At the very beginning of the twentieth century a huge cache of ancient manuscripts was discovered in a Buddhist cave complex near the desert town of Dunhuang in China. Famously, the monk who guarded the caves, Wang Yuanlu, was persuaded by the archeological explorers Aurel Stein and Paul Pelliot to sell them a large portion of the manuscripts so they could be acquired by the British and French governments. The manuscripts, which were in a variety of languages, the most common of which were Chinese and Tibetan, are now held at the Bibliothèque Nationale and the British Library, and it is the Tibetan group that will concern us here. The antiquity of the Tibetan Dunhuang manuscripts, which are estimated to date from the eighth and ninth centuries AD, has made the Paris and London collections especially valuable to historians studying this significant period of Tibetan history when the Tibetan kings extended the borders of their country far into Central Asia, to scholars studying the origins of Tibetan Buddhism, and to linguists studying the early development of the Tibetan language.
The original number of Tibetan manuscripts in the library cave
What Stein and Pelliot took from the cave
Suivant des informations que nous avons reçues postérieurement à cette lettre, M. Pelliot a pu aquérir définitivement tous les documents chinois, brahmī, ouïgours, tibétains dont il y est parlé, à l’exception des kia-pan du Kandjur, dont il rapporte cependent trois volumes.Pelliot’s volumes, like the large pages from the miscellaneous bundles acquired by Stein, turned out to contain only numerous copies of sūtras from the Prajñāpāramitā group, especially the largest, the Śatasāhasrikā.. Along with these volumes Pelliot took many more bundles of Tibetan scrolls, the majority of which, like the scrolls taken by Stein, contained the repeated text of the Aparimitāyur-nāma sūtra. He also took most of the remaining miscellaneous manuscript bundles, many of which contained Tibetan pothī pages, and about a third of the remaining bundles of scrolls.  Pelliot also discovered more Tibetan manuscripts, which he guessed to be from the thirteenth to fourteenth centuries, in two caves in the northern part of the Dunhuang cave complex decorated with Tibetan style murals with tantric subject-matter. 
What was taken afterwards
The Dunhuang Tibetan manuscripts which remained in Gansu
At last, in 1919, the provincial government of Kansu, having heard that a traveller had bought many Buddhist scriptures in Tibetan, sent an inspector to examine the situation on the spot. In a Buddhist cave to the south side of the second floor of the three-storied building, the inspector found 94 bundles of scrolls in Tibetan weighing 405 chin and 11 stacks of sheets between wooden boards weighing 1744 chin. He left 90 bundles and moved three bundles of scrolls and ten stacks of sheets to a school in Tunhuang, carrying away one bundle and one stack of scriptures to Lanchow, where they were placed in the provincial library. It is not superfluous to emphasise that the weight of the scrolls remaining at Tunhuang was probably more than one ton, no doubt exceeding that of any other collection. Fujieda’s source, Jiang Liangfu, did not make any reference to the source for this report, and Fujieda could add nothing further to this information. According to this report the official found 94 bundles of scrolls and eleven pothī volumes with wooden boards; he moved ten of the volumes and three scroll bundles to a school in Dunhuang, and one volume and one scroll bundle to Lanzhou. 
The numbers of the volumes and bundles given in this report conflict with Stein’s and Pelliot’s accounts. As we have seen, Stein counted 80 bundles of Tibetan scrolls and eleven Tibetan volumes when he arrived. Many scroll bundles were taken by Stein and Pelliot, and three pothī volumes were taken by Pelliot. Fujieda argues that the “volumes” mentioned in the passage above were merely a portion of the contents of a single large volume. However, as I will show below, there is reason to doubt this account and to believe that Pelliot did indeed come away with three of the original eleven volumes.
In 1928, all of the Tibetan manuscripts remaining in Dunhuang were moved to the local cultural institute, where they remained until the Communist revolution of 1949. After the revolution, some manuscripts were brought to the newly-founded Dunhuang Academy (Dunhuangwenwu Yangjiusuo), an organisation established to conserve the caves and foster research on their contents, while a few others were sent to provincial museums in Gansu. The great majority of the Tibetan manuscripts, a collection still comparable in size to those of Paris and London, were placed in the new provincial museum for the town of Dunhuang, the Dunhuang Museum (Dunhuangxian Wenhuaguan).
