Seeing Blue.

Where does blue begin and end?

There's a meme that seems to come around again and again on the internet. It is that if a language has no word for a concept then that concept must be absent in that language. This naive reading has been applied to the colour blue for example. Some people noticed that ancient European writers, particularly the ancient Greeks, had a limited colour palette in their writing. Indeed many modern languages are rather lacking in colour terms. Until the 1540s there was no word for the colour orange in English, which is why we call people with ginger hair "red heads". This does not mean that we could not distinguish the colour of blood from the colour of ginger hair. It only means that they were in the same colour category. And when we did name the colour orange, we named it after the fruit, not the other way around. However, it seems journalists love this idea that the Ancient Greeks could not see blue and the idea lumbers around like a zombie eating brains: it gets knocked down, but is quite difficult to kill and reduces IQs.

Colour words do not correspond to objects or entities. Colours are broadly defined categories of perception. Categories are mental and linguistic structures that help us to organise how we perceive the world. We can use the category name to talk about all the members of a category at once without having to use tedious lists of inclusions and exclusions. This is usually possible because we interact with all members of a category in the same way.

In George Lakoff's powerful model of thinking about categories we define categories towards the middle of a taxonomical hierarchy and by relationship to a prototype. So dog seems like a "natural" category, whereas for every day use: mammal is too broad and includes too many non-dog examples that need to be excluded; while spaniel is too narrow because it leaves out too many dog examples like terrier. Dog as a category works because there are consistent ways that we interact with dogs that are common to all dogs and different from other common pets or wild animals. And also because this interaction is not something personal, but common to other people in our language group. Sometimes pet is a more convenient category: when renting out a house for example. Though we think of categories defined by forms or functions, one of the most important defining properties is how we interact either in fact or potentially with the entities.

When we think of 'dog' as a category we will have an internalised prototype that defines the category. And we judge other entities to be a member of our category to the extent that they resemble our prototype (this is an extension of Wittgenstein's family resemblances'). By definition some members may be more central and others more peripheral. Say our prototype is something like a German shepherd (left). we can acknowledge, as dogs themselves usually do, that both a chihuahua and a great dane are members of the category dog, despite their size. Similarly though a long muzzle is typical, we can acknowledge that dogs with mutated skulls that give them a squashed look (boxers, pugs) are still dogs. On the other hand despite being furry, carnivorous, quadrupeds, no kind of cat is is a member of the category dog. In Cambridge there is a couple who take their cat out on a lead. But even a cat on a lead is not a dog.

However, the prototype is not fixed or absolute. It is relative to many things, not least of which is how we interact with the category. With respect to dogs, a farmer or a hunter may think in terms of a working animal, a pet owner in terms of companionship, and so on. On the other hand in India dogs are often semi-domesticated urban scavengers - neither pets nor workers, but barely tolerable vermin. In some cultures dogs are seen as food.

It's possible for there to be doubt about membership at the periphery. Is a wolf a member of the dog category? Is a fox? The wild dog is another peripheral case: it looks like a dog, but we interact with it as a wild animal (to which category it belongs with wolf and fox) rather than as pet or worker. There is no upper or lower limit on how many categories we employ or the extent to which they overlap.

navy

royal

cobalt

azure

sapphire

beryl

electric

sky

turquoise

cerulean

teal

cyan

Our terms for colours are categories also. Typically for an English speaker the prototype for blue is the sky. This can get complicated because in England the sky is more often grey than blue, and when it is blue, it's often a very pale and washed out blue compared to where I grew up (about 15 degrees of latitude closer to the equator, about 1000ft above sea level, and with much less pollution). In some cultures lapis lazuli or the throat of a peacock are prototypes (the latter is important in India for example).

Other languages, including many living languages define their categories differently. And research has shown patterns in how languages categorise colours. Many languages for example put blue and green in one category. In ancient Chinese the word 青 qīng meant both blue and green, but also black. In this sense it appears to be similar to the Sanskrit śyāma which can mean black, dark, dark shades of blue or green. Used of people it refers to a dark complexion. So in fact, Śyāma Tārā is not Green Tārā, but Dark or Swarthy Tārā despite the fact that she is routinely depicted in bright hues.

