In October 2022, the National Gallery of Art in Washington announced that one of its Vermeers was not a Vermeer. The painting, Girl with a Flute, a small panel of a young woman in an extravagant hat, had for decades been catalogued, with caution, as a work by the master of Delft. After years of technical analysis, the museum concluded that it had been painted by someone in his circle, a studio collaborator, and changed the record: today it is listed as a work from the studio of Johannes Vermeer, and Vermeer appears only as a related artist. They did not identify who did paint it. Only who did not.
A few weeks after the Rijksmuseum in Amsterdam announced the list of works for its major Vermeer retrospective, the exhibition that brought together twenty-eight of the little more than thirty paintings attributed to him. And there was 'Girl with a Flute', included without qualification in the list of works by Johannes Vermeer. The Dutch museum did not use 'attributed to', nor 'workshop of', nor a question mark. It listed it as a Vermeer.
When asked about the contradiction, the head of painting at the Rijksmuseum and curator of the exhibition, Pieter Roelofs, responded with a remarkable statement. He said that in Amsterdam they viewed matters differently than in Washington, that the painting was lent to them as a 'non-Vermeer' but that they would hang it as a Vermeer, and concluded: the doubt will vanish during the flight across the ocean. The painting crossed the Atlantic ceasing to be one thing and arriving as another, without anyone having touched it.
The interest lies not in the anecdote, however amusing it may be. It is that this case resembles others of dubious attribution and is, in reality, the opposite. When a museum does not know if a sculpture is ancient or a forgery, the problem is a lack of evidence: there is no data to allow for a decision. Here, there is an abundance of data. The National Gallery conducted an exhaustive technical study, compared the painting with another sibling work from its own collection, examined pigments, layers, and brushwork, and published the results in a peer-reviewed scientific journal. The conclusion was that the person who painted the work knew Vermeer’s methods intimately and used his same materials, yet did not reach his level of delicacy and skill. The hand was that of someone trained in the workshop, not that of the master.
And with those same proofs on the table, the other museum maintained the attribution. The reason Roelofs gave is more subtle than it appears: the National Gallery had reached its conclusion with the knowledge and comparisons at its disposal, which were those of its own collection, and the Rijksmuseum, which was assembling almost every Vermeer in the world, could compare it against a much broader corpus. That is to say, he did not dispute the technical data. He disputed what they should be compared against. And that is a difference of judgement, not of measurement.
The symmetry of the scene is deceptive. The two positions are not supported by the same weight of evidence. The National Gallery published a detailed, signed, peer-reviewed study, available for anyone to scrutinise. The Rijksmuseum declared its disagreement and pointed to a comparative criterion, yet it has not published—or at least I have not found—an equivalent scientific paper that responds point by point to that analysis. To this, one must add a detail of no small significance: the painting is the property of the National Gallery, and the Rijksmuseum was the institution requesting the loan. One institution makes a ruling on its own work with a published study; the other maintains the opposite on a gallery label and in a press conference. They are two incompatible rulings, yes, but not two truths of the same standing.
Almost four years later, the situation remains unchanged. If you search for the painting today on the Washington website, it is a work from the studio of Vermeer. If you search for it on the Amsterdam website, it is by Vermeer. The same panel, the same young woman, the same hat, two official authors depending on where you enquire. And no consensus in sight: there are specialists who accept the attribution and specialists who do not, and no one has published the work that settles the matter.
We tend to imagine attribution as an almost forensic procedure. Pigments are examined, measurements are taken, comparisons are made, and science dictates who the author is. However, data does not perform attribution on its own. Someone must decide what to compare with what, how much weight to give to a clumsy brushstroke versus an identical pigment, whether a genius’s off day explains more than the hand of a diligent disciple. That decision is a judgement, and two competent judgements based on the same data may diverge. What cannot diverge is the painting: it remains exactly what it was before either of the two museums spoke.
Roelofs was joking, but his remark is truer than he intended. The doubt did not vanish during the flight, obviously. What changed upon crossing the ocean was not the painting: it was who held the authority to declare whose it was.
On the open conversation
This text is based on a verified case: in 2022, the National Gallery of Art withdrew the attribution to Vermeer of 'Girl with a Flute' and reassigned it to his studio, and months later the Rijksmuseum exhibited it and continues to present it as a work by Vermeer, with an express statement of disagreement from its curator. I maintain that this case is not one of undecidability due to a lack of evidence, but one of divergence of judgement despite abundant evidence. And I avoid presenting it as a draw: I point out that the two positions are not documented with the same weight (one published a peer-reviewed study; the other declared its discrepancy), and that ownership of the work belongs to one of the two institutions. If anyone wishes to intervene from the fields of art history, conservation, or the practice of attribution, this notebook remains open.
Sources
National Gallery of Art, record for Girl with a Flute: 'Studio of Johannes Vermeer'; 'Once cautiously attributed to Johannes Vermeer, Girl with a Flute was more likely painted by a studio associate of Vermeer' (announcement of 7 October 2022).
M. E. Wieseman, A. Libby, E. M. Gifford and D. Anchin, 'Vermeer’s Studio and the Girl with a Flute', Journal of Historians of Netherlandish Art 14.2 (2022), peer-reviewed technical study.
Rijksmuseum: press release of 1 November 2022 with the list of 'works by Johannes Vermeer' for the Vermeer exhibition (10 February to 4 June 2023); its page for the work labels it 'Johannes Vermeer, ca. 1669-1675'.
Statements by Pieter Roelofs to Het Parool, reproduced by DutchNews (1 November 2022): 'it is lent as “non-Vermeer” but we will hang it as Vermeer. The doubt will disappear during the flight across the ocean'; 'in Amsterdam we see things differently than in Washington'. Coverage of the debate in The Art Newspaper (10 February 2023).
No scientific publication from the Rijksmuseum has been located that responds specifically to the National Gallery’s study on this painting. In 2026, both institutions maintain incompatible attributions.