University, Santa Clara, CA 95053, USA
Religions 2020, 11(3), 136;
Received: 1 February 2020 / Revised: 13 March 2020 / Accepted: 14 March 2020 / Published: 18 March 2020
(This article belongs to the Special Issue The Society for Tantric Studies Proceedings (2019))


This essay will explore the challenges presented by transgressive rituals, particularly the secret and wisdom-consort consecrations found in the Mahāyoga and Yoginī tantras. In particular, it examines how Atiśa Dīpaṅkara Śrījñāna aided in the dissemination of these traditions to Tibet during the eleventh century, in part through encouraging the enactment of transgressive rituals via internal visualization. I will do so through the exploration of a largely unstudied work by Atiśa, his commentary on a meditation manual (sādhana) attributed to the mahāsiddha Lūipa, “The Realization of the Cakrasaṃvara”, Cakrasaṃvarābhisamaya. Through an examination of this work, I will argue that Atiśa played an important role in facilitating the acceptance of the Yoginītantras in Tibet, during a time when tantric traditions were subject to a considerable amount of scrutiny.

One of the defining features of the tantras is their antinomian rhetoric, namely passages describing violent or erotic rituals that challenged traditional Mahāyāna ethical norms. Such rhetoric is found in the earliest tantric Buddhist works dating to the seventh century,1 and rose to prominence in the Mahāyoga and Yoginītantras, that were composed in India beginning in the eighth century.2 This antinomian rhetoric inspired considerable controversy, as evidenced in the apologetic defenses contained in the scriptures themselves,3 as well as in historical records.4
The antinomian discourse in the Buddhist tantras also presented a challenge to those who advocated their dissemination to, and translation within, other cultural contexts, and concerns about this discourse led to several attempts to block their transmission in Tibet and East Asia, or to facilitate this through translations that bowdlerized or eliminated transgressive content.5 One such context was the region of Western Tibet in the late tenth and eleventh centuries. Some of the then relatively new Mahāyoga and Yoginītantras, especially the Hevajra Tantra, were circulating in Tibet circa the mid-tenth century. The king of the state of Guge in Western Tibet, King Lha-bla-ma Ye-shes-’od, doubted the authenticity of these new teachings. His eulogy, written by his nephew Zhi-ba-’od, explains his doubts as follows: “Furthermore, the hidden meaning of secret mantra was vitiated, and was further corrupted by the practice of the rites of ‘sexual union’, ‘deliverance’ and the ‘tshogs offering’. To find out whether these practices were correct at all, the lo-tsā-ba Rin-chen bzang-po was sent to Kashmir.”6 He thus sent the young Tibetan monk Rin-chen bzang-po (958–1055 CE) to India, to ascertain whether these new and disturbingly transgressive texts were considered orthodox. Rin-chen bzang-po arrived in Kashmir in 975 CE, and learned that these texts were considered orthodox, at least by the Buddhist pandits living there. In fact, with the help of his Kashmiri mentor Padmākaravarman, he translated the Cakrasaṃvara Tantra, one of the more transgressive Yoginītantras, into Tibetan.7
Despite this assurance, not all skeptics in Tibet were assuaged, and the royal family in Guge continued their efforts to reform Buddhism in Western Tibet during the eleventh century. Most famously, the King Lha-bla-ma Ye-shes-’od’s dying wish was that the great Indian scholar Atiśa Dīpaṃkara Śrījñāna (982–1054 CE) be invited to Tibet, a wish that his nephew, Byang-chub-‘od, attempted to fulfill. Atiśa Dīpaṃkara Śrījñāna was a learned Buddhist scholar who was then abbot of Vikramaśīla.8 While en route to Tibet, Atiśa, no doubt informed of skepticism concerning the orthodoxy of the tantras held by some members of the Guge royal family, composed his very well known verse text, the Bodhipathapradīpa, “A Lamp for the Path to Awakening”, a concise introduction to the Buddhist path of practice. In this work, he addressed one of the most controversial aspects of the new Mahāyoga and Yoginītantras, the sexual rite of consecration, particularly the second “secret consecration” (guhyābhiṣeka) in which the presumed male Vajrācarya enters into a sexual union with a female consort, and then anoints the initiates with the consecrated, mixed sexual fluids produced by this union. This is then followed by the third consecration, the prajñājñānābhiṣeka, which I like to translate as the “gnosis of the consort consecration”, in which the male initiate is then instructed on the art of union with the same consort. At the end of the text, he comments on these initiations briefly, arguing that the monks or nuns should not receive them, since they would entail the violation of the pratimokṣa vow of celibacy.9 In his auto-commentary on this text he explains this suggested prohibition as follows:
Regarding consecrations there are two types: those applicable to householders, and those applicable to the celibate. Those applicable to householders include everything taught in the tantras, while the celibate from amongst those should avoid the secret and gnosis of the consort consecrations. Why should they avoid those two? Celibacy is understood to be one of the virtues which occurs as a point of doctrine, in reliance upon the Buddha’s teaching. Those two consecrations are regarded as not being in accordance with the practice of celibacy. Hence it is said that the two consecrations would bring about the end of celibacy, and the end of celibacy would be the end of the Buddha’s teaching. And by its ceasing the continuum of merit making would be broken. Since from that basis there would arise innumerable non-virtuous people, the celibate should thus avoid those two.10
This is a text that was written for a specific audience, his prospective hosts, the royal family in Guge.11 The kings of Guge were renowned for their interest in the renewal of Buddhism in Tibet, and thus presumably held the traditional belief that Buddhism depended upon the monastic sangha, which traditionally served as the teachers and spiritual leaders for Buddhist communities, as well as the idea that monastic identity hinged on celibacy, the most important characteristic distinguishing them from the laity. However, Atiśa’s solution to this problem had an unfortunate consequence; it seemingly excluded the monks from full participation in what at the time was considered, in Indian Buddhist tantric circles at least, the highest and most efficacious system of practice. This, I argue, was not a sustainable state of affairs; it would not be easy for a Buddhist community to sustain the monastic sangha if the monks were excluded from practice of the “highest” teaching.
There was in fact another solution to this problem that developed in Indian tantric Buddhist communities. This solution called for the visualization of the more problematic, antinomian ritual practices involving sexuality or violence. This is simply another example of a venerable practice in Indian religions, namely the interiorization of ritual practices. Additionally, it has prestige in this context, since visualization is one of the central practices of tantric Buddhist traditions. It would not be controversial to argue that visualization practice is one of the defining marks of tantric Buddhist traditions. Since the eighth century at the very latest, the practice of visualizing awakened beings in their divine mandala abodes has been identified as a key practice that in and of itself can lead to awakening, an idea that is advanced in many of the tantras, such as the Sarvarahasya-nāma-tantrarāja, which describes “buddhayoga” as follows: “Meditate on oneself as a Buddha via the five yogas, since Buddhahood arises from concentration and gnosis. Therefore, the yogī devoid of buddhayoga will not attain awakening.”12
