Until the Dalai Lama received the Nobel Price in 1989, Buddhism had been largely portrayed as a peaceful religion. But since the 1990s, several articles and books have appeared that critically investigated this image. By now, western scholars and the public have a much more balanced view of the history of Buddhism and the violence that was a part of it. Unfortunately, not all contributions to this new critical assessment have applied the necessary care and fairness. In this posting, I would like to investigate two recent and influential contributions to the study of violence in tantric Buddhism. Both portray particular kinds of Buddhist tantric practices as involving “real” violence; both conclude that people are really getting killed here. While I do not want to argue that such a thing—killing of people in the context of mantra—did not happen in the history of Buddhism, I do want to point out that the sources that were used for these studies not only do not allow that conclusion but are actually prime examples for texts that leave no doubt that the ritual activities are only imagined (: text) and that this kind of imagining has a particular training purpose (: context).

How could such a misjudgment have come about on the part of the authors? I will show that this is mainly due to mistranslations of the texts and to a lack of cultural context.

Interestingly, Jacob Dalton’s recent book, The Taming of the Demons (2011),^{♦ 1} provides – despite its many attempts to prove that the texts he investigates speak of real actual violence – the best arguments for the contrary.^{♦ 2} In his introduction, Dalton first concedes that “[c]onclusive proof of early tantric ritual killings may never be found; indeed, it is unclear what such evidence would even look like” (p. 3). But then he speaks throughout his book of “ritual killing of actual people” (p. 3), “a Buddhist rite for human sacrifice” (p. 4), “the [bestowed] right to enact violence” (p. 14), and that “several details suggest that a live person may be intended” [to be killed] (p. 77).

Dalton knows that later Tibetans were adamant about killing “not being performed on a live person, and their ritual manuals bear this out” (p. 4). His thesis is that the old 9th century (or earlier) Tibetan texts found in the famous library cave of Dunhuang do not “bear this out,” but rather are explicit in stating instructions for actual killing of living persons.

His chief evidence for the alleged existence of actual violence in these early rites is contained in two texts from Dunhuang. One of them (PT 840/1), which he presents in Appendix C of his book, he translates thus (p. 212, my square brackets):

“When citta [the consciousness] is led forth by the mudrā, imagine that it is grabbed with ʻja’ʼ (bzung bar bsam),” … “imagine he [sic!] is led (’dren par bsam),” and “imagine that he [sic! ] is sent (bton par bsam),” [i.e., to nirvana].^{♦ 3}

This passage of the text extends over five and a half lines in his translation. The object that is “led,” “grabbed,” and “sent” remains the same: the consciousness (citta) of the enemy. Consciousness has the neutral gender, and thus “it” is grabbed—surely “grabbed” in a metaphorical sense. But then, suddenly, in that same passage, Dalton switches to the male gender when he translates “he is led” and “he is sent” to nirvana. What justifies this switching from consciousness to a real person—for which there is no indication in the Tibetan text? Was the wish to prove that this is an instance of killing of a real person the father of this not so subtle slip? Moreover, even if one could label this as violence, one could not state more clearly that it is entirely imagined violence (note the Tibetan verb bsam in all quotes).

The second example (PT42/ITJ419) is presented as his supposedly strongest evidence of the entire book for the possibility that a real and human victim is actually to be sacrificed. It begins with the following line (in my translation):

Imagine that a demonic iron ax that one holds aloft performs the activities of the Lady of Death.

Dalton translates “holding aloft the demonic iron axe, imagine that one performs the activities of the Lady of Death.” The Tibetan text says sta re thogs pa gcig, lit. “an ax held aloft.” The gcig in this syntactic position must refer to the ax; it cannot mean “one.” Thus, it is not oneself, the performer of the ritual that one imagines to carry out the destructive activities; it is the ritual ax. Therefore, the violence is thrice removed: (1) It is imagined, (2) it is carried out by the ax, (3) and that ax is not a real ax, but a ritual simulacrum. In exactly this manner the text continues (my translation):

Then^{♦ 4} imagine a dark blue syllable KRONG atop the head [of the being that is liberated]. … From the DRONG (!) atop the head, many razor-[like] weapons cleave and chop, the consciousness emerges, and it is offered to the principal [deity]. (…) Having made all these visualizations (bsam ba’i rnams) clear in his mind, the master must revive the consciousness and hold it. He throws it on top of the mandala and then has to examine the signs.^{♦ 5}

In previous passages, the enemy to be liberated, termed throughout the text “the object of compassion,” is placed in the middle of the mandala, facing west, and smeared with white mustard seed paste. Is this a real person, a real body of a living being? Even if that should be the case—for which there is no evidence—it is clear that everything else is imagined: A syllable is imagined atop the head, from that appear visualized weapons, and it is imagined that the consciousness appears and is offered to the visualized deity. The text makes this as explicit as one could wish: “Having made all these visualizations (bsam ba’i rnams) clear in his mind”—thus, the ritual master is visualizing the activity. Other than the visualized weapons there are no real weapons mentioned in the text through which a head could be chopped off. Also, what else but a visualization would be the emerging consciousness? Tibetans do not have a concept according to which any part that could be chopped out of the head would be seen as the consciousness.

