After I win the game I’ll tell you what the rules were, or A new Rembrandt from 1632

A painting that is ignored in the first three volumes of the Corpus of Rembrandt paintings has now been attributed to Rembrandt by the leader of the Rembrandt Research Project. His arguments give Schwartz reason to investigate his methods and ask what criteria should be addressed in judging the authorship of old master paintings. He first tells the story of how the painting was bought and sold since 1970.

Since the appearance of the first volume of the Corpus of Rembrandt paintings in 1982, the world has grown accustomed to seeing owners of de-attributed Rembrandts taking it on the chin from the Rembrandt Research Project. A well-known case is that of the Dutch industrialist Sidney van den Bergh, who in the course of selling his Head of an old man was told that the Project did not accept it as an authentic Rembrandt. Because of this, the American industrialist Alfred Bader was able to buy it for about one-fourth the price van den Bergh had been asking until then.

Over the past year and a half [since November 1997], the equally interesting reverse process has been taking place. Ernst van de Wetering, who now heads the Project on his own, has contradicted the negative (published or unpublished) judgment of the RRP on five works. In spring 1996 he re-attributed to Rembrandt a self-portrait in the collection of Queen Elizabeth and the aforesaid Head of an old man, making Bader’s painting saleable for much more than he paid for it. At a Rembrandt conference in Melbourne on 4 October 1997 van de Wetering contradicted the opinion of the former leader of the Project Joos Bruyn that The Polish rider might be by Willem Drost, and in Sydney on 7 October 1997, van de Wetering argued for the re-canonization of a painting of a young woman in an American collection (C61). Yesterday, on 4 November 1997, in the Rijksmuseum, he came out for Rembrandt’s authorship of a painting he believes is a self-portrait by Rembrandt from 1632. It forms an interesting case for the interplay between connoisseurship and the monetary aspects of collecting.

The painting in question was bought at auction at Sotheby’s in London on 8 April 1970 by the Paris art dealer J.O. Leegenhoek, a native of Bruges. It came from the collection of the heirs of the Hon. Henry Robert Brand (later 2nd Viscount Hampden), who was reported to have bought it at the Vinot sale in Paris on 29 January 1891. Never having been published, it was not taken seriously by the auction house, which called it a “Portrait of Rembrandt” by “REMBRANDT.” Had they believed in it, it would have been described as a “Portrait of the artist” by “REMBRANDT HARMENSZ. VAN RIJN,” and they would have advertised it with much fanfare. It was knocked down to Leegenhoek for a mere 650 pounds. The dealer was convinced that the painting was genuine, but no one else shared his faith. Rather than sell it for a too-low price, he took it home and gave it to his wife. From 1970 to 1982 it hung in the Leegenhoek kitchen on the Avenue Kléber and from 1982 to 1996 in the apartment above the new Leegenhoek gallery on the quai Voltaire. According to Mme Leegenhoek, it was seen in the course of time by all the important French museum officials as well as French and foreign colleagues in the trade. None of them believed it to be by Rembrandt.

In 1977, in the course of examining all the potential Rembrandt paintings in the world, the Rembrandt Research Project looked at the small half-length, a panel measuring a mere 21.8 x 16.3 centimeters. The members involved were Ernst van de Wetering and Bob Haak. Van de Wetering now claims that he “argued – though with some reservations – in favour of its authenticity” but that Haak and the other three members rejected it. (Mme Leegenhoek’s recollection is different. She thinks that three members were in favor and two against, but that the two carried the day.) The painting is not mentioned once in the massive volumes 1-3 of the Corpus.

At the request of the RRP, the Hamburg specialist in the dating of wood panels, Peter Klein, examined the work at that time. According to van de Wetering, he did not hear from Klein about the results until about three years ago. The findings were remarkable. Klein had discovered that the wood of the panel came from the same tree as that used for Rembrandt’s portrait of Maurits Huygens, a documented work whose authenticity has never been doubted. In combination with other positive indications, van de Wetering now became completely convinced of the authenticity of the work. He communicated this to the Leegenhoeks and requested permission to re-examine the panel. This was granted, and the re-examination indeed took place about two years ago. Early in 1996 van de Wetering began approaching Dutch museums and art-history journals for the public presentation of his discovery.

In the meanwhile, word was getting around. At a New York dinner table, a Dutch art collector who lives abroad heard enough to lead him to approach the Leegenhoeks with an offer. They accepted in March 1996. At the same time, the New York dealer Otto Naumann got wind of the affair and tried to buy the painting on behalf of Bader, with whom he has successfully bought other Rembrandts, including the much larger and more majestic Portrait of Johannes Wtenbogaert, which they sold to the Rijksmuseum for 10 million dollars. However, Naumann went to the wrong Leegenhoek – the son of the old couple, who was taking over the family business but did not own his parents’ private collection. When the two generations compared notes, Mme Leegenhoek told me sadly, they discovered that the old folks had sold the painting for less than half the price the son could have gotten for it. Mme Leegenhoek has not yet recovered from the ensuing consternation, or from the feeling that the price she received – in excess of a million and a half dollars – was unfairly low. Her husband has since died, in his nineties. The prominent Dutch art dealer who represents the buyer considers the price to have been “daringly high” under the circumstances.