Does this mean that those languages which lump blue in with other colours lack a concept of blue? Not necessarily. Because even blue is a broad category. I can distinguish many shades of blue, from cyan to navy, but I don't have words for all these colours. Similarly I can distinguish many shades of green from the almost yellow green of new spring leaves, to the dark blue-green of New Zealand jade. Think about all the distinctions of colours on a typical paint sampler that we have no words for, but for which arbitrary names have to be invented for marketing purposes. We also have at least one word for a colour that is made up, indigo. When Newton was describing the colours of the rainbows he created with prisms he wanted their to be seven colours to fit in with an alchemical scheme and so invented the colour indigo. What Newton called blue is what today we'd call cyan, and what he called indigo is deep blue like ultramarine or cobalt blue. In fact most English speakers shown swatches of these colours would call them both blue.

As Lakoff explains in his book on categorisation, Women, Fire and Dangerous Things, those languages that have four colour terms will have black, white, red and one of either yellow, blue or green (p.25). Now it seems that Ancient Greek was a four colour language.

"Empedocles, one of the earliest Ancient Greek color theorists, described color as falling into four areas, light or white, black or dark, red and yellow; Xenophanes described the rainbow as having three bands of color: purple, green/yellow, and red." (Ancient Greek Color Vision)

This fits the pattern noticed by colour perception research. The Greeks used four colour terms, roughly, white, black, red and yellow. So when Homer uses the phrase "wine dark sea" or describes the sky as "bronze", he is employing categories that are much broader than we currently use in English. In fact modern English has eleven basic colour categories:

"black, white, red, yellow, green, blue, brown, purple, pink, orange and gray."

This does not stop us seeing blueish green, yellowish red, reddish purple and other colours for which we have no name or category. Categories are as broad as are useful to us. And often colours are difficult to categorise. Blue-green colours for example may appear to be in different categories to different people. But there is no evidence to suggest any anatomical differences between speakers of languages with four or less colour terms and those with eleven.

Now colour perception is a feature of our particular sensory apparatus. We've seen recently with the example of "that dress" how the background against which we see something and the colour of the light illuminating it, affect how we perceive it. But vision does have an objective component because the physiology of it is the same for everyone. Light of particular wavelengths hits our retina and activates patterns of the three (sometimes four) kinds of colour sensing cone cells. Each of the cells responds to different frequencies of light.

The peaks of these curves are the same in all humans. This means that where languages have the same colour terms they tend to agree on where in the spectrum the prototype for that category lies. I presume this has applied at least since anatomically modern humans. Now of course turning the signals from our cone cells into the experience of colour is a process that happens in our brains. But it's not arbitrary. For people who are not colour blind the brain is set up for blue cone cells to respond to the same frequency of light. If I shine light with a frequency of 500 nm in your eyes, you'll perceive this in more or less the same way as every other human being regardless of language and culture. Linking the experience to a word is a function of language, but the ability of the language to translate the experience into words is always limited. People with four cones describe a far more vivid palette of colours (What it's like to see 100 times the colors you see). Some animals have cones sensitive to different wavelengths. In particular bees can see much shorter wavelengths - well into what we call the ultraviolet. While snakes can detect much longer wavelengths in the infrared (though not with their eyes)

Now, the story goes that because some languages lack a word that corresponds to the English word blue, and they treat what we call blue as a member of broader colour category, that this means that the speakers of that language could not see blue. This is like saying that because the English lack a word for schadenfreude that they do not enjoy the misfortunes of others, whereas in fact the laughing at the misfortunes of others is very popular here (it is perhaps the most important theme of English humour). So why does this suggestion keep surfacing?

The idea about the Greeks not being able to see blue can be traced to the 19th century British Prime Minister and amateur philologist William Gladstone. He published a long and highly regarded study of Homer's epics and noticed that Homer's colours did not match ours, the "wine-dark sea" being one of the well known examples (wine being reddish-purple in our language, a colour we never associate with the sea). Others joined in. More recently the idea that how we use language reflects how we perceive the world is called Linguistic Relativism. It is also known as the Whorf-Sapir Hypothesis because theories about it were postulated (separately) in the early 20th Century by linguistic Benjamin Lee Whorf and his teacher Edward Sapir (amongst others). Whorf in particular was interested in the way that grammar divided the world up into entities and activities. He discovered that some Indigenous American languages seem to not make the same kinds of distinctions. On the basis of this he hypothesised that these differences in grammar might affect how we see the world at a very deep level. How would the world appear to us, for example, if we did not divide it up into nouns and verbs. What if we only had verbs for example, if everything was seen as a process? Whorf asked is the world really is divided up into objects

Linguistic relativism comes and goes in the media. Every few years some journalist comes across Whorf or some other author and writes a piece about it. I should add that Whorf's essays make very good reading (they were collected into a book, Language, Thought and Reality, MIT Press, 1956). The "Greeks couldn't see blue" meme is a popular version of this and one can find many variations on the theme, on the internet, including a few other attempts to debunk it.

