III · THE WORK · ART HISTORY
The Cave Paintings of Dunhuang
▶ Listen · narrated
Buddhist pilgrims carved temples into a cliff face for a millennium. Then the route shifted, the caves were abandoned, and a hidden library stayed dark for eight centuries.
At a glance
- Location
- 25 km (16 mi) southeast of Dunhuang, Gansu province
- Number of caves
- Approximately 500 temples
- Active period
- 366 CE to 14th century
- Library Cave sealed
- 11th century
- Library rediscovered
- 1900
Look closer
Flying apsaras with scarves as the only sign of motion
The painted ceilings are populated by celestial beings called apsaras, moving through an undefined sky. They have no wings. Instead, long ribbons of cloth trail from their shoulders and arms, curving in arcs that read as flight paths. The scarves do all the work: without them the figures would be static. With them, the entire surface becomes airborne. The convention appears across centuries of painting at the site, a continuous visual language for weightlessness.
Central Asian and Chinese styles in one wall
Dunhuang sat at the junction of trade routes connecting China, Central Asia, India and Persia. The paintings record this. Early caves show figures with pronounced shading and modelling, techniques associated with Indian and Central Asian Buddhist art. Later work adopts flatter planes and flowing line more typical of Chinese painting. Some caves contain both approaches on facing walls, painted decades or centuries apart as new patrons commissioned additions to older shrines.
Oxidised lead white now read as dark grey
Some figures have faces and hands that appear shadowed or even black. This is a chemical change, not an iconographic one. Lead white pigment, exposed to hydrogen sulphide from oil lamps or organic decay, converts to lead sulphide, which is dark grey to black. The painters intended pale skin. What we see now is an inversion: highlights have become shadows, and the tonal structure of the original image has been reversed by time.
The story
Dunhuang is an oasis town in Gansu province, positioned where the northern and southern branches of the Silk Road converge before entering China proper. In 366 CE, according to later inscriptions, a monk named Lè Zūn had a vision of a thousand Buddhas in golden light and began to carve meditation chambers into the cliff face east of the town. Other monks followed. Over the next thousand years, approximately five hundred caves were excavated, plastered, and painted.
The site was not a monastery in the usual sense. It functioned as a place of pilgrimage, a centre for copying and translating Buddhist texts, and a location where merchants, officials and aristocrats could sponsor the creation of a cave as an act of merit. The paintings served devotional and didactic purposes: they depicted episodes from the life of the Buddha, illustrations of sutras, mandalas, and protective deities. Styles shifted as dynasties changed and as artistic influences moved along the trade routes. The earliest surviving work dates to the fifth century and shows clear connections to the art of Gandhara and Central Asia. By the Tang dynasty the painting had become more distinctly Chinese in its treatment of space and line.
Construction continued until the fourteenth century, but the site's importance had already declined. The Silk Road trade routes shifted. Dunhuang became less central. At some point in the eleventh century, a chamber now called Cave 17 was filled with manuscripts, paintings on silk and paper, and other documents, then sealed behind a wall and painted over. Why this was done is not recorded. Suggestions include protection from an anticipated invasion, a respectful storage of worn-out sacred texts, or the closure of a monastic library no longer in use.
The wall stayed intact for eight hundred years. In 1900 a Taoist monk named Wang Yuanlu, who had appointed himself caretaker of some of the abandoned caves, noticed a crack. He enlarged it and found a small room stacked floor to ceiling with scrolls and bundles. The chamber contained approximately fifty thousand manuscripts and printed documents, along with paintings and textiles. The texts were in Chinese, Tibetan, Sanskrit, Sogdian, Khotanese, Uyghur and other languages. They included Buddhist sutras, Confucian and Taoist works, contracts, letters, calendars and music scores. The earliest dated document is from 406 CE; the latest from 1002 CE.
Word of the discovery spread slowly. Between 1907 and 1914, foreign expeditions arrived: the Hungarian-British explorer Aurel Stein, the French sinologist Paul Pelliot, the Russian scholar Sergei Oldenburg, and the Japanese expedition led by Ōtani Kōzui. Each removed large quantities of manuscripts and art. Stein took approximately seven thousand complete manuscripts to the British Museum. Pelliot selected another six thousand for the Bibliothèque nationale de France. The removals were conducted with Wang's cooperation, in exchange for modest payments, but without formal permission from Chinese authorities. By the time officials in Beijing learned the scale of what had been taken, much of the collection had already left the country.
