25 December 2020

Modern Interpretations of the Khandhas: Intro and Rūpa

Thanks to a benefactor I was able to purchase a copy of Tilmann Vetter's The Khandha Passages, which came out in 2000. The book is largely what it says, i.e. a collection of Pāli passages related to the khandhas in a single volume. It came out in the same year as Sue Hamilton's book Early Buddhism: A New Approach (2000). In this essay (in two parts), I aim to summarise and critique aspects of Vetter's overview of the khandhas (2000: 19-82) and to compare and contrast with Hamilton's overview (2000: 72-84). The aim is to create a synthesis based on these two indepth, nonsectarian studies. In part one I'll introduce the topic and cover rūpakkandha.

Vetter is aware of (cites) Hamilton's earlier book on the khandhas, Identity and Experience (1996) but he makes little use of it and (surprisingly) says that there has been no comprehensive study of Pāḷi khandha passages. Identity and Experience is an edited version of Hamilton's doctoral dissertation and looked at every occurrence of the khandha in the four Nikāyas. Vetter includes the uses of khandha in the Vinaya and perhaps this is what he means by "comprehensive". Pāli as a language and the ideas conveyed in it both exist in layers. A comprehensive account would include comments on the diachronic as well as synchronic uses of the word. Hamilton, with some justification, treats the Nikāyas as a single chronological unit; Vetter, with less justification, treats the Vinaya as part of the same chronological unit. While there is minimal evidence for stratification in the Nikāyas, there is some. Unfortunately, Vetter proposes a chronological development of the relevant terminology but does not attempt to tie this to the chronological development of the Nikāyas. 

In some places the five khandhas are only four (Vetter 2000: 15, n. 12), i.e. rūpa, vedanā, saññā, and saṅkhāra (e.g. AN. 4.16). Also, the five khandhas appear to have been unknown to the original compilers of the Dīghanikāya or Aṅguttaranikāya. Vetter mentions these two facts in passing and moves on, since they are incidental to his project. However, such discrepancies really ought to catch our attention and motivate us to explain them. 

In Pāli, there are two terms for the five items collectively: khandha and upādānakkhandha. Sometimes these two words are synonymous, though sometimes they are opposed (e.g. Sn 22.48). However, upādānakkhandha is by far the most common term. Vetter takes this to mean that the collection of five khandhas came first (sans-label), then were labelled upādānakkhandha, which was later abbreviated with khandha (2000: 19). This is of course possible but the rationale for it is weak. What if the kandha/upādānkkhandha is a sectarian difference? Or geographical? 

Vetter overlooks or ignores Richard Gombrich's observations about the word khandha when used in words such as aggikhandha and dukkhakhanda (1996: 67-9). Gombrich placed the term within the context of an extended cognitive metaphor: EXPERIENCE IS FIRE (Cf. Lakoff and Johnson). Buddhist texts make broad use of this metaphor to characterise sensory experience and the khandhas are another way of talking about experience. And in the context of fire, upādāna takes the meaning "fuel". The khandas, in this view, are the fuel that supports the fire that is experience. A key to this is the Ādittipariyāya (SN 35:28), in which the Buddha is made to say "everything is on fire" (sabbaṃ ādittaṃ). And what is "everything", it is the sense faculties and sense objects (i.e. rūpa qua appearance rather than substance), i.e. sensory experience is on fire (compare this to the Sabba Sutta, SN 35.23 and the subsequent suttas). From Hamilton we know that "dukkha is not contingent to experience. Rather, one cannot have experience that is not dukkha" (2000: 68) and "dukkha is better understood as the fact of experience" (71). Note that Gombrich was Sue Hamilton's PhD supervisor and he was strongly influenced by her conclusions about the khandhas

Hamilton notes: 

"The term [khandha] is not one used by any of the other religious teachers of the day, and they are hardly explained in any coherent way anywhere in the Sutta Piṭaka: there is no text which gives a full and clear account of what is being referred to by the term khandha. (2000: 70. Emphasis added)

Nevertheless, khandhas are mentioned frequently in early Buddhist texts and form an essential part of later Buddhist metaphysics (especially in accounts of Prajñāpāramitā). I outlined the etymology and meaning of word khandha (Skt. skandha) in an essay in 2013: Pañca-skandha: Etymology and Dynamics. The etymology is obscure, but seems to relate to the idea of "branch" and, in particular, the way the torso branches into limbs and the arm branches into five fingers. Hamilton (1996) notes that, in Vedic, skandha means "trunk" as in the trunk of a tree, but she provides no references for this. I have also proposed that the word prapañca relates to this branching of the arm into five fingers. I often refer to pañcakkandha as "the five branches of experience".

Following his assumed pattern of historical development, Vetter opts to define the individual khandhas first (20-73) then the terms upādānakkhandha and khandha (73-82). He notes that outside of this context several of the terms have other uses and are defined accordingly. This is common in Buddhist literature. Sometimes context is all important in deciding how a word is being used. 

Vetter relies on the Khajjanīya Sutta (SN 22.79) for his definitions because it is "the only passage in Vinaya- and Sutta-piṭaka where an attempt has been made to 'define' all five items" (19). It will soon be apparent why this source is problematic, but in advance Vetter notes that the sutta is "late" according to his understanding of the development of the terminology (it uses the term upādānakkhandha). It's not clear what "late" means here (since Vetter has already asserted that the use of upādānakkhandha predates the use of khandha) and, as I noted, this is not connected with any of the descriptions of the chronological stratification of the suttas. This brings to mind Jan Nattier's quip about textual stratification:

"If I like it, it's early; if I specialise in it, it's very early; if I don't like it at all, but it's in my text, it's an interpolation." (2002: 49)

Vetter gives some obvious caveats—the text could be late, it is stylistically different, and definitions are inconsistent (19). Still, when it comes down to it, Vetter starts with this text and is much less critical than I would like him to be.


Vetter's definition of rūpa is the shortest of his book, which is focussed more on saṅkhāra and viññāna. I think this is a little unfortunate because some important nuances are lost in the process. Vetter (20) notes that rūpa is often defined with respect to the verb ruppati "harm, suffer". This definition is confused, precisely because of the passage found in the Khajjanīya Sutta which relates the five nouns that make up the khandhas to five verbs ostensibly from the same root. Abbreviating, we get:

Ruppatīti kho, bhikkhave, tasmā ‘rūpa’nti vuccati... Vedayatīti kho, bhikkhave, tasmā ‘vedanā’ti vuccati... Sañjānātīti kho, bhikkhave, tasmā ‘saññā’ti vuccati... Saṅkhatam abhisaṅkharontīti kho, bhikkhave, tasmā ‘saṅkhārā’ti vuccati...Vijānātīti kho, bhikkhave, tasmā ‘viññāṇa’nti vuccati. (SN III.86-7)
It harms, monks, therefore it is called "form", it feels, therefore it is called "feeling", it perceives, therefore it is called "perception", it constructs the constructed, therefore it is called "construct", it cognizes, therefore it is called "cognition."

Even with only my English translation anyone can see that something is wrong here. There is a pattern and the rūpa passage does not conform to it. The reason is that someone has misread the text. It is simply incorrect to derive rūpa from √rup "to break, to harm", which has a indicative form ruppati (Skt rupyati; causative ropayati). Compare a similar passage in the Large Prajñāpāramitā Sūtra:

tathā hi śāradvatīputra yā rūpaśunyatā na sā rūpayati | yā vedanāśunyatā na sā vedayati | yā saṃjñāśunyatā na sā saṃjānāti | yā saṃskāraśunyatā na sābhisaṃskaroti | yā vijñānaśunyatā na sā vijānāti | (Karashima et al. 2016, 21r–v).

“Therefore Śāriputra, what is empty of form does not form; what is empty of feeling does not feel; what is empty of perception does not perceive; what is empty of willing does not will; and what is empty of cognition does not cognise.”

Here, instead of a verb from √rup we find the denominative of rūpa, i.e. rūpayati "appears, forms". The expected Pāli form—rūpayati or rūpeti—does not occur in the Nikāyas. Examples of denominatives from Warder's Introduction to Pali (316-7) include sukheti "he is happy" (from sukha "happy"), tīreti "he accomplishes"  (from tīra "river bank, shore"), and udāneti "he speaks joyously" (from udāna "a joyous utterance") This tautological definition of rūpa does fit the pattern noted above: "it appears, therefore it is called appearance."

It's not just that rūpa is unrelated to ruppati but that the definition of rūpa in the Khajjanīya Sutta makes no sense. Vetter calls the Pāḷi etymology "doubtful" (20) but it's just wrong and it is not clear why he prevaricates on such an obvious mistake. However, this text does a lot of heavy lifting in his definition of rūpa. In fact, Vetter's understanding of rūpa as "body" appears to be based mainly on the Khajjanīya Sutta where, having mistakenly equated rūpa with ruppati, "harming", the text continues to ask "by what is it harmed? (kena ruppati). The answer is: 

sītenapi ruppati, uṇhenapi ruppati, jighacchāyapi ruppati, pipāsāyapi ruppati, ḍaṃsamaka-savātātapa-sarīsapa-samphassenapi ruppati. (SN III.86)

It is harmed by cold. It is harmed by heat. It is harmed by hunger. It is harmed by thirst. It is harmed by contact with flies, mosquitoes, wind, sun, and snakes.

Note that Bodhi (2000: 915) has opted to translate ruppati here as "deformed". One can see that he is trying to pun on rūpa as "form" but this falls flat in my view. Clearly this text suggests that rūpa means "body" and Vetter appears to rely on this text as his main support. But we've just shown that the author of this text was confused. This text is wrong about the definition of rūpa as ruppati (rather the rūpayati). The issue of what is harmed is not relevant given that "harm" is the wrong word. The question is what appears? In which case, why should we take this text seriously, let alone as definitional? Even with scripture, one has to evaluate the quality of sources. Not all suttas were created equal. 

Worse, Pāli has a number of words for body and Vetter says:

"[rūpa] was probably preferred to other words for body such as kāya, deha or sarīra, because it showed the body as attract to the eye, and as such causing attachment." (2000: 20. Emphasis added)

The etymological confusion notwithstanding, this is a poor explanation. This is a central Buddhist doctrine. Vetter has to guess at the meaning here precisely because this important and oft-used term is not defined by early Buddhists themselves and is not drawn from other religious teachings (unlike saññā, saṅkhāra, and vijñāna). But I think here he has guessed wrong. 

Rūpa doesn't primarily mean "body" and, as Vetter notes, there are a number of words that are consistently used in this sense (kāya, deha, sarīra) although curiously "body" is a derived sense the case of kāya "collection" and deha means "moulded". The etymology of sarīra is not clear. Generally, rūpa means "outward appearance" and early Buddhists used words for body when they meant body. For example, the sense of touch has body (kāya) as its organ and tangibles (P. phoṭṭhabbā; Skt. spraṣṭavya) are its objects.

In sensory terms, rūpa is to the eye what sound is to the ear, i.e. it is not the matter or substance of an object but the visual stimulus that reaches the eye considered independently of the object. Even though we may say "I hear a conch", we know this is shorthand for "I hear the sound of a conch". We don't hear the object, we hear the sound it gives off. Similarly rūpa as sense object is the visual impression that the object gives off, not the object itself. This aspect of Buddhist epistemology is thoroughly confused across the board of Buddhist exegesis. 

Of course, the context here is different. It is possible for rūpa to mean something else, but it strikes me as unlikely that it means "body" in any context. Vetter's argument is based on misreadings and guesses which does not inspire confidence. 