The Gansu collections
|Dunhuang Museum (Dunhuangxian Wenhuaguan)||8780||224|
|Dunhuang Academy (Dunhuangwenwu Yangjiusuo)||42||43|
|Jiuquan Museum (Jiuquanxian Wenhuaguan)||0||19|
|Zhangye Museum (Zhangyexian Wenhuaguan)||0||1|
|Wuwei Museum (Wuweixian Wenguanhui)||7||0|
|Lanzhou Library (Lanzhou Tushuguan)||1117||30|
According to Huang, the great majority of these manuscripts contained the Śatasāhasrikā-prajñāpāramitā sūtra and the Aparimitāyur-nāma sūtra. In addition, he mentions the Pañcaviṃśatisāhasrikā, Aṣṭasāhasrikā, and Sañcayagāthā.  Some Dunhuang Tibetan manuscripts are held elsewhere in China as well. There are between 200 and 300 Tibetan Dunhuang items at the National Library of China, in Beijing, in both scroll and pothī form.  Certain other institutions in China hold a few Tibetan Dunhuang manuscripts; generally these are Aparimitāyur-nāma sūtra scrolls and a few stray pothi leaves. They are the Beijing University Library, Shanghai Library, Shanghai Museum, and Tianjin Arts Museum. The Beijing University Library acquired most of its Tibetan manuscripts in the 1950s from collectors, and this is also likely to be the case with the other institutions listed above.  Finally, In Taiwan, the National Central Library of Taipei has four Tibetan Dunhuang Buddhist scrolls, all containing the Aparimitāyur-nāma sūtra.
Conclusions: the fate of the original contents of the library cave
The Pelliot tibétain collection contains over 3800 large-format leaves of the Śatasāhasrikā-prajñāpāramitā sūtra (excluding leaves in fragmentary form). When held compressed together, about 80 leaves of this kind of Dunhuang pothī fit into an inch, and, according to Stein, the original volumes were just under one and a half feet (18″) high. So the most we might expect each volume to have contained is 1440 leaves. If we estimate that the wooden cover-boards were each an inch in depth, then the number would be closer to 1280. Therefore, the number of pages in the Paris collection is exactly what we would expect if Pelliot did take three whole volumes.
Furthermore, we can reasonably assume that the 1117 leaves at Lanzhou are indeed the single volume which was said to have been taken there, lacking a hundred or so pages which may be the pages in the Dunhuang Academy. The 8780 leaves in the Dunhuang Museum look very much like seven volumes. Thus it seems most likely that out of the original eleven volumes, three went to Paris, one to Lanzhou, and the remaining seven to the Dunhuang Museum. This, of course, contradicts the version of the report we have from the official who visited Dunhuang in 1919. If our version is correct, it means that all of the eleven volumes are accounted for, and accessible to scholars.
The contents of these eleven volumes seem to be an early proto-canonical collection of Prajñāpāramitā sūtras. Unfortunately, contrary to Pelliot’s hopes, they are not in sequential order; the pages in the two largest collections, the Bibliothèque Nationale and the Dunhuang Museum, are very disordered. Many of the pages carry page numbers and letters indicating the original volumes to which they belonged, and these are often non-sequential. Furthermore, there is repetition of the same volume and page numbers on different pages containing different text. This suggests the presence of more than one sūtra collection, or multiple drafts of a single collection. Many of the pages in the Bibliothèque Nationale are extensively corrected, and Marcelle Lalou believed that these were draft versions which formed the basis for the later, corrected pages which are found in the same collection. 
The reports of Stein, Pelliot and the Chinese official also conflict regarding the scrolls, and it is more difficult to arrive at a conclusion in this case. The Paris and London collections contain by far the largest number of scrolls known to us. The majority of the scrolls contain the Aparimitāyur-nāma sūtra; these are of various lengths because the text of this short sūtra may be repeated many times on the same scroll. The Paris collection and London collections hold about 950 such scrolls each. The scrolls containing Prajñāpāramitā texts are generally incomplete and thus also vary greatly in size. There are about 600 of these scrolls in the Paris collection, and far fewer, around 100, in the London collection. The Paris and London collections also hold a number of Tibetan scrolls containing other material of both Buddhist and secular character. 
There are 212 Tibetan scrolls in from Dunhuang in St Petersburg, 202 of which are the Aparimitāyur-nāma, and the other 10 the Prajñāpāramitā-hṛdaya.  The scrolls were originally packed in bundles. Stein counted 80 bundles of Tibetan scrolls, and took 30 away with him. Pelliot took around a third of what was left, and Oldenburg took several more. Yet the Chinese official is said to have found 94 bundles of scrolls. It certainly seems, judging by the number of scrolls which are known to us, that Stein underestimated the number of Tibetan bundles. If the official’s report, as we have it, was correct, then, at around a dozen scrolls per bundle (Stein’s estimate), there should be over 1100 scrolls still in China. However, our numbers so far give us just over 300. Even if the Beijing National Library contains another 300 scrolls, this is still only half of the number one would expect. One possiblity is that these last 94 bundles were smaller, or had already been diminished by the theft of individual scrolls from bundles. A second is that most of the 90 bundles which the official is said to have left in the cave were stolen, and if still in existence, might be in various private collections. A third is that the version of the official’s report which we have maybe incorrect in its enumeration of scroll bundles as well as pothī volumes.