The dispersal is now a longstanding point of contention. The largest collections are held in Beijing, London, Paris and Berlin. Scholars have argued both that the manuscripts would have been destroyed or lost had they remained in situ during the upheavals of early twentieth-century China, and that their removal constituted straightforward looting enabled by the weakness of the Qing state. Both positions have evidence. What is not disputed is that the manuscripts are now scattered across institutions on three continents, and that this complicates their study.
The International Dunhuang Project, established in 1994, maintains a digital archive to reunite the collection virtually. The caves themselves are managed by the Dunhuang Academy, which has capped visitor numbers to slow deterioration caused by humidity and carbon dioxide from breath.
Why it mattered then
Dunhuang mattered because it was a threshold. For travellers entering China from the west, it was the first major settlement after crossing the desert. For those leaving, it was the last. The caves served pilgrims, but they also served people who needed the protection of the Buddha before or after a dangerous journey. Sponsoring a cave was an act of devotion, but it was also a declaration of safe arrival or a petition for safe passage. The library's contents suggest the site was also a centre for the production and translation of texts. Buddhist sutras were copied and recopied. Translations were made from Sanskrit into Chinese and from Chinese into Central Asian languages. The presence of secular documents—contracts, letters, administrative records—indicates that the caves were embedded in the practical and commercial life of the oasis, not isolated from it. Dunhuang was a working religious site in a working trade town, and the manuscripts reflect both functions.
Why it matters now
The Mogao Caves are now among the most studied Buddhist sites in the world, not despite the dispersal of the library but in part because of it. The manuscripts reached institutions with the resources to conserve and photograph them, and scholars in multiple countries began to work on texts that would otherwise have remained unknown. This does not resolve the ethics of how they left China, but it describes what happened afterward. The caves also matter as a record of sustained artistic production over a very long period in one location. Few sites show a thousand years of continuous painting. The stylistic shifts at Dunhuang map onto changes in political control, trade patterns and religious practice across Central Asia and China. The paintings are evidence not just of what Buddhists believed but of how images moved along trade routes and what happened when artistic conventions from different regions met. The conservation challenges are ongoing. The pigments are fragile, the plaster is crumbling in places, and the caves were never designed for the humidity generated by large numbers of visitors. The Dunhuang Academy has introduced timed entry, replica caves for tourists, and environmental monitoring. The question is whether the site can remain accessible without being destroyed by that access.
The surprising detail
The Library Cave was rediscovered because Wang Yuanlu was clearing sand from the corridors and noticed that one section of wall sounded hollow when tapped. He was not an archaeologist or a scholar. He was a Taoist monk from Hubei who had arrived at the site around 1900, found it neglected, and decided to restore it. He funded his work by soliciting donations from travellers and local officials. When he found the manuscripts, he understood they were old and valuable, but he had no means to conserve them and no authority to prevent their removal. Stein's first acquisition from Wang cost one hundred and thirty pounds sterling. Pelliot paid less. The sums were not trivial to Wang, but they were modest by the standards of what was being acquired. The disparity was not lost on later observers.
What is disputed
The reason Cave 17 was sealed is not documented, and the explanations offered—protection from invasion, respectful disposal of damaged texts, closure of a library—are scholarly inferences rather than recorded facts. The date of sealing is given as eleventh century based on the latest dated manuscripts inside, but the actual event was not recorded. The story of Lè Zūn's vision in 366 CE comes from much later inscriptions at the site, not from contemporary sources, and should be understood as a foundation legend rather than verified history.
Remember this
A thousand years of painting in five hundred caves, and a library sealed in the eleventh century, opened in the twentieth.
Test yourself
The Library Cave was sealed in the eleventh century and not reopened until 1900. Why does the fact that it was sealed—rather than simply abandoned or forgotten—matter to how we interpret what was inside?
Because sealing the chamber was an intentional act that required effort and planning. An abandoned room might contain whatever was left behind when people stopped using it: broken furniture, worthless scraps, accumulated debris. A sealed room contains what someone chose to preserve or protect, even if we do not know their reasoning. The contents of Cave 17 were placed there deliberately, the entrance was bricked up, and the wall was plastered and painted over to match the surrounding corridor. That sequence of actions suggests the manuscripts were valued, whether as sacred texts too worn for use, as a collection requiring protection, or as a library being closed in an orderly way. The sealing converts a storage space into a time capsule. It also means that what we found in 1900 is more likely to represent a coherent collection from a specific moment—the eleventh century—rather than a random accumulation. The interpretive problem is that we do not have the record of why it was sealed, only the physical evidence that it was.
Go deeper
Image: Zhangzhugang, CC BY-SA 4.0. Licence: CC BY-SA 4.0. Source.