This idea was expressed in a number of tantras composed circa the eighth and ninth centuries, and the argument that visualization practice was a key requisite to awakening was made by key intellectual figures from this era, including Buddhaśrījñānapāda and Nāgārjuna, founding figures in the Jñānapāda and Āryā schools of tantric exegesis, respectively.13 Following their exalted example, Tsong Khapa, writing many centuries later in Tibet, argued that deity yoga, meditatively visualizing oneself as an awakened being, is in fact a defining mark of the Vajrayāna path.14 Given the centrality of visualization practices in tantric Buddhist circles, it should not be surprising that some Buddhists saw this as a natural way to resolve the challenge to Buddhist identity presented by the antinomian rituals described in the tantras.
We see this solution in the work of Abhayākaragupta, who lived shortly after Atiśa, from approximately 1064–1125 CE. Like Atiśa, he also served as a teacher at Vikramaśīla in Northeastern India.15 Of particular relevance here is his work the Vajrāvalī, “The Diamond Array”, a manual for the initiation rites of a wide range of tantras. In this work, when describing the secret consecration, Abhayākaragupta uses verbs such as “visualize” (bsam par bya)16, implying that the liturgy is, at least in part, visualized. Abhayākaragupta also addresses the question of whether the sexual aspects of the rite are to be performed with an actual partner or consort, typically referred to as an “actual consort” (karmamudrā) in this literature, or with a “gnostic consort” (jñānamudrā), that is, a consort who is visualized.17 It turns out that the manner in which initiation should be bestowed depends upon the candidate’s qualifications. Abhayākaragupta comments as follows:
There is no infraction even if a monk experiences the limits of the bliss of the two organs with an outer consort who is seen in a dream, if seen with discernment of the habitual tendency of the bliss of union that is so experienced.” One should understand this as a teaching on monastic discipline. However, in the case of an alternate instance, with respect to an evil person, the secret and consort gnosis consecrations should be bestowed with a gnostic consort (jñānamudrā) on such occasions. If there are no evil persons, then an actual consort (karmamudrā) only [is acceptable]. A monk who, while faithful, is not steadfastly intent upon reality should be prohibited from receiving consecration with an outer wisdom-consort.18
Abhayākaragupta clearly here sets up a hierarchy of disciples, the idea being that only the most advanced disciples, who fully understand the nature of reality, are qualified to practice with an actual human partner. For such disciples, due to their advanced level of understanding, there would be no transgression or sin. Lesser disciples who lack this high degree of realization should not practice the actual ritual and should practice its internally visualized version only. Here, Abhayākaragupta follows an old Mahāyāna idea that advanced practitioners are beyond ordinary social rules and ethical norms, and can transgress them, so long as their motivation is selfless and compassionate.19 This also provides Buddhist institutions with a way to teach and engage in tantric practice in a non-transgressive fashion, the manner that is appropriate for most “ordinary” practitioners. It also resolves the dilemma that Atiśa’s solution created by excluding monks from the “highest” path of practice. It should also be noted that elsewhere, in his Harvest of Esoteric Instructions, Abhayākaragupta classifies disciples into five ranks, with the “lesser” type doing the outer rituals only, while the four higher ranks engage in internalized Creation and Perfection stage contemplations related to the outer rite.20
If we only take into consideration Atiśa’s A Lamp for the Path to Awakening and his auto-commentary thereon, we might come to the conclusion that the Abhayākaragupta’s solution to the problem of transgressive ritual, its visualization, was newly developed in Indian Buddhist circles following Atiśa’s departure for Tibet. However, this was certainly not the case. This is because evidence for this solution is also found in Atiśa’s oeuvre, albeit somewhat less prominently than in Abhayākaragupta’s. This suggests that Atiśa’s argument in A Lamp for the Path to Awakening, that only the laity should receive the secret and gnosis of the consort consecrations, does not fully and accurately represent Atiśa’s approach to the teaching of the tantras. Instead, it seems that this recommendation was in part a rhetorical trick or strategy, an act of upāya intended for a specific audience, namely Atiśa’s Tibetan patrons, for it is undercut by his other works in Tibet, including his translation of the Abhidhānottara and Yoginīsaṃcāra tantras as well as his composition of a commentary on a famous sādhana text, “The Realization of Cakrasaṃvara”, Cakrasaṃvarābhisamaya, attributed to the great Mahāsiddha Lūipa. In Atiśa’s commentary, entitled prosaically “Analysis of The Realization”, abhisamayavibhaṅga, he takes a much more subtle approach to the issue of the second and third consecrations, which points toward the Tibetan’s ultimate way of integrating these practices into a religious milieu, dominated by monastics and monasticism.
Since Lūipa’s Cakrasaṃvarābhisamaya is a meditation manual (sādhana) and not a commentary on initiatory rituals (abhiṣekavidhi), there is actually no direct discussion of the four consecrations in this text. As a result, I believe that any references to them made by Atiśa is particularly meaningful, since such commentary is not motivated by the root text itself, but rather by extratextual or contextual concerns. Moreover, as Atiśa himself was a monk, and his commentary on the sādhana would presumably be intended for his dedicated monastic disciples, we would expect that he would omit any mention of the second and third consecrations, if he genuinely intended that monastics not receive them at all. As it turns out, this is not the case.
In his Analysis of The Realization commentary, Atiśa discusses the initiatory rituals twice, once briefly at the start of the text, when he discusses the adept’s qualifications, and later toward the end of the text when he relates a detailed visualization of the four consecrations. I included a portion of the latter in my dissertation years ago, but at the time I did not fully appreciate the significance of this text and missed the reference to the four consecrations contained within it.21 I will thus include here a full translation of this passage, as I believe it sheds light on Atiśa’s strategy for facilitating the acceptance of the Yoginītantras in a conservative Tibetan context.
The first mention of them occurs in his commentary to the first line of Lūipa’s sādhana, which reads as follows: “First, the Lord of Yoga should contemplate for some time the four divine abodes.”22 The four divine abodes—loving kindness (maitrī), compassion (karuṇā), sympathic joy (mudita), and equanimity (upekṣa)—are classical Buddhist contemplative categories that thus root the practice to very venerable Buddhist precedents. For Atiśa, this first line indicates the type of practitioner who should be engaged in this practice, which is a person well-grounded in classical Mahāyāna ethical training, and who has received the requisite tantric empowerments. He comments as follows:
