Portrait of a Girl

An oil portrait of a young woman on a cradled oak panel, found by an appraiser among stacks of pictures in the attic of a Camden, Maine estate, catalogued as after Rembrandt against an estimate of ten to fifteen thousand dollars, and sold on 24 August 2024 for a hammer of 1,175,000 US dollars. The press treated it as an attic lottery ticket. It is something considerably more interesting: a picture with a documented nineteenth-century history that nobody reported.

It was published as a Rembrandt by Wilhelm von Bode, reproduced as a Rembrandt in the Klassiker der Kunst survey of 1908, catalogued as a Rembrandt by Hofstede de Groot in 1914, and hung as a Rembrandt in the Mauritshuis from 1893, on loan from Abraham Bredius, that museum's former director and the foremost Rembrandt authority of his generation, who owned it. The Mauritshuis catalogue of 1910 records it as signed with the monogram RHL. The 2024 lot record describes it as unsigned. No published argument for the demotion has been found, and nobody has examined the corner where the monogram was recorded.

The attribution remains open and this record leaves it open. Gary Schwartz rates the chance of autograph authorship as extremely large and proposes testing it against an accepted Rembrandt in Vienna; Volker Manuth leans against; Betsy Wieseman explains why the question is structurally hard, since Amsterdam painters of the 1630s were working to erase the very traits attribution depends on. All of them say the same thing about method: examine the object first. No authentication body has ruled, and none of the routine technical work has been reported. No ruling has been published as of 2 August 2026.

What it is

A small oil portrait of a young woman on a cradled oak panel, seen half length without hands, turned slightly to the left with her face to the front. She wears a close-fitting black dress, a broad white millstone collar and a white diadem cap over smooth light brown hair, against a plain light grey ground, with the light falling full from the left.

The auction catalogue that offered it in 2024 described it in one line: Portrait of a Girl, oil on cradled oak panel, unsigned, label verso from the Philadelphia Museum of Art, lent for a 1970 exhibition, in a hand carved Dutch gold frame, from a Camden, Maine estate. It gave a sight size of 20 1/2 by 16 1/2 inches inside a frame measuring 28 1/2 by 24 1/2 inches overall. It gave no artist beyond the phrase at the head of the lot, which was after Rembrandt Harmenszoon van Rijn, and it carried an estimate of ten to fifteen thousand dollars.

That one line is the whole of what the market was told. It omits everything set out below.

The attribution is not new, and it is not a press invention

The story reported in September 2024 was that a picture nobody had ever heard of had turned up in an attic and might be a Rembrandt. That is not what happened. This panel was published as a Rembrandt, repeatedly, by the leading authorities of the day, for at least the first quarter of the twentieth century.

It appears in Wilhelm von Bode's eight-volume Rembrandt of 1897 to 1905 as number 52. It was reproduced as a full-page plate in Rembrandt: des Meisters Gemaelde in 643 Abbildungen, the Klassiker der Kunst survey published in Stuttgart in 1908 under Adolf Rosenberg and Wilhelm Reinhold Valentiner, captioned Bildnis eines jungen Maedchens, dated about 1631, and located at The Hague. It is catalogued as number 577 in the Royal Picture Gallery's own abridged catalogue of 1910. And it is number 853 in Cornelis Hofstede de Groot's Rembrandt volume of 1914, the standard critical catalogue of its generation, where it is dated about 1630 to 1631.

Four separate publications, none of them hedged. In each of them the picture is simply a Rembrandt. The 2024 catalogue heading of after Rembrandt is therefore not the survival of a cautious old opinion. It is a demotion, and no published argument for that demotion has been produced.

Bredius owned it

This is the fact that changes the weight of the case, and it is absent from every press account of the sale.

The 1910 catalogue of the Royal Picture Gallery at The Hague, the Mauritshuis, lists entry 577 as Portrait of a young girl, on oak, 0.564 by 0.445 metres, and records it as lent by Dr A. Bredius in 1893. Abraham Bredius, who lived from 1855 to 1946, was the director of that museum from 1889 to 1909 and the most consequential Rembrandt authority of his era; the catalogue he published in 1935 is still cited as a standard reference by its author's bare surname.

So the man who ran the Mauritshuis, and who spent his life deciding what was and was not a Rembrandt, bought this panel, kept it in his own collection, and hung it in his own national gallery for at least seventeen years under Rembrandt's name. Hofstede de Groot, who disagreed with Bredius about a great many attributions, agreed about this one. Bode had already accepted it.