Sue Hamilton also translates rūpa as "body"; however, she adds: "It is not the matter of the body qua matter that is relevant, but that one's body is the physical locus of one's experience" (2000: 72) and "A central feature of the body is that it is the locus of the senses, which further emphasises that what is being referred to here is the living functioning body and not just its substance" (73). 

There is a question here regarding the nature of experience as understood by moderns and ancients. See for example my 2012 essay, The 'Mind as Container' Metaphor. It seems entirely natural in the modern Anglosphere to think of experience as occurring in a "theatre", as being internal to us and contained in our minds. This also a cognitive metaphor: THE MIND IS A CONTAINER. What I argue in that essay is that Pāli texts lack this metaphor. I have not seen any evidence that they had the metaphor of the body as container either, certainly not of experience. Here I think Hamilton's approach is problematic. It seems so obvious and intuitive to us what kind of thing that sensory experience is. But our views are socially conditioned and not as natural as we think. I'm not convinced by the "locus of experience" idea. But I also think it's an open question: my blog posts are fairly thorough but not peer-reviewed. I aim for them to be topics of serious discussion, not definitive statements.  

Part of the reason for taking rūpa to concretely mean "body" is that it is said to be composed of body parts. In Buddhist meditations, such as found in the Satipaṭṭḥāna Sutta (MN 10) or the Kāyagatāsati Sutta (MN 119), one contemplates rūpa in the body by imaginatively noting all the constituent parts including things like viscera and bones (one cannot directly experience one's viscera or bones, for example, so the practice must be imaginative). But note here that it is kāya (literally "the collection") that is treated thus, not rūpa. We've already seen that Buddhist definitions of jargon terms are highly context dependent. So the idea that a definition or understanding  of kāya is directly applicable without any caveats to a definition of rūpa in a whole different context is dubious at best. Both authors have made a rather devious manoeuver here and have hidden it with hand waving. 

Vetter and Hamilton (at much greater length) also observe that rūpa is defined as the four great elements, i.e. catumahābhūta: paṭhavī, apo, tejo, vāyu or  earth, water, fire, and wind. Here it is definitely rūpa in the context of the khandhas that is defined and we are back on track. Still, as Noa Ronkin points out: 

"[the elements] are not meant to give an account of matter as constitutive of external, mind-independent reality" (2005: 56). 

Rather the words are used metaphorically for aspects of experience. Earth represents solidity; water, fluidity and cohesion; fire, heat; and wind, movement. The temptation to read the mahābhūtas as synonymous with the Greek four element theory is strong, but early Buddhists were not theorising about metaphysics as the Greeks were. Strangely, given the thesis of her book, Hamilton notes the metaphorical use but seems to stop short of acknowledging the implications of it - i.e. that kāya does not appear to mean "body" in this context. It may at best mean the experience of body or embodiment. 

Generally speaking, then, rūpa is not an ontological term; it is an epistemic term. It refers to the visual appearance of objects, not to objects per se, nor to the matter they are made of (on which subject the suttas are silent). And most dictionaries seem to take this view also. On the other hand, some Pāli suttas clearly do use rūpa as a ontological term and consider it to mean "body", although this is associated with a mistaken folk etymology of ruppati. And this definition has crept into dictionaries as well. 

Vetter's definition of rūpa as "body" is an inauspicious beginning because it means that he leans into the idea of rūpa as "substance" rather than rūpa as "appearance". In fact he specifically rejects the definition as "visible form" (20). Where the word rūpa is equated with "body" is precisely where it is confused with Pāli ruppati in exegesis. And this mistake is not repeated in some other occurrences outside the Theravāda tradition. 


It's worth repeating that the khandhas are never clearly defined in the early Buddhist texts. One of the partial definitions is either based on a misreading (rupyati "harm" for the denominative rūpayati "appear") or it is a tautology "appearance is that which appears". Either way this is not helpful. That scholars conflate definitions and overlook context when it is convenient is not helpful either. Both Vetter and Hamilton make the issue seem more clear than it is. The one thing it seems rūpa cannot be is the body as substance. Whether Hamilton's reading of body as "locus of experience" is what was intended is moot, but I have doubts how this would fit the context since the relevant cognitive metaphor is missing from Pāli as far as I can tell.

Another problem I have with rūpa qua body, is how anyone can think that body makes a set with vedanā, saññā, saṅkhārā, and viññāna. The body and four terms concerned with mental functioning. Some Buddhists make an analogy with the term nāma-rūpa "name and appearance". However, as I pointed out in 2011 in my essay, Nāmarūpa, this term too is vague and can't really serve as an anchor for other definitions. The Pāli tradition is conflicted as to what constitutes nāmarūpa, a term borrowed from late Vedic literature. Furthermore, Buddhists more characteristically used a threefold division of the person into body, speech, and mind (kāya, vāca, citta) and speech does not easily fit into the khandha analysis (note the threefold division does not occur in Vedic literature, but it does occur in pre-Buddhist Zoroastrian literature). In addition, vedanā, saññā, saṅkhārā, and viññāna are all epistemic terms (in Buddhism) and if there is any pattern then rūpa should also be an epistemic term, i.e. appearance rather than substanceThe set of khandhas cannot possibly accommodate "all phenomena" even if many Buddhists consider this plausible based on a very old tradition. Even if we follow Hamilton and reduce the scope of early Buddhist interest to just experience (as opposed to reality), there is still an explanatory gap between rūpa and the other four khandā (and missing vāca).

How do we know we have a body? Because we feel it and see it. Note that when these two channels of information get out of sync we have out-of-body experiences as Thomas Metzinger described in The Ego Tunnel. Feeling and seeing are sensory experiences. So our awareness of our body, as with all other instances of intentional awareness is subjective by definition. Moreover, one of the most common altered states that occurs early in meditation is that our sense of embodiment is affected. The body may feel very large, for example, or very light. Or we may lose our sense of being located in space. As we go deeper we may lose our sense of being extended or bounded in space. 

Later Buddhist traditions notwithstanding, I don't think "the body" can be what is meant by rūpakkhandha. I keep coming back to the basic idea that rūpa is to the eye what sound is to the ear. Clearly rūpakkhandha is some kind of abstraction or metaphor (perhaps a metonym) based on this idea. Lacking a clear definition, Buddhist traditions drifted into various interpretations based on the needs of sectarian doctrinal entailments. Making sense of rūpakkhanda may well be impossible given what we have to work with, at least in terms of first principles. The idea that it must make sense is plausible enough and I'm sympathetic to it, but it should be not become a defence of procrustean efforts to force the definition to fit. Similarly, I'm sympathetic with attempts to reverse engineer what kind of thing rūpakkhanda might be based on the hermeneutics of later, but still ancient, exegetes. But we cannot discount the possibility that they were just as much in the dark as we are. And we have to take into account that in the ancient world exegetes were firmly embedded in sectarian doctrinal systems. 

This level of confusion of terminology, definition, and application so close to the centre of Buddhist orthodoxy is fascinating. And it makes me wonder why I seem to be almost alone in being fascinated by it. I'm reading two professional academics on the subject (writing 20 years ago!), both of whom were trying to smooth over the inconsistencies and present what's in the suttas as a coherent discourse. But that is the job of theologians or religieux not the job of academics! Why are they not studying the content of the texts critically? This seems very weird to me. 

Next up the other khandas, but I haven't even started yet so don't hold your breath. 


Attwood, J. 2018. "Defining Vedanā: Through the Looking Glass." Contemporary Buddhism, 18(3): 31-46. https://doi.org/10.1080/14639947.2018.1450959
Bodhi. 2000. The Connected Discourses of the Buddha. Wisdom. 
Gombrich, Richard F. 1996. How Buddhism Began: The Conditioned Genesis of the Early Teachings. London, Athlone. Reprint New Delhi: Munshiram Manoharlal, 2007. 
Hamilton, Sue. 1996. Identity and experience: the constitution of the human being according to early Buddhism. London: Luzac Oriental.
Hamilton, Sue. 2000. Early Buddhism: A New Approach. London: Routledge.
Lakoff, G. & Johnson, M. (2003). Metaphors We Live By. New Ed. University of Chicago Press.
Ronkin, Noa. 2005. Early Buddhist Metaphysics the Making of a Philosophical Tradition. Routledge.
Vetter, Tilmann. 2000. The Khandha Passages in the Vinayapiṭaka and the Four Main Nikāyas. Wien Verlag der Österreichischen Akademie der Wissenschaften.

20 November 2020

That Chariot

The simile of the chariot is a well-known Buddhist explanation. In ancient India, explanations often take the form of analogies and, in this case, the analogy makes use of a simile: X is explained by pointing out that X is like Y; where Y is assumed to be understood. These analogies often drew on apparent parallels in nature or agriculture, such as the rivers, plants, animals (and animal husbandry), seasons, ploughing. But occasionally the analogy would draw on technology and the simile of the chariot is one of these.

The locus classicus of the simile is the Vajirā Sutta (SN 5:10). The set up is that Māra is trying to torment a meditating bhikkhunī, Vajirā, with questions about the origin and nature of "this being" (ayaṃ satto). His taunt is presented as a verse
Who made this being? Where is the maker?
Where did this being arise, where will it cease?
Of course, these might seem like innocuous questions, but in a worldview that incorporates the idea that no phenomenon can persist, it becomes more difficult. Vajirā realises it is Māra and replies in verse:
Māra, what makes you harp on about your theory of 'a being'?
No being is found in this pure collection of constructs.

Just as the combination of parts is called 'a chariot';
So, when khandhas exist, conventionally we say "there is a being."

For only suffering is produced; suffering persists, and ceases.
None other than suffering is produced, none other than suffering ceases.
When I commented on the simile of the chariot in 2009, I made two points. Firstly, when we read this text through reductionist Buddhist paradigms, we have to interpret this as saying that the whole is simply the sum of its parts; not less, but not more, either. Where there is the right assemblage of parts (aṅgasambhārā) there is a chariot. But the chariot is not in the parts individually or collectively. We say there is a chariot when the parts are assembled, but this is a mere conception. Similarly, while the khandhas exist (khandhesu santesu) we say there is "a being" but it is no more than convention (sammuti). Although the sutta does not insist on it, our preference for methodological reductionism forces Buddhists to conclude that, in reality, there is no chariot, and therefore there are no beings, eitherThe idea that beings don't exist either is a grammatical contradiction: be-ings are quint-essential examples of exist-ence).

This is a banal point; of course, when you disassemble a chariot into its components you destroy the chariot, there is no longer a chariot: the method determines the outcome (by destroying the object of study). Note that the text tells us that the arrangement (sambhārā) is important. But knowing this doesn't really move me. My thoughts go more like this:
I'm driving around on my chariot and a Buddhist monk flags me down. He says "You know that your chariot doesn't exist, right?"

I'm confused. "So what is this I'm driving around on?" I ask.

The monk answers, "Ah yes, but if you disassemble the chariot, there is no chariot."

To which I reply, "Venerable One, do not disassemble my chariot. I have places to go and people to see."
A philosopher might naturally segue here into a discussion of the Ship of Theseus, which I also wrote about in 2017 (Ship of Theseus).
The monk says, "Don't worry I won't touch your chariot, but time will. All things must pass."

And I say, "That's true, I can always make new parts and repair my chariot."

The monk smiles, a knowing smile, and says, "Sure, but will it be the same chariot?"