With respect to wisdom there are two, conceptual and non-conceptual wisdom. One establishes a mentality according with these two wisdoms, and, from the perspective of practice, meditate upon their non-dual integration. From the perspective of the four consecrations that are the means for achieving this, the jar consecration achieves the conventional truth, while the ultimate truth is achieved with the gnosis of the consort consecration for which the secret is a preliminary. The fourth [consecration] achieves the realization of their non-dual integration. They are attained in dependence upon and through the kindness of the guru. The characteristics of the adept who thus succeeds are [as follows]: First, [he] has stabilized the perfected spirit of awakening which arises from the power of compassion, and has obtained through the grace of the guru the consecrations, the jar, secret, and so forth.23
Of particular interest to me here is his indication that the qualified practitioner should have progressively received the tantric consecrations, which are portrayed as essential steps on the path to awakening, with the two transgressive consecrations correlated to the ultimate truth. There is no indication here that they should be received only by householders, as he argued in his A Lamp for the Path to Awakening. Here, he makes no such qualification, suggesting that this distinction is not relevant in this context, an implication reinforced by his second reference to the consecrations in this text.
This second instance is a detailed visualization of the consecration rites that occur later in his commentary. They are a distinctive aspect of this work and are not a direct commentary on Lūipa’s text. They occur as follows in Atiśa’s Analysis of The Realization:
Next visualize the body maṇḍala. One should contemplate Vārāhī in the secret lotus, and Heruka on one’s crown. In one’s heart is Dākinī, in the throat Lāmā, in the navel Khaṇḍarohā, in the forehead Rūpiṇī, and the [goddesses] such as Pracaṇḍā in the twenty-four places such as the head and so forth. Visualize that in one’s mouth is Kākāsyā, in the right nostril Ulūkāsyā, in the rectum Śvānāsyā, in the left nostril Śūkarāsyā, in the left ear Yamadāḍhī and Yamadūtī in the right, and in the right eye Yamadaṃṣṭriṇī, and Yamamathanī in the left. Then visualize a sacred thread composed of the oṃ kara kara [portion of the root mantra]. Visualize the atoms of the twenty-four bodily elements as the twenty-four heroes. Then create the six insignia from the six [hero armor mantras], oṃ ha and so forth. Create a necklace from Vajravārāhī’s long mantra. Create the skull garland from [Heruka’s essence mantra], oṃ śrī vajra he, etc., earrings from [Heruka’s] seven-syllable [quintessence mantra], the elephant hide from the Mother’s essence, the sash from the heroine’s armor mantra. Use the mantras of each of the ḍākinīs to[create] the skulls in each of their hands. [Create] the skull crest with the Mother’s quintessence.24
With the blessings of body, speech and mind, the blessings of the aggregates and so forth, the armor of the armor mantras, and the light rays of one’s heart, invite, worship and mix with the gnosis hero. Create the pride of being non-dual with him through the oṃ yogaśuddhaḥ mantra. Then invite and worship the thirty-seven [goddesses]. Next, those deities dispel obstructing forces with protective utterances. The heroines sing about the great bliss of play in consecration bestowal, the great bliss of the network of ḍākinīs, the great bliss which grants great pleasure, the great bliss of entreaty, songs such as “Alalaho, the great bliss supreme, the great bliss of the ḍākinīs’ net! Alalaho!” The heroes utter the benediction.25
Imagine that the five, the four mothers and Vārāhī, [descend] from the sky holding the jar filled with the ambrosia of gnosis, and that they bestow [the jar] consecration with their hands. Then, the four mothers dissolve into Vārāhī, and bestow the secret consecration with the spirit of awakening (bodhicitta) of equipoise with Śrī Heruka. Then, Heruka takes Vārāhī as his consort (mudrā), and through being equipoised their winds dissolve. Relying on that, contemplate the experience of the natural (sahaja). Then you, a child of the clan, unite with the consort (mudrā) as Heruka, and, depending on that, meditate on clear light, that wisdom which is attained in visionary experience. This is the very essence of the Transcendence of Wisdom (prajñāpāramitā) which is the purity of the three consciousnesses,26 and which is liberation from birth due to the non-existence of body, speech, and mind. This is the ultimate truth that has the characteristic of always appearing completely luminous like the moon, sun, fire, and jewels. Regard [everything] with the eye of the wisdom and gnosis [consecration], the vision that is beyond the objectification of the other. In this way, do not see anything in and of itself, but see the clear light. And while there is no sort of causation at work with this sort of clear light, conventionally, see the thirty-seven deities from mere wind-mind clear and complete like a reflected image, colored like a rainbow, and distorted like [the image of] the moon in water. Regard them as caused, and since they arise, they are conventional. The integration which does not divide the two truths, the great spirit of awakening, is the fourth [consecration].27
Then imagine that the consecration deities enter into you, and that they are individually sealed. Then worship as before, understanding that the objects of worship, the worshipper, and the worship are all birthless. Then the ambrosia is tasted, meaning that from yaṃ and raṃ there is wind and fire, and from a the skull, within which, from go, ku, da, ha and na appear the five fleshes, and from vi, mu, śu, ma and ra, the five ambrosias. Above these are the syllables hūṃ, oṃ, trām, hrī and a, which are the five clans. Cover it with a lunar disk arising from a. Above that is oṃ āḥ hūṃ, the three clans, above which is a nine-pronged vajra arising from the moon disk, marked with a hūṃ garlanded with light rays, the six great secrets. Then imagine that the light radiating from the four syllables draws forth the gnostic ambrosia of the body, speech and mind of all Tathāgatas, and dissolves into each.28
In Atiśa’s understanding of this meditation manual, the tantric consecrations ideally effect one’s transformation into an awakened being. Of particular significance is his connection of the transgressive second and third “secret” (guhyābhiṣeka) and “gnosis of the consort” (prajñājñānābhiṣeka) consecrations to the realization of the ultimate truth. It involves a symbolic re-conception of the initiate, using the very same sexual symbolism and substances, the red and white “drops” of sexual union, through which the individual is generated at and following conception.29 This gnosis involves the realization of the clear light gnosis of nonduality, followed by the generation of the subtle body, which in the context of advanced yogic practice is usually referred to as the “magic body” (māyākāya or māyādeha), which is a purified mind-made body envisioned in a divine form, composed of the deities of the maṇḍala.
