Two thousand years ago, Buddhist monks rolled up sutras written on birch bark, stuffed them into earthen pots, and buried them in a desert. We don’t know why. They might have been disposing of sacred trash. Maybe they were consecrating a stupa. If they meant to leave a gift for future members of the Buddhist community—a wisdom time capsule, so to speak—they succeeded; and they could never have imagined how great that gift would turn out to be.
Fragments of those manuscripts, recently surfaced, are today stoking a revolution in scholars’ understanding of early Buddhist history, shattering false premises that have shaped Buddhism’s development for millennia and undermining the historical bases for Buddhist sectarianism. As the implications of these findings ripple out from academia into the Buddhist community, they may well blow away outdated, parochial barriers between traditions and help bring Buddhism into line with the pluralistic climate of our times.
Sometime probably around 1994, looters unearthed 29 birch bark scrolls somewhere in eastern Afghanistan or northwest Pakistan, an area once known as Gandhara—a Buddhist cultural hotspot during the early Christian era. The scrolls appeared on the antiquities market in Peshawar, having weathered the same turbulent political climate that would lead to the Taliban’s demolition of the Bamiyan Buddhas. The British Library acquired them in 1994.
The scrolls arrived rolled up, flattened, folded, and disintegrating. Curators carefully unpacked and examined them. They found the script indecipherable, the language unusual. Suspecting that they might in fact be written in the forgotten language of Gandhari, they immediately sent a photograph to Richard Salomon, a professor of Sanskrit and Buddhist studies at the University of Washington, one of a handful of early Buddhist language experts worldwide who could read Gandhari.
The news soon came that the birch bark scrolls were the oldest Buddhist manuscripts known. (Now called the British Library Collection, these scrolls are in the process of being translated by the Early Buddhist Manuscript Project, a team of scholars under Salomon’s direction.) The initial find was followed by several others throughout the following decade. Today there are at least five collections worldwide, comprising roughly a hundred texts and several hundred text fragments dating from the first century B.C.E. to the third century C.E. The Gandharan collections are not only the oldest extant Buddhist manuscripts but also the oldest surviving manuscripts of South Asia, period. They reach back into an era when the oral tradition of Buddhism probably first began to be written down.
Preliminary inventories and initial translations reveal that many texts are Gandhari versions of previously known Buddhist material, but most are new—including never-before-seen Abhidharma (Buddhist philosophy) treatises and commentaries, and stories set in contemporary Gandhara. The collections contain the earliest known Prajnaparamita (Perfection of Wisdom) texts and the earliest textual references to the Mahayana school, both first century C.E. Taken together, these scrolls and scroll fragments are a stunning find: an entirely new strand of Buddhist literature.
According to experts in Gandhari, the new material is unlikely to reveal earth-shattering facts about the Buddha. And don’t expect big surprises in terms of new doctrine either—no fifth noble truth is likely to be found. But the discovery of a new member in the Buddhist canonical family has profound implications for practitioners. It settles the principal justification for long-standing sibling rivalries among Buddhist traditions, and it does so not by revealing a winner but by upending the cornerstone—a false paradigm of history—on which such rivalries are based.
Buddhist tradition maintains that after his awakening, the Buddha taught for some 45 years throughout eastern India. Among his disciples were a few, including his attendant Ananda, who had highly trained memories and could repeat his words verbatim. It is said that after the Buddha’s death, his disciples gathered at what we now call the First Council, and these memorizers recited what they had heard. Then all the monks repeated it, and the single and definitive record of the “words of the Buddha” [buddhavacana] was established. Thus was the Buddhist canon born.
Or was it?
Every school of Buddhism stakes its authority, and indeed its very identity, on its historical connection to this original first canon. Buddhists of all traditions have imagined that our texts tumble from the First Council into our own hands whole and complete—pristine—unshaped by human agency in their journey through time. This sense of the past is deeply ingrained and compelling. If our texts don’t faithfully preserve the actual words of the Buddha in this way, we might think, how could they be reliable? Isn’t that what we base our faith on?