Being a pragmatist, I reply "As long as I can ride around on it, I don't care whether it is the same or not. Why would I?"
My interpretations of the Ship of Theseus problem is that the traditional focus is on the wrong level. This is partly because in the classical problem the ship is now a monument, rather than a ship. Some of the variations on the problem do involve the ship as a ship but even they focus on something called "the identity" of the ship. My focus is not on identity because identity is subjective. Identifying is something humans do, not an intrinsic property of objects. The Ship of Theseus does not identify itself as the Ship of Theseus. Still, it is a ship, despite having all the parts replaced, and that seems to me to be a more significant thing to notice. It probably gets left out of the presentation of the problem because we cannot argue about it. No one argues that having replaced all the parts with identical parts the ship is now a chariot. If you replace all the parts with identical (or nearly identical) replacement parts and the integrity of the whole is preserved, the ship is still a ship, because the arrangement of the parts is stable over some period of time. No one is suggesting that the time period is infinity or that its stability is independent of time. Some people do argue that for God, but we're talking about ordinary objects like chariots and people. And it is obvious, even in the European intellectual tradition, that everything changes

Identity is not vested in the parts or the whole of an object, but in the minds of observers. We may say that humans tend to identify (and identify with) persistent structures, even if structures ultimately lose their integrity. I am Jayarava, I was ordained in 2005, in the intervening 15 years I have changed a lot, but there is a complex of persistent structures that are identifiable. Old friends that I have not seen for some years still recognise me, despite the addition of a full beard and receding hairline. Even though they might have known me as Michael in the past, old friends and family have all adjusted to the new name. Though when my colleague Jayarāja lived in Cambridge we were occasionally confused. 

Orthodox Buddhists argue, along with many modern physical scientists, that the arrangement is not real. But this is an ideological point, not a logical point. Any useful definition of "real" has to include the arrangement itself or what I call the structure as opposed to the substance. Without structure there is nothing but undifferentiated substance, nothing even recognisable as part of a chariot, let alone minds capable of recognising. No world, no chariot, no beings. Some claim that this is what reality is like, but I think they are confusing the cessation of experience with the nature of reality - a philosophical odyssey that never arrives at Ithaca. The arrangement (of parts) is real. It has to be real or "real" has no meaning. 

The second point I made in 2009, which is lost in almost every use of this text, is that it says: "only suffering is produced, persists, and ceases" (dukkhameva hi sambhoti, dukkhaṃ tiṭṭhati veti ca) and, in fact, "Nothing other than suffering is produced, nothing other than suffering ceases. (Nāññatra dukkhā sambhoti, nāññaṃ dukkhā nirujjhati). 

If you take this to be a metaphysical statement, that is if you take it literally and think that arising is a description of how everything works, then you are left with a conundrum. Why would the text say that dukkha is the only thing that arises and ceases and nothing else? After all, the text admits that there are times when khandhas exist (khandhesu santesu) I don't recall seeing this discussed when the chariot comes up.

We have learned from Sue Hamilton to take this kind of material as epistemic. Here, dukkha is synonymous with "experience". And the khandhas are the factors, or even the apparatus, of experience. The result is our world (loka), not to be confused with the world. As an epistemic problem, we take this to mean that arising and ceasing is only concerned with the arising and ceasing of sensory experiences (including the mind sense). Similarly, when the Sabba Sutta (SN 35.23) says that everything (sabbaṃ) is the six sense organs and six sense objects and that any other definition of everything is wrong, it is not asserting an Idealist ontology. It is delimiting a sphere of interest: only experience is of interest. And this is because in North India, around the time of the second urbanisation, religieux discovered bringing sensory experience to a halt through what we now call "meditation".

With respect to this last point, we need to make a distinction between objects and sense objects. For example, a conch shell is an object. It has a distinctive structure and substance: a tapering cone twisted into a helix, made largely of calcium carbonate. And it has an appearance (rūpa) which we now relate to the light reflected from it. In Buddhist thought the conch itself is not the object of our senses: the sense object is the appearance (rūpa) not the substance (dravya). This is similar to a distinction made by Immanuel Kant between the "thing-in-itself" (ding an sich) or noumenon and our experience or phenomena. But it goes back to Plato's "theory of ideas" and the simile of Plato's Cave: the idea that what we experience is like a shadow cast on the wall of a cave by real objects. In the long Platonic tradition, we can never get at the ding an sich. As I have noted previously, Bryan Magee argues, in his book on Schopenhauer, that reality must be utterly different from what we experience. In my essay Buddhism & The Limits of Transcendental Idealism, I pushed back against this view on the pragmatic basis that my experience can't be so different from reality or I would bump into things and get lost a lot more often (if only metaphorically). I would add now that when we meet someone who is delusional we very soon understand that they are "out of touch with reality" and they do bump into things and get lost (if only metaphorically). 

Conceptualising Objectivity and Subjectivity

We've seen that pudgalavāda was a view that emphasised the idea of a person being a persistent structure of khandhas (Early Buddhist Heterodoxy: Pudgalavāda). It seems to me, as it seems to any observer, that I am something more than an illusion or a concept, even when I admit to being less than a soul (ātman, jīva, satva, puruṣa). I appear to exist but this does not—contra Buddhist orthodoxy—mean that I exist in any absolute sense. Indeed, I become increasingly aware of my own mortality. The pudgala was emphatically not an ātman, rather it was a persistent assemblage of skandhas, i.e. a structure or system of skandhas strongly connected to each other and only weakly connected to others. Keep in mind also my argument that the pudgala was devised in order to solve a problem created by the application of dependent arising outside the domain of experience: it meant that karma could not function as required. For karma to be meaningful, for my actions to have consequences for me, there had to be long term continuity. But dependent arising denies any kind of continuity. Despite Buddhist rhetoric, continuity is not a concept we can do without when trying to make sense of our world. Buddhists themselves routinely employ this concept in moral instruction. 

As an alternative solution to the problem of action at a temporal distance, pudgalavāda is intrinsically interesting and Buddhism is much the poorer without it. Which is not to say I think pudgalavāda solved the problem, just that it was different and interesting and Buddhism is the less for the lack of plurality we now see. 

We now have ways of thinking about experience and continuity that can fill that gap without resorting to explanation by analogy. So let's start by delving into the subjective/objective distinction. Experience is ontologically subjective, but this does not mean I don't have experiences. Experience is epistemically real because we can have knowledge about experience and the processes that create experiences. This is a distinction I have drawn from the philosophy of John Searle (see a long series of essays written in 2016 beginning with Searle on Consciousness & Implications for Buddhism).

Experience is all subjective, but we can distinguish two broad classes of experience. In one channel of experience, we have information related to our physical senses and body which contribute to a representation of the physical world. In the other channel, we have information that is not related to objects but which mainly concerns our self and includes various forms of memory and conscious mental activity.

Around the age of four, we work out that other people have minds of their own. By this age, we can infer from interacting with them, that other minds are like our minds. If we fail to go through this developmental stage then we are at a severe disadvantage in dealing with the world. Social interactions themselves are experience but still even young children work out that the experience of social interaction leads to the conclusion that other minds exist. A few people go through this phase of development only to abandon the idea as adults, usually because of a superficial encounter with certain types of academic philosophy, but we don't have to take them seriously. The fact that we can interact with other minds via experience is perhaps the most wonderful thing about being alive. Not only that but this ability is not limited to humans. Animals also display behaviour consistent with being aware of other minds, and humans can also interact with the minds of animals. Each animal has its own characteristics and within each species, there are different personalities.

Comparing of notes about experience tells us that the first channel—our physical senses—is much the same for everyone, i.e. we all inhabit the same world. Not only do our senses confirm each other, but reports from others confirm our observations. There is a world of objects that are apparent to all observers. The same methods tell us that our internal experience is more or less unique to us, though it still has some broad similarities. The internal experience is still sense experience but without external objects that are accessible to others. Early Buddhists made this same distinction between physical (kāyika) and mental (cetasika) experience.

So within experience itself, there are experiences that relate to the external, objective channel and some that are related to the internal, subjective channel.

A Mind-Independent World

An important question is: If all we have is experience, how do we know that there is a mind-independent world? After all, we can infer that all experience is the result of our brains processing electrochemical signals produced by specialised cells when they interact with objects and respond to various physical characteristics: pressure, resistance, texture, temperature, acceleration, physical vibrations, and electromagnetic vibrations. We can imagine, for example, each of these types of information being simulated. Indeed, via so-called "virtual reality" we can crudely simulate a world which is not reality.

In other words, if we stipulate that all knowledge comes from sensory experience, how do we establish the existence of a mind-independent world? The first thing to admit is that we have to do so indirectly. We infer such a world. Indeed, we infer other minds as well. Some might argue that this is too flimsy a basis for knowledge, that what results is more like belief than knowing. 

The world that I infer from experience and comparing notes with others is a very reliable guide. The world I move around in appears to be quite stable. Indeed, having lived in my present residence for seven years, I can find my way around in the dark, muscle memory guides my feet. My trip to the supermarket is not based on guesswork or luck but on past experience. The physical world is not a mystery to me, it has regularities. Moreover, when things go wrong, when I bang into something, cannot find something, when I get lost, it is not because the world itself has changed, it is because of my mind.

The mind undoubtedly makes mistakes. The world does not. Objects don't move around at random. They follow patterns that we intuitively grasp, even without physics. My house is not No.15 one day and No.3 another, it is not in this street one day and that avenue on another. My city is not in East England today but in Yorkshire tomorrow. And so on. The basic facts of our world do seem to be facts

We know that naive Realism cannot be the case because of what we know about our own physiology, but the level of detail with which we can describe the world on different scales is beyond the ability of any one person to comprehend these days. Moreover, we can use inferred knowledge to rule out certain types of world. We do not live in a world with magic, for example. Matter and energy are bound inexorably to follow certain patterns. Based on what we can infer to be the case, from possibilities we can eliminate by counter-example, there can be no doubt that there is a mind-independent world. 

Could that world be a mental projection or a simulation? Sure, it could be, but there is no reason to believe that it is. The physicality of the world is really quite compelling. An Idealist gains no explanatory leverage over the world of experience by adopting that position. Rather, they are left trying to explain how we people can have coordinated experiences. Imagine three people singing in harmony. If the sounds they make are not physically independent of their minds, then one has to propose some immaterial connection between them that does exactly the same job as physical sound. Since we can infer the existence of sound as pressures waves in air from all kinds of angles, this makes the idea of sound as a mere idea rather weak. Sound as a physical manifestation of vibrating physical vocal cords resonating in body cavities and then being converted into nerve impulse by the ear is not a simple explanation. It's complex. But it is quite a lot simpler and more complete than any Idealist explanation of how three people can sing in harmony.

But it goes a lot deeper than just explaining this one instance. We have generalised explanations of sound that apply across the board. And these are informed by other explanations of related phenomena like other kinds of waves and other kinds of sensory experience. Generalised explanations of waves apply not only to sound waves but also to light. Phenomena such as constructive and destructive interference apply to waves in water, in air, in earth, in light, and in spacetime. Idealists have never developed such depth, they merely claim that the depth is not physical but mental.

The same thing applies for the simulation argument. If we are living in a simulation then the simulation is no less complex than reality as we experience it: no fewer orders of magnitude (roughly 100 in mass, length, and energy), no fewer elements (be it atoms, or smaller building blocks). The machinery to support such a complex simulation would necessarily be more complex and older than the universe it simulates. Such a thing is clearly possible, but we don't gain anything by adopting this explanation. The universe makes less sense, rather than more. Explanations that leave us with less knowledge and clarity defeat the purpose of explanations.

In the billions of experiments done over 450 years by millions of scientists, let alone the trillions of observations by billions of human beings over hundreds of millennia, none has given us reason to believe we live in a simulation. If we do live in a simulation then it is perfect to the limits of our ability to test it across 100 orders of magnitude of mass, length, and energy. In which case how is that different from a real world?

