Nine Rock-Cut Buddhist Caves at Bagh in Dhar District Hold Painted Courts, Dancers and a Royal Cavalcade That Sir John Marshall Judged Close to Ajanta, and Joined India's UNESCO Tentative List on 29 September 2026

Nine caves cut into a sandstone cliff above the Baghni river at Bagh, in Dhar district of Madhya Pradesh, hold wall paintings of a kind that survives in very few places in India. On 29 September 2026 Akashvani News reported that the caves had joined India’s UNESCO Tentative List, which reached 78 sites with that round of additions. India’s submission, dated 8 July 2026, singles out Cave 4, known as Rang Mahal, a hall carried on 38 pillars with 28 monks’ cells around it, and describes its murals as continuing the Ajanta tradition. A place on the Tentative List is the first step toward a nomination and is not World Heritage status.

Sir John Marshall, who wrote the main chapter of the 1927 monograph on the caves, counted wall paintings in only three groups of Indian cave temples: Ajanta, Bagh and Sigiriya in Ceylon. On artistry, he wrote, “there is little to choose between the pictures of Bagh and Ajanta.”

What is cut into the cliff

Marshall’s description is still the standard one. The cliff rises about 150 feet above the river and is the only outcrop of sandstone in an otherwise basalt region, a point India’s submission repeats. The nine caves spread over about 750 yards of frontage and are not all side by side. Above the sandstone lies a thick band of claystone, and Marshall blamed its weight and the moisture seeping from it for most of the damage.

Cave 2, popularly the Pandava’s cave, is the best preserved. It has a square hall with cells on three sides, a pillared portico and, at the rear, a stupa chapel (a shrine built around a dome-shaped Buddhist monument) whose vestibule walls carry carved figures of the Buddha with attendants. Cave 4 is larger. According to Marshall its central hall is 94 feet across and has 38 pillars and 28 cells, figures the submission repeats. The friable sandstone shaped the plan. Four extra columns stand in the middle of the halls of Caves 2 and 4, and in Cave 4 their shafts are built of small cut stones, probably because of faults in the rock, Marshall wrote. Caves 7, 8 and 9 had already collapsed by 1927. M. B. Garde, Superintendent of Archaeology in Gwalior State, judged the group the work of Mahayana Buddhists.

The Pandava name goes back at least to 1818, when Dangerfield of the Bombay Military Establishment visited and became the first to describe the caves to antiquarians, in Garde’s account. According to Saurabh Saxena in the heritage journal Puratattva, Dangerfield counted four caves, reported that local people credited the five Pandava brothers with excavating them, and published in 1820. Dr E. Impey counted nine in 1857, and Colonel C. E. Luard published photographs and plans in 1910.

The painted wall of Caves 4 and 5

The best-preserved paintings are not inside a cave. They cover the back wall of a verandah, more than 220 feet long, that once ran in front of Caves 4 and 5. The roof of that verandah has fallen, so the upper half of the wall is open to the weather.

J. Ph. Vogel, who wrote the iconographic chapter of the 1927 volume, described the surviving scenes one by one. Two women sit in a pavilion, one covering her face in grief. Four men sit in earnest talk in a garden, the senior of them wearing a jewelled headdress. Figures fly out of clouds above a group of women musicians. Two groups of women play drums, sticks and cymbals around a dancer. A cavalcade of at least seventeen horsemen rides under a parasol, the central rider on a horse hung with yak-tail plumes. An elephant procession follows, its leading rider holding a blue flower.

Marshall wrote that the paintings are tempera, not fresco as had often been stated. S. Paramasivan’s 1939 study, as Saxena summarises it, found the pigments mixed with a binding glue and applied to a dry prepared surface. The rough first coat was local iron-rich earth mixed with gravel, lime and jute or hemp fibre, according to Marshall. India’s submission lists the pigments as yellow ochre, red ochre, terre verte (a green earth), lapis lazuli, carbon and lime. Marshall thought the first coat was less carefully made than at Ajanta and that this weakness explained part of the decay. He added that nothing left at Bagh equals the famous Bodhisattva of Ajanta. Bagh’s wall, he wrote, gives the impression of one planned scheme, while most of Ajanta looked to him to have been added piecemeal by successive donors.

The submission puts the case for Bagh in its own terms. It says the murals are more intimate in scale than Ajanta’s and give more room to courtly and social scenes, and it compares them with Sigiriya, Mogao and Dambulla. It also separates Bagh from the prehistoric rock art of Bhimbetka, inscribed in 2003, calling Bagh a monastic complex with murals set in a specific religious and historical context.

Nobody has settled what story the wall tells. Vogel wrote in 1927 that the paintings were too fragmentary to identify and probably came from a jataka or avadana (a tale of the Buddha’s past lives or of a pious deed). Saxena’s survey lists later readings. Monika Zin proposed the story of King Mandhata in 2001, Anupa Pande read the scenes through Asvaghosa’s poem on Nanda and his wife Sundari in 2002, and Meena Talim read them differently in 2002 and again in 2014.

