Dunhuang documents consist of about 40 thousand documents from 5th to 11th century. The documents were discovered in Mogao Cave 17, in Dunhuang, China, by the Daoist monk Wang Yuanlu in 1900. At that time, many foreign explorers visited Central Asia and especially Dunhuang: the Hungarian-British archaeologist Aurel Stein in 1907, followed by the French Sinologist Paul Pelliot in 1908. Both brought back to Europe thousands of documents that they bought from the monk. Although scattered in many countries, the documents are available on the International Dunhuang Project website and the Gallica website of the French National Library; however, except the digital images and bibliographic data, there is no further information on the constituent materials (paper, ink, dyes etc.)

This priceless treasure represents an invaluable resource that led to the creation of a new research field called Dunhuang studies, in order to contribute to a better knowledge of the evolution of paper over 6 centuries. To date the manuscripts, in the 1990s, Prof. Akira Fujieda, a Japanese scholar adopted codicological analysis, focusing both on paper and morphology of the manuscripts, and on the shape of characters written (e.g. Clerical script (隷書), Regular script (楷書), etc.). Only preeminent documents were deliberately taken into account by him and thus ignoring many manuscripts. As a result, 5 classes were determined according to the historical periods, that is Northern dynasty (386-581 CE), Sui dynasty (581-618), Early and High Tang dynasty (618-765), 765-786 and Tibetan Empire and Guiyi Circuit (786-1036) (Fujieda, 1999).

In addition to Prof. Fujieda’s study, we investigate more details of paper using nondestructive scientific analysis on more than 400 Chinese Dunhuang manuscripts from the Pelliot collection and the Stein collection, and collected various data using a high-resolution digital microscope (Keyence VHX-1000) together with visual checks. Information contained in colophons (date, title of manuscripts) was also collected. A manuscript title is useful for categorization of the manuscripts. Analysis results show differences that can be criteria for differentiate paper. We developed the database (http://www.afc.ryukoku.ac.jp/pelliot/index.html), including scientific analysis data such that microscopic images from the Pelliot collection, as part of new digital archives for old documents.

As we obtained new data by scientific analysis and visual check, we can define new classes, A2, A3, B1 and C1, besides Fujieda’s classes, A1, B2 and C2, as follows; A1: Fujieda’s class from Northern Wei (北魏 (386-534)) and Western Wei (西魏 (535-556)). Paper is Ma-shi (麻紙) including hemp or ramie with 4~6 laid lines/cm, and with clerical like script of northern dynasty style. On the other hand, A2: paper from southern dynasty, Liang (梁 (502-557)) and Chen (陳 (557-589)), is high quality Ma-shi, and have finer laid lines, 8~9 laid lines/cm, paper width is 49~50 cm and well dyed, and is written sutra with clerical like script of southern dynasty style. Moreover, A3: new class paper from Northern Zhou (北周 (556-581)) is not Ma-shi but Cho-shi (楮紙) including mulberry paper (B. papyrifera, M. alba, etc.), and they have around 6 laid lines/cm, and with clerical like script. B1: new defined class paper from Sui (隋 (581-618)) is Cho-shi with 6~8 laid lines/cm and paper width is narrow, 41~43 cm. Few paper include rice starch. But B2: Fujieda’s class from Sui. Paper is Cho-shi with about 6~7 laid lines/cm, paper width is wide, 50~53 cm, and well dyed, and is written sutra with clerical like or regular script of southern dynasty style. C1: paper in this new small class from early and high Tang (初唐 (618-712), 盛唐 (712-765)) is similar to the ones in B1, that is Cho-shi with 6~8 laid lines/cm and paper width is 37~44 cm. Some paper include rice or millet starch. C2 is also Fujieda’s class from early and high Tang. Paper in C2 is Cho-shi and high quality Ma-shi with fine laid line, about 8~10 laid lines/cm, paper width is wider than the ones in C1, 45~51 cm, and well dyed, and is written sutra with regular script.

As mentioned above, scientific analysis data is very useful for Dunhuang studies, for example, the data improved Fujieda’s classification. We developed the database, Scientific Analysis of Pelliot Collection, digital archives, including such data.