By far the simplest explanation is that we live in a universe of objects and process, or matter and energy.  And we experience this world indirectly through our senses and virtual models our brain creates. We persist with other avenues of thinking about the world for the joy of using our imaginations (such as creating fiction) or the pleasure that some of us take in arguing or because of a predisposition to believe that everything happens for a reason - also known as the teleological fallacy. Arguments for living in a simulation are topologically the same as arguments for a Creator God or intelligent design. They are attempts to explain - not what we are, or what the world it, but why we are here and why there is a universe in terms of assigning responsibility (ultimately, therefore, the teleological fallacy is a moral argument).


I started off talking about reductionism in Buddhism and the consequences of ideological reductionism, i.e. paradox and internal contradictions. Reductionism per se fails because we live in a constructed universe. There is substance but there is also structure. Methodological reductionism destroys structure as a starting point and thus furnishes us with no knowledge of structure. And thus reductionist explanations fall well short of completeness. Which is not to say that we do no need reductionist explanations. The substances from which we or the world are made and the question of ultimate substance are intrinsically interesting and learning about substance and lower-level dynamics has been useful. It's just that this approach is incomplete. Human beings are made of atoms, for example, but we easily cannot get from atoms to sociology, and sociology is also intrinsically interesting. Arguably we cannot get from atoms to sociology even in principle.

As powerful as the chariot simile is in Buddhist rhetoric, if we adopt an exclusively reductionist approach it can be an incredibly unhelpful way of thinking about being and beings. Had we kept pudgalavāda alive we might eventually have come to terms with structure, but we didn't and we haven't. All modern forms of Buddhism are relentlessly reductive. Modern Buddhists are obsessed with non-existence: with anātman and śūnyatā routinely understood in metaphysical terms. Buddhists frequently claim to know, or at least to teach (often in the absence of personal experience), the "true nature of reality". Not just "reality" or the "nature of reality", but the "true nature of reality". However, this rhetoric almost always involves an eliminativist twist. 

Nāgārajuna, for example, asserts not only that karma and rebirth do not ultimately exist, but that agents and patients do not ultimately exist. Karma is merely a merciful lie that enlightened beings use to trick the unenlightened into enlightenment. I've never been entirely comfortable with this metaphysics or the rhetoric that accompanies it. It sounds like nonsense and the arguments that emerge from it are often mean spirited.

My complaint about Madhyamaka is that it appears to be confused about the distinction between epistemology and metaphysics. The state of contentless awareness is taken to be ultimate reality and the result is a truly awful philosophy full of contradictions and hand waving. I would complain less except that one can easily find ancient Buddhists who did not make this mistake, who understood the cessation of sensory experience to be a source of knowledge about the nature of experience rather than a doorway to a transcendental reality.

Methodological reductionism is appropriate in some contexts. In meditation we break down experience and focus on smaller and smaller parts of it, we see it in finer and finer detail. We do not always see, but we can infer from ancient Buddhist texts, that the aim was to bring experience to a halt and dwell in contentless awareness. The surprise is that in the absence of experience there is still something. In the absence of experience there is still awareness, and when experience commences there is a memory of contentless awareness and there are downstream consequences. In my view, this was the Buddha's insight (and not only his, sorry).

The end of sense experience is not the end of life and, for those who undergo cessation, life beyond death may well seem to make sense. Indeed, many who have undergone the cessation of experience express conviction about extra-corporeal and post-mortem existence, although this is difficult to reconcile with a reality in which nothing, ultimately, exists, or things exist in the kind of reductionist, merely conventional sense that many people draw from the Vajirā Sutta.

What's left out of almost every account of the chariot is the last verse: that only dukkha arises and nothing other than dukkha. If we take all the verses of the poem it forces us to think again about our metaphysical commitments. This may be why it is simply left out of discussions, starting with the restatement of the chariot simile in the Milindapañha.

But this all reflects the nature of experience. It does not reflect the nature of reality except that, in reality, human beings are capable of undergoing the cessation of experience without any consequence loss of basic awareness. Or, in other words, human beings are capable of contentless, non-intentional, awareness. But this awareness is pure subjectivity unrelated to the objective world. One cuts off from both channels of sensory information. To use this purely subjective state to make inferences about reality is misguided and it leads to bad philosophy (such as Madhyamaka).

Worse than bad philosophy, the whole mess is all too often framed in magical thinking. The Buddhist trikāya is not exactly like the Christian Trinity of Father, Son, and Holy Spirit, but it might as well be. The Buddha is all too often a messianic figure, a saviour, a god (in practice and deed if not in name). Buddhism is a religion like any other religion, promising salvation from whatever it is that we desire to be saved from. In East and South Asia, Buddhists want to be saved from chaos and oppression, in the West we want to be saved from ourselves.

We can reliably infer the existence of a real world of matter and energy, existing in layers of increasingly complex structure. The nature of reality involves both substance and structure. Illusions, like identity, are products of human minds. Buddhism offers us no good reason to doubt this account of reality. But it does offer us a practical approach to attaining contentless awareness and a vision of being honest, kind, and generous in the service of making the world a better place. Indeed, it tells us, if we only listen, that pursuing contentless awareness leaves us better placed to be honest, kind, and generous; and attaining it more or less guarantees a high level of honesty, kindness, and generosity. But it also makes the methodological point that taking on honesty, kindness, and generosity as practices contributes to both personal and societal well-being and to progress in pursuing contentless awareness. 

However, I have to also say that, in my experience, people who attain contentless awareness don't like it when I question their personal philosophy; they often come to definite conclusions about the metaphysical significance of their attainment, and they are often unwilling to contemplate other conclusions. For those who undergo cessation, the state has a hyperreal quality to it and subsequently can become the sole authority that that person will acknowledge. Cessation tends to cement a view of reality, usually based on prior indoctrination, that is impervious to argumentation. For Gary Weber, for example, the Advaita-Vedanta spiel has become an unquestioned and unquestionable dogma. For Buddhists, the spiel is different but equally dogmatic. I see the magical thinking that is so prevalent amongst religieux as a barrier to social and personal progress. It seems to be ironic that anyone mired in magical thinking should insist that they have a monopoly on reality.

The other catch is that achieving cessation of sense experience and contentless awareness is not something everyone can do. The people who can achieve contentless awareness are a bit like concert pianists who have to be born with long fingers, have early opportunities to learn and practice, have a fascination with the process almost to the exclusion of other interests, and the determination to persevere with it through adversity. People who lack any one of the factors may still become good musicians but they won't make it as concert soloists. Still, as a community, we can create the conditions to help realise the potential of individuals. And we all seem to benefit from the pursuit as well as from the presence of people for whom honesty, kindness, and generosity are the default mode. 

It seems that the techniques we use to pursue contentless awareness and the ability of that state to transform our minds are relatively independent of doctrine or belief. Where I think Buddhist views about reality hamper us is in communicating about the practicalities and aims of what we offer. And where we hold unorthodox, religious views about reality we put very strong limits on who will listen to us. My epistemic hermeneutic is one way around some of these difficulties. The simplest thing would be for Buddhists to substitute the word "experience" where we would normally use "reality". Information about contentless awareness is already escaping from our context and becoming secularised. It is going the same way as "mindfulness". 


09 October 2020

The Extended Heart Sutra: Overview

This post is a first attempt to sum up my close reading of the extended Heart Sutra that spanned eight posts, over which I worked my way through the distinctive features of the different versions of the extended Heart Sutra, noting down and trying to explain differences, omissions, additions, and errors. One of the principle unanswered questions about the extended Heart Sutra is: what language was it composed in? Which raises methodological questions about how would we assess this. We also want to know when the extended text came into existence and who was responsible for it. 

I should stress that I have only consulted versions of the text in Sanskrit, Middle Chinese, and canonical Tibetan. At this point I have not consulted the secondary literature. As such, my conclusions are limited to philological points, except where my more broadly based historical research on the standard text clearly applies to the extended text as well. On the other hand, I'm not aware of any existing philological studies of the extended text, except for Silk's critical edition of the Tibetan variants. And Silk's work is a critical edition of the Tibetan, not a critical history. There are, of course, religious studies of the text in Tibetan by the Dalai Lama and others, but their methodology is very different from mine. Religious commentaries seek to justify beliefs (i.e. emotions about ideas). Donald Lopez's two books on the commentaries preserved in Tibetan by "Indian" authors are also on the long version, but Lopez takes the tradition on its own terms. The books offer a reading of the tradition, but little or no insight into the kinds of problems that interest me here (i.e. how such a tradition became established).

Of particular interest is the Manuscript kept at the Hasedera Temple, which was the foundation for both Müller's edition, the edition by Vaidya which is now widely available on the internet (e.g. DBSC), and Conze's critical edition. As far as I can tell, the Hasedera manuscript has not been digitised or published in any form since Müller (1884). Müller himself was working from hand copies made for him by Japanese Buddhists and everyone since then is working from Müller's edition (although this is not always acknowledged - by Conze, for example). I could not find any information about where the mss is now.

The extended Heart Sutra has passages added at the beginning and the end of the standard text. Pragmatically, what the extensions do is tell the story, in formulaic terms, of how Avalokiteśvara came to preach the Heart Sutra to Śāriputra and, almost as importantly, how it was received. Such extensions were presumably the result of internal pressure to make the Heart Sutra more authentic, since it lacked all of the features that are traditionally associated with authenticity. It is sometimes suggested that this was an Indian concern, but the Chinese had exactly the same concerns and were more concerned, because so many indigenous Buddhists texts were in circulation. There is little or no evidence that the Heart Sutra was even known in India.

The extensions infer that the narrator is Ānanda (the "me" in "thus by me heard" evaṃ mayā śrutam); tell us the occasion and place of the discourse, that the Buddha is present, and that the speaker speaks with the anubhāva or empowerment of the Buddha. At the end, the interlocutor praises the teaching, and the Buddha praises the speaker. And in Chinese the audience promise to faithfully practice the teaching. Without these features the Heart Sutra was not a sūtra. The fact that the Heart Sutra was considered to be a sūtra was down to the forging of a Sanskrit text and document, and the attribution of the Chinese to Xuanzang as a translation (See Attwood 2019, 2020).

I refer to the unknown composer of the original digest text (chāo jīng 抄經) we call Xīnjīng 心經 as the Author. Although we think the Author was Xuanzang, this is, strictly speaking, unproven and it is possible that it was one or more other people. I refer to the person who translated the text into Sanskrit as the Translator. And now I have to introduce the idea of the person who redacted the text to produce the extended recension of the text, whom I will call the Redactor

It is also useful to make a threefold distinction between work, text, and document. As I said in Heart Sutra: Work, Text, Document (2019):
In his recent article, Silk (2015:205-6) draws out a threefold distinction first made by Chaim Milikowsky. First we have the Work, which is the author's or editor's product. This may only exist conceptually and never have been committed to words. Or the author may have attempted to put it in words and be more or less satisfied that the result, but still consider this as inferior to their conception of the Work. A presentation of the Work in words is a Text. A single Work may generate multiple Texts; for example, one story that is told many times, but with minor differences each time. No single Text is the "original" in this case, because the Text is not the Work. Lastly a Document is some physical instantiation of a Text. Typically, in studying Buddhist manuscript cultures, we are faced with multiple Documents representing multiple Texts. This is certainly in the case of the Heart Sutra.
I think we have clear evidence of their being at least two Redactors: one responsible for Recension Two (T 252) and one for Recension One (all other texts) (I started this convention but I wish I'd used the numbers the other way around since T 252 looks like the first recension to me). Where necessary, I will refer to Redactor One and Redactor Two. Redactor One might have been Fǎyuè 法月, who is credited with the translation of T 252. Redactor Two might have been possibly Prajñā (Bānruò 般若) and his Chinese collaborator Lìyán 利言. At least one scholar has argued that Prajñā, in fact, composed the works for which he was credited with being the translator.