At the request of the RRP, the Leegenhoeks shipped the painting to Amsterdam for closer examination. This led to another drama. Since the French art authorities did not believe it was a Rembrandt, the Leegenhoeks thought it safe to export it without asking the permission that is required for a “bien culturel.” However, it was anticipated that once the RRP announced its new opinion, the French customs and tax people might take a jaundiced view of the export. The buyer and sellers, advised by heavy-duty lawyers, agreed to return the painting to France and to the (strictly defined) custody of the older Leegenhoeks in order for it to be exported properly. This required examination by the experts of the Louvre, who were also in a position to block the export license if they felt the painting should remain in France. One of those experts was Jacques Foucart, curator of the Ecole du Nord. “The painting is good,” he told me. “We examined it in the laboratory; it’s a ruin but it’s genuine. But we couldn’t make a bid. They placed an absurd valuation on it – 30 or 40 million francs, as I recall. And we already have two other self-portraits from 1633 that are much better.”

Back to 4 November 1997 and to the Rijksmuseum. Ernst van de Wetering introduces as a new Rembrandt self-portrait from the year 1632 a diminutive half-length of a young man in a black hat, white collar and black coat, his body turned half to the right, his gaze passing us on the left. The panel is about one-tenth the size of Rembrandt’s other self-portraits of the 1630s. Although it has been known to van de Wetering and the Rembrandt Research Project since 1977, it was omitted from the Project’s Corpus of Rembrandt paintings. In such a case – a work in a unique format that was formerly considered so unlikely to be by Rembrandt that it did not even merit a rejection – one expects powerful new arguments to back up the attribution.

These are the arguments advanced by van de Wetering:
1. Investigation of the wooden panel on which the portrait is painted has revealed that it came from the same tree as Rembrandt’s undoubted portrait of Maurits Huygens, which is also dated 1632. Van de Wetering: “There can be no reasonable doubt that the Maurits Huygens and the present painting come from the same – Rembrandt’s – studio.”
2. Several compositional features were changed in the course of painting. VdW: This “exclude[s] the possibility that the present painting is a copy; it has to be a prototype.”
3. The signature was put on while the paint was still wet, and it has a spelling – Rembrant – that the artist only used in 1632. VdW: “The rarity of ‘Rembrant‘ signatures adds to the likelihood of the authenticity of the signature.”
4. Only between 1631 and 1634 did Rembrandt paint himself in formal clothes, as in the new work. VdW: “The costume […] adds to the likelihood of the painting being an autograph work.”

How compelling is this evidence? As van de Wetering himself admits, “Each of the arguments presented can be disputed with more or less success.” Since he does not explain how, perhaps I may be allowed to try.
1. Support: While the RRP has found various works painted by the master on wood from the same tree and linen from the same bolt of cloth, it has also found examples of the opposite: related planks and pieces of cloth painted by different hands. In all such cases, the RRP assumes that the non-Rembrandts came from the artist’s studio. However, part of the reasoning behind that is the evidence of the support, a circular argument. If we wish to be prudent, we must await the results of true random samples of Dutch paintings tested for the relationship between similarities in support and studio origins. Until such work is carried out, any conclusions based on the mere fact of such a similarity must be considered provisional.
2. Changes in composition: The Corpus of Rembrandt paintings contains several works which underwent changes of composition – for example, The good Samaritan in the Wallace Collection and The parable of the workers in the vineyard in St. Petersburg – and which are nonetheless called pupils’ copies by the RRP. This weakens van de Wetering’s claim that the new painting “has to be a prototype.”
3. The signature: Wet-in-wet signatures of characteristic form can be found in any number of paintings rejected by the RRP. Three such works, portraits dated to the years 1632-33, actually have the spelling Rembrant. Until this is accounted for, the signature on the new work adds no weight to the attribution.
4. The costume: This argument depends on the identification of the sitter as Rembrandt, but that is open to dispute. A second version of the painting under consideration, which van de Wetering calls a copy, was published in various catalogues between 1897 and 1969 without anyone ever having called it a self-portrait. Until this anomaly is explained, and in light of the ambiguous identity of several other possible self-portraits from the 1630s, it would seem wise to refrain from giving that designation to the new work. If the painting is a formal portrait of someone else, the costume would have no bearing on the attribution.