That does not make the picture a Rembrandt. Bredius was famously wrong about at least one very large thing, and the standards of 1893 were not the standards of today. But it does mean the burden of argument sits differently than the press coverage implied. The question is not whether anyone has ever thought this was a Rembrandt. It is why the people who thought so were wrong.

The signature that is no longer there

The 2024 lot description says unsigned. The historical record says otherwise, twice.

The Mauritshuis catalogue of 1910 records entry 577 as Signed: RHL. That monogram, standing for Rembrandt Harmenszoon of Leiden, is the form the artist used in the years around 1630, which is precisely the period to which every early cataloguer assigned this panel. Hofstede de Groot in 1914 describes a monogram at the upper right, reading it as an R with the lower bar of the L worn away.

Either the mark was there and is now gone, lost to cleaning, abrasion or overpaint at some point in the intervening century; or it was never there and two independent authorities recorded what they expected to see; or it is still there and was not looked for in 2024. Each possibility is testable, and none of them has been tested in public.

This is the single sharpest unresolved question about the object, and it is the one no coverage of the sale raised.

The measurements corroborate the identification

A reader is entitled to ask whether the panel sold in Maine is really the panel Bredius lent to The Hague, or merely a similar picture. The measurements answer it.

Hofstede de Groot gives the oak panel as 22 1/2 by 17 1/2 inches, which is 57.2 by 44.5 centimetres. The Mauritshuis gives 56.4 by 44.5 centimetres. The two agree. The Klassiker der Kunst plate of 1908 prints the height identically at 0.564 but gives the width as 0.415, and it is that figure, 56.4 by 41.5 centimetres, which has been carried into modern structured databases.

The 1908 width is impossible. The 2024 lot record gives a sight size of 20 1/2 by 16 1/2 inches, or 52.1 by 41.9 centimetres, being the area of panel visible inside the frame rebate. A sight width of 41.9 centimetres cannot be cut from a panel 41.5 centimetres wide. It can be cut comfortably from a panel 44.5 centimetres wide, leaving about 1.3 centimetres covered on each side, with about 2.2 centimetres covered top and bottom. The 2024 object is dimensionally consistent with the panel catalogued in 1910 and 1914, and the widely repeated 41.5 centimetre width is a printing error of 1908 that should be retired.

Confirmation also comes from the plate itself. The 1908 reproduction and the 2024 photograph show the same picture: the same tilt of the head, the same asymmetry in the collar, the same fall of light, the same dark braided sleeve entering at the lower right.

How a Philadelphia picture reached a Camden attic

The label on the reverse of the frame is a Philadelphia Museum of Art slip, typed, naming the lender as Cary W. Bok of Camden, Maine, giving the artist as Rembrandt and the title as Portrait of Girl, in connection with an exhibition in 1970. Scrawled on it in faded pen is a note that the picture had formerly been attributed to Aelbert Cuyp, which matches Hofstede de Groot's record that London dealers took it for a Cuyp when it was with M. H. Colnaghi in 1893.

Cary William Bok, who died on 29 December 1970, was the son of Edward Bok, the Dutch-born editor of the Ladies' Home Journal, and of Mary Louise Curtis, sole heir to the Curtis Publishing fortune, who died on 4 January 1970. The family's connection to Camden was long established; Mary Louise Curtis Bok had given the town the land for its public library in 1916.

Three things therefore happened to this picture in a single year. Its owner of four decades died in January. It passed to her son. It went to Philadelphia on a museum loan slip bearing his name. And by the end of December the son was dead too. After that the record stops for fifty-four years, until an auctioneer walked into the attic.

How the panel travelled from Bredius in The Hague to the Boks in Philadelphia is not documented anywhere consulted. Press accounts state the family had owned it since the 1920s, which is chronologically plausible given Edward Bok's Dutch birth, his known ownership of a Frans Hals, and his death in 1930, but no bill of sale, dealer record or shipping document has been published.

What each position rests on

Four distinct positions were stated publicly in the fortnight after the sale, and it is worth separating what each one actually rests on.

The auction house. Thomaston Place catalogued the lot as after Rembrandt and has never claimed to have authenticated it. Kaja Veilleux, its founder and the appraiser who found the panel on a house call to the Camden estate, has described the discovery but has published no attribution argument. A year later, in August 2025, the firm's own publicity referred to the picture only as having been linked to Rembrandt. This position rests on the object's appearance and on the verso label.