The headings I used were:
And I will continue with this structure.


I worked with eight sources: T 252, 253, 254, 255, 257; Conze's Sanskrit edition; and the two Tibetan recensions described by Silk in his 1994 critical edition. There is still some work to do on some of these texts.

As Silk (1994) pointed out, Conze's Sanskrit edition is chaotic and unreliable. None of the Sanskrit sources is in pristine condition and some of them are very badly corrupted. However, even though we need to have a better edition, my experience of studying British Library Manuscript EAP676/2/5 suggests the chances are that examining more Nepalese manuscripts would be fruitless. Describing EAP676/2/5 required 142 footnotes on errors, omissions, and additions for a 280 word text. Even the best of the Nepalese mss is significantly degraded by repeated copying without error checking.

I would like to collect copies of all the canonical Chinese versions but don't have access to the necessary resources. Also, we lack a clear understanding of the attributions and dates traditionally assigned to the Chinese translations. This is important, given the chequered history of the Heart Sutra, in which none of the attributions or dates turns out to be reliable. I would place this above a new Sanskrit edition in importance.

The Heart Sutra may be an interesting candidate for the application of computerised phylogenetic techniques for analysing relatedness, though I'm not sure how such techniques cope with multiple languages. And I don't have the knowledge required to do it. I do have a preliminary stemma diagram prepared manually based on my own work and preliminary comments about the Dunhuang manuscripts by Ben Nourse.

NB: R1 & R2 are reversed here because I made this diagram for an article I'm preparing that covers this territory for publication.

The node labelled "Hṛdaya with added negations" is a notional intermediary. It may not have existed as a separate text and is not extant. But it does seem that most of the extended texts don't have the extra negations and neither does the Xīnjīng. Similarly, I don't know that there was a prototype of the Sanskrit extended text distinct from some actual document that formed the template for the rest. But in terms of diagramming the history of the text it's useful to posit these notional versions of the texts.


The nidāna is one of the most important features that mark the authentic sūtra and that is missing from the standard Heart Sutra. The nidāna begins with the immortal words "Thus have I heard" (Skt evaṃ mayā śrutam; Ch. Rúshì wǒwén 如是我聞). The phrase is traditionally understood to indicate that the narrator is Elder Ānanda, even for Mahāyāna sūtras. The nidāna then continues in the usual fashion to name the place at which the discourse was delivered and who was present: "At one time the Bhagavan was staying... (Skt ekasmin samaye bhagavān... viharati sma ) together with a great congregation of bhikṣus and a great congregation of bodhisatvas (mahatā bhikṣusaṃghena sārdhaṃ mahatā ca bodhisattvasaṃghena). The term bodhisattvasaṃgha (púsà zhòng 菩薩眾) is rare but not diagnostic for my purposes. 

Although the nidāna is considered evidence of authenticity by early medieval Buddhists, we also know that many texts lacked a nidāna. For example the Pāli version of the Burden Sūtra (SN 22:22;) lacks a nidāna while the Chinese versions both have one: Samyuktāgama 73 (T 2.19a15-19b1); Ekottarāgama 25.4 (T 2.631c11-632a5). Greg Schopen has detailed how the Mūlasarvāstivāda Vinaya preserved in Chinese contains a set of guidelines for adding a nidāna to a sūtra where one is missing (Schopen 2004). This suggests that the practice was routine. The nidāna of the extended Heart Sutra shows variations that can only be deliberate redactions (as opposed to accidents).

There is considerable variation in the nidāna of the extended Heart Sutra documents:

T 253 T 257 T 252
如是我聞:一時佛在王舍城耆闍崛山中,與大比丘眾及菩薩眾俱。 如是我聞:一時,世尊在王舍城鷲峯山中,與大苾芻眾千二百五十人俱,并諸菩薩摩訶薩眾而共圍繞。 如是我聞:一時佛在王舍大城靈鷲山中,與大比丘眾滿百千人,菩薩摩訶薩七萬七千人俱,其名曰觀世音菩薩文殊師利菩薩彌勒菩薩等,以為上首。皆得三昧總持,住不思議解脫。
Thus have I heard. At one time, the Buddha (佛) was in Rājagṛha on Vulture Peak Mountain, along with a great congregation of bhikṣus together with a congregation of bodhisatvas. Thus have I heard. At one time, the Bhagavan (世尊) was in Rājagṛha on Vulture Peak Mountain, along with a great congregation of 1250 bhikṣus together with many bodhisatva mahāsatvas and together they circumambulated. Thus I have heard: At one time, the Buddha was in Rājagṛha on the mountain of Gṛdhrakūṭa, together with a great bhikṣusaṃgha of 100,000 and 77,000 bodhisatva mahāsatvas in all, those named Avalokiteśvara Bodhisatva, Mañjuśrī Bodhisatva, Maitreya Bodhisatva, were the leaders.* All had attained the samādhi of always remembering, and abided in inconceivable liberation.

*Possibly Maitreya was "at the head" (yǐ wéi shàng shǒu 以為上首), cf. Conze's translation of Pañcaviṃśati: "and Maitreya the Bodhisattva, the great being, at the head of many hundred thousands of niyutas of kotis of Bodhisatvas." (1975: 38).

What stands out is that, despite being also set in Rājagṛha and, despite being constrained by the conventions of Buddhist sūtra composition, T 252 is considerably different from the other versions of the extended sūtra (whether in Sanskrit, Chinese, or Tibetan) and continues to diverge. Any introduction that was added was bound to include certain phrases and references to the Buddha (by convention), while Avalokiteśvara and Śāriputra appear in the standard text and must be the main protagonists. Apart from these constraints, T 252 stands alone in giving exaggerated numbers of participants and naming bodhisatvas. By comparison the latest translation has added some detail to what is essential the same passage. As we will see in the next section, the story of T 252 takes a distinct path to get to the standard Heart Sutra.

The Buddha's Samādhi

Conze's Sanskrit text describes how the Buddha taught a dharma teaching (dharmaparyāyaṃ bhāṣitvā) named “profound illumination” (gambhīrāvabhāsaṃ nāma) and then entered a meditative state (samādhiṃ samāpannaḥ). The Hasedera Manuscript (which has strongly influenced other Sanskrit editions) truncates this to "At that time the Bhagavan entered a samādhi named deep understanding"* (tena khalu samayena bhagavān gambhīrāvasaṃbodhaṃ nāma samādhiṃ samāpannaḥ).
* reading avabodha for avasambodha which is not in any of my dictionaries.
It appears to be this text that is translated in T 253, 254.
T 253: "At that time, the Buddha, the Bhagavān, entered the samādhi named Vast and Extremely Profound" (時佛世尊即入三昧,名廣大甚深。).
The Tibetan recensions appear to have a hybrid, e.g. TibA.
de'i tshe bcom ldan 'das zab mo snang ba zhes bya ba chos kyi rnam grangs kyi ting nge 'dzin la snyoms par bzugs so //
and at that time the blessed one was entered into the concentration of the preaching of the Dharma called “profound illumination”
Note that TibB has what appears to be an eyeskip error at this point. The phrase chos kyi rnam grangs (dharmaparyāya) has been shuffled forward into the middle of the phrase zab mo snang ba. The result is comprehensible but less consistent with the other texts.

The word dharmaparyāya is present in Conze sources Na, Nb, and Ne (with a variant reading); Ca, Cd, and Ce which makes it a minority reading. However the Tibetan equivalent, chos kyi rnam grangs, is present in both TibA and TibB. A translation of the word is also present in T255 (fǎ zhī yì mén 法之異門) and 257 (xuān shuō zhèng fǎ 正法宣說). Thus it appears to belong.

T 253 and 254 call the samādhi "extensive" guǎng dà 廣大 (mahat, vaipulya) and "profound" shèn shēn 甚深 (gambhīra). T 255 and T 257 do a better job of conveying the Sanskrit, i.e. "profound (shèn shēn 甚深) illumination (míng liǎo 明了)" and "profound (shèn shēn 甚深) illumination (guāng míng 光明)".

There is some confusion here. It is possible that an earlier version represented in the translations T 253 and 254 left off dharmaparyāya and it was added later. But it is equally possible that it was simply left out of the early Chinese translations. 

Note that the narrative in T252 is again very different. Rather than the Buddha' entering a samādhi we find:
All had attained the samādhi of always remembering, and abided in inconceivable liberation.
It is de rigueur to mention samādhi at this point in a Mahāyāna sūtra. In Recension One the Buddha entering samādhi is a required detail to explain why he does not speak in the core part of the text (the standard Heart Sutra). But the story arc in T 252 is different. The mention of samādhi is really a final detail in the description of the audience. However, both recensions use this moment to introduce the main protagonist of the story.

Enter Avalokiteśvara

One of the primary purposes of the introductory extension is to introduce the only actor with a speaking part, i.e. Avalokitśvara bodhisatva. There is a distinction to be made here. When Śāripūtra is mentioned in the introduction he is called āyuṣman śāriputra or "Elder Śāriputra". The word āyuṣman literally means "one who possesses life" or something like "one who has lived". The word is invariably placed before his name as a title. When the word ārya "noble" is used in conjunction with the name Avalokiteśvara its occupies the same position. By contrast, bodhisatva is invariably placed after the name of the figure and thus is not a title or epithet but an adjective or predicate. Bodhisatva means "one whose nature or essence" (satva) is awakening or enlightenment (bodhi). The spelling is consistently satva in Prajñāpāramitā manuscripts and is a distinctive Buddhist variant that is usually obscured by editors who tacitly revert to the classical spelling sattva as though it were an error. Elder Śāriputra is a bhikṣu. Noble Avalokiteśvara is a bodhisatva.

James Apple notes that prefixing ārya to a name is a relatively late practice:
"The prefix ārya appears in the opening salutation of Indian and Tibetan sūtras and is a late Indian Practice that begins to occur, as far as I can currently tell, in the works of [6th Century Mādhyamika] Bhāvaviveka." (Apple 2015: 4-5 n.5).
At this point in Recension One, Avalokiteśvara is practising the deep practice of Prajñāpāramitā. This is what gives him the essential insight into the nature of experience that he then communicates to Śāriputra, using the words from the opening paragraph of the standard sūtra; in other words he sees that the five skandhas are all absent. Except that in Sanskrit he sees them as svabhāvaśūnya "lacking a essence". This interpolation of svabhāvaśūnya is discussed further below.

We can infer that Avalokiteśvara is practising the yoga of nonapprehension (anupalambha-yoga). This involves using meditative techniques to cut oneself off from the world of sensory experience by deliberately and systematically withdrawing attention from the senses. The result is the cessation (nirodha) of sensory experience and this leaves the meditator dwelling in the absence of sense experience (śūnyatā). I refer to this state of the absence of sensory experience as Absence (capitalised). In English we more often encounter the word translated as "emptiness" and emptiness is frequently reified as "reality" or, worse it is hyper-reified as "ultimate reality". The result is the received strain of nihilistic anti-rational metaphysics of Madhyamaka.

Using my hermeneutic, we can look at Absence (śūnyatā) from two angles. Firstly, in Absence (śūnyatāyām) there are no mental events rising into awareness and passing away. There is some form of conscious awareness but it has no object (in philosophical jargon it is not intentional). Hence the Heart Sutra goes on to say that "in emptiness there is no colour, sound, smell, etc ... through the yoga of nonapprehension". In Buddhist jargon: in Absence dharmas do not arise or cease. If we take this state to be reality then we may be tempted to say that no dharmas ever arise in reality. But this causes us problems because when we are not in this state, dharmas constantly arise and cease (except in deep sleep).