I find Atiśa’s description of the visualized consecration rites particularly interesting. Atiśa’s strategy in this commentary is brilliant. He takes visualization practice, which for centuries has been invested with tremendous soteriological potency in Buddhist circles, and uses it to bring about what must have been one of his major goals in Tibet. This is facilitating the transmission of tantric Buddhist traditions to Tibet, by assuaging doubts about the transgressive aspects of these traditions. In this case, this is done by relegating the transgressive practices to the safe, internalized arena of tantric visualization, rather than the messy and open performance of the actual ritual practices.
When Atiśa wrote this text, circa the mid-eleventh century, visualization practice had long been invested with tremendous soteriological potency by tantric Buddhists. It is also a convenient place for domesticating the transgressive consecratory rituals. Following Indian precedent going back to Vedic times, the transgressive rites could be safely handled by the highly trained mind of the yogī or yoginī and subjected to a metaphorical distillation process, in which the ritual potency is retained, while the messy spectacle of the actual physical rituals is discarded. The adept thus visualizes the rite, conducted by deities in their pure mandalic abode. While Atiśa does not state this directly, this depiction of the consecrations as visualized implies that the actual performance of these rites, in a manner that would problematize monastic vows of celibacy, are not actually required, and that they can be safely practiced in the internal arena of visualization, which monks, by virtue of dedication to spiritual exercises such as visualization practice, are especially qualified to practice.
There is evidence that Indian Buddhists developed a flexible teaching strategy, teaching tantric rituals in different manners, depending on the nature of the disciple. As Sakurai has earlier brought to our attention,30 Vāgīśvarakīrti’s Saṃkṣiptābhiṣekavidhi “Concise Consecration Manual”, likewise discusses the question of the second and third initiations and who should be receiving them. Vāgīśvarakīrti was a Newar student of Nāropa who was alive in circa the late tenth to early eleventh century, and with whom a number of prominent Tibetans studied, including Marpa Chos-kyi blo-gros (1012–1097)31 and ‘Brog-mi Shākya Ye-shes. The latter figure, ‘Brog-mi, lived circa 990–1074, and reportedly travelled to India for study around 1008 CE, where he remained for somewhere between nine to eighteen years. During this time, he studied at Vikramaśīla, and one of his teachers was Vāgīśvarakīrti, who was the southern gatekeeper at the time.32 This would place Vāgīśvarakīrti at Vikramaśīla most likely before Atiśa, who lived at Vikramaśīla from at least 1037 CE, when Nag-tso travelled there to invite him, until 1040 CE, when he departed for Tibet.33 While we do not know if the two actually met, it seems almost certain that Atiśa would have been aware of Vāgīśvarakīrti’s work, especially given Atiśa’s fame as a master of tantric literature.34
Vāgīśvarakīrti, in his “Concise Consecration Manual”, discusses the secret and gnosis of the consort consecrations. He also, interestingly, discusses whether or not monks should receive these initiations. His comments occur as follows:
However, with respect to those monks renowned for single-minded asceticism distinguished by [their adherence to] the disciples’ vehicle, there is no point in consecrating them with mantra, mudrā and so forth. Yet for those who have the disciple [monastic] vows and who have given rise to the spirit of awakening in the Mahāyāna, consecration with mantra and mudrā is efficacious, provided that there is no secret consecration and so forth. However, for those who have given rise to the spirit of awakening in the mantric method who also have the disciple [monastic] vows, it is efficacious for them to receive the consecrations in their entirety. That [sort of practitioner] will be [like] Vajrasattva chastising Mārā. But if this were the case, how would the monastic vow not be impaired? They wouldn’t be, since, it is said that without abandoning the monastic vows, the vows are conceived differently in the three vehicles, the outer, inner and secret.35
Here, we see a quite nuanced discussion of the question of whether monks should receive the tantric initiations. He recommends that those ordained in the more conservative “Disciple Vehicle” (śrāvakayāna, nyan thos kyi theg pa) traditions and who are committed to asceticism not receive tantric initiation at all. However, those monks who are so ordained, yet also engaged in Mahāyāna training, may receive them, provided that they not receive the more transgressive second and third initiations. However, he also indicates that monks fully engaged in Mahāyana and tantric practice may receive them, without specifying in exactly what manner they would do so, although his comments seem to imply the actual practice, which would not necessarily impair the vow of celibacy, due to the ameliorating effect of the higher “inner” Mahāyāna and “secret” Mantrayāna vows that supercede them.
Through an act of rhetorical jiujitsu, Atiśa or his predecessors in India successfully neutralized a challenge to monastic authority. That is, a physical, sexual practice, which threatened the authority of Buddhist monastics insofar as it was portrayed as a “highest teaching”, was evidently transformed into an exercise of visualization that monks and other renunciants are most capable of successfully practicing, thus contributing to the maintenance of authority of Buddhist monastic elite.
The question that remains is why Atiśa recommended that monastics not receive the second and third initiations in his A Lamp for the Path to Awakening. One possibility is that the practice that Abhayākaragupta describes—of the majority of practitioners engaging in the transgressive rites via visualization only, with the actual performance of them reserved for exceptional practitioners—developed in North Indian monastic centers during the eleventh century. While this might have been the norm by the end of the century when Abhayākaragupta was active, it may not yet have been fully established in the earlier part of the century, when Atiśa was immersed in this context. In this case, Atiśa’s work would reflect a development toward this compromise. Another possibility is that this norm was already established by the early eleventh century, and that Atiśa found it politically necessary in A Lamp for the Path to Awakening to erect a barrier between erotic practices, real or imagined, and the monastic sangha, a barrier which turned out to be firm only with respect to the actual performance of these rites, and not their visualized enactments.
The meditation Atiśa describes is very similar to what practitioners are instructed to visualize in contemporary Tibetan enactments of the abhiṣeka ceremony, in which one is instructed to visualize the acts of sexual union central to the second and third consecrations, while symbolic substitutes are employed on the level of actual physical practice; in the case of the act of anointment in the second consecration, yogurt colored with a red liquid is usually used in place of the mixed male and female sexual fluids. Given the very clear description of the visualized enactment of these rituals in Abhayākaragupta’s work, it is clear that this solution was initially developed in Indian monastic centers such as Vikramaśīla and then was transmitted to Tibet. Atiśa played a role in this development, facilitating the acceptance of internal visualized enactments of transgressive rituals, even while on the surface opposing their practice by ordained Buddhists.