A copper plate names the monastery

The one document that names the monastery is a copper plate. Saxena says it was unearthed in 1929 in the debris of Cave 2. V. V. Mirashi printed it in 1955 in the Corpus Inscriptionum Indicarum as the Bagh Cave plate of Maharaja Subandhu. It records the grant of a village, issued from Mahishmati, for worship of the Buddha, an alms-house, repair of the broken parts of the vihara, and clothing, food, medicine, beds and seats for monks arriving from every direction. In Mirashi’s translation, as quoted by Walter Spink, the grant is made to the monastery called Kalayana, which Mirashi glosses as “the Abode of Art,” and Dattataka is named as the man who caused it to be built.

India’s submission associates the caves with a Buddhist monk named Dataka. Mirashi’s introduction describes Dattataka as a private individual who had a cave cut and decorated with paintings. A piece of the plate has broken off, and only the name of the month, Shravana, survives of its date.

How old the caves are

Garde wrote in 1927 that there was not a single inscription on the caves themselves. He read one painted letter under the horsemen as sixth or seventh century on its shape, and classed the caves with the late but not the latest viharas at Ajanta. Marshall read a pilaster pattern at Caves 4 and 6 as one in fashion in the seventh and eighth centuries and took it to confirm the date reached on other grounds. The Victoria and Albert Museum’s catalogue, describing a tracing of a Bagh figure, also gives the sixth and seventh centuries. James Burgess, according to Saxena, put the caves at 450 to 500 CE and the paintings in the sixth century.

The Subandhu plate carries the later argument. Another grant by the same king, the Barwani plate, is dated in year 167 of an era it does not name. R. R. Haldar read the era as Gupta, which gives 486 CE. Mirashi argued for the Kalachuri era, which would put the grants in the early decades of the fifth century. Spink uses 486 and argues that a plate funding repairs shows the caves were finished well before then. In his account work began about 462 and ended by about 478, and Bagh and Ajanta were started within a few years of each other. Brill’s description of his 2017 book calls this the “short chronology.” The submission, by contrast, calls the murals “post-Ajanta” and dates the caves to about the fourth to sixth centuries.

The submission also says about 36 inscriptions run from 358 to 487 CE, and that most begin with the word Valkha, the name of a kingdom. It does not say which texts it counts. The dated texts of the Valkha kings that scholars have published are copper-plate charters. Daniel Balogh counts 37 known charters in a 2026 paper in the Bulletin of SOAS, with years from 47 to 134 and no era named on the plates. He follows Ramesh and Tewari, who published a 1982 hoard of 27 plates found near Bagh, in reading the years as Gupta era (from about 319 CE, the era of the Guptas), and dates the Valkha kings to the late fourth and early fifth centuries. He notes that Mirashi preferred the Kalachuri era, from about 249 CE. Whether Valkha was Bagh itself is also open. Balogh calls Bagh a very likely site, while Ramesh and Tewari proposed Balkhar, about 75 kilometres to the east-southeast.

Indian artists copied the wall in the 1920s

By the 1920s the paintings were fading, and the Gwalior State archaeology department hired copyists. Marshall’s monograph names seven: Nandalal Bose, Asit Kumar Haldar and Surendra Nath Kar of Calcutta, A. B. Bhonsle and B. A. Apte of Bombay, and M. S. Bhand and V. B. Jagtap of Gwalior. A table in the book credits each artist with particular plates. Saxena dates the Calcutta team’s work to 1920 and 1921 and the Bombay team’s to 1923 and 1924. The copies were shown in the museum at Gwalior, with duplicates in the British Museum. The preface says the late Maharaja of Gwalior planned the copying with Colonel Luard, and that the copies were exhibited at the British Museum in the autumn of 1925 before the India Society published the book.

Bose had been to Bagh before. The V&A’s catalogue entry for a tracing of a girl’s head, copied from a Bagh fresco, says he visited in 1910 with Lady Herringham and two other students. Bose is also the artist whose Santiniketan team decorated the pages of the handwritten Constitution of India.

What was done to save the originals

The Archaeological Survey of India began cleaning the paintings in 1954, Saxena writes, citing its annual Indian Archaeology reviews. Smoke, tar and oil had baked onto the surface. Later reports describe years of chemical treatment and a study of the salts that seeping water deposited on the walls. Lifting of paintings began in 1979 to 1980. Three pieces of about a square metre each were detached from the eastern wall of Cave 4 by the Italian strappo method, which peels off the paint layer alone, and the work on that wall was finished by 1981 to 1982. Outlook Traveller adds that a second method, which takes a slice of the plaster as well, was also used. The detached panels are now in the archaeological museum at Gwalior and in the small museum at Bagh.

Ritu and Phanikant Mishra wrote in Marg in 2002 that thirty years of restoration had brought little improvement against continuing seepage, as Saxena quotes them. India’s own submission says the site’s integrity is partly compromised by geological instability, seepage through claystone bands and weathering, and that several caves have collapsed. The Dhar district administration’s page calls Cave 4 the most important of the surviving caves. Another early painted cave in the Ajanta tradition, far to the south, is the Jain shrine at Sittanavasal.

As of 2 November 2026, UNESCO’s page lists the Bagh Cave paintings as reference 7043, submitted by India’s Permanent Delegation, under criteria (i), (ii) and (iii). Press Trust of India noted on 30 September that a site must be on the Tentative List before it can be considered for the World Heritage tag.