But as we’re about to see, history works otherwise. And having a view more in line with the facts here frees us from chauvinist views and gives us grounds for respecting differences between and within diverse Buddhist schools. As for undermining our basis for faith, not to worry. To get in line with the facts, we’re not going to abandon Manjushri’s sword of wisdom. We’re going to use it.
I first heard about the Gandharan manuscripts while living in Germany in 2009, when I attended a lecture on early Buddhism by Professor Salomon, who was visiting from Seattle. The complex details of the talk he delivered left me mystified—at that point the technicalities of early Indian philology stood as a dense forest I hadn’t yet entered. But I was curious about those scrolls. I wanted to understand what this new literary tradition meant for Buddhist practitioners like me.
While searching online, I found a 2006 talk by Salomon in which he first unveiled for a general audience the importance of translators’ findings. Toward the end of that talk, my attention became riveted. As Salomon was explaining, scholars had traditionally expected that if they traced the various branches of the tree of Buddhist textual history back far enough, they would arrive at the single ancestral root. To illustrate this model, he pointed to a chart projected on the screen behind him. The chart showed the Gandhari canon as the potential missing link along an evolutionary ladder—the hypothetical antecedent of all other Buddhist canons. “This is how someone who began to study this [Gandharan] material might have thought the pattern worked.”
As scholars scrutinized the Gandhari texts, however, they saw that history didn’t work that way at all, Salomon said. It was a mistake to assume that the foundation of Buddhist textual tradition was singular, that if you followed the genealogical branches back far enough into the past they would eventually converge. Traced back in time, the genealogical branches diverged and intertwined in such complex relationships that the model of a tree broke down completely. The picture looked more like a tangled bush, he reported.
Here is where I clicked Rewind: these newly found manuscripts, he declared, strike the coup de grâce to a traditional conception of Buddhism’s past that has been disintegrating for decades. It is now clear that none of the existing Buddhist collections of early Indian scriptures—not the Pali, Sanskrit, Chinese, nor even the Gandhari—“can be privileged as the most authentic or original words of the Buddha.”
It is odd how matters enacted on the wide stage of history can sometimes present themselves immediately in the close corners of personal life. I am a Mahayana practitioner; my partner practices in the Theravada tradition. The challenge of accommodating differences in the Buddhist family is an occasional cloud that hovers over our dinner table. What Salomon was saying seemed to indicate a new way of viewing and working with sectarian clashes at whatever level they might occur.
Puzzling out whether (and how) the discovery of a new Buddhist literary tradition could undermine sectarian sparring would lead me deep into the foreign terrain of academic Buddhism. In the months to come, I would follow a trail from one expert to another across college campuses from Seattle to Palo Alto. I pored over stacks of papers looking for insights. In the end, when it all came clear, I understood why the process had been so difficult. I had to assimilate new facts. I had to let go of some cherished beliefs. But what really made it hard was that also I had to identify and change a fundamental background picture I had about the nature of Buddhist history within which I construed those beliefs and assimilated those facts. I had to cut down the genealogical tree. And that was not easy, because I was sitting in it.
Actually, it isn’t just historians of Buddhism who are finding flaws in convergence-to-a-single-root pictures of the past. The evolutionary tree model of origins is also under the axe in biology and other scholastic fields. For some time there has been a broad trend of thinking away from tree models of history, Salomon later told me. In the academic study of early Buddhist history, Salomon says, this model had been gradually being discredited. But, he says, these scrolls were “the clincher.”
Because early Buddhism was an oral tradition, tracking any Buddhist text back in time is like following a trail of bread crumbs that ends abruptly. So for us looking to the past, a critical moment in history occurred when Buddhists started writing down their texts rather than transmitting them orally. That is when the Buddha’s words moved into a more enduring form.
Pali tradition reports that Buddhist monks in the Theravada tradition started writing down texts in about the first century B.C.E. The manuscript record in Pali, however, doesn’t begin until about 800 C.E. But the Gandhari manuscripts date from as early as the first century B.C.E. If monks were writing in one part of India, they could likely have been writing in other parts of India as well—so this would seem to add credence to the Pali claims.
If we were looking for a single ancestral root of all Buddhist canons, the moment the teachings got written down would be the first possible point in time we could find their physical record. So when these Gandhari scrolls appeared, dating to the earliest written era of Buddhism, scholars hoped they might turn out to be that missing link. They zeroed in on the Gandhari literature that had known versions in Pali, Sanskrit, and Chinese to see how texts preserved in Gandhari related to other early Buddhist texts. Comparing individual texts across canons, they noticed something startling and surprising, “although in retrospect,” Salomon admitted in his lecture, “it should have been expected, and it makes perfect sense.”
Salomon described what happened when he compared the Gandhari version of one well-known Buddhist poem, the Rhinoceros Sutra, to its Pali and Sanskrit versions. He found that the sequence of verses and their arrangement were similar to the Pali. The specific wording of the poem, however, was much closer to the Sanskrit. Salomon couldn’t say whether the Gandhari was more closely related to one or the other version (as it would have to be if one were the parent). It was closely related to both, but in different ways. In other words, the texts were parallel—and different.
This kind of complex linking showed up again and again when scholars compared Gandhari texts with their versions in Pali, Sanskrit, and Chinese. Texts had close parallels to one, two, and sometimes all three of the other language versions. Looking then at the group as a whole, they ascertained that this new corpus of Gandhari material was a parallel to, and not an antecedent of, the other canons—not the missing parent, but a long-lost sibling.
We now know that if there ever was a point of convergence in the Buddhist family tree—the missing link, the single original and authentic Buddhist canon—it is physically lost in the era of oral transmission. We have not yet found, and probably will not ever find, evidence for it.
But even more significant is what we have found: that is, difference. These scrolls are incontrovertible proof that as early as the first century B.C.E., there was another significant living Buddhist tradition in a separate region of India and in an entirely different language from the tradition preserved in Pali.
“And where there are two, we are now on very solid ground in suggesting there were many more than two,” says Collett Cox, a professor of Sanskrit and Buddhist studies at the University of Washington and the co-director of the Early Buddhist Manuscript Project. A single partial Gandhari Buddhist manuscript predated these modern finds—a version of the Dharmapada discovered in 1892. The fact of one extant manuscript in the Gandhari language suggested, but couldn’t prove, that Gandhara had once had a rich literary tradition. In the same way, there are other indicators—such as monuments and inscriptions—in other parts of India suggesting other potentially literate early Buddhist cultures. “We don’t have any texts from them,” Cox says. “But we now are on very solid ground in saying they probably had texts too. Where there are two [traditions], there are probably five. And where there are five, there may have been fifteen or twenty-five.”
Cox suggests that “rather than asking the question what single language did the Buddha use and what represents the earliest version of his teachings, we might have to accept that from the very beginning there were various accounts of his teachings, different sutras, and different versions of sutras transmitted in different areas. At the very beginning we might have a number of different sources, all of whom represent or claim to represent the teaching of the Buddha.” Cox emphasizes that the Gandharan Buddhism is clearly not a “rebel offshoot” of the Pali canon but its own entirely localized strand—unique, but not unrelated. Early Buddhists in different regions shared many texts in common. Clearly, Buddhist monks of different language traditions in early India were in contact, and they traded ideas and influenced each other in complex ways.
If a multiplicity of traditions is what we have now, and as far as the record goes back in time, multiplicity is what we’ve always had, maybe we’re not finding a single root of Buddhism because there wasn’t one in the first place. Sometimes not-finding is, after all, the supreme finding.
“Nobody holds the view of an original canon anymore,” Oskar von Hinüber, one of the world’s leading scholars of Pali, told me.
