The Ardabil Carpet

Two carpets, not one. A matching pair woven in Safavid Iran and dated 1539-40, designed to lie side by side and fill a square room. One is in London and is complete. The other is in Los Angeles and is not, because before either was sold its outer borders and part of its lower field were cut away and used to mend the one now in London.

Almost every popular account of the Ardabil Carpet describes a single object. The V&A's own catalogue record gives the twin the wrong inventory number in one place and the right one in another. This record separates the two carpets, states which is which on every plate, and records what was taken from which.

There are two Ardabil carpets, and most accounts describe a hybrid of both

The object usually called the Ardabil Carpet is the one in the Victoria and Albert Museum, museum number 272-1893. It is complete, about ten and a half metres by five and a third, and it is the carpet reproduced on posters, in textbooks and in the Jameel Gallery. It is not the whole story, because it was made as one of a matching pair, and the other carpet still exists.

The second carpet is in the Los Angeles County Museum of Art, accession number 53.50.2, a gift of J. Paul Getty in 1953. LACMA's own catalogue calls the two a matched pair whose designs are virtually identical apart from small variations in the ligatures of their inscriptions. The V&A calls the Los Angeles carpet the twin, notes that it is only partially intact, and records that fragments of both carpets exist in other public and private collections.

What follows is written to keep them apart. Where this record says the London carpet it means 272-1893. Where it says the Los Angeles carpet it means 53.50.2. Every picture in the gallery below names which of the two it shows.

What was taken from the Los Angeles carpet, and what it still has

This is the fact the record exists to get right. Before the South Kensington Museum bought the London carpet in 1893, the dealers had a complete carpet to sell and a damaged one, and they made the complete one out of both.

LACMA's curatorial note, revised in 2024, states it plainly: sometime before the V&A acquired its Ardabil in 1893, the outer borders and a section of the lower field are believed to have been removed from the Los Angeles carpet in order to repair the one now in London, and the Los Angeles carpet was subsequently given a new outer border. The V&A's record says the same thing from the other end, that pieces were taken from the Los Angeles carpet to repair the border of the London carpet, and that the London record carries a diagram in which dotted lines mark the patches.

So the Los Angeles carpet lost its outer borders, lost a section of the lower field, and lost most of the corner quarter-medallions, which LACMA describes as only partially preserved. It did not lose its central medallion. The great yellow sunburst with its sixteen radiating pendants is still there, both mosque lamps are still there, and the inscription panel naming Maqsud of Kashan and the year 946 is still there. A published claim that the Los Angeles carpet has no medallion is wrong, and it is wrong in a way that would make every plate in this record unusable if it were repeated here.

What the Los Angeles carpet has instead of its original border is a replacement, and the carpet is smaller in both directions: 718.8 by 400.1 centimetres against 1044 by roughly 532. The difference is not only the missing border. Part of the field went too.

Made as a pair, for a room that was square

The V&A gives the reason the carpets are the size they are. Each measures approximately ten metres by five, and the two were designed to lie side by side and occupy, together, a square space under the dome of the Jannat-sara, one of the new buildings Shah Tahmasp added to the shrine complex at Ardabil.

That is a design argument as much as an archaeological one, and it carries a consequence for the reading of the whole object. A pair of carpets that only makes sense as a square is a pair made for a specific floor, not for a market. The V&A notes that the chamber would have served as a royal reception hall, and that Tahmasp used it in 1544 when he received the exiled Mughal emperor Humayun. Humayun walked over these carpets, went home with Safavid court artists in his entourage, and Persian court design entered Mughal India through that visit.

The proposal is not unopposed, and the objection is a physical one rather than a stylistic one. The measured floor of the domed chamber usually identified with the carpets does not comfortably take two carpets of this size laid flat, and at least one survey has concluded that they could not have been placed there unfolded. A record that repeats the Jannat-sara story without the objection is repeating a museum label, not the evidence.

