Working with the earliest surviving Tibetan documents, it’s impossible not to be aware of differences between the way things are presented in traditional Buddhist histories and what we see in the manuscripts. Having done my doctoral research on Dzogchen,  I’ve always been interested in the divergence between the traditional image of early Dzogchen and the picture that emerges from the manuscript sources.

My first attempt to deal with this divergence was an article called “The Early Days of the Great Perfection” back in 2004 (which you can download here). In the first half of that article I tried to follow the way the contexts and usage of the word Dzogchen itself developed over time. This approach showed Dzogchen first appearing as the culmination of the meditative practice of deity yoga (the visualization of a deity and recitation of his or her mantra) around the 8th century. And then in the 9th and 10th centuries, Dzogchen became a way of contextualizing deity yoga in terms of nonconceptuality, nonduality and the spontaneous presence of the enlightened state.

One of the objections to this view of the gradual evolution of Dzogchen is the ‘nine vehicle’ system of the Nyingma school. This Tibetan way of organizing the Buddha’s teachings builds on a ‘three vehicle’ system from India, which comprised the vehicles of the śrāvakas, pratyekabuddhas and bodhisattvas. To this are added three vehicles of ‘outer’ yoga, and three vehicles of ‘inner’ yoga, making nine. The top three vehicles are Mahāyoga, Anuyoga and Atiyoga. Dzogchen is located at the very top of this system, within the ninth vehicle of Atiyoga. If Dzogchen was always a separate vehicle, then the idea of its primary role ever having been as a mode of practising deity yoga seems far-fetched.

So, in that same article, I tried to trace the the evolution of the term Atiyoga as well. The earliest instance of the term that I found was in an 8th century tantra called Sarvabuddhasamāyoga, one of the earliest of the yoginī tantras. In one part of the tantra, the stages of ritual practice are laid out, starting with Yoga, and then proceeding to Anuyoga and Atiyoga:

Through Anuyoga the bliss of all yogas is practised,
And through Atiyoga the true nature is fully experienced.

In this tantra there seems to be an association of Anuyoga with yogic bliss, and Atiyoga with a realization of the nature of reality via that bliss. This ties in with the three stages of deity yoga described in a work attributed to Padmasambhava: development (kye), perfection (dzog) and great perfection (dzogchen).

In another tantra, the Krṣṇayamāri, we have four stages of yogic practice: Yoga, Anuyoga, Atiyoga and Mahāyoga.  Here Atiyoga is the penultimate stage, below Mahāyoga. In any case, in these Indic sources there is no sense that Atiyoga is anything like a vehicle. Instead it is a stage or aspect of yogic practice.

*  *  *

Even in Tibetan sources, we don’t see Atiyoga identifed as a separate vehicle before the 10th century. Instead it is characterized as a ‘mode’ (tshul) or a ‘view’ (lta ba) to be applied within deity yoga. Here’s an example: in the 9th century treatise, The Questions and Answers on Vajrasattva we have the following explanation about the right way to practise deity yoga:

In the ultimate deity yoga no subject or object is perceived. Because there are no difficulties or effort, this is the highest deity yoga.

A note written underneath the second line says that this is “an explanation of the view of Atiyoga.” That is to say: Atiyoga is still at this point a way of practising deity yoga. (The manuscript, by the way, is IOL Tib J 470.)

*  *  *

So when did Atiyoga become a vehicle? Moving on to the 10th century, there are a couple of texts from Dunhuang which do set out early versions of the nine vehicle system. Yet even here, though we see the beginnings of the standard distinctions between Mahāyoga, Anuyoga and Atiyoga, these three are not yet called ‘vehicles’. The texts carry on presenting Anuyoga and Atiyoga as modes of Mahāyoga practice, without any specific content of their own.

As far as I know, the first sign of Atiyoga becoming a vehicle is in the work of the great scholar of Tibet’s “dark age”, Nub Sangyé Yeshé. But even in his work, this seems to be a tentative first step. In Nub’s Armour Against Darkness (written in the late 9th century) he treats the yogas of Mahā, Anu and Ati as systems (lugs) representing modes (tshul) of practice, and not  as vehicles. In fact they are specifically characterized as the lower, middle and higher divisions of a single vehicle.