Now the master “must revive (sos) the [visualized] consciousness and hold it. He throws it on top of the mandala and then has to examine the signs.” As Dalton himself has noted a few pages before (p. 79), in the Guhyasamāja Tantra, i.e., the canonical work from which this ritual manual derived, the thing that is beheaded is, in fact, a paper effigy. Moreover, in a footnote (p. 239, n. 40) he states: “The interlinear notes to the same lines in the tantra make the use of a paper effigy still more explicit (ITJ438, 53v5: ri mor bris pa de’i mgo gcad par byas pa).” The interlinear note says: “The head drawn in the picture is chopped off.” Isn’t thus the most probable scenario that the “object of compassion” is represented in the described ritual by a picture drawn on a piece of paper or by any other effigy, such as one made of dough? The signs that are to be examined after the effigy is thrown onto the mandala are in accordance with that: The effigy faces into one particular direction, it lands “face down,” or it may “split at the crown” (which indicates an effigy made of dough).^{♦ 6}

Despite all this evidence and virtually no evidence to the contrary, Dalton insists in his translation that what is thrown onto the maṇḍala is a real head. Moreover, he describes the above ritual procedure with these words:

Certainly the most violent text to emerge from the library cave ofDunhuang is a ritual manual for the performance of the notorious “liberation rite” (sgrol ba). Many early Mahayoga writings from Dunhuang and elsewhere mention the liberation rite, but none is so explicit or detailed as this manual. … The instructions have the victim brought in and placed at the center of the ritual altar so that he faces west; the weapon is blessed and the victim purified, before being beheaded with an axe. Finally, the position in which the head comes to rest is interpreted to determine the rite’s success. The description is unusual, as it supplies so much detail yet makes no mention of an effigy.

To repeat: The iron ax is a ritual instrument that is clearly not used to chop off any heads (“imagine[!] that a demonic iron ax that one holds aloft performs the activities”); the “weapons” that arise from the syllable are visualized, and whatever activity they perform can therefore only be imagined; according to the text, it is not the head that is thrown onto the mandala, but the consciousness; and although the ritual text itself does not mention an effigy, the tantra from which it is derived and its interlinear notes do. Dalton’s above description of the ritual is forced. If this is his “most violent” and most explicit example, his thesis that the old 9th century (or earlier) Tibetan texts from Dunhuang are explicit in stating instructions for actual killing of living persons is failed.

My second example is an article by David Gray (esp. pp. 248 ff.).^{♦ 7} In it, he refers to an interesting passage in Atishaʼs Abhisamayavibhanga.^{♦ 8} Atisha discusses here destructive means in a ritual text. Gray concludes from his translation of the passage that Atisha “justifies” and “legitimate[s] such [violent] actions.” Moreover, Gray states: “This justifies violence by those who have controlled their minds, and are thus not motivated by the passions, but rather by the cool calculus of compassion (…).” Apart from his rather eccentric formulation “cool calculus of compassion,” which is obviously a contradiction in terms, and other strange expressions (“Atisha finds solace”), Gray has, in fact, not done justice to the subtlety of Atishaʼs argument. For my discussion I would like to begin the passage Gray investigates two lines earlier because these lines provide the context for Atishaʼs reply (all translations are mine):^{♦ 9}

འོ་ན་ཀུན་རྫོབ་ཏུ་གནོད་བྱ་གནོད་བྱེད་མེད་དོ་ཞེ་ན་ནི་རྒྱུ་འབྲས་ལ་སྐུར་པ་ཡིན་ལ།
Query: Are, conventionally speaking, the one who is going to be harmed and the harm doer inexistent? Reply: That would be to disavow cause and result!