How they were made

Warp and weft are white silk. The pile is wool, in ten colours: dark red, red, light red, yellow, green, dark blue, blue, light blue, black and white. The knot is asymmetrical and open to the left, at an average of 5,300 knots per square decimetre, which the V&A gives as 340 knots per square inch. The warp is Z2S, depressed, at 126 threads per decimetre, with three shoots of paired weft threads after each row of knots.

The V&A is careful about the knot count in a way that matters here. It records that the count varies throughout the structure, that 340 per square inch is an average, and that the Los Angeles carpet has been recorded at an average of 350 per square inch. Near-parity in knot density across two carpets is technical evidence, not connoisseurship: it supports the accepted proposal that both were woven by the same team at the same time. LACMA goes further and reports the plausible proposal that the two were woven simultaneously on back-to-back looms.

That parity is reported to be contested. Rexford Stead's 1974 monograph for the Getty is cited in later writing as publishing counts that do not agree with the museums', giving a lower density for the London carpet than for the Los Angeles one and arguing from the difference that the two were not woven side by side. This record could not read Stead at source and does not reproduce his figures. What is recorded here is that a disagreement about the knot counts exists in the literature, that it bears directly on the simultaneous-weaving claim both museums make, and that it needs checking against the book.

Wool for the pile rather than silk is a deliberate choice and not a cheap one. The V&A points out that wool holds dye better than silk does, which is why a carpet meant to read as ten distinct colours across fifty square metres is knotted in wool on a silk foundation rather than in silk throughout.

An older V&A gallery panel estimated about four years to weave the pair, and inferred a royal commission from the cost that implies. LACMA says each would have taken several years. Neither museum claims a documented contract.

Nobody agrees how many knots are in it

The total knot count of the London carpet is one of the most quoted figures in Islamic art and there is no agreed number. Published totals range from roughly twenty-five or twenty-six million at the low end to well over thirty million at the high end, depending on which density is used and whether the figure is taken over the whole carpet or the field alone. LACMA's own public writing has put its smaller carpet at about fifteen and a half million.

Every one of those totals is a multiplication, not a count. Nobody has counted the knots in either carpet. The correct way to read the figure is as a derived quantity that inherits all the uncertainty of the density it was derived from, and the density is explicitly an average over a structure the V&A says varies throughout.

The design is a book cover the size of a room

The field is dark blue. At the centre is a yellow medallion, a shamsa, with sixteen satellite ovals radiating from it. A quarter of the same medallion is repeated in each of the four corners. Above and below the centre, on the long axis, hang two red mosque lamps. The rest of the ground carries two independent systems of spiralling leafy scrolls laid one over the other, one with thick dark red stems, the other with thinner cream stems.

The V&A traces the vocabulary directly: designs of this kind were produced for manuscript illumination and bookbinding in the same period and the same court workshops, and the carpet composition is derived from them. That is why it reads as a single vast illuminated page rather than as a repeating textile pattern. It is also why the medallion-and-corner scheme is usually associated with Tabriz, the Safavid capital, where the court design studio was.

The lamps carry the argument. The V&A reads them as an image of divine light, referring to the Safavid claim of descent from the Prophet, whose nature is described in terms of light, the Nur-Muhammadi. Shah Tahmasp was completing an expansion of the shrine of Shaykh Safi al-Din at the time these carpets were made, and the shrine was the basis of his family's claim to rule: descent from a saint, and beyond him from the Shia Imams and the Prophet. A carpet with lamps of divine light hanging over the floor of that shrine's reception hall is a dynastic argument in wool.

The optical correction almost nobody notices

The two lamps are not the same size, and neither are the oval pendants nearest to them. This is not damage or repair. A V&A wall panel of 1998 records it as a deliberate correction for foreshortening.