Van de Wetering deals with this situation in a manner of his own. “Each of the arguments presented can be disputed with more or less success. In conjunction, however, they reinforce each other in such a way that the evidence amounts to what in the field of art history comes closest to actual proof.” I wonder about that. Why should four arguments each of which is debatable in itself reinforce rather than weaken each other? However one judges the proofs and my devil’s-advocate objections to them, van de Wetering certainly overplays his hand when he writes that “this accumulation of converging arguments renders it unnecessary to use the type of arguments connoisseurship is bound to rely on – arguments based on opinions regarding the style and quality of a painting.” If one wants to attribute this painting to Rembrandt unreservedly, as van de Wetering does, then one cannot escape judgments of style and quality. And whereas the new painting bears all the stylistic marks of a Rembrandt from the early 1630s, it falls short in terms of the standard of quality maintained until now by the Rembrandt Research Project itself. Compared to the portrait of Maurits Huygens or the self-portrait in Glasgow, both from 1632, the new work lacks definition, focus, and finish and contains some disturbingly inferior passages, including the entire face. Van de Wetering blames this on the condition of the work. Much of the original surface is abraded. If this is so, one may ask, how can we ever be sure that the missing brushwork was by Rembrandt? Does the painting improve in quality if we imagine that it was?

Ernst van de Wetering’s move, along with his other recent reversals of RRP attributions, should bring about a renewed discussion of all the many considerations involved in the assessment of artistic authorship: materials, techniques and signatures; studio practice, commercial copying, collaboration and patronage; likeness and iconography; ideas about personal identity, intellectual property and copyright; documentation and provenance; condition and restoration; style and quality; myth, hype, deceit and market mechanisms; reception, reputation, authority and consensus. These features line up consistently only in a small number of Rembrandt paintings. Most are a mixture of contradictory indications. If one insists nonetheless on arriving at a categorical attribution – “Definitely by Rembrandt” or “Definitely not by Rembrandt” – one has to allow some criteria to weigh above others. Until now the prevailing criteria were style and quality. Van de Wetering would like to get away from these subjective measures and still come up with categorical attributions. To do this, he looks for “converging” indications of other kinds. To me, this seems like wanting to have your cake and eat it too – proffering a set of ad hoc arguments as a definitive judgment.

In the meanwhile, the art market is faced with a dilemma. By the standards implicit in van de Wetering’s un-de-attribution of heretofore rejected Rembrandts, a far larger number of de-attributed paintings must now be taken with renewed seriousness. By my estimate, well over a hundred such works are now back in play. The amounts involved are astronomical. Until two years ago, the new Rembrandt was not worth much more than the two or three thousand guilders paid for it in 1970; after van de Wetering’s change of mind it was sold for two-and-a-half million. Although not every rejected Rembrandt is worth as little as this painting was, potential gains of several hundred million guilders have now been created. Before trying to cash in on this opportunity, it might be advisable to await the clarification of the new criteria for authorship implicit in van de Wetering’s recent attributions, and for the reactions of his colleagues and of the trade. Rembrandt connoisseurs like to say breezily that every generation needs its own new image of the master and his work. This might be fine for the livelihood of art historians, but who wants to bequeathe such art-historical funny money to one’s heirs?

© 1997 by Gary Schwartz. Published in Dutch and German translation as:
“De herijking van een Rembrandt,” Het Financieele Dagblad, 5 November 1997, p. 2 and “Auteurschap versus kwaliteit in de nieuwe Rembrandt,” Het Financieele Dagblad, 13 November 1997, p. 2

“Holz vom selben Stamm: der Trend hat sich umgekehrt: wieder ein Neuzugang zum Werk Rembrandts,” Frankfurter Allgemeine Zeitung, 6 November 1997, p. 41

Posted on the Schwartzlist 21 July 2010.

P.S. On 19 September 2005 the self portrait of 1632 was put up for sale at the Noortman gallery in Maastricht, to take the place of Aeltje Pietersdr. Uylenburgh. (See Schwartzlist 240.) The price is $10 million, the equivalent of about 60 million French francs in 1996. I don’t know if Mme Leegenhoek is still alive. If she is, I hope this news does not reach her.

Responses to

8 October 2005: Received from reader Ralph Lieberman

I am in receipt of #241 and read the back issues you attached. I agree with your evaluation of the weight of the arguments offered in favor of the putative Rembrandt self-portrait and it reminded me of a discussion I had a long time ago with a colleague who is an anthropologist. He told me that social scientists find entirely laughable the way people in the humanities present arguments: “there is an excellent probability that this is true, in which case the argument in favor of that is much strengthened, which lends support to my claim that what happened was ….” The only rational way to think about it, he insisted, is to halve the probability with each new level. If A could be true and if so it supports the argument for B, and if B is true it means there is strong evidence for C, by the time you have gotten to C there is in fact only a one in eight chance that it is correct. A string of maybes, or even of probablies, gets weaker and weaker the longer it goes on, but humanities types imagine that a possibility, even a probability, survives undiminished through any number of stages of hypothesis.