However, quite a bit of research has shown that because of the physical apparatus of seeing there is no room for relativistic effects in colour perception. All humans see colour in the same way, even though different languages categorise colours in different ways. Every (normally sighted) human being is capable of seeing millions of colours, most of which we don't have names for (which is where categories come in handy). And all this commonality is true of subsets with variations on the the normal pattern: people with four cones see similarly to each other; people who are red-green colour blind all see the same shades of grey and so on. In other words the research disproves idea that having no word for blue means one cannot see the colour blue. So basically the whole "can't see blue" thing comes down to a failure to read the research on colour vision.

Ironically if you do a simple image search on "Greece" the predominant colours in the results are white and blue, the colours of the modern Greek flag.

~~oOo~~

12 Apr 2015. See also: 3-D Map Shows the Colors You See But Can’t Name. Wired.

16 Feb 2016. See also, Bogushevskaya, Victoria. (2015). Qīng (青) in Chinese: when and why it means ‘green’, ‘blue’ or ‘dark’/‘black’, in Thinking Colours Perception, Translation and Representation [Edited by Victoria Bogushevskaya and Elisabetta Colla]. Cambridge Scholars Publishing, 26-44.

28 Mar 2017. 'Why don't Americans have a name for the color 'light blue?' Study finds unique color terms used in Japan, US.' Phys.org.

10 Sept 2025. "Large-scale color biases in the retinotopic functional architecture are region specific and shared across human brains." Journal of Neuroscience. 8 September. e2717202025; . https://doi.org/10.1523/JNEUROSCI.2717-20.2025.

 

See also Kavanagh, Katie. (2025). "My blue is your blue: different people’s brains process colours in the same way. A machine-learning tool can predict what colour a person is looking at when trained on the brain activity of others." Nature Articles.

Rebirth in the Ṛgveda

Most modern discussions of the afterlife in the Vedas say that rebirth/reincarnation is not found in the Ṛgveda. Conventionally speaking, the first mention of rebirth in India literature is thought to be in the Bṛhadāraṇyaka Upaniṣad (4.4). This text was composed somewhat before Buddhism and probably in or near the kingdom of Kosala (within reach of the śrāmaṇa religions). The idea of karma in relation to rebirth is introduced by a king (kṣatriya) and many scholars take this to mean that the Brahmins adopted rebirth from a śrāmaṇa group.

Joanna Jurewicz has shown that despite the conventional understanding that there is evidence for belief in rebirth in the Ṛgveda. This essay will walk through her discovery and comment on the relevance for Buddhist rebirth. Jurewicz's discovery begins with by revisiting a well known passage in the Ṛgveda. In RV 10.16.5 we find this passage. I cite Wendy Doniger's translation to show the conventional understanding of this stanza.

áva sr̥ja púnar agne pitŕ̥bhyo yás ta ā́hutaś cárati svadhā́bhiḥ |
ā́yur vásāna úpa vetu śéṣaḥ sáṃ gachatāṃ tanúvā jātavedaḥ ||

Set him free again to go to the fathers, Agni, when he has been offered as an oblation in you and wanders with the sacrificial drink. Let him reach his own descendent, dressing himself in a life-span. O knower of creatures, let him join with a body. 

Here the dead person is treated like a sacrificial offering (āhuta) to the fire (agni). Agni transforms offerings into smoke and wafts them up to the sky where the devas live. Jurewicz makes the point that in Vedic eschatology the fathers or ancestors also dwell in the sky or heaven.

Jurewicz points about that pitṛbhyaḥ can either be dative and ablative and that all translators to date have read it as dative (to the fathers). But really there is no apriori reason not to read it as ablative (from the fathers).

Jurewicz analyses the verb ava√sṛj according to the principles of George Lakoff. She points out that the concrete meaning is 'untie' as in 'untie a bound captive, or a tethered animal'. Abstractly this can refer to forgiveness for wrongdoing. In a ritual context untying the animal means to sacrifice it, as the victim is bound to a post before being killed. The verb ava√sṛj can also refer to releasing an arrow. Jurewicz speculates that the bow-string might be seen as binding the arrow which is then release from captivity when the archer looses it (though 'to [let] loose' an arrow is also a metaphor in English and I think refers to the right hand hold (back) the bow-string and the arrow). Another sense of ava√sṛj is the releasing of the waters by Indra (RV 10.133.2, 8.32.25). In this usage the preverb ava takes its most obvious meaning of 'down'. When Indra releases the waters, rain pours down and rivers flow down from the mountains (an image also found in the Pāḷi Canon). Thus there is a strong argument for emphasising the reading of "release him down from the fathers again."

