The Massacre of the Innocents

An oak panel of about 1611, bought around 1700 by the Prince of Liechtenstein together with the Samson and Delilah now in the National Gallery in London, catalogued in 1767 under a name almost nobody recognises and by 1780 as the work of a Rubens assistant, sold out of the princely collection in 1920 to an Austrian family, and lent from 1923 to a monastery in upper Austria by an owner who disliked it. In 2001 George Gordon of Sotheby's saw it and gave it back to Rubens. On 10 July 2002 it made 45 million pounds at the hammer, 49.5 million with fees, then the highest price ever paid at auction for a work of art in the United Kingdom and for any Old Master anywhere.

This is the best documented case there is of an auction house creating value by changing an attribution, and it inverts the usual sleeper story. Here the house that stood to earn the commission made the reattribution before the sale, so the gain went to the elderly consignor rather than to a dealer who had spotted something in a saleroom. The finder took no risk and captured no margin; the seller captured all of it.

The pair matters as much as the picture. Both Rubens paintings bought from the Forchondt brothers around 1700 were reduced in the same Liechtenstein inventories to the same assistant, Jan van den Hoecke. One was recovered in 1929 by Ludwig Burchard; this one waited another seventy-two years. And the attribution of the picture that authenticated this one has been under sustained attack for four decades. This record leaves both positions standing.

What it is, and the error that keeps being copied

An oak panel, 142 by 183 centimetres and 1.9 centimetres thick, weighing 93.9 kilograms, showing about twenty figures in a compressed shallow space: soldiers seizing and killing infants, mothers clawing at them, a dead child in the foreground, a classical architecture of columns and an arch closing the scene at the back.

It is on panel. That has to be said flatly because the most widely consulted free reference on the picture states that it is oil on canvas, in the same infobox in which it records a thickness of 1.9 centimetres and a weight of 93.9 kilograms, neither of which describes a canvas. The Art Gallery of Ontario, which owns it, captions it oil on panel. The Antiques Trade Gazette, reporting the sale on the day, called it a superbly preserved panel painting. Support is not a detail. It governs dendrochronology, the single most powerful dating tool available for a Flemish picture of this date, and a record that gets it wrong forecloses the strongest evidence there is. It is also the test that was applied to the companion picture in London, where dendrochronology confirmed a date in the right period.

The condition is the second thing worth knowing, and it is unusual. Panels of this size and age are normally thinned, cradled, cut down or transferred. This one was described in 2002 as superbly preserved, which is part of why the price went where it did. An attribution argument is much easier to win on a surface that has not been repainted.

Why a painter in Antwerp in 1611 painted a massacre

The subject is from the Gospel of Matthew, chapter 2, verses 16 to 18: Herod, deceived by the Magi, orders the killing of every male child in Bethlehem under two years old. It is a few verses of text and it produced one of the most physically violent subjects in Christian art.

Rubens had returned to Antwerp in 1608 after eight years in Italy, where he had looked hard at Caravaggio, and he painted this within three or four years of coming home. Antwerp in 1611 was two years into the Twelve Years Truce that had frozen the war with the northern provinces. The city had been the site of real massacre: in a single year more than eight thousand of its citizens had been killed by Calvinist and Catholic forces alike as Spain fought to hold the Netherlands. A hundred miles north, the rebel Prince Maurits had commissioned Cornelis van Haarlem to paint the same biblical subject for the town hall at Haarlem, as propaganda about Spanish atrocity. Antwerp remained Catholic and became a centre of Counter-Reformation thought, and Rubens painted the subject for a Catholic city that had been on the receiving end of the thing depicted.