Gary Schwartz, for. Schwartz, a Rembrandt scholar of long standing, put the chance that Rembrandt painted it as extremely large. His argument is comparative: he points to a portrait of a woman of 1632 in the Academy of Fine Arts in Vienna, accepted as Rembrandt, and says of the resemblance that he is amazed that people accept one and simply dismiss the other. He proposed the obvious remedy, which is to take this panel to Vienna and stand it beside the other. He also said plainly that nobody should give a definitive opinion without studying the object.

Volker Manuth, against. Manuth, of Radboud University and author of the 2019 Taschen Rembrandt, said by email that he had more doubts about the attribution than not. He attached the same condition: no attribution without thorough investigation of the original.

Betsy Wieseman, on why the question is hard. Wieseman, who heads Northern European paintings at the National Gallery of Art in Washington, made the structural point rather than a verdict. In the 1630s a great many painters in Amsterdam were deliberately trying to look like Rembrandt and to suppress their own habits of hand. The features by which a modern eye separates artists are exactly the features those painters were working to erase.

The Philadelphia Museum of Art. The museum could not confirm which exhibition, if any, the panel was lent for, and said that a loan slip does not verify a work. That is correct and it is important: a registrar's label records what a lender called a picture, not what a curator concluded about it. The 1970 label is evidence of the family's belief and of the object's location. It is not evidence of authorship.

Two further figures are quoted in the coverage giving values conditional on authentication, of about five million dollars and about fifteen million dollars. Those are commercial estimates of an outcome, not opinions on authorship, and should not be read as either.

The price, in three figures

Reports gave the result as 1.2 million dollars, as 1.4 million and as 1.5 million. All three are traceable to one hammer.

The auction house's own lot record shows the panel sold on 24 August 2024, as lot 2363 in the second day of the Summer Grandeur 2024 sale, for 1,175,000 dollars against an estimate of 10,000 to 15,000. Bidding opened at 5,000, ran past sixty bids and drew 123 bidders. The conditions of sale provide for a buyer's premium of up to 25 per cent, which is the standard tiered arrangement where a lower rate applies in the room and a higher rate to online platform bidding.

Apply 20 per cent to 1,175,000 and the total is 1,410,000, which is the figure in the auction house's own press release and the source of every 1.4 million headline. Apply 25 per cent and the total is 1,468,750, which is the figure that reached the encyclopaedias and produced the 1.5 million headlines. The 1.2 million reports are the bare hammer, rounded.

The hammer is 1,175,000 US dollars. The premium-inclusive total is either 1,410,000 or 1,468,750 depending on how the winning bid was placed, and the house has not stated which. No source consulted reports the sum in any other currency; every conversion in circulation is derived and depends on the date applied. The buyer is described by the auction house as a private European collector and in parts of the trade press as an anonymous private collector in the United Kingdom; the two are not inconsistent but neither is confirmed.

What would settle it, and what has not happened

No authentication body has ruled on this panel, and no ruling has been published as of 2 August 2026. That is the finding, and it should be stated as flatly as that.

There is no body left that could issue one with the old authority. The Rembrandt Research Project, whose word decided these questions for four decades, wound up at the end of 2011 and formally closed in 2014 with roughly a quarter of the material still unexamined. Its archive went to the Netherlands Institute for Art History as the Rembrandt Database. Attribution now proceeds by individual scholarship and by the technical departments of museums and auction houses, which is why the two named scholars who spoke in 2024 both refused to commit without examining the object.

What would settle it is not connoisseurship but a programme of work, and it is the same programme that re-attributed The Adoration of the Kings between 2021 and 2023: dendrochronology of the oak to date the felling and, in the best case, tie the board to others Rembrandt used; X-radiography and infrared reflectography to expose the underdrawing and any changes of mind; paint sampling and pigment analysis; and a direct examination of the upper right corner to determine whether the monogram recorded in 1910 is present, lost or was never there. Every one of these is routine. None of them has been reported.

The panel is in private hands in Europe. Its owner has no obligation to commission any of it, and every incentive to choose the moment. Until then this record holds the attribution open, because the evidence is genuinely open, and because a picture that four authorities called a Rembrandt and one auction house called a copy is worth more to a reader as an unresolved question stated accurately than as a verdict nobody has earned.