The Clark Sickle-Leaf Carpet

A small seventeenth-century Persian carpet, under three metres long, that a Washington museum sold in 2013 for the highest price ever paid for a carpet at auction. It came from the collection of William Andrews Clark, the Montana copper senator, and the museum that sold it was dismantled the following year.

The figure reported everywhere, 33,765,000 dollars, is the price with buyer's premium. It is not the hammer, and it is not what the museum received either. Three different numbers describe one afternoon and this record keeps them apart.

What it is

A knotted pile carpet in wool, silk and cotton, catalogued by Sotheby's as a sickle-leaf, vine scroll and palmette vase-technique carpet, probably Kirman, southeast Persia, measuring approximately 8 feet 9 inches by 6 feet 5 inches, which is 2.67 by 1.96 metres.

That size matters to the story. This is not a palace carpet. It is a small rug, smaller than a great many domestic rugs sold every week, and it made more money than any carpet has ever made at auction.

Its design is the one it is named for: sickle-shaped leaves, vine scrolls and palmettes, with two cypress trees running vertically through the pattern and a pair of blue and white cloudbands marking the central axis. Structurally it belongs to the group known as vase-technique carpets, and that group is the reason the object is important rather than merely expensive.

What a vase-technique carpet actually is

The name misleads. A vase-technique carpet is not a carpet with a vase on it. It is a carpet built a particular way, and the class was defined by the structure rather than by the ornament.

The defining feature is the weft. Between each row of knots run three shoots rather than the usual two. The first and third are wool and are woven at high tension; the second is at low tension and is normally silk or cotton. The result is that the warps are markedly displaced, one set pushed well behind the other, and the knot is a Persian knot open to the left.

This is why the group holds together as a group when the designs do not. Carpets that look nothing alike are the same class of object because a loom in the same tradition made them the same way, and it is why a small rug can sit in the same category as very large ones. It is also the reason the attribution to Kirman is as firm as it is: the structure travels with a place and a workshop tradition, and the pattern does not.

The definition is owed to May Beattie, who did the structural analysis that turned a loose stylistic grouping into a technical one. Naming a class of carpets after a motif and then defining it by its weft is an accident of scholarly history that the field has never bothered to correct.

The rarest pattern in the group, and the only red one

Within the vase-technique carpets, the sickle-leaf design is the rarest pattern of all. Sotheby's states the point that gives this particular carpet its standing: of the surviving sickle-leaf pieces known, the Clark carpet appears to be the only one with a red ground.

The rest are blue. The Gulbenkian carpet in Lisbon, the closest complete comparison, has a deep blue ground and a mirrored design in more typical Safavid proportions. The Clark carpet's rose-red field is, on the present state of knowledge, unique in its class.

Uniqueness of that kind is exactly what a record price is made of. A collector buying the finest example of a rare pattern is buying one thing; a collector buying the only one in a colour is buying another, and the second is not substitutable at any price.

Where the leaves came from

The sickle leaf is not a Persian invention. It descends from the Ottoman saz leaf, the curling, feathered, serrated lancet leaf that appears in imperial Ottoman design collections around 1550, and Safavid weavers took it and reworked it.

On this carpet the leaves curl, split and serrate around horizontal and angled palmettes, and are internally decorated with flowering vines, frequently split into two colours. Smaller leaves in contrasting colours sprout from the larger ones and curl the opposite way. The pattern is asymmetric across the field, with three split medallions at one end and two large half-palmettes with quarter rosettes at the other.

That asymmetry is evidence about what the object originally was, and it is dealt with below.

It is probably half of something

The pattern does not resolve on its own. Bold palmettes run along the central vertical axis and half-motifs are cut by the field edges, and the design reads as one half of a mirrored composition that would, doubled, produce the proportions a Safavid carpet of this class normally has.

Arthur Upham Pope proposed a specific explanation: that it was woven for a throne dais, a takht, so that the Shah seated on it would appear to sit within the great carpet. The idea stuck hard enough that when the carpet was exhibited in 1976 it was labelled The Corcoran Throne Rug.

It is a proposal, not a finding, and this record keeps it in that register. What is not in doubt is that the carpet as it survives is compositionally incomplete in a way that asks for an explanation, and Pope's is the only one anyone has offered.

The garden, and the trees that are not decoration

Two cypress trees pierce the pattern vertically, which is uncommon in sickle-leaf designs; among the comparanda only the Burrell and Victoria and Albert fragments carry them.

The cypress is not filler. It is indigenous to Persia, it is long-lived, and on that basis it became a symbol of immortality and was attached to the Zoroastrian god Mithra. Sotheby's catalogue essay places the whole composition in the tradition of the Persian walled garden, the paradise park that runs back to Cyrus the Great in the sixth century BC, and reads the carpet as a garden in which trees and branches issue ripe fruit and flowering blossom.

A carpet is a floor. A garden carpet puts a paradise under the feet of whoever is standing on it, and if Pope is right about the throne dais, the man standing on this one was making a claim about himself.

The price, and which figure is which

The carpet was lot 12 at Sotheby's New York on 5 June 2013, sale N09012, against an estimate of 5 to 7 million dollars. The figure reported in every account since is 33,765,000 dollars, and it is a premium-inclusive total, not a hammer price. The Associated Press said so at the time, in the sentence everyone quoting the number since has dropped: the price includes the auction house premium.

This is the single most common error made about the object. A price database entry, a press report and a dealer's comparable will all repeat 33.765 million as though it were the price the bidder called. It is not. The bidder called a lower number, and the difference is the auction house's commission.

