Krishna Lifting Mount Govardhan

A larger than life sandstone Krishna carved about 600 for a man-made cave on the mountain of Phnom Da in southern Cambodia, cut down on site in the early 1900s to be reused as the armature of a Buddha, sold in Paris in 1920 by the Cubist dealer Leonce Rosenberg to Adolphe Stoclet, and displayed in the music hall of the Palais Stoclet in Brussels for nearly fifty years under the name the dancing prince.

Fragments dug up at Phnom Da in 1935 were shipped to Brussels, failed to fit, and were thrown into a neighbour's garden, where they held up a water cistern and edged a flower bed until a Cleveland curator recognised the pleats of a Phnom Da sampot carved into a thigh in 1977. Cleveland used eight of them to build a nearly complete figure in 1978 and sent nine to Phnom Penh in 2005, where six were attached to a different Krishna.

Almost all of that was wrong. Between 2014 and 2018 conservators in Cambodia, France and Ohio established by three-dimensional scanning and petrography that six pieces in Cleveland belonged to the Phnom Penh statue and two pieces in Phnom Penh belonged to the Cleveland one. Cambodia transferred a hand, strut and panel to Cleveland in 2015; Cleveland took its own 1978 restoration apart; and both sculptures now stand with fewer limbs and the right ones.

A god who is the mountain he lifts

The subject is a single episode. In the Harivamsha and the Vishnu Purana, the boy Krishna persuades his cow-herding village to divert the annual offerings from Indra, god of rain, to the nearby Mount Govardhan, on the ground that the mountain is what actually feeds them. Indra sends a storm; Krishna lifts the mountain like an umbrella and holds it over the village and its cattle for seven days until Indra concedes.

What makes the Cleveland figure more than an illustration is where it stood. It appears to have been installed in a man-made cave part way up the northern slope of Phnom Da, so that the god held up the mountain he was carved inside. Phnom Da rises out of the flood plain of the Mekong delta, a landscape of rice fields and thousands of miles of canals whose whole economy is the management of too much water and too little. A congregation walking into that cave in the year 600 was walking under the shelter in the story.

Carved from one block, and the consequences of that

The Phnom Da figures are monolithic: each was cut from a single block, with an arch or stele of stone left behind the figure to carry the weight and, in several cases, a long tenon below the feet to seat it in a pedestal. The Phnom Penh Krishna has a 43 centimetre tenon, the Wat Koh Krishna a 27 centimetre one, and their catalogued heights are given twice for that reason, with and without the tenon.

Monolithic carving is why this object exists as a problem. A figure with free-standing arms carved from the same block as its body will break at the arms, and every one of the eight gods of Phnom Da has been found in pieces. It is also why the fragments are interchangeable at a glance and not in fact: five heroic two-armed figures carved in the same stone, in the same workshop, at the same moment, produce arms and legs that look like each other and belong to exactly one of them.

Eight gods, and how the site was emptied

The sanctuaries of Phnom Da are the earliest monumental Hindu temple complex recorded in Cambodia, dated by inscriptions to the reign of Rudravarman between 514 and 539, and the site holds the earliest surviving Khmer stone temple. Eight monumental sculptures of Hindu deities have been recovered from it: two Krishnas lifting Mount Govardhan, Krishna's brother Balarama, Rama, Vishnu, the dual god Harihara, and others. Every one was found broken.

The evidence is that the Cleveland Krishna was actively worshipped for about seven hundred years. Around 1300, when Thai forces took southern Cambodia, the eight figures were toppled from their pedestals, most probably by people who wanted the consecration gold and gems buried beneath them. The site was then largely abandoned. Veneration did not stop everywhere: gold was applied to the face of the Harihara now in Paris in a ritual of renewal no earlier than the 1500s.

The caves were visited by a French civil servant in the 1880s. The archaeologist Henri Parmentier of the Ecole francaise d'Extreme-Orient recorded his own visit in June 1911. By then a Buddhist community from Vietnam had settled at the site in the first decade of the century and was reusing the temples and the broken sculpture, and the report published in 1911 indicates that the head and body section had been taken to France the previous year. It also records the detail that matters most and is quoted least: before removal the torso was already being cut down to serve as the body armature for a new seated Buddha. The object was saved from destruction by being stolen, and both halves of that sentence are true.

Rosenberg, Stoclet, and the dancing prince

The head and torso reached Paris and were sold in 1920 by Leonce Rosenberg to Adolphe Stoclet, a Brussels financier, and his wife Suzanne. Rosenberg is remembered for something else entirely: he was the dealer of the Cubists, the man who took on Picasso, Braque, Leger and Gris at the Galerie de l'Effort Moderne. A pre-Angkorian Khmer torso passing through his hands in 1920 is a fact about how the Paris market of that decade worked, in which the newest European painting and the oldest Asian sculpture were traded by the same people to the same clients.

The Stoclets put it on a stage in the music hall of the Palais Stoclet, the Josef Hoffmann and Wiener Werkstatte house on the Avenue de Tervueren that also holds Klimt's Stoclet Frieze, and called it their dancing prince. It was published soon after the purchase, and by the late 1920s art historians internationally were describing the quality of what survived of the carving. The sculpture's reputation was made while it was in a private drawing room, by scholars who were invited in.

The fragments in the flower bed

In 1935 Henri Mauger of the Ecole francaise d'Extreme-Orient found further sandstone fragments, arms, legs, feet and a hand, near the caves at Phnom Da. On the belief that they would complete the Stoclet torso they were shipped to Brussels; the museum's account puts the arrival in 1937 and counts seventeen fragments belonging to four different sculptures, while a published account of the find gives fourteen pieces recovered in 1935. Those numbers have not been reconciled in print and this record does not reconcile them either.

The Stoclets asked their neighbour, the sculptor Marcel Wolfers, to attach them. After several attempts the result did not satisfy Stoclet and the project was abandoned; the pieces were left in the Wolfers villa's garden. New residents, who did not know what they were, used them to support the installation of an underground water cistern and used one to edge a flower bed.

Michele Leon-Stoclet, who had inherited the sculpture from her grandparents, died suddenly in 1967. Sherman E. Lee, then director in Cleveland, spent eight years securing the torso, which the museum bought from the estate in 1973 as a fragment with no arms and no legs. In 1977 the curator of Indian and Southeast Asian art, Stanislaw Czuma, went to Brussels after hearing of the forgotten pieces and recognised the distinctive pleats of a Phnom Da sampot carved into a thigh lying at the edge of the flower bed. Seventeen pieces were dug out of the garden with the owners' consent and sent to Cleveland.

The 1978 restoration, which was excellent and wrong

In 1978 and 1979 Cleveland's conservators attached eight of the seventeen pieces to the torso, producing the impression of a nearly whole figure standing on two feet. They drilled new holes in the fragments and set steel rods in epoxy; missing sections of the legs were filled with epoxy putty and metal casings. That reconstruction stood in the galleries for about forty years and is the version of this sculpture that a generation of visitors and textbooks knew.

It was done to the best standard of its moment and it was documented properly. The museum's present conservator has said the department has good documentation from that time, including photographs and notes on each step. What the 1978 team did not have was any means of testing whether a leg that fitted was the right leg. Nine pieces were left over because they seemed to belong to two other sculptures from Phnom Da in the National Museum of Cambodia, and in 2005 Cleveland sent those nine to Phnom Penh, where conservators attached six of them to a second, limbless Krishna Lifting Mount Govardhan that had been brought from Phnom Da in 1944.

So by 2005 two museums on opposite sides of the world each had a plausible, published, photographed Krishna, and each was wearing parts of the other.

How the mistake was caught

From 2014, as communication between the two conservation departments increased, Bertrand Porte of the Ecole francaise d'Extreme-Orient and Sok Soda of the National Museum of Cambodia concluded that the fragments had been distributed wrongly. The instrument that settled it was not the eye. It was three-dimensional scanning, medical computed tomography, and petrographic analysis of the stone.

Petrography established that the largest of the pieces sent to Phnom Penh in 2005, a hand with a strut and a section of supporting panel, had been carved from the same block as the Cleveland torso. High-resolution three-dimensional scans of both the sculpture in Ohio and the fragment in Cambodia were then produced and compared. The final count is exact and unflattering to both earlier campaigns: two of the pieces held in Phnom Penh belonged to the Cleveland Krishna, and six of the pieces held in Cleveland belonged to the Phnom Penh Krishna.

This is a rare thing in provenance work. Almost every attribution dispute in this archive is argued from documents, style or absence. Here the question of what belongs to what was answered by the rock.

The transfer, and what kind of transfer it was

On 30 October 2015 the Ministry of Culture and Fine Arts of the Kingdom of Cambodia and the Cleveland Museum of Art announced that the Ministry had authorised the transfer to Cleveland of the fragment with the hand, strut and supporting panel. Kong Vireak, director of the National Museum, said that reuniting the fragment with the Cleveland Krishna would present a new and enhanced opportunity for the public in the United States to see one of the great artistic accomplishments of the Khmer people. William Griswold, Cleveland's director, described the transfer as another aspect of the museum's cultural cooperation with Cambodia.

Neither statement says whether the fragment was given, lent, exchanged or transferred outright, and the press release does not name signatories. That is not a small omission. It is the difference between Cambodia lending Cleveland a piece of Cambodian state property and Cambodia ceding title to it, and a reader doing diligence on this object cannot settle it from the published record. The transfer operates under a cultural cooperation agreement between the museum and the Ministry whose terms have not been published either.

What is clear is the direction of travel in both directions. Cleveland sent nine fragments to Phnom Penh in 2005 and more in 2020; Cambodia sent one back in 2015. In a field where the word restitution normally means a one-way transfer after a dispute, this is something else: an exchange conducted between conservation departments to put objects back together, with the ownership question left in the background.

Taking a restoration apart

The sculpture left the galleries in November 2017 and was in the objects conservation laboratory for almost exactly four years. The full project ran more than seven, and the museum notes the symmetry that the 1978 reconstruction was exactly forty years before the team returned to it in earnest in 2018.

Removing the 1978 work meant excavating epoxy putty, brass and steel until the original stone surfaces appeared for the first time in four decades. What went back is less, not more: the figure lost a thigh, two calves and the feet that had been added in 1978 and 1979, so it no longer stands on its own legs. What it gained is the large upper section carrying the left hand and the top of the stone niche, which is the part that actually shows the god holding the mountain.

The engineering problem this created was substantial. Body and upper section together weigh about 830 pounds and had to be aligned and suspended with the four stone pieces kept physically separate while reading as one sculpture. Case Western Reserve University's engineering department helped test epoxy and steel combinations strong enough to carry the load while keeping the intervention as reversible as possible, and the work drew in the Cleveland Clinic for medical computed tomography, an industrial X-ray facility, structural engineers and a steel fabricator.

The reversibility is the point. The 1978 campaign drilled into ancient stone. The 2021 campaign was designed on the assumption that it too will be found to be wrong.

What the exhibition did with it

Revealing Krishna: Journey to Cambodia's Sacred Mountain ran at the Cleveland Museum of Art from 14 November 2021 to 30 January 2022. It showed the two restored Krishnas together for the first time, with Balarama from the National Museum of Cambodia and the Harihara from the Musee national des arts asiatiques-Guimet in Paris, both lent for the first time, plus five sculptures from Angkor Borei. The four sculptures that travelled are four of the eight gods of Phnom Da; the rest were too fragile to move.

All eight were then reunited in the only way now possible, as life-size three-dimensional models built from photogrammetry and lidar scanning and projected on individual scrims. Visitors wearing Microsoft HoloLens 2 headsets were taken through a mixed-reality reconstruction ending inside the cave temple, with the sculpture restored to its original appearance including gold jewellery. A final gallery ran an audio-visual timeline of the eight sculptures' dispersal, narrated by Angelina Jolie and Loung Ung. A different version of the exhibition, focused on the Cleveland sculpture, was shown at the Smithsonian's National Museum of Asian Art in Washington from 30 April to 18 September 2022.

It is worth naming what the digital work is doing here, because it is not decoration. Every one of the eight gods is in a different place and several cannot be moved. A three-dimensional model is currently the only form in which the group that stood together on one mountain can be seen together, and it is also the instrument that proved which limb belonged to which body. The same technology that reunites them virtually is what dismembered the 1978 fiction.

What has never been published

The museum's own provenance list runs Leonce Rosenberg, Paris, about 1910 to 1920; Adolphe Stoclet, Palais Stoclet, Brussels, 1920 to 1949; Michele Leon-Stoclet, Brussels and Barcelona, 1949 to 1967; the Cleveland Museum of Art, 1973 to the present. The sequence numbers in that list are not consecutive, which suggests entries between Rosenberg and Stoclet and between Michele Leon-Stoclet and the museum that are not shown.

What Cleveland paid the Stoclet estate in 1973 has not been published, and there was no auction, so there is no hammer and no premium. Who carried the torso from Phnom Da to France about 1910, and under what authority, is not documented; accounts describe an anonymous removal and one popular account attributes it to a French sailor, which this record treats as unsupported. No French colonial export permit has been published, and the Ecole francaise d'Extreme-Orient's own archaeologist recorded the removal only after the fact.

The object is not the subject of a claim. Cambodia has recovered a great many sculptures from American collections in the last decade, mostly through litigation and criminal investigation, and this is not one of them. It is the case where the two museums simply worked on the problem together, which is why it belongs in this archive: it is the counter-example that the restitution debate rarely produces.