This research received no external funding.


This paper was originally presented at the Society for Tantric Studies conference in Flagstaff, Arizona in September 2019. I am indebted the conference organizers, as well to the helpful feedback I received from both conference participants and Religions peer reviewers. I am solely responsible for any faults that remain.

Conflicts of Interest

The author declares no conflict of interest.


  1. Primary Sources

    Abhayākaragupta. Śrīsampuṭatantrarājaṭīkā-āmnāyamañjarī-nāma. To. 1198, D rgyud ‘grel vol. cha, 1b-316a.
    Abhayākaragupta. Vajrāvalī-nāma-maṇḍalasādhana. To. 3140, D rgyud ‘grel vol. phu, 1b-94b.
    Atiśa Dīpaṅkaraśrījñāna. Abhisamayavibhaṅga. To. 1490, D rgyud ’grel vol zha, 186a–202b.
    Atiśa Dīpaṅkaraśrījñāna. Bodhimārgadīpapañjikā. To. 3948, D dbu ma vol. khi 241a-293a.
    Atiśa Dīpaṅkaraśrījñāna. Bodhipāthapradīpa. To. 3947, D dbu ma vol. khi 238a-241a.
    Buddhaśrījñānapāda. Samantabhadra-nāma-sādhana. To. 1855, sDe-dge rgyu ‘grel vol. di, 28b-36a.
    Lūipa. Cakrasaṃvarābhisamaya. To. 1427. D rgyud ’grel vol. wa, 186b–193a. Sanskrit edited in Sakurai 1998.
    Nāgārjuna. Piṇḍīkṛtasādhana. To. 1796. sDe-dge rgyu ‘grel vol. ngi, 1b-11a.
    Sarvarahasya-nāma-tantrarāja. To. 481, sDe-dge rgyud ‘bum vol. ta, 1b-10a.
    Vāgīśvarakīrti. Saṃkṣiptābhiṣekavidhi. To. 1887, D rgyud ’grel vol. pi, 184a–190a.
  2. Secondary Sources