The second angle is to say that although all mental states arise in dependence on conditions, this state is only attained when all conditions for the arising of mental states have ceased (though withdrawing attention). Absence itself then is not a conditioned mental state. It is without conditions. A conditioned mental state is saṃskṛta and thus śūnyatā is asaṃskṛta. That is to say śūnyatā is a synonym of nirvāṇa "extinction".

Something to keep in mind is that where the Heart Sutra says Shì zhū fǎ kōng xiāng, bù shēng bù miè 是諸 法空相 ,不生不滅 "All dharmas are marked with emptiness; not born, not dying" this is based on a mistranslation by Kumārajīva. The original Sanskrit says: yā śūnyatā na sā utpadyate, no nirudyate "that Absence does not arise, does not cease". In other words the original subject of this passage was not all dharmas, but the absence of all dharmas. Absence does not arise because it occurs only when all sensory experience has ceased. 

However, the story in Rencension Two is very different. Here, Avalokiteśvara has a dialogue with the Buddha. He announces his desire to give a teaching on the Heart of Prajñāpāramitā (般若波羅蜜多心), the Bhagavan praises this urge to teach and then takes a back seat, but not in samādhi. Now Avalokiteśvara enters a samādhi in which he "sees" the absence of essence in the five skandhas (照見五蘊自性皆空。). And then he addresses Śāriputra.

Quirks in this part of the text

T 253 and 254 have an extra phrase: lí zhū kǔ è 離諸苦厄 "apart/removed from all suffering and misery" means much the same as dù yīqiè kǔ è 度一切苦厄 "transcended all suffering and misery" in standard text. This phrase is absent from all extant Sanskrit manuscripts and the Tibetan canonical versions. It is present in the Fangshan Stele (the earliest physical evidence of the Heart Sutra) so cannot be a late edition. So it seems likely that it was simply left out of the Sanskrit translation, but whoever translated T 253 and T 254 from Sanskrit added it (back) in.

Zhàojiàn 照見 doesn't really make sense as two standalone characters: zhào 照 means "shining, radiant; illuminate, make visible; reflection" and jiàn 見 is the usual verb "to see". This has led to some very odd translations such as "illuminatingly sees", where zhào 照 functions as an adverb. This is far from satisfactory and there is no consensus on how to translate it. If the two characters are a binomial reflecting the Sanskrit vyava√lok then we still have a problem because the conclusion doesn't fit the premise.

I would like to propose a solution (suggested by the Sanskrit translation), which is that zhàojiàn wǔyùn jiē kōng 照見五蘊皆空 is, in fact, two phrases: zhàojiàn wǔyùn 照見五蘊 "[he] examined the five skandhas" and jiē kōng 皆空 "all void". The latter is minimal and requires us to interpolate much that we would naturally spell out in English or Sanskrit. Firstly, it is implied that the skandhas were void or absent. Secondly, the verb "to see" is implied by the context of the initial zhàojiàn 照見. If one looks, then one sees something. That something was seen as a result of looking can remain implicit in written medieval Chinese. Knowing that Sanskrit did not have this kind of flexibility, the Translator has to specify both the action "he examined" (vyavalokayati sma) and the result "he saw" (paśyati sma).

Note that the Sanskrit translation of the standard text substitutes what should be śūnyatā with svabhāva-śūnya. I will say more about this below.

T 255 complicates matters somewhat: guānchá zhàojiàn wǔyùn tǐ xìng xījiē shì kōng 觀察照見五蘊體性悉皆是空. Guānchá 觀察 would seem to be a synonym of zhàojiàn 照見, while jiē kōng 皆空 is expanded out to tǐ xìng xījiē shì kōng 體性悉皆是空 "self-nature, without exception, is absent". Here tǐ xìng 體性 conveys svabhāva. T 254 and T 257 also pick up the Sanskrit idea that the five skandhas are void of svabhāva. It's possible T 255 intended this but a character got dropped because the text only has characteristic (xìng 性) but there are no text critical notes in the Taishō so this is speculative. T 253 has the same text as the standard version, i.e. jiē kōng 皆空.

The present participle caramāṇa is not used in either Aṣṭasāhasrikā or Pañcaviṃśatisāhasrikā, probably because it is the ātmanepada ("middle voice") form of the present participle and √car is usually parasmaipada (indicative), so that the expected present participle is carant (nominative singular caran) and this is the form we find throughout the two main Prajñāpāramitā sūtras. Although note that caramāṇa is found in the Ratnaguṇasamcayagāthā (not translated into Chinese till the 11th Century), the Mahāvastu, and in many Pāli suttas (although the use of the ātmanepada conjugations generally in Pāli is minimal).

We can now come back to the issue of the substitution of svabhāvaśūnyan for śūnyatā. In fact, the phrase svabhāvaśūnyan is absent from the Pañcaviṃśatisāhasrikā as a whole. If one is in the śūnyatā samādhi one does not (cannot) see skandhas because they are absent (śūnya). They cannot be examined. This is explicit in the core passage of the Heart Sutra:
是故,空中—無 色、無受、想、行、識... —以無所得故。
Shì gù, kōng zhōng—wú sè, wú shòu, xiǎng, xíng, shì... —yǐwúsuǒdégù
Therefore: in emptiness—there is no form; no feeling, no thought, volition, awareness... —through the yoga of nonapprehension

In the list of eighteen kinds of śūnyatā we find the following terms
16. Absence of non-being (abhāva-śūnyatā).
17. Absence of being (svabhāva-śūnyatā).
18. Absence of non-being and of being (abhāvasvabhāva-śūnyatā).
Here svabhāva is not being used in the Madhyamaka sense but as a contrast to abhāva "non-being". This is to say that svabhāva appears to be used in the sense of sabhāva "with bhāva" and abhāva means "without bhāva". In another list of four kinds of śūnyatā, svabhāva "self-being" is contrasted with parabhāva, literally "other-being"

But Pañcaviṃśatisāhasrikā is constantly telling us not to take these as metaphysics; for example:
abhāvasvabhāvaśūnyatā abhāvasvabhāvaśūnyatāyāṃ na saṃvidyate nopalabhyate, abhāvasvabhāvaśūnyatāpi yāvad adhyātmaśūnyatāyāṃ na saṃvidyate nopalabhyate. (PvsP1-2: 146)
"... the absence of being and non-being does not perceive or apprehend the absence of being and non-being nor are the other kinds of absence perceived or apprehended."
In other words, whether or not something exists cannot be distinguished when all sense experience has stopped because all information about the world outside brain has stopped registering. Although the verbs saṃvidyate and upalabhyate are routinely given an ontological gloss in translation (especially by Conze) - the former, in particular, is often translated as "exists" - both are (based on etymology) epistemic terms related to the use of one's sensory apparatus. The reason that something is na saṃvidyate "not discovered" or nopalabhyate "not apprehended" in this context is that sense experience has ceased, not because they don't exist. But on the other hand, if they did not exist, how would we apprehend them in the first place? 

Similarly with respect to the list of four kinds of absence, the Upadeśa (T 1509) says (using Lamotte's reconstruction of the Sanskrit):
  • bhāvo bhāvena śūnyaḥ "being is absent from being"
  • abhāvo ‘bhāvena śūnyaḥ " non-being is absent from non-being"
  • svahāvaḥ svabhāvena śūnyaḥ "one's own being is absent from one's own being"
  • parabhāvaḥ parabhāvena śūnyaḥ "the being of others is absent from the being of others"
It is apparent that by introducing an element of Madhyamaka metaphysics into the Heart Sutra, the Translator (Ch → Skt) has hit a bum note. This is an important consideration since it is normally assumed that Prajñāpāramitā and Madhyamaka are intimately associated. In fact, the two sects have very different ideas and there is almost no crossover.

This also means that the translation of 照見五蘊皆空 as vyavalokayati sma pañcaskandhāṃs tāmś ca svabhāvaśūnyān paśyati sma is incorrect. The first clause seems about right but what Avalokiteśvara saw was more like tāṃś ca sarvaśūnyān paśyati sma "... and he saw that they were all absent". This is consistent with the practice of Prajñāpāramitā being synonymous with the yoga of nonapprehension (anupalambhayoga) and both with the cessation of sense experience. 

Śāriputra's Question

The puerile attitudes towards arhats in some Mahāyāna sūtras are generally absent in Prajñāpāramitā. The principle representative of the Prajñāpāramitā perspective is Elder Subhūti. Elder Śāriputra is deeply interested in Prajñāpāramitā and raises the kinds of questions that we might expect someone trained in another Buddhist tradition to ask. In the Large Sūtra, just before the teaching from which the core passage is drawn, Śāriputra asks the Buddha:
kathaṃ yujyamāno bhagavan bodhisatvo mahāsatvaḥ prajñāpāramitāyāṃ yukta iti vaktavyaḥ. (PvsP1-1: 61-2)
Engaging in what way, Bhagavan, is a bodhisatva mahāsatva to be called 'engaged in prajñāpāramitā?'
This generic or abstract use of the word bodhisatva is typical of the Large Sutra. In creating the Heart Sutra, the Author has placed the teaching in the mouth of Avalokiteśvara who, in Chinese Buddhism, is a cult figure particularly associated with saving people from disasters. Hence, in the introductory extension, rather than the usual abstract questions, Śāriputra addresses his question to Avalokiteśvara whose answer then segues into the core passage of the standard Heart Sutra.

The question Śārputra asks in the Sanskrit Heart Sutra is
yaḥ kaścit kulaputro vā kuladuhitā vā asyāṃ gambhīrāyāṃ prajñāpāramitāyāṃ caryāṃ cartukāmas tena kathaṃ śikṣitavyaṃ?
That son or daughter of the community desiring to practice this deep perfection of gnosis, how should they train?
As we will see, one of the Chinese texts mistakes kulaputra for a vocative, i.e. that Śāriputra is addressing Avalokiteśvara as kulaputra. Also note that in Sanskrit the question is framed with both male and female communards (kulaputro vā kuladuhitā vā)

Again Recension Two has a different approach. Avalokiteśvara has already announced his intention to speak. In this passage he tells Śāriputra to get ready, and Śāriputra says “Indeed, Great Purifier, I am ready, preach it. Now is the right time" (唯,大淨者!願為說之。今正是時。) .

We begin to notice that the form of names used in the texts is variable. The conventions established by Kumārajīva were:
  • Avalokitasvara: Guānshìyīn 觀世音 
  • Śāriputra: Shèlìfú 舍利弗,
Kumārajīva's translation is based on the older form of the name, i.e. Avalokitasvara (see Nattier 2007 and Karashima 2016). Using the new form, Avalokiteśvara, Xuanzang translated Guānzìzài 觀自在. He also modified Śāriputra: Shèlìzi 舍利子. Even so, Guānshìyīn 觀世音 or Guānyīn 觀音 are still the most common forms of the name in China, just as Kumārajīva's translations are still in use and were not superseded by Xuanzang's.

T 252 and T 253 have 舍利弗, the Kumārajīva spelling of Śāriputra, while T 257 has the Xuanzang spelling, 舍利子. All Chinese texts have the Xuanzang spelling of Avalokiteśvara 觀自在, except for T 254 which has Guānshìyīnzìzài 觀世音自在 which is a hybrid of Kumārajīva's 觀世音 and Xuanzang's 觀自在.