However we read pitṛbhyaḥ it raises the question of why the text asks Agni to do this again (punar). "Send him to the the fathers again" or "release him from the fathers again." One obvious reading is that the poet conceives of this happening repeatedly, i.e. that he believes that one is born, dies and goes to the ancestors repeatedly.

The other padas of the verse support this reading. Although śeṣa  (literally 'remainder') can mean what is left after the fire has burned out, in the Ṛgveda it definitely also means 'offspring'. In addition āyuḥ refers to a human life. In her consideration of the word svadhā, which qualifies the movement of the dead (ta ā́hutaś cárati svadhā́bhiḥ "The sacrificial offering proceeds with svadhā"), Jurewicz says "Most scholars in their translations choose words denoting will, right or autonomy." Doniger, by contrast relates svadhā to soma, the drug laced liquid imbibed as a stimulant and ladled onto the fire as an offering. However, Jurewicz argues that the main idea being conveyed here is 'contradictoriness'. This captures the sense that the unmoving dead body is none-the-less able to travel (as smoke from the fire) up to heaven.

Jurewicz proposes the alternate reading of the verse:

"Release him to his fathers and again down from them, who, poured into you, travels

according to his will. Let him who wears life come to his offspring. Let him join his

body, Jatavedas!"

On this reading the verse is quite clearly a reference to rebirth. Jurewicz's sensitivity to the nuances of the language allow us to see this verse in a new light. This does not exclude the other readings seen by other translators. Language does this. It covers a wide range of possibilities that are apparent to the community of speakers and which are often lost in translation. What we assume about the context may influence how we make editorial and translation decisions.

In my next essay I will explore the role of Yama in discovering the pitṛloka. Yama had to find his way there, and since the way there is the correct performance of the śrāddha or funeral rights, then this suggests a memory of the adoption of rebirth eschatology, probably in India, since Vedic speakers do not share this idea with anyone other Indo-Europeans.

The discovery of rebirth this early in Indian history is important. Because it means that if Vedic speakers did adopt rebirth from outside their immediate cultural sphere, as seems likely, they did so very early in their time in India. A broad consensus places the composition of the Ṛgveda in the time period 1500-1200 BCE. To my knowledge no one has suggested that śrāmaṇa culture can be found in the Vedic āryavarta or homeland in the Western Ganges Valley, nor is there evidence for śrāmaṇa culture at this early stage. It also suggests that some kind of rebirth eschatology was widespread in the Ganges Valley when the Vedic speakers arrived. And what was introduced in the Bṛhadāraṇyaka Upaniṣad was not rebirth per se, but the linking of rebirth to good (puṇya) and evil (pāpa) actions.

To my mind this is a blow to those who argue that rebirth was foreign to the Buddha and interpolated into Buddhism. An interesting parallel is the Brahmanical social institution of varṇa or class, that develops into jati or the system of caste. Buddhist texts often treat varṇa as foreign. There are many arguments about how and to whom it might apply. Brahmins insistence on being the best caste are regularly undermined. There are actual arguments about caste. No such arguments occur with rebirth as the focus. Similarly there are discussions about the role of Brahmā in the world. But rebirth is simply a background idea that is never challenged. There are those who explicitly reject any kind of afterlife (e.g. Prince Pāyāsi) but they are treated as misguided and untrustworthy.

When we discuss the appearance of Indic languages and culture in India we used to speak of invasions. The invasion theory has long been untenable. Most likely small, related bands of Indic speakers—Vedic and related dialects—began moving into the sub-continent and were assimilated by the existing population, leaving little genetic trace. In the language of the Ṛgveda we see many loan words from the Dravidian language family and quite a few from Munda (or Austro-Asiatic), structured in such a way as to suggest that Munda speaking people were met first. See Witzel (1999). This trickle of incomers from Iran and now Afghanistan has continued to the present, sometimes overshadowed by invasions proper (Alexander of Macedonia, the Kushans, etc). Rebirth like many features of Indian languages appears to be a regional feature of the Indian subcontinent. A few relics of the migration have been preserved, such as the retentions of the scheme of assessing morality by actions of the body, speech and mind, a scheme which comes from Zoroastrianism. 