  3. Bühnemann, Gudrun. 1992. Some Remarks on the Date of Abhayākaragupta and the Chronology of His Works. Zeitschrift der Deutschen Morgenländischen Gesellschaft 142: 120–27. [Google Scholar]
  4. Davidson, Ronald M. 1995. Atiśa’s A Lamp for the Path to Awakening. In Buddhism in Practice. Edited by Donald S. Lopez Jr. Princeton: Princeton University Press, pp. 290–301. [Google Scholar]
  5. Davidson, Ronald M. 2002. Indian Esoteric Buddhism: A Social History of the Tantric Movement. New York: Columbia University Press. [Google Scholar]
  6. Davidson, Ronald M. 2005. Tibetan Renaissance: Tantric Buddhism in the Rebirth of Tibetan Culture. New York: Columbia University Press. [Google Scholar]
  7. Decleer, Hubert. 1997. Atiśa’s Journey to Tibet. In Religions of Tibet in Practice. Edited by Donald S. Lopez Jr. Princeton: Princeton University Press, pp. 157–77. [Google Scholar]
  8. Gray, David B. 2001. On Supreme Bliss: A Study of the History and Interpretation of the Cakrasaṃvara Tantra. Doctoral dissertation, Columbia University, New York, NY, USA. [Google Scholar]
  9. Gray, David B. 2005. Eating the Heart of the Brahmin: Representations of Alterity and the Formation of Identity in Tantric Buddhist Discourse. History of Religions 45: 45–69. [Google Scholar] [CrossRef]
  10. Gray, David B. 2007. The Cakrasamvara Tantra: A Study and Annotated Translation. New York: AIBS/CBS/THUS [Columbia University Press]. [Google Scholar]
  11. Gray, David B. 2009. On the Very Idea of a Tantric Canon: Myth, Politics, and the Formation of the Bka’ ’gyur. Journal of the International Association of Tibetan Studies 5: 1–37. [Google Scholar]
  12. Gray, David B. 2013. “Imprints of the ‘Great Seal’”—On the expanding semantic range of the term of mudrā in eighth through eleventh century Indian Buddhist literature. Journal of the International Association of Buddhist Studies 34: 421–81. [Google Scholar]
  13. Gray, David B. 2015. Translation at the Limits of Buddhist Discourse: The Politics of the Translation of Esoteric Buddhist Scriptures. In In Vimalakirti’s House: A Festschrift in Honor of Robert A. F. Thurman on the Occasion of His 70th Birthday. Edited by Christian Wedemeyer, John Dunne and Thomas Yarnall. New York: American Institute of Buddhist Studies, pp. 73–104. [Google Scholar]
  14. Stephen Hodge, trans. 2003, The Mahā-Vairocana-Abhisaṃbodhi Tantra with Buddhaguhya’s Commentary. London: RoutledgeCurzon.
  15. Translated and Edited by Jeffrey Hopkins. 1977, Tantra in Tibet. London: George, Allen & Unwin, Ithaca: Snow Lion Publications.
  16. Karmay, Samten. 1980. The Ordinance of Lha Bla-ma Ye-shes-’od. In Tibetan Studies in Honour of Hugh Richardson: Proceedings of the International Seminar of Tibetan Studies, Oxford 1979. Edited by Michael Aris and Aung San Suu Kyi. Warminster: Aris & Phillips, pp. 150–62. [Google Scholar]
  17. Lo Bue, Emberto F. 1997. The Role of Newar Scholars in Transmitting the Indian Buddhist Heritage to Tibet (c. 750-c.1200). In Les habitants du toit du monde: Études recueillies en hommage à Alexander W. Macdonald. Edited by Samten G. Karmay and Philippe Sargant. Nanterre: Societé d’Ethnologie, pp. 629–58. [Google Scholar]
  18. Mori, Masahide. 2009. Vajrāvalī of Abhayākaragupta: Edition of Sanskrit and Tibetan Versions, Vols. 1 and 2 (Buddhica Britannica XI). Tring: The Institute of Buddhist Studies. [Google Scholar]
  19. Sakurai, Munenobu. 1996. Indo Mikkyō Girei Kenkyū: Kōki Indo Mikkyō no Kanjō Shidai. Kyoto: Hozokan. [Google Scholar]
  20. Sakurai, Munenobu. 1998. Cakrasaṃvarābhisamaya no genten kenkyū. Chizan Gakuho 47: 1–32. [Google Scholar]
  21. Wedemeyer, Christian K. 2007. Āryadeva’s Lamp that Integrates the Practices, Caryāmelāpakapradīpa, The Gradual Path of Vajrayāna Buddhism According to the Esoteric Community Noble Tradition. New York: AIBS/CBS/THUS [Columbia University Press]. [Google Scholar]
See, for example, chapter four of the Mahāvairocanābhisaṃbodhi Tantra, which contains descriptions of violent rituals for the desctruction of enemies; it is translated in (Hodge 2003, pp. 153–58).
For an overview of this history, see (Davidson 2002, especially chapter 4).
For example, see the apology found in the Mahāvairocanābhisaṃbodhi Tantra, chapter four, following the above mentioned violent rituals. The text here has Vairocana “predicting” that “faithless beings of little intelligence” will criticize this work, arguing that it was “not what was spoken by the Buddhas, but it belongs to the non-Buddhists!” (Hodge 2003, p. 157). The author of this text clearly expected that it would be subject to criticism by some Buddhists.
Examples of these attempts are discussed at some length in (Gray 2015).
Regarding the story of Atiśa’s journey to Tibet see (Decleer 1997).
For a translation of this text see (Davidson 1995, p. 301).
My translation from Atīśa’s Bodhimārgadīpapañjikā, 290a.7–b.4: de la dbang ni rnam pa gnyis te/khyim pa’i phyogs la brten pa dang/tshangs par spyod pa’i phyogs la brten pa’o//khyim pa’i phyogs la brten pa gang zhe na/ji snyed rgyud las gsungs pa thams cad do//tshangs par spyod pa’i phyogs la brten pa gang zhe na/de dag nyid las gsang ba dang/shes rab ye shes spangs pa’o//de ci phyir de gnyis spangs she na/‘di ltar sangs rgyas kyi chos la brten nas dge ba ji snyed cig ‘byung ba de dag thams cad nib stan pa gnas pa las ‘byung yin la/bstan pa gnas pa yang tshangs par spyod pa kho la ltos shing/dbang bskur ba gnyis ni tshangs par spyod pa’i mi mthun pa’i gnas su mthong ba’i phyir ro//de bas na dbang bskur gnyis ni tshangs par spyod pa zad par byed pa yin la/tshangs par spyod pa zad par byed pa yin la/tshangs par spyod pa zad na sangs rgyas kyi bstan pa nub par ‘gyur zhing/de nub pas bsod nams mngon par ‘du bya ba rnams rgyun chad par ‘gyur la/gzhi de las dge ba ma yin pa dpag tu med pa ‘byung ba’i phyir de gnyis tshangs par spyod pa rnams la spangs so zhes gsungs so/.