The bodies are studied, not invented. Rubens worked from ecorche figures, anatomical models with the skin removed, which is why the musculature of the soldiers reads as it does. He was also a man of a particular intellectual formation. His brother Philip had been a pupil of the Leiden neostoic Justus Lipsius, and at almost exactly the moment of this picture Rubens painted the four of them, himself, Philip, Lipsius and Jan Woverius, seated under a bust then believed to be Seneca. A painting of the slaughter of children, made by a man whose household reading was Lipsius on constancy under public calamity, in a city that had buried eight thousand of its own, is not horror staged for its own sake. It is a Catholic subject about innocent suffering, painted where innocent suffering had lately been a municipal fact, by someone who had a philosophical vocabulary for it.

The catalogue history, which is the attribution history

The sequence of catalogue entries on this panel is the record. Each successive description is a dated statement of what the best-informed people around the object believed, and the errors are as informative as the corrections.

Around 1700 the Forchondt brothers, Antwerp dealers, sold two Rubens paintings to Johann Adam Andreas I, Prince of Liechtenstein, whom they knew through his father Karl Eusebius. One was this Massacre. The other was the Samson and Delilah now in the National Gallery in London. Both received the Liechtenstein family seal, and drawings of 1815 show them hanging side by side in the Garden Palace in Vienna.

In 1767 Vincenzio Fanti catalogued the collection and gave the Massacre to Franciscus de Neve the younger. In an anonymous Liechtenstein inventory of 1780 it was given instead to Jan van den Hoecke, one of Rubens's assistants, and described as after Rubens. That is the wording that mattered: not a wrong master, but a demotion from invention to repetition. A picture described as after Rubens by an assistant is, in market terms, a copy, and it is filed, priced, lent and forgotten as one.

The same thing happened to its companion. When the Samson and Delilah was in the Liechtenstein collection in the eighteenth century, its painter too was identified as Jan van den Hoecke. Two autograph Rubens paintings, bought together, hung together, and reduced together to the same assistant by the same cataloguing apparatus. The Samson was sold in Paris in 1880 and recognised again by Ludwig Burchard in 1929. The Massacre had to wait another seventy-two years.

It was sold out of the princely collection to an Austrian family in 1920 and lent in 1923 to Stift Reichersberg, an Augustinian monastery in upper Austria, where it stayed for close to eighty years. The consignor in 2002 was an eighty-nine-year-old Austrian woman who had inherited it and had always disliked it, which is the stated reason it was in a monastery rather than in her house.

The copy that stood in for it, under a third wrong name

While the panel hung in the monastery, the composition itself was not lost. It was known from a large workshop version, oil on canvas, 162.5 by 234 centimetres, which the Royal Museums of Fine Arts of Belgium bought in 1902 from Edmond Huybrechts of Antwerp as a work by Antoon Sallaert.

So for most of the twentieth century the situation was this. The autograph panel was in an Austrian monastery catalogued as Jan van den Hoecke after Rubens. The workshop canvas was in a Brussels national museum catalogued as Antoon Sallaert. Between them the composition carried three names and none of them was Rubens. The Brussels picture is still in Brussels and is now given to the workshop of Rubens; it did not move, and its catalogue entry changed instead.

Two objects, one design, and the whole difference in their fates is a line of text in a catalogue.

What George Gordon did in 2001

George Gordon joined Sotheby's in London in 1980 in Old Master Drawings and moved to Old Master Paintings in 1990. In 2001 he saw the Massacre and concluded that it was by Rubens himself.

The argument was comparative rather than technical. The Samson and Delilah, painted around 1609 to 1610 and by then in the National Gallery in London, had come out of the same Liechtenstein collection and had hung beside this picture in Vienna. Set the two together and the handling, the palette and the compression of figures against a shallow architectural back wall belong to the same hand at the same moment. David Jaffe of the National Gallery is credited with recognising the connection as well, on the same basis. Sotheby's has since described the picture as having been unknown for nearly 250 years, which dates the loss from the 1767 catalogue rather than from the 1780 inventory.