The reasoning it sets out is precise. Weaving began at one end. Dignitaries would sit at that end, on the carpet or on low cushions, looking down its length against the direction of the pile, which is where the colours read best. From that seat the far lamp is further away and would appear smaller, so the far lamp was woven larger. Seen from the position the designer intended, the two lamps and their pendants appear the same size.

It is one of the few pieces of evidence anywhere in the object about how it was actually meant to be used, and it comes from the object itself rather than from a document.

How they left Iran

The V&A's object history is more specific than the popular account. Ziegler and Co of Manchester, a trading firm with offices at Tabriz and at Sultanabad, now Arak, bought the carpet in 1888 from the shrine of Shaykh Safi at Ardabil, either directly or through the Tabriz-based carpet dealer Hildebrand Stevens, and repaired it heavily. In 1889-90 the shrine authorities completed new restoration work on buildings of the complex. The V&A puts the inference carefully: arguably the carpet was sold in order to fund that renovation.

The version of that story usually told is more colourful and less supported. It holds that an earthquake damaged the shrine and that the carpets were sold to pay for the repairs. The earthquake is the weak link. It is a later tradition rather than a documented event tied to this sale, and the V&A itself does not assert it as fact; it says arguably. This record follows the museum and does not turn a plausible motive into a recorded cause.

By 1892 the London carpet was on display in the showroom of Vincent J. Robinson and Co Ltd at 34 Wigmore Street. Edward Stebbing, Robinson's managing director, published a monograph on it the same year, The Holy Carpet of the Mosque at Ardebil, and The Times ran a piece headed An Extraordinary Carpet on 26 May 1892. The object was being made famous before it was sold, by the man selling it.

Two nineteenth-century travellers had recorded carpets at the shrine before any of this: Keith Edward Abbott in a Foreign Office narrative of 1843, and W. R. Holmes in Sketches on the Shores of the Caspian in 1845. Those sightings are the reason the Ardabil attribution is taken seriously at all, and they are also the reason it is not certain, because a traveller describing large carpets in a shrine is not the same as an inventory naming these two. Eighteenth-century inventories of the shrine's contents survive and have been studied; what they do not do is name a pair of carpets that answers to these.

What the museum actually paid, and the figure that is usually quoted

The V&A's catalogue record and the V&A's own public article do not agree, and both are current.

The catalogue record for 272-1893 gives the sequence in detail. John Edward Taylor, a wealthy private collector, brought the carpet to the Director's attention over the summer of 1892. In January 1893 the Museum offered 1,500 pounds. That was not enough for Edward Stebbing at Robinson's, and Taylor volunteered to raise the shortfall from private donors. By March 1893 the money was still short, and after consulting two of its Art Referees, William Morris and Frederic Leighton, the Museum raised its offer to 1,750 pounds. Stebbing accepted. The donors named in the accession register are Taylor, who guaranteed 250 pounds, Morris, who offered 20 pounds, A. W. Frank, E. Steinkopff, and in the register's phrase other gentlemen.

The V&A's public article on the carpet gives the price as 2,000 pounds in March 1893. This record follows the catalogue entry at 1,750 pounds, because it is the account that cites the accession register and reconstructs the negotiation, and it flags the discrepancy rather than averaging it. A reader who finds 2,000 pounds elsewhere has not found a different sale.

The 20 pounds Morris put in is worth holding on to. The most influential designer in Britain contributed roughly one per cent of the purchase price, and his real contribution was the opinion, not the cheque. His written assessment for the Museum called the carpet of singular perfection, and logically and consistently beautiful, and it is the sentence that bought it.

The twin's American chain

In 1892, before the London sale closed, Vincent Robinson sold the second carpet abroad. The buyer was Charles Tyson Yerkes, the Chicago and London traction magnate, and the price was 80,000 dollars. The V&A adds a condition to the sale that is worth stating exactly as the museum does: apparently on condition that the carpet should not return to Britain.