The miscellaneous bundles
The collection of Tibetan manuscripts in the Dunhuang Museum, along with the significant smaller stores at the Dunhuang Academy and Lanzhou Library, rivals those of Paris and London. However, because Stein and Pelliot took all of the miscellaneous bundles before the Chinese arrived, the Tibetan manuscripts in the Chinese collections are much less varied in subject-matter. Aurel Stein and Paul Pelliot obtained the most varied, and in certain respects the most interesting, sets of manuscripts, leaving behind only those regular groups stored together in scroll bundles and pothī volumes. The textual contents of the latter are, as Stein suspected, largely repetitive. On the other hand, the importance of the proto-canonical itself material should not be underestimated, for it is still of great value to the study of early Tibetan Buddhism.
 Stein, Serindia, vol. II, pp. 822-823.
 Stein, Serindia, vol. II, pp. 919. It is worth noting that the Stein collection also includes many Tibetan manuscripts from other locations in Central Asia (in particular, the sites of Miran and Mazar-Tagh) which Stein acquired during his excavations. Some of these, especially those which Stein acquired during his second expedition, are of equal or greater antiquity to the Dunhuang library cave collection. The paper manuscripts from these sites have been recently catalogued by Tsuguhito Takeuchi: Old Tibetan Manuscripts from East Turkestan in the Stein Collection of the British Library (3 vols. ), The British Library, London, 1997-2000.
 Stein 1921, vol. II, pp. 919.
 Pelliot 1908, pp. 507-508. A photograph, much reprinted, shows Pelliot at work studying the scrolls by candlelight in the library cave. The photograph appears to show one of the large pothī volumes at the bottom of the pile of scrolls in the background. However, there does not seem to be a cover on top of the pages, and because one cannot see the bottom of the volume it is difficult to estimate the dimensions of the volume. I have not seen any other photographs showing the pothī volumes in situ.
 Pelliot 1908, p. 529 n. 1.
 On these manuscripts see Lalou 1954, 1957 and 1964.
 This information is provided in Stein 1921, vol. II, p. 823, n. 3. Stein estimated that there were 1130 bundles of all types in the cave in the first place, and 860 remaining after he had left. Pelliot estimated that he had then taken about a third of what remained, which would leave over 500 bundles in the cave after Pelliot’s departure (Stein 1921, vol. II, p. 827).
 Pelliot 1908, p. 529, n. 1. Alongside Tibetan the manuscripts included Chinese, Mongolian, Uighur, and a few in brāhmī script. There was also some Tangut printed matter. The northern caves explored by Pelliot were numbered by him 181 and 182. His description of these caves (though not the manuscripts found there) are in Grottes de Touen-houang carnet de notes de Paul Pelliot, Centre de recherche sur l’Asie centrale et la Haute Asie – College de France, Paris, 1992, vol. XI. 6, pp. 32-39.
 Some Tibetan manuscripts may have been taken as well. This was the opinion of the curators at the National Library of China, who stated that their collection of Tibetan Dunhuang manuscripts was acquired at this time. The government’s acquisition of the scrolls was not wholly successful; many were stolen shortly after their arrival in Beijing in 1910. The theft of many Chinese scrolls from the Ministry of Education in Beijing is discussed in Rong Xinjiang, ‘The Li Shengduo Collection : original or forged manuscripts?’, in Whitfield 2000. It was previously thought that all of these scrolls had been stolen en route to Beijing.
 In Japan there are sixteen institutions known to hold Dunhuang or other Central Asian manuscripts, mainly from the Otani collection and other private collections. Some of these institutions certainly hold Tibetan items. (Fujieda 1966, part 1, pp. 9-11; Whitfield 1998, pp. 1-2; Susan Whitfield, “Japanese Collections of Dunhuang and Silk Road Manuscripts”, IDP News 10 (Spring 1998), pp. 2-5. )
 Stein 1928, p. 355, 357-358. At this time Stein also bought a number of scrolls from traders in Dunhuang city. Their presence in the hands of these traders was taken by Stein as evidence of the carelessness of those who carried out the order of the Chinese government to transfer the scrolls to Beijing. A number of these third expedition scrolls have recently been identified as forgeries.