In the Tibetan texts we see some variation with the name Śāriputra, i.e. TibA shā ri'i bus; TibB shā ra dwa ti'i bus. The form Śāradvatīputra does occur in many Mahāyāna texts including the Gilgit manuscript of the Large Sutra. I'm not sure why.

All texts use the term kula-putra, "a son of the community" (Tib. rigs kyi bu; Ch. shàn nán zǐ 善男子) and some include the feminine kula-duhitṛ "a daughter of the community" (Tib. rigs kyi bu mo. Ch. shàn nǚrén 善女人). There is a tendency to see this as a generic term for a high status person. The Chinese translation shàn nán zǐ 善男子 literally means "good male child", where shàn 善 is the standard translation of kuśala.

In a Buddhist context kulaputra is often translated along the lines of "good", "gentle", or "nobly born". A kula is any coherent collection of animals: "herd, troop, flock, etc"; or people "race, family, community, tribe, caste, clique, fraternity, etc." Context suggests that the community in question is the Buddhist community made up of four saṃghas: bhikṣu, bhikṣūnī, upasaka, and upasikā. Such definitions as I can find emphasise that the term is used for one's social inferiors: a teacher calls a pupil kulaputra, but not the other way around. It means "one of the flock". And the prestige derives from Buddhists' attitude of Buddhist exceptionalism. That said the use of it does seem to drift. In my notes I adopted the translation devotee.

As I said above, there is a mistake in Śāriputra's question in T 253:
Kūlaputra, if there is a desire to genuinely learn the deep Prajñāpāramitā practising, how should they study the practice?
The passage is punctuated in CBETA as though kulaputra is a vocative, i.e. 善男子!meaning Śāriputra is addressing Avalokiteśvara as kulaputra. What we see more often is what we find in Sanskrit and Tibetan, that the two of them discuss the kulaputra in the third person, in the abstract. Or what we see in T 252, the teacher addressing the pupil as kulaputra. This appears to be a translation error, in which the translator has misunderstood their source, most likely a Sanskrit text and encoded this misunderstanding in Chinese.

Avalokiteśvara Preaches

With respect to kulaputra, R1 texts now answer the question as if Śāriputra had asked about both sons and daughters (kuladuhitṛ) of the community (Ch. shàn nánzǐ 善男子, shàn nǚrén 善女人). The only text that did phrase the question that way was the Sanskrit. Recall that R2 doesn't use kulaputra/kuladuhitṛ in this way and that Śāriputra doesn't ask a question. Variations in the names continue. T 253 in particular has both of the two different ways of writing Śāriputra in succession: Shèlìfú 舍利弗 and Shèlìzi 舍利子.

Avalokiteśvara opens his teaching by repeating the phrase in the question (that is also found in the opening paragraph of the standard version), i.e.
yaḥ kaścic chāriputra kulaputro va kuladuhitā vā asyāṃ gambhīrāyāṃ prajñāpāramitāyāṃ caryāṃ cartukāmas tenaivaṃ vyavalokitavyam
"Whichever son or daughter of the community who desires to perform this profound paragnosistic practice, therefore he should observe in this way.
The problem that I pointed out in Attwood (2012) holds here as well, i.e. vyava√lok is a transitive verb, so what is the devotee supposed to observe. The answer is obvious from the context, they are supposed to observe the five branches of experience (pañca skandhāḥ). Conze failed to grasp this in his edition of the standard text and does not see it here either. 

There are, however some major variations that are important to helping to make sense of the Tibetan versions. The first notable variant is that Ce (from Feer's 1866 polyglot edition) has śikṣatavyam "he should train" instead of vyavalokayitavyam "he should observe". 
  • Ce śāriputra kulaputro vā kuladuhitā vā asyāṃ gambhirāyāṃ prajñāpāramitāyāṃ varttakāmas tenaiva śikṣitavyaṃ || yaduta pañcaskandhāḥ svabhavaśūnyāḥ || katham svabhāvaśūnyāḥ ||
Words based on √śikṣ are common in Prajñāpāramitā. In T 257 the character xué 學 suggests that the translator might have had a text from the same lineage as Ce.

The other thing is that Nb, Ce, and Jb (Hasedera ms.) both add a phrase concerning the five skandhas
  • Nb yaḥ kaścit kulaputrā vā kuladuhitā vā asyā gambhirāyāṃ prajñāpāramitāyāṃ caryā catrukāma tenaiva vyavalokayitavyaṃ || pañca skandhān svabhāva śunya vyavalokatitavyaṃ ||
  • Ce śāriputra kulaputro vā kuladuhitā vā asyāṃ gambhirāyāṃ prajñāpāramitāyāṃ varttakāmas tenaiva śikṣitavyaṃ || yaduta pañcaskandhāḥ svabhavaśūnyāḥ || katham svabhāvaśūnyāḥ ||
  • Jb yaḥ kaścic chāriputra kulaputra vā kuladuhitā vā gambhīrāyāṃ prajñāpāramitāyāṃ caryā cartukāmas tenaivaṃ vyavalokayitavyam | pañcaskandhāḥ | tāṃś ca svabhasūnyān samanupaśyati sma |
The Chinese and Tibetan texts all have this additional clause, but in each case it appears to be better integrated than the Sanskrit, where the tense of the verb frequently clashes with the rest of the phrase. E.g.
TibB. shā ri'i bu rigs kyi bu 'am / rigs kyi bu mo gang la la shes rab kyi pha rol tu phyin pa zab mo spyad pa spyod par 'dod pa des / phung po lnga po de dag ngo bo nyid kyis stong par yang dag par rjes su mthong ba de ltar blta bar bya ste /
“Śāriputra! Whichever gentle son or gentle daughter desires to practise the practice the profound perfection of wisdom, he [sic] remarks that those those five aggregates are inherently empty, and should observe thus:”
However, the Tibetan texts are almost certainly translated from a Sanskrit source. Note the "he remarks" in Silk's translation contrasting with the subject "Whichever gentle son or gentle daughter". This could be a quirk of Tibetan.  

Jb has just tacked on the final part of the first paragraph of the standard text without alteration, leaving the final verb in the past tense so that it clashes with the rest of the sentence, particularly the future passive participle: "it should be observed by him... he perceived...". 

Where the Sanskrit adds this phrase it separates the verb (vyavalokayitavyam) from its object (pañcasakandha) by giving pañcaskandha in the wrong case (nominative plural instead of accusative plural) and often by adding a daṇḍa between them. I showed in "Heart Murmurs" (Attwood 2015) that this was an error. The phrase is vyavalokayitavyam pañcasakandhām "he should examine the five branches of experience." In order to be consistent with this (and with the intent of the Xīnjīng) the final phrase should have to be altered to tāṃs ca sarvaśūnyān dṣṛṭvāyam "and should see that they are  all absent".

None of the extant Sanskrit texts manages anything sensible and there is a great deal of variation. This suggests that scribes who could read Sanskrit were trying to make sense of the text as they copied it and took different approaches, but none looked to the Xīnjīng. However, in this passage all of the Chinese versions have replaced "he examined the five branches of experience" (zhàojiàn wǔyùn, jiē kōng 照見 五蘊皆空) with the Madhyamaka-inspired "they should contemplate the Five Skandhas as empty of self-nature" (Yīng guān wǔyùn xìng kōng 應觀五蘊性空). Also note the change in the verb from "examines" (zhàojiàn 照見)  to "should contemplate" (yīngguān 應觀; 應 having a similar sense to the future passive participle). This change argues against the extensions being made in Chinese, since it is the Sanskrit standard text (Hṛdaya) that introduces the svabhāvaśūnya where we expect śūnyatā from the Xīnjīng
This brings us to the end of the opening extension. The text now settles into the standard Heart Sutra, although the extra negations are almost universal in the Skt texts. And absent from all of the Chinese translations. We come back to the text after the dhāraṇī.

The Buddha's Endorsement
There is a small anomaly in T 253
T 253 般若波羅蜜多行,應如是
T 254 般若波羅蜜多行,應如是學
Skt. prajñāpāramitāyāṃ caryāyāṃ śikṣitavyaṃ
The second xíng 行 is probably a scribal error for xué 學, perhaps an eye-skip. Although note that TibA and TibB leave out "the practice of" completely:
TibA de ltar... bslab par bya'o "in that way... he should train"
TibB de ltar bslab par bya'o // "he should train in that way"
But in Tibetan, the question was asked in these terms (in Paragraph G).

T 257 adds a whole extra phrase: "If able to recite this Prajñāpāramitā dhāraṇī" (ruò néng sòng shì bōrěbōluómìduō míngjù 若能誦是般若波羅蜜多明句). Having studied the many ways that Chinese authors translated vidyā in a Prajñāpāramitā context I guess that míngjù 明句 literally "bright verse" is yet another way of translating vidyā or dhāraṇī, or even vidyā-dhāraṇī. This phrase has no parallel in any other version of the Heart Sutra.

T 253:
即時世尊 從廣大甚深 三摩地起
jíshí shìzūn cóng guǎngdà shènshēn sānmódì qǐ
The Bhagavan having arisen from the vast and profound samādhi...
This expression with the verb in the final position seems to be a Chinese idiom (Chinese is typically SVO) rather than a Sanskritism (Sanskrit is usually SOV). In T 223: cóng zuò qǐ 從座起 "he rose from his seat" literally "from seat rising" (8.229c08); also jiē cóng zuò qǐ 皆從座起 "all rose from their seats" (8.230b12). And numerous other examples.

It is slightly peculiar that the Sanskrit suggests that the teaching would be approved of by all the tathāgatas and arhats (anumodyate sarva-tathāgatair arhadbhiḥ). The arhats, generally speaking do not approve of Mahāyāna teachings, especially in the 8th Century, by which time the distinction has become somewhat schlerotic. However, as we have seen, the chief proponent of Prajñāpāramitā is Elder Subhūti, the arhat. Many of the Sanskrit manuscripts omit reference to the arhats as do all the Chinese and Tibetan translations.

T 257 also inserts a phrase—"it is real, supreme, and final" (shì jí zhēnshí zuìshàng jiùjìng 是即真實最上究竟)—which combines some common Buddhist superlatives that are also used in the standard Heart Sutra: jiùjìng 究竟 is from 究竟涅槃 translated in the Sanskrit Heart Sutra as niṣṭhanirvāṇa "final extinction" but probably more like nirvāṇa-paryavasāna "concluding in extinction"; zhēnshí 真實 is from the expression at the end of the epithets passage zhēnshí bù xū 真實不虛 "[Prajñāpāramitā] is truly real and not in vain."

Nothing parallel to this material is found in Recension Two (T 252).


The final paragraph in both recensions sees the audience rejoice at the teaching, employing a pericope that is similar in both R1 and R2. The phrase, "together with everyone in the gathering" (ca sarvāvatī parṣad), is a little unusual but is followed by the utterly stock phrase about the world with its various kinds of beings (sadeva-mānuṣa-āsura-gandharvaḥ lokaḥ). Most of the mss include the bodhisatvas (te ca bodhisatvā mahāsatvāḥ ) but this phrase is omitted in Jb, TibA, TibB, and Nh.

It was notable that the opening mentioned the bodhisatva-saṃgha, but they are not mentioned here.

There are some differences between the Conze edition and the Hasedera Manuscript (Jb).
Conze: āttamanā āyuṣmāñc Chāriputraḥ āryāvalokiteśvaro bodhisatvo mahāsatvas te ca bodhisatvā mahāsatvāḥ sā ca sarvāvatī parṣat sadeva-mānuṣāsura-garuda-gandharvaś ca loko bhagavato bhāṣitam abhyanandann iti.