What did not happen is as important as what did. There is no published dendrochronological report on this panel, no technical dossier and no committee. The reattribution that produced the highest price ever paid for a rediscovered painting was, on the public record, an act of connoisseurship by one specialist standing in front of a picture in Austria, corroborated by a second scholar making the same comparison. That is the honest description of the mechanism, and it is worth stating plainly in a decade when attribution is increasingly discussed as though it were a laboratory procedure.

Hammer against premium, and the two dollar figures nobody separates

The sale was at Sotheby's, London, on 10 July 2002, in the Old Master Paintings sale. Bidding opened low, jumped from 3.8 million to 6 million pounds in one step, and then climbed in increments of a million. The lot was knocked down in the room to the Mayfair manuscript dealer Sam Fogg, bidding for Kenneth Thomson, second Baron Thomson of Fleet.

The hammer was 45 million pounds. The Antiques Trade Gazette, reporting on the day and using hammer prices as its house style, put it in the headline: the picture joined the world record holders at 45 million. The premium-inclusive total was 49.5 million pounds, which is the figure the BBC and almost everyone since has reported, usually without saying which of the two it is.

The two dollar figures in circulation are those same two numbers converted, and they are routinely quoted as though they were rival estimates of one price. HENI records the sale at 69.69 million dollars. Other contemporary reporting gave 76.8 million. At the exchange rate of July 2002, 45 million pounds is about 69.7 million dollars and 49.5 million pounds is about 76.8 million. The first is the hammer, the second is the total. Neither is wrong and they are not the same measurement.

The fee is therefore about 4.5 million pounds, or ten per cent of hammer. That is consistent with the Sotheby's London schedule then in force of 19.5 per cent on the first 70,000 pounds and 10 per cent above it, which on a 45 million pound hammer produces 13,650 plus 4,493,000, a premium of 4,506,650 and a total of 49,506,650 pounds. That schedule is DERIVED here from the two published figures and has not been checked against a printed Sotheby's premium chart for July 2002, and no source consulted gave the total more precisely than 49.5 million.

One comparison sizes the event. The Rubens auction record before this sale was 7.5 million dollars, set by Portrait of a Man as Mars in January 2000. This hammer exceeded it by roughly eight hundred and thirty per cent.

The inversion: who captured the value

A sleeper, in the ordinary sense, is a picture a dealer buys cheaply because he knows something the cataloguer did not, and the gain accrues to the finder while the seller learns about it in the newspaper. That is the distributional complaint Ben Lewis has made in print, and it is the reason the practice is argued about at all.

This case runs the other way and is the strongest available answer to the complaint. The reattribution was made by the auction house that was going to sell the picture, before it was catalogued, in the ordinary course of preparing a consignment. Sotheby's therefore raised the value of an object it was about to sell on behalf of an elderly woman who had inherited it, disliked it, and lent it to a monastery to be rid of it. The entire difference between a van den Hoecke copy and an autograph Rubens went to her. Sotheby's took its commission on the larger number, which is a fee for work done rather than a margin on a purchase.

No finder bought this cheaply, restored it at his own expense and resold it. There is no injured consignor and no one to write a resale right for. When the practice is defended, this is the example, and it is the reason a record of it belongs beside the ones where a dealer paid five hundred pounds in a regional saleroom.

The yardstick is itself under attack

An attribution built on a comparison is only as strong as the picture it is compared with. The Massacre was given back to Rubens because it matched the National Gallery's Samson and Delilah. That picture has been contested for as long as the National Gallery has owned it.

It was commissioned by Nicolaas II Rockox, lord mayor of Antwerp and a personal friend of the painter, to hang above a seven-foot mantelshelf in his house, which is why the composition is built to be seen from below. It was sold for charity when Rockox died in 1640 to a purchaser who is not recorded. A Samson and Delilah bought by Hans-Adam I of Liechtenstein in 1700 is probably it. It went through Paris in 1880, was recognised by Ludwig Burchard in 1929, and was bought by the National Gallery at Christie's in 1980. That price is reported both as 2.5 million pounds and as 5 million dollars; at the 1980 rate those are the same figure in two currencies, not two prices.

Euphrosyne Doxiadis, an artist and scholar of the Fayum portraits, argues that the picture differs in detail from copies made in Rubens's lifetime, that it does not use the layered glazing Rubens had mastered, and that its provenance cannot be documented between 1641 and 1929. The detail most often cited is at the edges: Frans Francken the Younger's Banquet at the House of Burgomaster Rockox, painted between 1630 and 1635, shows the picture above the mantelpiece with Samson's foot wholly inside the frame and three soldiers in the doorway, where the National Gallery painting is cropped at both sides and has five. Michael Daley of ArtWatch UK and the independent researcher Kasia Pisarek have made related arguments. In September 2021 the Swiss company Art Recognition, with Carina Popovici, trained a classifier on Rubens paintings and reported a 91 per cent probability that the Samson and Delilah is not by him. Against all of that, dendrochronology of the panel confirms a date in the right period, and the attribution is accepted by most of the scholarly community. The National Gallery maintains it.

Iconotheca does not adjudicate. It records the structural fact, which no side disputes: the reattribution that produced the highest price ever paid for a rediscovered painting was anchored on a picture whose own authorship is denied by named scholars and by a published machine analysis. If the Samson and Delilah were ever downgraded, nothing physical about the Toronto panel would change and every argument made about it in 2001 would have to be restated from a different starting point. The two came out of the same collection, bought together from the same dealers, hung together in Vienna and demoted together to the same assistant. Whatever is decided about one is evidence about the other.

Kenneth Thomson, and where the panel went

Kenneth Thomson, second Baron Thomson of Fleet, inherited a newspaper and information business from his father Roy Thomson and was for a period the richest man in Canada. He collected on a scale that produced a named collection rather than a house full of pictures, and he had already committed that collection to the Art Gallery of Ontario in Toronto.

After the sale the panel was lent to the National Gallery in London, where it hung near the Samson and Delilah that had authenticated it. Thomson died in 2006. In 2008, as the AGO reopened after its rebuilding, the picture was transferred to Toronto, where it carries the accession number 2014/1581 in the Thomson Collection.

In 2018 and 2019 it went back to Antwerp, on loan to the Rubenshuis, the house in which it was painted, for the first time since it left the city. It returned to Toronto for Early Rubens, an exhibition of the artist's work of 1609 to 1621 organised with the Fine Arts Museums of San Francisco, which opened at the AGO in October 2019 with fifty paintings, prints and drawings, alongside the AGO's oil sketch of the Raising of the Cross and loans from the Metropolitan Museum, LACMA, the Rijksmuseum, the British Museum and the Getty.

The AGO publishes the panel through an open image viewer, which means a student anywhere can examine the surface at high magnification without asking anyone's permission. For an object that spent eighty years hanging in a monastery under the wrong name, that is the most consequential change of all.

What an attribution is worth, measured on one composition

Three objects carry this design and their histories are close to a controlled experiment in the value of a name.

The panel in Toronto was catalogued as a copy by an assistant and hung in a monastery because its owner did not want it. Recatalogued as Rubens, it made 45 million pounds at the hammer in a single afternoon, and it is now the centrepiece of a national gallery's collection, lent to the artist's own house in Antwerp and published in high resolution to anyone who wants it.

The canvas in Brussels was bought in 1902 as an Antoon Sallaert, is now given to the workshop of Rubens, and has stayed exactly where it was. Its catalogue entry changed and nothing else did.

The second Massacre, painted by Rubens between 1636 and 1638, twice the size and also on panel, has been in Munich since 1706 and has never been in doubt. It has therefore never been news, has never made a price, and is the only one of the three whose story consists of hanging on a wall.

Nothing about the physical object in Toronto changed between 1780 and 2002. Its price changed by a factor that no measurement of the object could produce, because what was bought in 2002 was not the panel. It was the sentence attached to it.