The interlocutor asks if the person who is going to be harmed, namely oneself (attacked by the demon or enemy) and the harm doer (namely the demon or enemy who will have to be repelled) are, conventionally speaking, inexistent. (Alternatively the one who performs the rite could be the harm doer and the demon or enemy the one to be harmed. This does not make a difference in the logic of the argument.) Atisha denies their inexistence because that would mean to disregard the teaching of cause and result on the level of conventional reality. In other words, conventionally speaking, both oneself and the enemy/demon do exist. The next query follows:

ཡང་ན་ནི་ཐ་དད་ཀྱི་ཚུལ་གྱིས་བཟློག་པ་ལ་འགལ་བ་མེད་དོ་ཞེ་ན།
Query: Is there not a contradiction concerning a repelling by way of [seeing the enemy] as something separate?

This question concerns the teaching that all appearances are only the mind. If that is the case, is the enemy not also nothing but oneʼs mind? Is it, therefore, not a mistake to see him as a separate entity? Now follows the part that Gray has presented in his article, again in my own translation:

ཀུན་རྫོབ་ཏུ་རྒྱུ་འབྲས་མེད་པ་མ་ཡིན་གྱི། འོན་ཀྱང་ཞེ་སྡང་གིས་བསླང་བའི་སྦྱོར་བས་བཟློག་པར་བྱ་བ་མ་ཡིན་ཏེ། ཀུན་རྫོབ་ཉིད་དུ་ཡང་བྱམས་པ་ལ་སོགས་པའི་གོ་ཆ་དང་དོན་དམ་པར་སྐྱེ་མེད་དུ་ཤེས་པས་སོ། ད་ཡང་ཅི་ཞེ་ན། ཀུན་རྫོབ་ཏུ་ཆོས་ཐམས་ཅད་སེམས་ཙམ་ལས་གཞན་མ་ཡིན་ཏེ། དེ་བས་ན་སེམས་ལོག་པར་གཡོ་བ་ཉིད་བདུད་དང་བགེགས་ཡིན་ནོ། དེ་ཡང་ངན་སོང་སྦྱོར་བའི་རྒྱུད་ལས། ཇི་སྲིད་ཡིད་ཀྱིས་ལོག་གཡོ་བ༎ དེ་སྲིད་བདུད་ཀྱིས་སྤྱོད་ཡུལ་ལོ༎ ཞེས་གསུངས་སོ༎ སེམས་གཉེན་པོ་ལས་འཁྱར་བ་ཉིད་བདུད་ལ་སོགས་པ་ཡིན་ནོ༎
Reply: Conventionally, cause and result do not fail [lit. “are not inexistent”]. However [even though the enemyʼs harm is experienced as a real threat] it is wrong to repel [harm doers] by the application of a vengeful motivation because conventionally [one wears] the armor of love, and so forth, and ultimately one understands [harm doers] to be without birth. How is that? Conventionally, all phenomena are nothing but merely mind, and, therefore, the very deviation from [the understanding that phenomena are only] the mind is the Mara and the demons [that one must repel]. As the Sarvadurgati Tantra says:^{♦ 10}

“Just as much as one wavers off with the mind,
just that much is the sphere of Mara.”

The very straying from [the understanding that all phenomena are] the mind, which is the antidote, is Mara and so forth.

Gray concludes from this:

Atisha here invokes the Yogacara theory of the baselessness of imputations of independent existence to phenomenal reality in order to deny the external reality of the demons who are the targets of the sadhanaʼs ritual violence. (…) This argument seems a bit inconsistent; if the demons do not in fact exist as independent entities, what need is there to insist that their destruction should be performed with a compassionate motivation?

Gray says: Since appearances are only mind, the demon doesn’t exist, and, thus, to say that he is destroyed with compassion is inconsistent. However, the passage translated from Atishaʼs text so far only pertains to a first and conventional part of his argument. Here, however, Atisha does not at all “justify” or “legitimate” a violent killing of a victim. On this conventional level, he instructs to apply love as an antidote to hatred and, still conventionally, to understand that the phenomenon perceived as the enemy is nothing but oneʼs mind. He also briefly alludes to an absolute level that is without the arising of such conventional phenomena like enemies and demons, but this will be only discussed later. Here, on the conventional level, one applies antidotes: love and the understanding that all phenomena are one’s mind.

One of the most important requirements for understanding this kind of ritual is that this is a training. As in all Mahayana practices, love for all sentient beings is cultivated at the beginning. Then, based on that, when there is a disturbance, one applies first of all that loving-kindness, and secondly one trains one’s understanding that all phenomena are only mind. This is the typical and well-known Mahayana training as it is expressed, for instance, in the Thirty-seven Trainings of the Bodhisattvas:

(16) Even if someone whom we have taken care of like our own child,
is seeing us as an enemy, we only increase our love for him,
like a mother for her sick child.
That is the training of the bodhisattvas.