Jb: ānandamanā āyuṣmān śāriputraḥ āryāvalokiteśvaraśca bodhisatvaḥ sā ca sarvāvatī pariṣat sadeva-mānuṣāsura-gandharvaś ca loko bhagavato bhāṣitam abhyanandan||
Jb has ānandamanā आनन्दमना for āttamanā आत्तमना, which looks like a simple scribal error. Jb also omits te ca bodhisatvā mahāsatvāḥ, and omits garuda from the list of beings. Jb also omits the sandhi that affects āyuṣmān śāriputraḥ: the rule here is complex but there are two possible outcomes here: āyuṣmāñcchāriputra āryāvalokiteśvaro or the more comprehensible āyuṣmāñ śāriputra āryāvalokiteśvaro

Unfortunately, the notation in Conze's (1967) edition completely falls apart at this point. E.g. the citation labelled "na" doesn't exist in Conze's notes. Some of his lettered citations are both the beginning and the end of a passage, some are adjacent to words but the notes deal with phrases. Numbered citations go up to 61, but the notes only go up to 58. I have found at least one variant omitted (note d, Nb tasmā tahi). Also Nb and Ce omit garuḍa but this is not noted by Conze.


The most striking thing, for me, is the sheer scale of variation in various documents instantiating the extended Heart Sutra text. The later manuscripts are overburdened with an accumulation of scribal errors but most also show signs of being deliberately edited. In a text of around 280 words, even small changes can be very important (even if it can take years of examination to see them). Passages have been added, omitted, and changed quite freely. So describing the extended Heart Sutra is no easy task. And while a broad outline of the stemma is not so difficult, including all the variations and horizontal influences (I'm trying to avoid the jargon "contamination" here) is beyond me at this stage. This is where computer-generated phylogenetic diagrams would be of use. 

T 252 is substantially different in every paragraph of the extensions despite also containing the standard Heart Sutra at its core. The protagonists are still the Buddha, Avalokiteśvara, and Śāriputra, but these are forced on any redactor by the conventions of Buddhist sūtra composition and the content of standard Heart Sutra. The differences between T 252 and the rest make it appear that this is a separate, possibly prior effort to extend the text (the traditional date is certainly earlier). The idea that the Heart Sutra was extended twice and no one noticed till now would be entirely in keeping with the history of the text and the history of scholarship on the text. I have found no evidence of who might have created T 252, but it may well have been the man credited with translating it, i.e. Fǎyuè 法月 or *Dharmacandra.

Lacking any working knowledge of Tibetan grammar I have not done a detailed comparison of the Tibetan recensions with the extant Sanskrit sources. But they are clearly a better fit with some than with others. Paragraph I of TibA and TibB suggests the hands of two different translators or redactors. The two read the conclusion of the extended opening in quite different ways. There is also the (to date) informal suggestion, from Ben Nourse, that different Tibetan manuscripts found at Dunhuang correspond somewhat to the two recensions.

These notes are preliminary to any attempt to tackle a critical edition. I also await a definitive study of the Dunhuang Heart Sutra manuscripts.

Language of Composition

If we are to establish what the language of composition was, what criteria would we use? With the Heart Sutra the task of deciding was relatively easy because of the copied passages - we could compare the Heart Sutra versions of the passage with the Large Sutra versions (which is what Jan Nattier did in 1992). The extensions, however, are a patchwork of pericopes. This is not unusual, since all Buddhist texts are to some extent modular (See Work, Text, Document), but it doesn't give us much purchase.

We could argue that the text which is more coherent and has the fewest grammatical errors is likely to be the original. On the other hand, most of the texts we use in this comparison have been edited and standardised for publication and/or canonisation. And all of them have been copied multiple times by scribes. Copying of manuscripts in greater India and its cultural sphere was often rather careless. Even after writing was widely used, sūtras were memorised and recited rather than read. It's likely the early transmissions of sūtras to China were memorised. Buddhists texts were often objects of religious devotion, never intended to be read or studied and thus careless copying didn't matter that much. In China, at least, scribes were literate and excellent handwriting was highly prized. The downside of this is that the educated scribe is more likely to casually "correct" a manuscript that they don't understand.

The Buddha's dharma teaching and samādhi is a point of departure for the R1 texts. It seems to me that the Sanskrit text which names the dharmaparyāya and leaves the samādhi unnamed is likely to be original. The Chinese texts don't mention a dharmaparyāya and this would be an odd detail to add in the circumstances. Or we could say that, since the Chinese R1 texts are in agreement, that the work that does not feature the Buddha giving a dharmaparyāya, and the presence of one in Sanskrit, is the oddity. However, the Tibetan texts, especially TibA, do seem to have both dharmaparayāya (Tib. chos kyi rnam grangs kyi) and samādhi (Tib. ting nge 'dzin). 

The Chinese texts, except for T 253, pick up on an editorial blunder that occurs in the Sanskrit standard text where svabhāvaśūnyan (zì xìng jiē kōng 自性皆空) is found where we expect śūnyatā (jiē kōng 皆空) based on the Xīnjīng. If one were to extend the Chinese text in Chinese this is not so likely. But note that the very next phrase in T 253 and 254—lí zhū kǔ è 離諸苦厄—is drawn from the Xīnjīng dù yīqiè kǔ è 度一切苦厄. 

T 253 treats kulaputra in Śāriputra's question to be a vocative, addressing Avalokiteśvara. This seems a very unlikely mistake to make if one were composing in Chinese. It looks like a translation error, a misreading of a Sanskrit text. Also with respect to the question in Para E, only Sanskrit asks in terms of both sons (putra) and daughters (duhitṛ) of the community (Ch. shàn nánzǐ 善男子, shàn nǚrén 善女人). Chinese and Tibetan texts phrase the question in terms only of sons. But they all have Avalokiteśvara answer the question in Para I in terms of both. Thus only the Sanskrit text is consistent in this case.

Another argument for a Sanskrit original for R2 is an awkwardness that occurs because of the use of the verb vyavalokayati. This works well enough in the standard Heart Sutra but when the Redactor tries to recast this verb in the standard form of a Prajñāpāramitā question, i.e. "how should the bodhisatva go about his business", where the activity is phrased using a future passive participle, the transitivity of vyavalokayati trips them up. 

For example, if the bodhisatva was to train (śikṣati*) in some form of Buddhist practice then the question would be kathaṃ śikṣitavyaṃ "how should he train?". And after the explanation Avalokitesvara might say "for this reason he should practice this way" tenaiva śikṣitavyaṃ (Ce, Feer's polyglot edition). Unfortunately, most of the Sanskrit manuscripts finished with "for this reason he should examine in this way" tenaivaṃ vyavalokitavyam. The reason it sounds so awkward is that the verb is transitive (Conze makes this mistake throughout his edition); that is, one cannot simply examine in the absence of something to examine. We saw above that the Translator filled out a Chinese phrase involving the implication that looking involves seeing. This is just the kind of adjustment a translator has to make moving between very different languages. But because the Translator chose a transitive verb (vyaavalokayati) this left the Redactor in a bind. In order to maintain consistency they had to have Avalokiteśvara use the awkward phrase tenaivaṃ vyavalokitavyam, something that would not have happened if the Sanskrit text were a translation of an idiomatic Chinese phrase. In this case the infelicitous Sanskrit also indicates that the Redactor of R1 was working in Sanskrit. 
*śikṣati is the desiderative of √śak "be able" and therefore literally means "the desire to be able" but it is used to indicate the training that a Buddhist practitioner (cārin) undertakes in Buddhist practices (cārya). 

So my first impression, which needs further scrutiny, is that the standard Heart Sutra was extended twice. The first time produced the text T 252 and since there is no evidence of it in any other language we can conjecture that it was made in Chinese (just like the Xīnjīng). This fits well with my revised history of the standard Heart Sutra as a Chinese digest text (chāo jīng 抄經) which was "authenticated"  based on the misperception that it was a translation from Sanskrit by Xuanzang and a forged Sanskrit "original" (actually a translation from Chinese). 

The standard Heart Sutra was extended a second time, probably in Sanskrit, although with varying influence from the Chinese (especially in T253), which was then translated into Chinese (T253, 254, 257) and Tibetan (including some Dunhuang manuscripts and the canonical versions), and from Tibetan into Chinese (T 255).

I see two new conjectures emerging from this study. 1. The extended Heart Sutra exists in two recensions; and 2. analysis of the language of the documents and editions strongly suggests that R1 was redacted in Sanskrit, while the lack of evidence for other versions suggests that R2 was redacted in Chinese. Also, since T 252, the sole representative of R2, is the earliest dated version of the text, I have got the nomenclature wrong. T 252 represents Recension One, an early attempt to create a more authentic Heart Sutra (by 7th Century Chinese Buddhist standards of authenticity). However, R1 never caught on. Recension Two is a second, probably later, perhaps unrelated attempt at a more authentic Heart Sutra, this time produced from the Sanskrit standard Heart Sutra and gave rise to all the other Chinese and Tibetan translations as well as a line of Sanskrit copies. 



Apple, James B. 2015. "Redaction and Rhetoric in Mahāyāna Sūtras: The Case of the Jayamatiparipṛcchāsūtra." Indo-Iranian Journal, 58(1): 1-25.
Attwood, J. 2019. "Xuanzang’s Relationship to the Heart Sūtra in Light of the Fangshan Stele." Journal of Chinese Buddhist Studies, 32: 1–30. http://chinesebuddhiststudies.org/previous_issues/jcbs3201_Attwood(1-30).pdf
Attwood, J. (forthcoming). "The History of the Heart Sutra as a Palimpsest." Pacific World.
Conze, Edward (1948). "Text, Sources, and Bibliography of the Prajñāpāramitā-hṛdaya." Journal of the Royal Asiatic Society, April 80(1-2): 33-51.
Conze, Edward. (1967). "The Prajñāpāramitā-Hṛdaya Sūtra." In Thirty Years of Buddhist Studies: Selected Essays, Bruno Cassirer, pp. 147-167. Modified version of Conze (1948).
Karashima, Seishi. (2016) "On Avalokitasvara and Avalokiteśvara." In Annual Report of The International Research Institute for Advanced Buddhology at Soka University (ARIRIAB), vol. 20 (2017): 139-165.
Lopez, Donald S. (1988) The Heart Sūtra Explained: Indian and Tibetan Commentaries. State University of New York Press.
Lopez, Donald S. (1996) Elaborations on Emptiness: Uses of the Heart Sutra. Princeton University press.
Nattier, Jan. (2007) "Avalokiteśvara in Early Chinese Buddhist Translations: A Preliminary Survey." In Magee, W and Huang, Y.H. (Eds). Bodhisattva Avalokiteśvara (Guanyin) and Modern Society. Proceedings of the Fifth Chung-Hwa International Conference on Buddhism, 2006: 191-212. Taiwan: Dharma Drum Publishing.
Schopen, Gregory. 2004. "If you can’t Remember, How to Make it up, Some Monastic Rules for Redacting Canonical Texts." In Buddhist Monks and Business Matters, Still More Papers on Monastic Buddhism in India. Honolulu: University of Hawai’i Press. 395-407.
Silk, Jonathan A. (1994) The Heart Sūtra in Tibetan: a Critical Edition of the Two Recensions Contained in the Kanjur. Arbeitskreis für Tibetische und Buddhistische Studien, Universität Wien.
Jonathan A. Silk (2015) "Establishing/Interpreting/Translating: Is It Just That Easy?" Journal of the International Association of Buddhist Studies. Vol. 36/37: 205-225.
Related Posts with Thumbnails