The final verse of the Bodhimārgadīpa indicates that Atiśa wrote it in response to being “implored by Byang chub ’od”, the member of the royal family who invited him to Tibet. See (Davidson 1995, p. 301).
My translation from the Tibetan edition of the Sarvarahasya-nāma-tantrarāja; see To. 481, 2b.2-3: /sbyor lnga sangs rgyas bdag ces bsgom//ting ‘dzin ye shes las byung ba//sangs rgyas nyid phyir mdor bsdus pas//sangs rgyas rnal ‘byor med par ni//rnal ‘byor pa ni ‘tshang mi rgya/. Note that the idea of the “five” yogas in the first line appears to be an interpolation in the Tibetan translation. It does not occur in the Chinese translation (see T. 888, p. 537). Prof. Alexis Sanderson has also informed me that it does not occur in the extant Sanskrit text, which I have not had the opportunity to consult.
Both Nāgārjuna and Buddhaśrījñānapāda argued that visualization meditations can purify the mind of the conceptualizations (vikalpa) that bind us to cyclic existence, thus investing this practice with tremendous soteriological significance. They make this argument in their Piṇḍīkrama-sādhana (To. 1796) and Samantabhadra-nāma-sādhana (To. 1855), respectively.
Abhayākaragupta, Vajrāvalī 61b.5. I am indebted to one of my peer reviewers for bringing this work to my attention.
Regarding the meaning of the term of art mudrā in this context see (Gray 2013).
Abhayākaragupta, Vajrāvalī 67a.2-4: dge slong gi nyams su myong ba de’i sbyor ba’i bde ba’i bag chags sad pas rmi lam du mthong na rmi lam du mthong ba’i phyi rol gyi phyag rgya la dbang po gnyis kyi bde ba’i mthar thug par nyams su myong na yang ltung ba med do zhes ’dul bar bstan pa de rtogs par bya’o//gal te phyogs kyi bye brag skye bo ngan pa’i spyod yul du gyur na/de’i tshe de la yang ye shes kyi phyag rgyas gsang ba dang shes rab ye shes kyi dbang sbyin par bya’o//skye bo ngan pa med na ni las kyi phyag rgya kho nas so//de kho na nyid la mos pa mi brtan pa’i dge slong la ni dad pa dang ldan yang phyi’i shes rab ma’i dbang bzung ba bkag ste/. I am aware that the Sanskrit for this text was edited in (Mori 2009). However, as I have not been able to secure a copy of what is evidently a very rare edition, I am unfortunately unable to take into consideration the Sanskrit text for this work.
Regarding this idea see (Gray 2007).
For a translation of the relevant passage from his Śrīsaṃpuṭatantrarājaṭīkā-āmnāyamañjarī-nāma, see (Gray 2007, pp. 69–70).
See (Gray 2001, p. 144). In my initial translation of this passage I only recognized the text’s reference to one of the four consecrations.
(Sakurai 1998), 3: prathaman tāvad yogeśvareṇa caturbrahmavihāriṇā bhāvyam/; Lūipa, Cakrasaṃvarābhisamaya, fol. 186b: rnal ’byor pa’i dbang phyug gis re zhig dang por tshang pa’i gnas bzhi bsgom par bya’o.
Atiśa, Abhisamayavibhaṅga, 186a.4-7: /shes rab la gnyis te/dmigs pa dang bcas pa’i shes rab dang/dmigs pa med pa’o//shes rab gnyis kyis mthong ba dang mthun par de la blo ’jog pa dang/spyod pa’i sgo nas gnyis med zung ’jug tu bsgom pa’o//de yang sgrub byed dbang bzhi’i sgo nas te/bum pa’i dbang gis kun rdzob kyi bden pa dang/gsang ba sngon du song ba’i shes rab ye shes kyis don dam pa dang /bzhi pas gnyis med zung ’jug tur togs pa bsgrub pa’o//de yang bla ma la rag las pas de mnyes pa las thob pa’i phyir ro//de ltar bsgrubs pas sgrub pa po’i mtshan nyid du gyur pa ni/dang po snying rje’i stobs las byung//rdzogs pa’i byang chub sems brtan byas//bum pa gsang sogs dbang bskur ba//bla ma’i drin gyis yang dag thob/.
Atiśa, Abhisamayavibhaṅga, 197a.1-5: /de nas lus kyi dkyil ‘khor bsgoms te/gsang ba’i padmar phag mo/spyi gtsugs la he ru ka’o//snying gar mkha’ ’gro ma/mgrin par lā ma/lte bar khaṇḍa ro ha/dpral bar gzugs can ma/mgo bo la sogs pa gnas nyi shu rtsa bzhir rab tu gtum mo la sogs pa bsgom par bya’o//kha la khwa gdong ma/sna g.yas par ‘ug pa’i gdong can ma/bshang lam du khyi gdong ma/sna g.yon par phag gdong ma/rna g.yon par gshin rje brtan ma/g.yas par pho nya ma/mig g.yas par mche ba ma/g.yon par ’joms ma bsam par bya//de nas oṃ ka ra ka ra la sogs pa las mchod phyir thogs bsam par bya//de nas khams nyi shu rtsa bzhi’i phra rab rnams dpa’ bo nyi shu rtsa bzhir bsam par bya’o//de nas oṃ ha la sogs pa las phyag rgya drug bskyes par bya’o//rdo rje phag mo’i sngags ring las mgul rgyan bskyed par bya’o//oṃ śrī badzra he la sogs pas thod pa’i phreng ba bskyed par bya’o//yi ge bdun pas rna rgyan dang//yum gyi snying pos glang po’i pags pa dang/dpa’ mo’i go cha’i sngags kyis ska rags dang/mkha’ ’gro ma rnams so so’i sngags kyis so so’i phyag gi thod pa dang/yum gyi nye ba’i snying pos thod pa’i dbu rgyan dang /.