It is in the Lamp for the Eyes of Contemplation, which Nub wrote at the beginning of the 10th century, that he sometimes refers to Atiyoga as a vehicle. But he does so rather haphazardly. In his final summary of the differences between Mahāyoga and Atiyoga, he doesn’t call them vehicles (though he doesn’t call them modes either). In general the Lamp for the Eyes of Contemplation stands midway between the understanding of Mahāyoga, Anuyoga and Atiyoga as modes of esoteric yoga, and the understanding of them as independent vehicles.

*  *  *

So far as I have been able to tell, there is no reliable source before the 11th century for the classic presentation of the nine vehicles as vehicles. Though such a source may yet come to light, I suspect that Atiyoga was not widely and consistently treated as a vehicle with its own specific practices before that time. By then a context existed in which some people (in the newly emerging Nyingma tradition at least) accepted this definition of Atiyoga. And this same context allowed Dzogchen to be understood as more than a way of  doing deity yoga practice. It’s interesting to note, though, that even in the 13th century (and later) the idea of Atiyoga as a vehicle was controversial in other Buddhist schools. Sakya Pandita wrote in his Distinguishing the Three Vows that:

If one understands this tradition properly,
Then the view of Atiyoga too
Is wisdom and not a vehicle.

*  *  *

See also:

Early Dzogchen I: The Cuckoo and the Small Hidden Grain
Early Dzogchen II: An approach to tantric practice
Early Dzogchen III: The origin of Dzogchen

*  *  *

References:

This post draws heavily on an article published in 2004: “The Early Days of the Great Perfection” in Journal of the International Association of Buddhist Studies 27.1 (2004): 165–206. (You can download a PDF from the link at the beginning of this post, or the “Author” page of this site.)

I have also drawn on an article from 2008: “A Definition of Mahāyoga: Sources from the Dunhuang Manuscripts.” Tantric Studies 1 (2008): 45-88. (Not yet scanned, unfortunately.)

And on those two doxographical texts, have a look at Jacob Dalton’s “A Crisis of Doxography: How Tibetans Organized Tantra in the 8th-12th Centuries” in the Journal of the International Association of Buddhist Studies 28.1 (2005): 115–182.

Nub Sangyé Yeshé’s Armour Against Darkness can be found in the Rnying ma bka’ ma shin tu rgyas pa (v.93, pp.7-680). Its full title is: Sangs rgyas thams cad kyi dgongs pa ‘dus pa’i mdo’i dka’ ‘grel mun pa’i go cha lde’u mig gsal byed rnal ‘byor nyi ma.

And his Lamp for the Eyes of Contemplation is also in the Rnying ma bka’ ma shin tu rgyas pa (v.104, pp.575-1080): Sgom gyi gnad gsal bar phye ba bsam gtan mig sgron.

*  *  *

Finally, a bit of Tibetan and Sanskrit:

Here’s the passage from the Sarvabuddhasamāyoga tantra (P.8, 184-4-7): rjes su sbyor bas mchod byed cing/ rnal ‘byor kun gyi bde ba dag/ bdag nyid kun tu myang byed na/ shin tu sbyor bas ‘grub par ‘gyur/

The Sanskrit text of this verse is found in the ninth chapter of Āryadeva’s Caryāmelāpakapradīpa, which was kindly pointed out to me by Harunaga Isaacson: pūjyate ‘nuyogena sarvayogasukhāni tu/ samāsvādayamānas tu atiyogena siddhyati//

Here is the Sanskrit passage from the Kṛṣṇayamāri tantra (17.8, p.123):bhāvayed yogam anuyogaṃ dvitīyakam/ atiyogam tṛtīyam tu mahāyogam caturthakam//

The Tibetan is in P.103, 16-4-1ff: dang por sgom pa rnal ‘byor te/ gnyis pa rjes kyi rnal ‘byor yin/ gsum pa shin tu rnal ‘byor te/ bzhi pa rnal ‘byor chen po’o/