(20) When hatred, our enemy, is rampant,
we may defeat the outer enemies, but they still increase.
Thus, one’s mental continuum is to be tamed with the armies of love and compassion.
That is the training of the bodhisattvas.

(22) Whatever appears is our mind.
Its nature is primordially free from proliferation.
Knowing that, not to indulge in the characteristics of perceived objects
and perceiving mind is the training of the bodhisattvas.

Up to this point in Atisha’s text, there is no one killed or repelled, neither with anger nor with compassion. This is a training of the mind. And here now follows in the text the crucial line:

… the very deviation from [the understanding that phenomena are only] the mind is the Māra and the demons [that one must repel].

Who or what is the enemy? It is one’s own deviating from this training. Just as much as one deviates from that, Mara can operate. The ritual practice of repelling enemies is, up to this point, explained by Atisha as a training in love and the understanding that appearances, even if they are threatening, are only one’s mind. Within a ritual practice such as the one commented upon by Atisha, the training takes place within several “frames,” from the outside to the inside: renunciation (when one contemplates the four thoughts that turn the mind away from samsara), love, compassion, and the resolve for awakening, purity and emptiness (with the shunyata-mantras, etc.), and, when appearances are mentally deleted and once again arise from the syllables, the understanding that all appearances are only the mind.

It can hardly be stated in clearer terms than Atisha provides them here, that the enemy one has to remove is oneʼs inability to cultivate these “frames” in one’s meditative practice. Whatever happens within these frames is a training of the mind to realize the true nature. The very wavering from these frames is the Mara that one must (metaphorically) “kill.” Since there is no person here that one must kill—only one’s own shortcomings in the training—to call this a “justification for killing” is to misunderstand the whole character of this endeavor as Atisha describes it. The enemy that appears is obviously an instance of training, and, as I have understood from interviews, it can become a particularly powerful training when it is done under adverse conditions, i.e., when, conventionally speaking, one perceives a real threat.

Atisha also clearly states that it would be wrong on the level of conventional truth to attack a real enemy, because the enemy conventionally exists for a deluded mind, and to attack him would have karmic consequences for the attacker in agreement with the teaching of cause and result. Therefore, the antidote to hatred for an enemy perceived as real is “the armor of love.” Once the yogi or yogini has removed the gross obscuration of hatred, he or she should train in the manner described above to realize that the threat that is perceived is nothing but mind. Atisha nowhere says, as Gray claims, that the enemy must be killed with compassion. The “armor of love” is not to prevent the yogis from karmic consequences; its purpose is to train and tame the mind.

Atisha provides the above explanation on the level of conventional truth. What follows is his brief argument on the level of absolute truth, which has been briefly alluded to with the above line “and ultimately, one understands [harm doers] to be without birth.” The text continues:

དེའི་ཕྱིར་ངེས་བརྗོད་ལས། མི་གསོད་གསོད་པ་མེད་པ་ཡིན༎ གང་གིས་རང་སེམས་ཐུལ་བ་ཡིས༎ བདག་གི་སེམས་ནི་བཅིངས་པ་ཡིན༎ གང་ལ་གང་གིས་གསོད་པ་ཡོད༎ ཅེས་པ་དང།
(…) Thus the Abhidhānottara says:

“There is neither killing nor not-killing.
One who has tamed his mind
is binding his mind.
Who is killed by whom?”

Note that Atisha does not primarily argue that “harm doers have no birth” with a mere “intellectual” argument of emptiness, but that he embeds his teaching in the “practical” instruction of binding the mind by taming it. This refers to exactly the training and taming of the mind as it was explained above, i.e., by entering successively into renunciation, love, and so forth, up to the realization of “only the mind” within the ritual of liberation embedded in mahayogatantra sadhana practice. Within such a realization of “only mind,” one must indeed ask the question “who is killed by whom?”^{♦ 11} Atisha concludes the argument with a continuation of the quote and his conclusion:

ཡང་། བྱམས་པའི་གོ་ཆ་བགོས་པ་དང་༎ སྙིང་རྗེ་ཆོས་ཀྱི་གོ་ཆ་ཡིན༎ ཤེས་རབ་མཚན་ཆར་ལྡན་པ་ཡིས༎ ཉོན་མོངས་བདུད་རྣམས་བཟློག་པར་བྱ༎ བཀའ་ཡི་འཁོར་ལོ་སྲུང་བ་ཆེ༎ ཕུར་བུས་བགེགས་མེད་དངོས་གྲུབ་འགྱུར༎ སྲུང་བའི་ཆོ་ག་འདི་དག་གིས༎ སྒྲུབ་པ་པོ་ལ་བྱང་ཆུབ་སྩོལ༎ གཙོ་བོའི་བཀའ་ནི་བླང་བར་གྱིས༎ རྣལ་འབྱོར་པ་ནི་གར་གནས་པར༎ བགེགས་རྣམས་ཐམས་ཅད་མེད་པར་མཐོང་༎ ཞེས་གསུངས་པའི་ཕྱིར་རོ༎ དེ་ལྟར་མ་ཡིན་པར་བཀའ་ལ་ལོག་པར་རྟོགས་ཏེ་ངན་པའི་ལས་དང་ལྡན་པ་ངོ་བོ་ཉིད་ཀྱིས་འོག་ཏུ་འགྲོ་བ་ལ་ཕྱོགས་པ་ཉོན་མོངས་པ་ལ་སོགས་པའི་ཞགས་པས་བཅིངས་པ་ཟོ་ཆུན་ལྟར་འཁོར་བ་རྣམས་དང་ཁྱད་པར་མེད་པས་ལམ་ལ་གནས་པའི་རྣལ་འབྱོར་པ་ཡང་མ་ཡིན་ནོ༎
Moreover:

He wears the armor of love
and the armor of the Dharma of compassion.
Being endowed with discriminative knowledge as a weapon,
he should repel the afflictions, which are the Mara.
This wheel of instruction is a great protection.
With the kila it turns into the siddhi of freedom from obstruction.
With these armors of protection
awakening is bestowed on the adept.
Hold this principal instruction!
Wherever he abides, the yogi
will perceive all obstructions as not existent.

To understand the instructions not like that and, thus, wrongly [means that] one is not a yogi abiding on the path since one possesses bad karma, moves towards the lower realms through oneʼs very nature, is bound by the noose of affliction, etc., and is not different from samsaric beings, roaming [in samsara] like a water wheel.

Gray, however, concludes from the above passage:

This justifies violence by those who have controlled their minds, and are thus not motivated by the passions, but rather by the cool calculus of compassion, which calls for violence as a defensive strategy, that is, as a way preventing evildoers from committing greater acts of violence. This denial of the reality of violence …

Such a complete misunderstanding of the intention of Atisha’s texts is only possible through several translation mistakes and a biased reading. If there ever existed a Tibetan text that justified violence, then it is certainly not this one.

Notes
^{1. []↩} Dalton, Jacob P. 2011. The Taming of the Demons: Violence and Liberation in Tibetan Buddhism. New Haven: Yale University Press.

^{2. []↩} The qualities and problems of this book were reviewed by Matthew Kapstein, 2013, Review of The Taming of the Demons: Violence and Liberation in Tibetan Buddhism by Jacob P. Dalton (New Haven: Yale University Press, 2011). Harvard Journal of Asiatic Studies 73.1: 177–84; and Cathy Cantwell, 2014, Review article of The Taming of the Demons: Violence and Liberation in Tibetan Buddhism by Jacob P. Dalton. History of Religions 54.1: 106–12.

^{3. []↩} Lines 37–9 in the Tibetan text.

^{4. []↩} The first sentence quoted ends with the Tibetan particle las, and Dalton connects this to the Lady of Death: “from her” appears a syllable. However, that cannot be because such a function of the las particle is only possible when it is attached to a noun. In the text, it is attached to a verb, “imagine,” and in this case it simply continues the action.

^{5. []↩} Cf. Dalton 2011, 208.

^{6. []↩} It is difficult to explain that according to one sign the effigy “does not stop shaking.” But that is as difficult to explain for an effegy as for a real human head once it is chopped off.

^{7. []↩} Gray, David B. 2007. “Compassionate Violence? On the Ethical Implications of Tantric Buddhist Ritual.” Journal of Buddhist Ethics 14: 239–71.

^{8. []↩} Abhisamayavibhanga by Atisha, 188r f. Mngon par rtogs pa rnam par ’byed pa, D vol. 22, 185v7–202v3.

^{9. []↩} These and the following quotes are taken from Atishaʼs Abhisamayavibhanga, 188r f.

^{10. []↩} One of the more obvious problems of Gray’s translation is that he did not recognised the Sarvadurgati Tantra quote and has translated the words of the title and the quote as if they were part of Atishaʼs speech.

^{11. []↩} But Gray translates: “Yet those whose minds are bound kill one another,” which is syntactically impossible. He also overlooks that this is a question.