~~oOo~~

Bibliography

Jurewicz, Joanna. The Ṛgveda, ‘small scale’ societies and rebirth eschatology. Oxford Centre for Buddhist Studies. http://www.ocbs.org/lectures-a-articles-ocbsmain-121/63-the-rigveda-small-scale-societies-and-rebirth-eschatology

Witzel, Michael. (1999) Substrate Languages in Old Indo-Aryan (Rgvedic, Middle and Late Vedic).
Electronic Journal of Vedic Studies. 5(1): 1-67.
http://www.ejvs.laurasianacademy.com/ejvs0501/ejvs0501article.pdf

The Very Idea of Buddhist History

Uni of Washington

Alexander Wynne has been at the forefront of apologetics for taking the suttas at face value as historically accurate. He recently uploaded a copy of his 2005 article, The Historical Authenticity of Early Buddhist Literature, to academia.edu:

(2005) The Historical Authenticity of Early Buddhist Literature: A Critical Evaluation. Vienna Journal of South Asian Studies. 49: 35-70

Readers may know that there is a split in Buddhist studies. On one side are religious traditionalists and mainly British scholars (particularly Richard Gombrich and Wynne at Oxford) who see the early Buddhist texts as a more or less accurate account of Buddhist history. On the other side are religious sceptics (yours truly) and mainly American scholars (particularly Greg Schopen and Don Lopez) who don't think there is anything authentically historical in the suttas.

Wynne writes and argues well and is taking an active role in the debate. But he is far less critical than he ought to be of his sources and it's only by ignoring many of the problems that I'll set out below that he can stick to his conclusions. He places too much credence on the traditional narratives of Buddhism. 

We know this: there is a body of literature we associate with early Buddhism and the early phase of sectarian Buddhism. This literature is preserved, in a language we now call Pali, in major collections of manuscripts in Sri Lanka, Burma and Thailand, with minor collections in Laos, Vietnam and perhaps other places. Substantial parts of several other recensions are preserved in Middle-Chinese in China, Japan and Korea (the most influential modern edition of the Chinese Canon, the Taishō, is based on a recension preserved in Korean) both in manuscript form and in printed editions. Fragments of several recensions are preserved in Gāndhārī, Buddhist Sanskrit, and Tibetan collections.

There are clearly relationships between these recensions of the literature. Many of the actual sutta/sūtra texts are same or similar, though some are quite different. However at the level of Nikāya/Āgama collections, they only partially overlap. Even within the Pāḷi Nikāyas when a sutta is recorded twice in two different Nikāyas it is often different in significant ways. Similarly the various collections titled Vinaya only partially overlap (and to a lesser extent than the sūtras). The Abhidharma collections are all very different and even adopt different hermeneutic principles for interpreting the texts. Thus the collections seem to draw on a common body of texts, but to collate and interpret them differently. 

Scholars in favour of seeing this body of literature as "authentic" (a word we'll have to come back to and define) tend to portray the literature as fairly homogeneous and belonging to a specific region in space and time. They tend to accept the story told by the literature about it's own genesis. I find the reasons for doing so less and less convincing the more I read the literature. 

Dates

Here's the thing about dating. We can be reasonable certain about the date of Asoka because of his inscriptions. Mid 3rd century BC. All of the dates for Indian history in the first millennia BC are worked out (mostly guessed at) with respect to this single reference point. We assume that literature which does not know Asoka, was written before his time. The Pali Nikāyas and Vināya do not know Asoka. What's more they seem to describe a landscape of fortified cities and petty kings in which the later hegemons like the Kingdom of Magadha are just starting to flex their imperialistic muscles. Not only is there no sign of Asoka, there's no sign of the Mauryan dynast, Candragupta. The Magadhan capital is still Rājagṛha rather than Pāṭaliputta. Archaeology tells us that this form of geography began to emerge ca. 6th century BC. We've rediscovered ruins that correspond to the major civil and geographic reference points that seem to align with the Buddhist stories. The major exception of the on-going dispute over the location of Kapilavastu. If, as I suspect, the story of the Buddha's life is largely a later fiction, then the identity home town is also likely to be fictional - Kapilavastu is likely to have been a wattle & daub village. 

Given these points we assume that the early Buddhist texts were composed between ca. 6th-4th century BC. There is no way to date the texts absolutely. The oldest Buddhists texts are birch bark scrolls from Gandhāra dated roughly to the first century BCE. The oldest Pāli text is a metal plate from the 6th century, most of the manuscripts of the Pāli Canon are only a century or two old. We know from their texts that Brahmins of a slightly earlier period (associated with the exegetical Brāhmaṇa literature) thought the region the Buddha lived in was barbaric and an unsuitable place for Brahmins to live. None of the early Upaniṣads mention Buddhists and whereas Brahmins feature quite prominently in the early Buddhist texts. Many scholars think we can see hints at some knowledge of themes from the Upaniṣads in the early Buddhist texts. One major scholar, Johannes Bronkhorst, inverts this however and argues that the Upaniṣads must have been composed after the early Buddhist texts. Brahmins took some time to fully convert the central and eastern Ganges plain to their culture. In fact this happened after Asoka. But the early Buddhist texts record fairly frequent meetings with Brahmins and interesting phenomena like land grants by kings to Brahmins (where land = income). Buddhist texts also give a scattered, but overall fairly comprehensive, picture of the Brahmanical religion revolving around sacrificial fires and strong guru/disciple relationships. 