Atiśa, Abhisamayavibhaṅga, 197a.5-7: /de nas sku gsungs thugs byin gyis brlabs pa dang/phung po la sogs pa byin gyis brlabs pa dang/go cha’i sngags kyi go cha dang/snying ga’i ’od kyis ye shes sems dpa’ spyan drangs la mchod de bsre ba dang/oṃ yo ga śuddhaḥ sa gnyis su med pa’i nga rgyal bya’o//de nas yang sum cu rtsa bdun spyan drang la mchod de/de nas lha de dag gis kyang bka’ bsgo rnams kyis bgegs bskrad/dpa’ mo rnams kyis glu len/dbang bskur rol mo’i bde chen po//mkha’ ’gro dra ba’i bde chen po//dgyes chen stsol ba’i bde chen po//gsol mdzad bde chen po ste//a la la ho//mchog tu gyur pa bde chen po//mkha’ ’gro dra ba’i chen po//a la la ho/zhes pa’i glu dang/dpa’ bo rnams kyis bkra shis brjod pa dang /.
These are the three subtle states of consciousness that emerge from (or, in reverse order, lead to) the clear light consciousness, namely “luminance” (āloka), “radiance” (ābhāsa), and “immanence” (upalabdhaka). Regarding them, see (Wedemeyer 2007, pp. 95–96).
Atiśa, Abhisamayavibhaṅga, 197a.7-b.4: yum bzhi phag mo dang lngas nam mkha’ nas ye shes kyi bdud rtsis gang ba’i bum pa blangs nas/lag pas dbang bskur par bsam mo//de nas yum bzhi phag mo la thims ste/dpal he ru ka snyoms par zhugs pa’i byang chub kyi sems kyis gsang ba’i dbang bskur bar bya’o//de nas he ru kas phag mo mu dra gnang ste snyoms par zhugs pas rlung thim/de la brten nas lhan cig skyes pa myong bar bsam mo//de nas dpal he ru kas rigs kyi bu khyed kyis mu dra dang gnyis sprod pa la brten nas myong ba snang ba thob pa’i shes rab de ’od gsal bar sgoms shig /de ni rnam par shes pa gsum rnam par dag pa shes rab kyi pha rol tu phyin pa’i ngo bo nyid lus dang ngag dang sems med pa las dang/skye ba las grol ba/zla ba dang/nyi ma dang/me dang/nor bu ltar shin tu gsal ba rtag tu snang ba’i mtshan nyid can don dam pa’i bden pa de ni mthong ba gzhan gyi g.yul las ’das pa ste/shes rab dang ye shes kyi mig gis ltos shig /de ltar gang gi yang rang bzhin du ma mthong ba de ’od gsal ba mthong ba yin no//de ltar ’od gsal ba de’i don rgyu ’bras gang yang med mod kyi/’on kyang kun rdzob tu rlung dang sems tsam las lha sum cu rtsa bdun me long gi gzugs brnyan ltar gsal la rdzogs pa/’ja’ tshon ltar kha dog dang bcas pa/chu zla ltar sgro skur dang bral bar ltos shig /de ni rgyu rkyen la ltos nas skyes pas kun rdzob bo//de ltar bden pa gnyis mi byed pa zung du ’jug pa ni byang chub kyi sems chen po ste bzhi pa’o/.
Atiśa, Abhisamayavibhaṅga, 197b.4-198a.1: /de nas dbang bskur ba’i lha rnams bdag la zhugs par bsams la so sor rgyas gdab ste/snga ma ltar mchod bya mchod byed mchod pa rnams skye med du shes par byas la mchod pa bya’o//de nas bdud rtsi myang ste/yaṃ raṃ las me rlung a las thod pa de’i nang du/go ku da ha na rnams las sha lnga dang/bi mu shu ma ra rnams las bdud rtsi lnga’o//de rnams kyi steng du hūṃ oṃ tram hrī a ste rigs lnga’o//a las byung ba’i zla ba’i dkyil ’khor gyis kha dgab par bya’o//de’i steng du oṃ āḥ hūṃ gsum ste rigs gsum mo//de’i steng du zla ba la gnas pa’i rdo rje rtse dgu pa hūṃ ’od kyi phreng ba can gyis mtshan pa ni gsang ba chen po drug pa’o//de nas yi ge bzhi las ’od ’phros pas de bzhin gshegs pa thams cad kyi sku gsum thugs ye she kyi bdud rtsi bkug pa pa so so la ’thim par bsam mo/.
For more detailed descriptions of these consecrations see (Gray 2007, pp. 103–31).
Here, I refer to his 1996 work, Indo Mikkyō Girei Kenkyū: Kōki Indo Mikkyō no Kanjō Shidai, which I have unfortunately not been able to access or consult, although I understand that he discusses in this work the connection between Vāgīśvarakīrti’s and Abhayākaragupta’s comments on the second and third initiations. I am indebted to one of my peer reviewers for bringing Vāgīśvarakīrti’s work to my attention.
See (DeCleer 1997, pp. 159–62). We do not know, to my knowledge, how long Atiśa lived at Vikramaśīla, nor when Vāgīśvarakīrti departed the monastery or passed away; it is possible that he was there at the same time as Vāgīśvarakīrti, who would have been a senior contemporary.
Regarding this, see (Gray 2009, pp. 14–16), which translates a passage in Atiśa’s biography relating to his pride about his extensive knowledge of the tantras, and the way in which the ḍākinīs humbled him.
Vāgīśvarakīrti, Saṃkṣiptābhiṣekavidhi, 189a.1-4: ’on kyang dge slong ’ga’ zhig gcig tu nges par grags pa’i dka’ thub can nyan thos kyi theg pa lta ba gcig pa ste/de la ni sngags dang phyag rgya la sogs pa’i dbang bskur ba dbang med pa nyid do//yang gang zhig nyan thos kyi sdom pa dang theg pa chen por sems bskyed pa dang ldan pa ste/de la sngags dang phyag rgya la sogs pa la dbang yod de/de yang gsang ba’i dbang bskur ba la sogs pa med do//yang gang zhig nyan thos kyi sdom pa dang sngags kyi tshul la yang sems bskyed pa dang ldan pa ste/de la ni thams cad du dbang bskur ba la sogs pa’i dbang yod pa ste de nyid ni rdo rje ’dzin pas bdud la bsdigs pa la sogs pa bya’o//’o na de ltar na yang ji ltar na dge slong gi sdom pa mi nyams par ’gyur ram zhe na/mi ’gyur te/’di ltar dge slong gi sdom pa ma btang bar/phyi nang gsang ba’i theg pa gsum//sdom pa so sor gzung bar bgyi/. I understand that the Sanskrit for this text is edited in (Sakurai 1996), although unfortunately I have not been able to secure a copy of this work.

Share and Cite

MDPI and ACS Style

Gray, D.B. The Visualization of the Secret: Atiśa’s Contribution to the Internalization of Tantric Sexual Practices. Religions 2020, 11, 136.

AMA Style

Gray DB. The Visualization of the Secret: Atiśa’s Contribution to the Internalization of Tantric Sexual Practices. Religions. 2020; 11(3):136.

Chicago/Turabian Style

Gray, David B. 2020. "The Visualization of the Secret: Atiśa’s Contribution to the Internalization of Tantric Sexual Practices" Religions 11, no. 3: 136.

Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop