The Diamond Sutra

Seven sheets of paper, stained yellow, each printed from a carved wooden block and then pasted end to end to make a continuous roll approximately five metres long. Six carry the text of the sutra in Chinese; the first carries a printed illustration.

The frontispiece shows the Buddha Sakyamuni seated and preaching to the elder Subhuti, who kneels before him, surrounded by monks, guardian figures and attendants. It is a picture printed from a block, which is a considerable thing in 868: the image is not an afterthought or a hand-added decoration but part of the same industrial act as the text, cut, inked and pulled in the same way.

The scroll form matters as much as the printing. This is not a book in the sense of leaves bound at a spine. It is a roll, read by unwinding, and the colophon sits at the inner end, which is to say at the place a reader reaches last, after the whole text has passed through their hands.

The Diamond Sutra is a short work of the Perfection of Wisdom literature, in the Chinese translation associated with Kumarajiva. Its argument is that all appearances are provisional, that the categories a mind uses to grasp reality are themselves empty, and that clinging to any of them, including to the idea of merit, is the error.

It also does something more immediately relevant to this object: it repeatedly declares that the merit generated by copying and transmitting the text exceeds the merit of material offerings. That instruction is why the sutra survives in such quantity from Dunhuang, and it is why this particular copy exists in this particular form.

Wang Jie did not print one copy for himself. The colophon says the scroll was made for universal free distribution, on behalf of his two parents. Woodblock printing is here not a commercial technology looking for an application; it is a devotional technology, and it is doing exactly what the text told the donor to do. The world's earliest dated printed book is a religious giveaway, and the fact that mass reproduction was put to work for merit rather than for profit is among the most interesting things about it.

The British Library's phrasing is the earliest complete survival of a dated printed book, and every qualifier is load-bearing.

Printing on paper is older. The Korean Pure Light Dharani Sutra found in the Seokgatap pagoda at Bulguksa is generally placed in the first half of the eighth century but carries no date. The Japanese Hyakumanto darani, the million printed charms commissioned by Empress Shotoku, are datable by court record to around 770 but are single sheets rather than books, and their dating rests on the record rather than on the objects. Fragments of dated Chinese printing survive from before 868.

What no other object supplies is all four conditions at once: printed, a book, complete, and dated on itself. That is a narrower claim than the popular one, the world's oldest printed book, and this record uses the narrow version, because the popular version is what starts the arguments between the Chinese, Korean and Japanese claims and none of those arguments are about this scroll.

Cave 17 at Mogao, near Dunhuang in Gansu, is a small side chamber opening off the corridor of the larger Cave 16. At some point around the beginning of the eleventh century it was filled with manuscripts, paintings and textiles, tens of thousands of items, then walled up and plastered over. Why it was sealed is still argued: a response to an approaching threat, or the orderly disposal of a monastic library's superseded and damaged copies.

It stayed shut for roughly nine hundred years. On 25 June 1900 Wang Yuanlu, an itinerant Daoist priest who had appointed himself caretaker of the derelict Buddhist cave complex and was raising money to restore it, noticed a crack in the plastered corridor and opened the chamber.

What he did next is the part usually left out. He took specimens to the Gansu provincial authorities in the hope of selling them, or at least of proving that the caves were worth funding. The late-Qing administration did not act on the find. The governor-general's response was an order that the documents be locked up in the cave to prevent theft, with no money and no staff attached. That is the position Wang was in when Stein arrived.

Marc Aurel Stein, a Hungarian-born archaeologist working in the service of British India, reached Dunhuang in 1907. His first transaction with Wang Yuanlu ran between 21 May and 6 June. He obtained roughly ten thousand items for the equivalent of about 130 pounds sterling, framed as a donation towards the restoration of the temple.

Note what that does to the market history of this object. There is no price for the Diamond Sutra. It was not itemised, valued or negotiated over as an individual thing; it came out of the cave inside a bulk consignment, and it has never once been offered for sale in the twelve centuries since Wang Jie paid for it. DERIVED, and given only to show the shape of the transaction rather than any real valuation: 130 pounds across ten thousand items is roughly threepence an item in the money of 1907. That figure is arithmetic, not a price, and no source states it.

Stein could not read Chinese. The selection was made with the help of his Chinese secretary, and the sheer volume meant that neither man knew in detail what was being taken. Nobody present understood that the most important printed object in the world was in the consignment.

Paul Pelliot followed in 1908 and, being a first-rate sinologist, selected far more knowingly and took fewer items of higher scholarly value. In 1910 the Qing court finally ordered the remainder moved to Beijing, and material was pilfered en route. The dispersal was therefore not one event but at least three, and only the first two were foreign.

Chinese scholarship on the dispersal is not neutral and does not pretend to be; the standard framing treats the removals as theft dressed as purchase, and the literature carries titles to match. The argument is not that no money changed hands. It is that Wang Yuanlu had no title to convey and no authority to sell, that the sum was derisory against what was taken, that Stein represented an imperial power operating inside a collapsing state, and that a payment described as a donation to a temple is a transaction with the form of a gift and the effect of an export.

Against this stands a revisionist reading, associated with the historian Justin Jacobs, which treats Wang and the other Chinese participants as rational actors making defensible decisions in the circumstances they actually faced, rather than as dupes or traitors. Wang had told the authorities. The authorities had done nothing. He spent what he received on the caves.

Both readings can be held at once and this record holds them. The acquisition was legal by no clear law, was made in circumstances of gross power asymmetry, was preceded by a genuine and ignored attempt to interest the Chinese state, and produced the best-conserved corpus of Dunhuang material anywhere. None of those four statements cancels the others.

The scroll has been worked on twice in London, and the second campaign was in substantial part a response to the first.

Early twentieth-century practice at the British Museum was to back and mount Dunhuang scrolls to make them handleable and displayable, an intervention that stabilised objects in the short term and locked them into a structure later conservators judged inappropriate. The modern campaign was led by Mark Barnard of the British Library's conservation section and published with the sinologist Frances Wood in 2010. It ran for years - reported as seven years to 2010, and as a decade-long project completed in January 2013 - and involved reversing earlier treatment and remounting the scroll panel by panel.

The conservation is itself a provenance event, and it is the argument the holding institution makes for itself: that the object survives in this condition because of what London did to it and spent on it. That argument is real. It is also the standard argument of every retaining museum, and it is not a title.

The British Library's answer to the Dunhuang question for the last thirty years has effectively been the International Dunhuang Project: photograph everything, publish it free, and reunite in data what cannot be reunited in fact. Material scattered across London, Paris, Beijing, St Petersburg and Tokyo can be read side by side by anyone with a connection, which is more than could be said for it in any period when it was whole.

In October 2023 the British Library was hit by a cyber-attack that took down its online services. As of August 2026 the picture is split. The Library's own bl.uk item pages for named collection objects, including this one, still carry a notice that only a limited amount of digital content is available and the interpretative page for the Diamond Sutra does not resolve to actual content. But the International Dunhuang Programme itself - the specialist portal that actually hosts the digitised images and catalogue data, distinct from bl.uk's general site - is substantially restored: its record for Or.8210/P.2 loads with full catalogue metadata and a working image viewer showing 71 of 73 photographed folios, and IDP reports 154,919 catalogue entries and 114,332 digitised items in its database as of 1 January 2026. Wikimedia Commons file pages for British Library images still carry a standing warning that links to the Library's website may be broken while it recovers from the attack.

That is a fact about infrastructure and it is also a fact about the argument. A digital surrogate offered in place of return is a promise of permanent access, and permanent access turns out to be contingent on one institution's servers.

A small discrepancy runs through the literature and is worth pinning down, because it is the kind of thing that makes one record look like two objects.

The translation attached to the British Library's own image gives the 13th of the fourth moon of the ninth year of Xiantong. Other published renderings, including the standard encyclopaedic one, give the 15th day of the fourth month. Both are then converted to 11 May 868, and the conversion is not disputed.

This record reproduces both readings and arbitrates neither. It matters because a reader comparing two accounts should know they are looking at one scroll with one colophon and a variant transcription, not at two printings a fortnight apart.

There is no auction record, no insurance figure in the public domain, no probate value and no offer. The object has been in national ownership in Britain since 1907, first through the British Museum and then, from 1973, through the British Library, and British national collections do not deaccession into the market.

Anyone reaching for a comparable should be careful about what the comparison establishes. Manuscript and early-printing records at auction establish what the market pays for an object that can be sold. This one cannot be, which makes any figure a thought experiment rather than a valuation. The useful diligence statement is the opposite one: no Dunhuang material of this rank is legitimately available, and anything of comparable claim appearing on the market should be treated as a fragment, a forgery or a title problem until proved otherwise.