We don't know and cannot know if this is correct. The assumption by Gombrich and Wynn is that if the texts say something is so and we have no reason to doubt them, then we ought to take the texts as being accurate. I and others argue that we have every reason to doubt the texts and should never take them at their word, but should interrogate them to expose inconsistencies.

One of the major problems is that while these cities are known to exist and archaeological evidence for Buddhist activity is found in and around them, it all dates from many centuries after the period when we assume the Buddha to have lived. There's no clearly Buddhist archaeology that pre-dates Asoka. Given Buddhist's own stories this is quite surprising - we know for example that land was donated and structures created by rich followers. But all the structures discovered to date are from the period of Asoka or later. Attempts to interpret pre-Asokan dates are often obviously bogus.

Another problem is that by their own admission the texts were preserved orally for something like three centuries before being written down in Sri Lanka, ca. 100 BC. We don't know (I think) when the texts were written down in India. However we do have manuscripts of Buddhist texts from about the 1st century AD from Gandhāra, so at least by that point Buddhist texts were written. In some cases it's obvious from internal evidence that the texts were composed well after the period they purport to be an account of. Comparison of little details in the Chinese Canon suggests that it was closed to additions and emendation somewhat later than the Pali Canon, which perhaps suggests a later date for being committed to writing. This is a lot of conjecture however, and none of it testable with the present state of archaeology and paleography.

The texts represent an event horizon beyond which we cannot see much if anything with clarity. All dates, except Asoka are vague and/or speculative, if only because establishing relationships to Asoka is often an exercise in speculation. Any historical facts are smeared out and information is lost. We can say what kind of world is reflected in the texts, but we have a great deal of difficulty demonstrating a relationship between this story and reality. Contra Gombrich and Wynne I think we have every reason to doubt traditional narratives.

Cracks

My own recent work has been focussed on cracks in the facade of homogeneity and unanimity presented by many scholars. In particular I've tried to show that the idea that the Pali Canon represents a single tradition is not sustainable. The Pali texts are clearly collated from multiple oral lineages that have been inexpertly edited into several collections (nikāyas). Supposedly fundamental doctrines show a bewildering amount of diversity in the suttas. Details are often fudged or changed. And the approach to this by scholars has been to compose unifying narratives that gloss over internal inconsistency or try to demonstrate a linear development within the texts. I think this methodology is flawed from the start. Instead, I argue that it is precisely the flaws in our material that convey the most information about the history of the collection. We ought to be foregrounding flaws rather than explaining them away. And we have to consider that linear developments do not explain the divergences sufficiently well. 

My work on the Buddha's name for example shows that Siddhartha is a name attributed long after the time of Canon, and that if his family used a gotra name they used it extremely idiosyncratically and ahistorically (See What's in a Name pdf). If the head of the family was alive, he (i.e. Suddhodana) rather than his son ought to have been called Gautama. His son ought to have been Gautamaputra,  Gautamya, or some other diminutive (though of course they certainly did not speak or use Sanskrit!). On the other hand why were his mother and aunt called Gautamī? Gotras are exogamous marriage groups: Māyā and Prajāpati wouldn't have changed their gotra names on marriage, and nor could they have married within the Gautama gotra. And why are there so few other members of the Gautama clan in the Pāli? Other gotra names occur frequently. Gotra names are a tradition belonging to the strongly patriarchal Brahmins, though matrilineal names are not unknown, it's unlikely that the Buddha took his mother's gotra name (the matrilineal form Gautamiputra occurs in non-Buddhist contexts). Gautama is a gotra of considerable renown in the Vedic world. The patriarch, Gotama, was one of the poets who composed the Ṛgveda, and his descendents, the Gautamas, feature prominently as teachers and exegetes.  And yet the Śākyas were clearly not part of the Vedic milieu and at times the Buddha is portrayed as treating the Brahmins as crazy foreigners. The ascription of Brahmanical caste identity to the Buddha is almost certainly anachronistic, not to mention being somewhat variable. Though he is usually a Gautama, at least once he is said to be from the Āditya gotra. The Buddha's biography has been reworked to give him prestige in a Brahmin dominated world. This tells us that the biography was probably composed in a time of Brahmanical hegemony, which would place it after Asoka. And it happened in a way that was accepted and recorded by all Buddhists. This fact contradicts all other theories about the time period that the texts represent.