Consider why scholars might think this. First of all, there are certain practical difficulties of oral transmission in a time before digital recording. How could 500 monks have agreed on 45 years of the Buddha’s words?
Von Hinüber also points out that the sutras themselves record a deep and persistent quarrel between the Buddha’s attendant, Ananda, and Mahakasyapa, who presided over the Council and was the principal disciple at the time of the Buddha’s death. He suggests that it would be Pollyannaish to imagine that the Council (if it even occurred) was politic-free and harmonious.
“There are many indications that [the stories of the First Council] are not correct in the way of a historical report. But they tell us something that is interesting and important,” says von Hinüber. “Buddhists themselves were aware of the fact that at some point in history their texts must have been shaped by somebody into the standard form they now have, beginning Thus have I heard. Who this was, we don’t know.”
Interestingly, built into the traditional account of the First Council is the story of one monk who arrived late. He asked the others what he had missed. When they told him how they had formalized the Buddha’s teachings, he objected. He insisted that he himself had heard the Buddha’s discourses and would continue to remember them as he had heard them.
“This is a very important story,” says von Hinüber, “because it shows that Buddhists themselves were aware of the fact of diverging traditions.”
Religious orthodoxy wants to claim that one’s own tradition is the best. To do that, one needs to point to something unique to make it so. Having the sole true version of a singular truth is just such a foothold. And not only for Buddhists. Elaine Pagels, the scholar of religion who brought to light the Gnostic gospels, told Tricycle in 2005:
The Church father Tertullian said, Christ taught one single thing, and that’s what we teach, and that is what is in the creed. But he’s writing this in the year 180 in North Africa, and what he says Christ taught would never fit in the mouth of a rabbi, such as Jesus, in first-century Judea. For a historically-based tradition—like Christianity, and as you say, Buddhism—there’s a huge stake in the claim that what it teaches goes back to a specific revelation, person, or event, and there is a strong tendency to deny the reality of constant innovation, choice, and change.
The Buddhist canons as they exist today are the products of historical contingencies. They resound with the many voices that have shaped them through time. But orthodoxy requires the opposite, a wall you can’t put your fist through: singular, unchanging, findable truth. Buddhism’s textual root wasn’t singular, and it wasn’t unchanging. As it turns out, it wasn’t so findable, either.
“That’s the further step that we’re taking, to dispense with the idea of the original because that is a kind of pipe dream or figment of the imagination,” says Paul Harrison, a professor of religious studies at Stanford University and a member of the editorial board for the Schøyen Collection (another recently discovered collection of ancient Buddhist manuscripts). Harrison is also a translator. As such, he gives us a hands-on report of how texts weather the practicalities of translation. To the extent that we are still holding onto that tree model, Harrison is about to pull the last leaves from our hands. Translators used to be guided by the notion, he explains, that if you put enough different versions of a sutra together, kept the overlap, and eliminated all the variance, eventually you could reconstruct the prototype. “According to that model,” he says, “it’ll all narrow to a point. But basically what we are finding is that it doesn’t narrow to a point. The more we know, the more varied and indeterminate it is right at the beginning.” Trying to reconstruct the original version of any early sutra—the one that is unmediated, accurate, and complete—is now generally considered, in principle, futile. Indeed, Harrison asks, “What are you aiming at?” Looking for such an original is ingrained, essentialist thinking, he says.
He points out, “We often say, ‘Tibetan translation, Chinese translation, Sanskrit original. As soon as you say Sanskrit original, you drop back into that sloppy but entirely natural way of thinking, that this is the original so we can throw away the copies. But in fact, that Sanskrit original of whatever sutra is just again another version. So the idea that one of them is the original and all the others are more or less imperfect shadows of it has to be given up. But it is very hard to give it up. It’s almost impossible to give it up.” And the irony is not lost on Harrison, who adds, “This is what the teaching of the Buddha is all about.”