That condition, if it was real, is a piece of dealer behaviour rather than law. Robinson was about to sell the London carpet to a national museum as a unique object. A second, near-identical carpet arriving in London would have been inconvenient.

From Yerkes the carpet went to Joseph Raphael De Lamar, then to the dealer Joseph Duveen, then in 1938 to J. Paul Getty, who gave it to the Los Angeles County Museum of Art in 1953. Yerkes died in 1905 and his collections were dispersed at auction in New York in 1910; De Lamar died in 1918 and his were dispersed in turn. The second Ardabil carpet therefore passed through two full American estate dispersals before Duveen placed it with Getty, which is the reason its chain is better documented than most nineteenth-century carpet provenance.

Ten minutes on the hour, and twenty in every sixty

In 2006 the V&A built a case in the centre of the Jameel Gallery for the London carpet and set a lighting regime that is itself part of the object's history: the carpet is lit for ten minutes on the hour and ten minutes on the half hour, and is otherwise in near darkness, to preserve its colours.

Note the arithmetic, because it is usually got wrong. The gallery label says on the hour and half-hour, which is two ten-minute periods, twenty minutes of light in every sixty. The common paraphrase that the carpet is lit for ten minutes an hour halves the real exposure.

The regime is a public admission that the object is consumed by being seen. It also changes what visiting means. The most famous carpet in the world is dark for two thirds of the day, and the museum tells you so on the label.

It has almost never been on a floor

A carpet is a floor covering and this one has spent almost none of its museum life on a floor. For most of the twentieth century the London carpet hung vertically behind glass, which is how a textile that heavy is displayed when the alternative is people walking on it.

The 2006 Jameel Gallery case put it flat again, in a horizontal vitrine at the centre of the room. That is closer to the way it was meant to be seen than anything the museum had done before, and it is still not the way it was meant to be seen, because the designer's whole optical correction assumes a viewer sitting on the carpet at one end of it.

The name is a claim, not a fact

Ardabil is where the carpets are said to have come from. It is not where they are documented to have been made, and the strength of the shrine attribution has been questioned in print.

The critical case runs roughly as follows. The shrine provenance rests on a nineteenth-century dealer's account and on travellers' sightings of large carpets at the shrine, not on an inventory. The room the pair is supposed to have furnished may not take them. The eighteenth-century shrine inventories that survive do not obviously list them. And the man who first published the carpet at length, Edward Stebbing, was the managing director of the firm selling it, which gives the founding account of the object a commercial interest in its holiness.

Dorothy Armstrong's 2020 study in Iran treats the whole thing as a case of invention: how a Persian artefact was appropriated and transformed into the object the Western market and the Western museum needed it to be. That is an argument about reception, and this record keeps it in that register rather than presenting it as a finding about the sixteenth century. But it is the reason the words said to have come from appear in both museums' own descriptions, and a reader is entitled to know that the two most careful institutions in the field both hedge.

Where they are now

The London carpet is in Room 42, the Jameel Gallery, at V&A South Kensington, under the 2006 lighting regime, museum number 272-1893.

The Los Angeles carpet is on view in LACMA's David Geffen Galleries in a display titled Persianate Arts: The Ardabil Carpet and Manuscript Painting. LACMA's curatorial note was revised in 2024 and its Unframed blog ran the carpet as part of a 50 Works 50 Weeks series in April 2026. For most of the twentieth century the Los Angeles carpet was the one nobody photographed. It is now the centre of a named gallery display, which is the first time in its museum life that it has been presented as an object in its own right rather than as the other one.

Both carpets are easier to study from a desk than they have ever been, and unequally so. The V&A publishes an open IIIF manifest and a machine-readable record for the London carpet, which means anyone can zoom into its knots in a standard viewer without asking permission, and the Google Art Project capture of it runs to eight thousand pixels on the long side. No equivalent open manifest for the Los Angeles carpet was found. The carpet that was cut to repair the other one is still the harder of the two to look at.