The lower number can be reconstructed, and it is worth doing because the arithmetic checks itself. Sotheby's New York charged the buyer 25 per cent on the first 100,000 dollars, 20 per cent from 100,000 to 2 million, and 12 per cent on everything above 2 million. Run those rates over a hammer of 30,000,000 dollars and the total comes to 33,765,000 exactly. A hammer of precisely thirty million is what an auctioneer's bidding increments actually produce, and a premium schedule that converts a round number into the reported total to the dollar is not a coincidence.

So the hammer was almost certainly 30 million dollars and the premium 3,765,000. This record carries that figure marked DERIVED rather than reported, because the rate card has not been checked against a Sotheby's document. A reconstruction that shows its working is useful. A reconstruction that hides it is a fabrication.

There is a third figure, and it is the one the museum cared about. The Corcoran did not receive 33.765 million from this lot or 43.8 million from the sale as a whole: the 25 carpets made 43.8 million with premium, and the museum netted 38.4 million. The biggest number in the story never reached the seller.

What the record actually broke

The previous world auction record for a carpet was 9.6 million dollars, set at Christie's in London in April 2010. The Clark carpet made more than three times it in a single lot, and the result was reported as a record not only for a carpet but for any Islamic work of art.

It went to an anonymous telephone bidder after a contest reported as multi-bidder and lasting around ten minutes. That detail is worth more than it looks. A single determined buyer against a reserve produces a high price; several buyers holding on through ten minutes of bidding establishes that the price had depth, and that the next-best bid was not far below it. No underbidder has been identified, which is the piece of information a collector doing diligence would most want and the one auction houses least often give.

Where it came from before Clark

Sotheby's provenance section is one line long: Bacri Freres, Paris.

That is the Paris dealing house through which the carpet reached William Andrews Clark, and it is the only named owner before him. When Clark bought it, for how much, and what the firm knew of its history before that are not stated in the catalogue and are not established here. For an object that made thirty-three million dollars, the documented ownership chain before 1925 is a single dealer's name and nothing else.

Clark, and the wing he paid for

William Andrews Clark made a copper fortune in Butte, Montana, and became one of the richest men in the United States. He is also one of the least respectable of the great American collectors: his election to the Senate in 1899 was annulled after a bribery investigation, and he later returned to it anyway.

His collection went to the Corcoran Gallery of Art in Washington, reaching the museum in 1926, the year after his death, and the museum built a wing to hold it. The carpet stayed there for eighty-seven years and was exhibited regularly: Carpets for the Great Shah in 1948, The William A. Clark Collection in 1978, and Masterpieces: European Arts from the Collection as late as 2007. It travelled, to the Textile Museum in 1972, to Asia House Gallery in New York and the Fogg at Harvard for Shah Abbas and the Arts of Isfahan in 1973 and 1974, to Sheffield and Birmingham in 1976, to the Hayward Gallery in London in 1983, and to the Sackler in 2003.

This was not an object mouldering in a store. It was one of the best-published carpets in America, and the museum sold it anyway.

The deaccession, and the money it could not spend

The Corcoran consigned 25 classical sixteenth and seventeenth century carpets from the Clark collection to Sotheby's in 2013, to benefit future acquisitions.

The museum was in serious financial trouble at the time and the sale did not solve it, because it could not. Under the deaccessioning rules the American museum sector holds itself to, proceeds from selling works of art may be applied only to future acquisitions, never to operating costs. The Corcoran maintained that position publicly: 38.4 million dollars net, restricted to buying art, in an institution running out of money to open its doors.

That is the fact that makes this a provenance record rather than a market story. A museum sold its single most valuable object, cleared a world record, and was dismantled the following year, because the rule that stops collections being liquidated to pay bills worked exactly as it is designed to work.

In 2014 the Corcoran ceased to exist as an independent institution. Its collection went to the National Gallery of Art and its college of art and design went to George Washington University.

Its condition, in the auction house's own words

Sotheby's published a condition report, and it is more candid than the photographs suggest.

The pile is described as generally good and dense, ranging from a near-original tenth of an inch where it is closely shorn down to low knots and exposed foundation, particularly across the centre of the carpet. The dark browns have oxidised to the knots and foundation in places, and so has some of the deep red ground. There is scattered repiling to the dark browns and to the rose-red field, and scattered moth damage, mostly at the ends.

The listed repairs are small and specific: a reweave of about one and a half by one inch in the upper centre edge, one of about an inch square in the upper right corner, a repiled hole in the upper left corner, an inch-square reweave on the left edge, and a repaired hole a quarter of an inch across in the field. The sides keep remnants of a rose-red outer guard stripe with the warp cords partly secured by later overcasting; the ends have losses running into the light blue guard stripes.

The summary verdict is four words long and it is what a buyer pays for: sound fabric, flexible handle, dense robust weave, good saturated colour. A carpet four hundred years old that still handles well is rarer than a carpet four hundred years old.

Exporting it is not automatic

The catalogue carried a notice that most reporting of the sale ignored: a licence may be required to export textiles, rugs and carpets of Iranian origin from the United States, and buyers were told to enquire with the Office of Foreign Assets Control.

That is sanctions law, not cultural property law, and it is a different mechanism from the export controls that usually appear in provenance records. It does not protect the object or the source country; it restricts dealings in goods of Iranian origin. The practical effect is the same in one respect: a buyer of this carpet could not assume a free hand to move it across a border, and an anonymous buyer at 33.765 million dollars had a regulatory question to answer before shipping.

Where it is now

The buyer was not named. The carpet has not been reported on public display since the sale.

An object exhibited nine times between 1948 and 2008, in Washington, New York, Cambridge, Sheffield, Birmingham and London, has not been seen in public since the afternoon it set the record. That is the ordinary fate of record-setting objects and it is worth stating plainly: the sale that made it famous is also what removed it from view.