One problem with the traditional model of textual transmission, according to Harrison, is that it doesn’t take into account cross-influences—the very real cases of text conflation when scribes or translators might have (for example, when standardizing) copied features from multiple differing versions, thus producing a new version. He continues: “If everything just proceeds in its own vertical line, and there is no crossways influence, that is fine; you know where you are. But once things start flowing horizontally, you get a real mess. Having something old, of course, is valuable because you are more likely to be closer to an earlier form. But notice I’m careful to say now ‘an earlier form’ and not ‘the earliest form.’ A first-century B.C.E. [Gandhari] manuscript is going to give you a better guide to an earlier form than an 18th-century Sri Lankan copy will. But that’s not an absolute guarantee, just a slightly better one.”
Harrison says that not only is it physically unlikely that we could find an original Buddhist canon (because the teachings predated writing), but also it is theoretically impossible, according to the Buddha’s own teachings on the nature of reality. “It is pure anatmavada [the doctrine of nonself, non-essentialism]. We expect it [the original buddhavacana] to be the same—invariable and unchanging, kind of crisp and sharp at the sides all around.” That is, after all, the kind of canon that Buddhists who make historical claims to authenticity—and all Buddhist schools have traditionally made such claims and based their authority on them—believe their tradition possesses or other traditions lack: not a “one-of-many-versions” canon but “the real one.”
“It’s just not going to be like that,” Harrison says.
What would it mean to have “all the Buddha’s teachings?” Would it be every word he said? What about meaningful silences? Well, would it be what he meant then? When he said what to whom? About what? We can’t pin down the complete content of the Buddha’s teachings, nor can we isolate the teachings from their context. We can’t draw a hard line around them.
Neither can we draw a solid line around different schools. Harrison reports that looking backward in time, already by the first century C.E. boundaries between the Mahayana and non-Mahayana begin to blur. The Gandhari manuscripts probably reflect content of early monastic libraries, and the texts seem to have been intentionally buried. Mahayana and mainstream Buddhist sutras were recovered together and presumably buried together. Harrison believes that the monks who engaged in Mahayana practices were most likely Vinaya-observing; they likely lived in monasteries side by side practitioners of more mainstream Buddhism.
These first-century Mahayana texts in the new collections are already highly developed in terms of narrative complexity and Mahayana doctrine. They couldn’t be the first Mahayana sutras, Harrison says. “The earlier stages of the Mahayana go far back. The Mahayana has longer roots and older roots than we thought before.” (Not roots all the way back to the Buddha, though—Harrison agrees with the general scholarly consensus that the Mahayana developed after the Buddha.) Nonetheless, he says, “Probably lying behind these Mahayana texts there are others with much stronger mainstream coloration, where it is not so easy to tell whether it’s Mahayana or Shravakayana.” [Shravakayana means literally ‘the way of the hearers’; those who follow the path with arahantship as its goal.]
During this period of early Buddhism there were many different strands of practice and trends of thought that were not yet linked. “We could have the Perfection of Wisdom strand and a Pure Land strand and a worship of the Buddha strand, and all sorts of things going on,” Harrison remarks. Only later did these threads coalesce into what we now consider “the Mahayana.”
Harrison suggested we consider a braided river as a better metaphor than a tree for the historical development of Buddhist traditions. A braided river has a number of strands that fan out and reunite. “Its origin is not one spring, but a marsh or a network of small feeder streams,” he told me. According to this model, the Mahayana and Vajrayana “are merely downstream in the onward flow of creativity. They are activities similar in nature to early Buddhism—not radically different. And a lot of current in their channels has come all the way from the headwaters,” he says. “Whether it all has the single taste of liberation is another question.”
In such a picture of textual transmission—fluid, dynamic, and intermingled—where and how could one stake a territorial claim? Sectarian posturing is based on having the actual words of the Buddha—complete, stable, unmediated, and self-contained. Once all one can have is a complex of versions of the Buddha’s words—partial, changing, shaped, and commingled with other versions—in what sense would it be authoritative if one’s own version was bottled upstream or down?