 He seems not to have taken these from the library cave itself, but they almost certainly originated there (see the article by Fang Guangchang in Whitfield 2000). There seem to be 366 Chinese scrolls, and an unknown number of Tibetan scrolls containg 212 texts (see section 7).
 Fujieda 1966, pp. 13-14.
 Fujieda 1966, part 1, pp. 15-16.
Fujieda 1966, pp. 13-14.
 The English name of the institute has recently been changed from the more literal Dunhuang Research Institute to Dunhuang Academy.
 The manuscripts sent to the other provincial museums were token objects which were considered less important, or less likely to be genuine Dunhuang library cave manuscripts. This paragraph is a personal communication from Luo Huaqing, director of the Dunhuang Academy Exhibition Centre.
 According to the article, this institution also holds six Tibetan woodslips. These would almost certainly not be from the library cave, or even Dunhuang, but from another Silk Road site.
 Also known as Gansu Provincial Library (Gansu Sheng Tushuguan).
In July 2000 I travelled to Dunhuang, with the
support of the British Academy,
to examine the Tibetan manuscripts held there.
Thanks to the generousity
of Luo Huaqing, Director of the Dunhuang Academy
Exhibition Centre, I was
able to spend some time examining and
cataloguing the Tibetan manuscripts
held in the Academy’s library. There are in fact
47 scrolls here (a few
more than the 43 reported by Huang), all of
which, except one, contained
the Aparimitāyur-nāma sūtra. There were
87 large pothī
leaves (many more than the 42 reported by
Huang), all containing the Śatasāhasrikā-prajñāpāramitā;
some of these pages were in order but most were
not in a continuous sequence.
In the following list the manuscripts are
categorised by text. The only
exception to the two most common texts is a
single Heart Sutra.
The catalogue numbers are those of the Dunhuang
Large pothī leaves with two holes for string; page numbers in margin:
0014: 12 folios, chapter 30 0015: 2 folios, chapter 13
0672: 30 folios, chapters 22, 23, 43
0750: 43 folios, chapters 17, 18, 19
Scrolls, from 1 to around 30 panels each:0119, 0120, 0122-0126, 0144-0147, 0644-0646, 0672, 0677-0689, 0691-0693, 0729-0743, 0745, 0747, 0749, 0750.
Newly catalogued: W. 1, W. 2, W. 3, W. 4.
On exhibition: 0016, 0018.
The following scrolls have Chinese text on the verso: 0144 - 0147, 0692.
Scroll, 1 panel:00121
In addition to the above, the Academy holds at least 700 more Tibetan pothī leaves, and numerous fragments, none apparently older than the Yuan period (13th-14th centuries) and some certainly more recent than that. Some of these may have come from the northern caves where Pelliot found Tibetan manuscripts dating from a later period than those from the library cave.
 The Library staff state that the Tibetan scrolls were acquired along with the Chinese scrolls in 1910, however, some or all of them may have been been aquired when a portion of the Otani collection came to the Library (Susan Whitfield, “Count Otani’s Central Asian Expeditions”, IDP News 10 (Spring 1998), pp. 1-2. The Library is in the course of publishing all of its Dunhuang manuscripts in volumes of photographic reproductions. In the volumes that have been published, there are two Tibetan manuscripts, one of which is another large-format pothī containing a Prajñāpāramitā (BD00001).
 Personal communication from the curator of the manuscript collections at the Beijing University Library. This fact is also stated in Rong Xinjiang, ‘The Li Shengduo Collection:original or forged manuscripts?’, in Whitfield 2000. Rong adds the information that the manuscripts were purchased by Professor Xiang Da, the then Director of the Library.
 Wu Chi-yu, ‘A study of four Tibetan sūtras of Tun-Huang conserved in the National Central Library of Taipei’, in Studies on Tun-Huang, vol. II, Hong Kong, 1975, pp56-69.
 See, for instance, her catalogue entries for Pelliot tibétain 1336-1343 (Lalou 1961).
 Most of the Aparimitāyur-nāma sūtra scrolls in the British Library are catalogued under the number 310 in de la Vallée Poussin’s catalogue (now IOL Tib J 310). Most of the Prajñāpāramitā scrolls are under the number 109 (now IOL Tib J 109). At some point all of the scroll panels were separated and rebound as booklets, but the original number of scrolls can be reconstructed. There is currently no printed catalogue of the Aparimitāyur-nāma sūtra scrolls in the Bibliothèque nationale, which occupy the catalogue numbers Pelliot tibétain 3500-4450. This sequence include fragments as well as complete scrolls, especially from Pelliot tibétain 4100 to 4450. There are also many incomplete Prajñāpāramitā scrolls in the Pelliot collection, which in Lalou’s catalogue occupy the numbers Pelliot tibétain 1495 to 2105. Both the Paris and London collections also contain miscellaneous scrolls scattered throughout their catalogues.