I've also discussed the contradictory biographical traditions in the suttas (see The Buddha's Biography). There are at least two biographies of the Buddha. In one he is a unmarried youth when he leaves home and his mother is still alive. In another he is a man of 29 whose mother died in childbirth. The youth is found in the Pāli version of the Ariyapariyesana Sutta and the 29 year old in the Chinese counterpart of the same text. Both stories cannot be true and we have no objective way of knowing which is. All we have is a general historical principle that Buddhist stories become more elaborate over time (there is clear evidence of this in the accurately dated Chinese translations). Thus, we usually assume that a less elaborate version of a story is (relatively) earlier than the same story in a more elaborate version. 

Ambiguity

Thus the suttas are at best an ambiguous source of information. Sometimes consistent with an early period centred on the 5th century BCE and yet at other times with a period some centuries after this. Many details of history found in the Canon are also contradicted in the Canon. This lack of internal consistency is not simply overlooked or explained away, it is presented as the opposite, as strikingly self-consistent. For the average Buddhist, almost inevitably falling into confirmation bias, the endorsement of their views by scholars of considerable reputation makes it all the harder to objectively evaluate the information at our disposal. 

The situation is probably that the suttas do not represent one period of history, but were composed over several centuries and that earlier texts form a template for later texts. Everything we know about later Buddhist texts, such as the Perfection of Wisdom texts, tells us that Indian Buddhists never stopped retelling and embellishing their stories. This is clear for example where we have multiple Chinese translations over the centuries: later translations of the same text are inevitably longer and more elaborate, often accumulating several chapters at a time. My observation of comparing Chinese and Pali texts is that this happened to a lesser extent with early Buddhist texts as well (see my translations and notes on the Chinese Spiral Path texts). 

However, we also know that words, phrases, sentences, paragraphs and sometimes even whole texts were recycled time and again. It's usually impossible to know which was the first use of any repeating unit. The end result is that suttas are frequently formulaic and repetitive. Many "suttas" are merely fragments of larger units. This would seem to fit a situation in which a relatively small number of texts were expanded and elaborated by story tellers who both used standard vocabulary and themes, and embellished stories as they saw fit. Versions of the stories diverged, sometimes to the point of seeming like a distinct story though usually with the same characters in the same setting.

In support of this is the surprising ambiguity in the technical vocabulary of Buddhism. All too often key terms are vaguely defined, ambiguous, or simply confused. Important words seem to be unhelpfully defined in multiple ways, so that each time the word appears one must pause and consider the context in order to understand. Important basic terms like dharma, vijñāna, and nāmarūpa fit this pattern. If this tradition traces back to a single person, then he was a far from being a philosophical genius. Rather the confusion in terminology argues against a founder figure who defined a jargon. It argues for a jargon that drew on many external sources and developed over time, often in independent groups, and was then roughly pulled together by some encyclopedist urge of the kind we often associate with empires. This may point to the reign of Asoka as a time when Buddhism first coalesced from a number of small, scattered, independent religious communities into a more systematic religion. On the other hand the existence of multiple recensions of the Canon argues for unevenness even in the process of consolidation. The sheer scale of the material that had been produced at this point defeated attempts to organise it into an homogeneous collection. What's more, probably quite quickly, at least two competing collections were created in different languages including Gāndhārī and Sanskrit. Gāndhārī was an important regional dialect; while Sanskrit is the prestige language of the Brahmins that did not come into widespread use until the Brahmin hegemony was well established and Sanskrit the language of the royal courts. However Gandhāra was also a centre for Sanskrit study and learning, being the home of Pāṇini, the most famous grammarian in Indian (and world) history.

It's not simply words that are confused in the suttas. There is the fundamental incompatibility of karma and pratītyasamutpāda. The former demands effects long after conditions have ceased, and the latter forbids it. Karma in the suttas is eternalistic by the standards of the suttas themselves (a point which Nāgārjuna notices even if modern scholars do not); while pratītyasamutpāda would prevent karma from working at all. I've outlined a few of the approaches that emerged to deal this problem in the Abhidharma period and after. Such fundamental metaphysical problems have been routinely ignored in Buddhist studies. The situation is analogous to the problem of skewed reporting in drug trials. Studies sponsored by drug companies have tended to only publish results when the results were positive. If a new drug scored badly, the research was simply not published. In this case meta-analysis would show an overall benefit. This has been particularly true, for example, for antidepressant medications. Many such medications are no better than placebo on average (and sometimes worse) when you take into account unpublished negative studies. While some work has been done comparing sectarian thinking, it seems to gloss over the problems they were trying to solve when they came up with their different ideas. 