But I still wanted to drink my water bottled upstream even though I knew that kind of thinking no longer made sense. I couldn’t put my finger on what was bothering me. Finally, I looked inside my glass. What did I assume was in it? What do we imagine we have when we have the Buddha’s words?
We think that if we have the Buddha’s actual words we have his true intent. The whole edifice of sectarian claims based on history remained teetering on this.
Somehow we picture the Buddha’s true, single, unambiguous meaning encapsulated in his words like jewels inside a box, passed from one generation to the next like Grandmother’s heirlooms. But that’s not the way meanings or words work. Consider the following from the well-known scholar of religion Robert Bellah:
Zen Buddhism began in Japan at a time when strong social structures hemmed in individuals on every side. The family you were born to determined most of your life-chances. Buddhism was a way to step outside these constricting structures. Becoming a monk was called shukke, literally, “leaving the family.” We live in an almost completely opposite kind of society, where all institutions are weak and the family is in shambles. You don’t need Buddhism to “leave the family.” To emphasize primarily the individualistic side of Buddhism (especially Zen) in America is only to contribute to our pathology, not ameliorate it.
In India, “leaving the family” means “getting married.” To my Jewish grandmother, it meant “changing religions.” In the household where I was raised, it meant “going to college.” The very same words, spoken in a different context, have different meanings. The meaning of words is their use in context. A set of words stripped of their context is like playing pieces stripped of their board game. What would we have? Certainly it would be good to know what the Buddha said. To the extent that we share the conventions of 5th-century B.C.E. Indians, we might understand some of what he meant. If we increased the conventions we shared with them (say, by learning early Indian languages or by studying history), obviously we would understand more. But context is vast—an unbounded, interdependent web of connections. And it is dynamic, shifting moment to moment. Context is finished the moment it happens; then it is a new context. We really can’t recreate it. And even if we could, we still wouldn’t know exactly how the Buddha was using his words within that context, so we wouldn’t know exactly what he meant.
Just as our search for an original set of Buddha’s definitive words failed, and all we were left with were provisional versions, in the same way a search for the Buddha’s definitive meaning fails too. What we have are traditions of interpretation. But that’s not the kind of authority we imagine when we claim sectarian primacy. Sectarian authority claims assume solid essentialist ground. That type of ground is just not there.
When it comes right down to it, sectarian posturing contradicts the Buddha’s message as all traditions understand it. Those false pictures of history and language within which sectarianism finds a foothold are in turn rooted in another false picture—a picture even more pervasive and pernicious. That picture is an essentialist view of the nature of reality, which according to the Buddha’s doctrine of selflessness is the source of not just this but all our suffering—the wrong view that is the very point of Buddhism to refute.
The siblings in my family don’t have a single, same, enduring, essential feature in common that connects us to each other (or to our ancestors), nor do we need one. Anyone could pick us out of a crowd as related. I have my father’s nose and my aunt’s height; my sister has my grandmother’s hair and my father’s fast walk; my brother looks like my father and me. The traditions of the Buddhist family can dress, think, and practice differently and still be recognizable family members in exactly the same way in which the members of our own family are recognizably related to us.
All the siblings in my family are authentic members of my family. Because our identity doesn’t depend on our possessing some unchanging “common thing,” we don’t have to argue over who has more of it. If we understand identity in this way, all Buddhists are 100 percent Buddhist.
Letting go of our old assumptions about history and language shouldn’t make us uneasy. The views we’re challenging as we assimilate these new archaeological discoveries were never Buddhist to begin with. We’re not abandoning the basis for our faith; we’re confirming it. And in so doing, we open up the possibility to truly appreciate different Buddhist traditions as equal members of our Buddhist family.
Start your day with a fresh perspective
Thank you for subscribing to Tricycle! As a nonprofit, we depend on readers like you to keep Buddhist teachings and practices widely available.