 There are two other Tibetan items (nos. 213 and 214 in Savitsky’s catalogue), which contain a transcription of a Chinese text, and a fragment of an unidentified Tibetan text.
Jiang Liangfu. 1956. Dunhuang - weida de wenhua baozang. Shanghai: Shanghai gudian wenxue chubanshe.
Konow, Sten 1916. ‘The Aparimitâyuḥ Sūtra’, in Rudolf Hoernle (ed. ), Manuscript Remains of Buddhist Literature Found in Eastern Turkestan, Oxford: Oxford University Press.
Huang Wen-huang. 1978. ‘He xi tu fan wen shu jian shu’ [’A brief description of Tibetan documents in the Hexi corridor’], Wen Wu xii, pp. 59-63.
Lalou, Marcelle. 1939, 1950, 1961. Inventaire des Manuscrits tibétains de Touen-houang conservés à la Bibliothèque Nationale (Fonds Pelliot tibétain). (3 vols. ) Paris: Bibliothèque Nationale.
—. 1940. ‘Tun-Huang Tibetan Documents on a Dharmadāna’, Indian Historical Quaterly xvi. 2, pp. 293-298.
—. 1954. “Les manuscrits tibétains des grandes Prajñāpāramitā trouvés à Touen-houang”. Silver jubilee volume of the Zinbun-kagaku-kenkyusho, pp.257-261.
—. 1957. “Les plus anciens rouleaux tibétains trouvés à Touen-houang”. Rocznik Orientalistyczny Tom. 21, pp.149-152.
—. 1964. “Manuscrits tibétains de la Satasāhasrikā-prajñāpāramitā cachés à Touen-houang”. Journal Asiatique 1964, pp.479-486
Pelliot, Paul. 1908. ‘Une bibliothèque médiévale retrouvée au Kan-sou’, Bulletin de l’École Française D’Extreme Orient viii, pp. 510-529.
Savitsky, Lev S. 1984. ‘Tunhuang Tibetan Manuscripts in the Collection of the Leningrad Institute of Oriental Studies’, in Ligeti, Louis (ed. ). Tibetan and Buddhist Studies Commemorating the 200th Anniversary of the Birth of Alexander Csoma de Körös. Budapest: Akadémiai Kiadó.
Savitsky, Lev S. 1991. Opisanie tibetiskikh svitkov iz dunkhuana v sobranii instituta vostokovedeniia an SSSR. Moscow.
Skilling, Peter. 1997. ‘From bKa’ bstan bcos to bKa’ ‘gyur and bsTan ‘gyur’, in Ernst Steinkellner (ed. ). Tibetan Studies: Proceedings of the 7th Seminar of the International Association for Tibetan Studies, Graz 1995. Vienna: Verlag der Österreichischen Akademie der Wissenschaften.
Stein, M. Aurel. 1921. Serindia: Detailed Report of Explorations in Central Asia and Westernmost China. (5 vols. ) Oxford: Clarendon Press.
Stein, M. Aurel. 1928. Innermost Asia: Detailed report of explorations in Central Asia, Kansu and Eastern Iran. (4 vols. ) Oxford: Clarendon Press.
Takeuchi, Tsuguhito. 1994. ‘Tshan: Subordinate Administrative Units of the Thousand-Districts in the Tibetan Empire’, in Per Kvaerne (ed. ), Tibetan Studies: Proceedings of the 6th Seminar of the International Association for Tibetan Studies, Fagernes 1992. Oslo: The Institute for Comparative Research in Human Culture.
Takeuchi, Tsuguhito. 1995. Old Tibetan Contracts From Central Asia. Tokyo:Daizo Shuppan. Thomas, F. W. 1951. Tibetan Literary Texts and Documents from Chinese Turkestan, Part II: Documents, London: Royal Asiatic Society.
de la Vallée Poussin, Louis. 1962. Catalogue of the Tibetan Manuscripts From Tun-Huang in the India Office Library. Oxford: Oxford University Press. Whitfield, Susan (ed. ). 2000. Dunhuang Manuscript Forgeries. (The British Library Studies in Conservation Science 3. ) London: British Library Press.