Buddhist scholars seem loath to be critical of Buddhism. I notice the same phenomenon in Indian Philosophy generally: the field is still descriptive rather than critical. We don't go looking for major problems of the kind that I have outlined. We tend to cite previous descriptive work that has meticulously ironed out problems in our primary sources as proof of the overall coherence of Buddhist teaching. But even the first sectarian Buddhists did not find the metaphysics of early Buddhism coherent and invented new doctrines to deal with various problems. Written records of sectarian conflict on this issue emerged probably before the first millennium. And yet the question of why Buddhists felt the need to further explain their own doctrines and to argue quite so much as they did between schools of thought seems to be somehow underplayed. As though these ancient intellectual battles are of no contemporary interest. To my mind they are absolutely crucial to understanding the history of Buddhism.

Was the Buddha a Poor Philosopher?

An over-arching question here is: Was the Buddha a poor philosopher who could not articulate a coherent system of thought, even with the restrictions that he himself placed on epistemology? Was he like many of his contemporary disciples an anti-rationalist dedicated to subverting reasoned discussion? Or was the Buddhist tradition a syncretic mishmash from the beginning, in search of unifying narratives (like Siddhartha Gotama) from the beginning and only really finding sufficient coherence after some centuries had passed? Or is there some better answer to the philosophical mess we find in the suttas? 

To me Buddhism makes more sense if the story of the Buddha marks the culmination of a process of assimilating a wide range of cultural influences that is a new synthesis of religious ideas in India. In this we see the same kind of pattern in the advent of Tantra in the 6th century CE. Tantra is a grand synthesis of religious ideas, attitudes and practices that revitalising Indian religion, but older religious ideas are also conserved and propagated alongside. I argue that something like the Tantric synthesis happened in Indian in the 6th - 4th centuries BC, and that it took a long time for doctrines and narratives to settle into the familiar patterns. 

So where does this leave those of us who want to use the suttas as historical sources? It leaves us in a metaphorical minefield. There seem to be two main responses: smooth over and ignore the inconsistencies, or highlight and focus on them. The former argues that, at least to some extent the suttas are history. The latter that there is no history. Some straddle this divide. For example Anālayo has written cogently about the nature of oral traditions. Overall however I'm arguing for a change in method and a revision of all of what has passed for history to date.


Authenticity

I think the facts speak for themselves as long as we consider all of them. I said at the beginning that we would have to address the question of what authenticity means. In fact given the complexity of the problem I doubt a single definition of authenticity would suffice. The question is always authentic with reference to what? Is the Theravāda canon any more (or less) authentic than the other recensions just because by accident of history it survived seemingly intact?  Doesn't the massive amount of internal contradiction cast doubt on authenticity? Apparently for many scholars and traditionalists it does not. If the texts are an authentic record, then what are they an authentic record of? And more importantly when?

If the texts are written down centuries after the time they purport to represent, then are they authentic? Are the Pāḷi texts as authentic as history as, say Shakespeare? We presume they were composed orally and transmitted verbally for some centuries. Probably something like 12-15 generations passed if current guesses for the dates of the Buddha and the dates of writing are correct. There's no evidence that Buddhists used the mnemonic techniques that Brahmins used to accurately record the Vedas. It's just an oral literature passed on from generation to generation.

The different collections seem to reflect separate periods of relatively localised consolidation and stability after periods of sectarian diversity. In other words what precedes the period of canonisation is not unity, but diversity and conflict. What encourages us to think in terms of a big bang is the clearly fictional story of a founder. That story is clearly based in a period in which Brahmanical values are hegemonic, most likely some centuries after the life of the putative founder. We have no way of knowing whether this fiction is based on a true story. Nothing much is left when we strip out the clearly ahistorical details. At present we have no way to probe beyond the event horizon of the texts. And the date when any given text was composed is uncertain by several centuries at least, even if we do have a clue when it was written down.

The position we're in is analogous to trying to understand the history of Christianity when all we had to go on were Augustine's Confessions, Dante's Inferno and Milton's Paradise Lost. They are authentic in their own way and attempt to be true to the Christian Religion, but if we had nothing earlier we'd never reproduce the history that we do have. Our guesses would most likely be wide of the mark. 

~~oOo~~