The Man with the Golden Helmet

An old man's head and shoulders, turned to the right, in an iron gorget under an ornamental helmet loaded with gold paint so thick that it stands off the canvas while the face beside it sinks into shadow. For most of the twentieth century it was among the most reproduced pictures in Germany, hanging as a print, and in embroidery, in a very large number of ordinary homes. It was bought for Berlin in 1897 as a Rembrandt.

In 1986 the Gemaeldegalerie concluded that it is not. That conclusion is the subject of this record, and the precise wording matters: the museum withdrew an attribution. It did not find a fake, and the curator who announced the finding said so at the time in as many words. The picture was made in Rembrandt's lifetime by somebody who worked in his studio or was in close contact with him, and it is still one of the finest things in the gallery.

The examination that unmade it also found something that changes how the object is read. The canvas had been cut down on all sides, and large areas, chiefly in the face, had been filled in or painted over by restorers, under an ochre varnish that gave the picture the warm gallery tone by which generations had recognised it as Rembrandtesque. A judgement of authorship had been made about a surface that was substantially not the painter's.

What the picture is, before anyone argues about who made it

Oil on canvas, 67.5 by 50.7 centimetres. A life-size head of an elderly man, turned diagonally to the right, wearing an iron gorget and an elaborate helmet. The background is dark and gives nothing away. There is no signature and there never was one.

It is not a portrait. It is a tronie, the Dutch studio genre of a head study made for its own sake, usually of an anonymous or invented character in exotic or antique dress, painted for the market rather than for a sitter. That distinction does real work here, because a tronie has no commission, no client and no documentation, and a picture with no documentation can only be attributed by eye.

The helmet does almost all of the work. Its gold is applied in heavy impasto that catches light physically as well as pictorially, so that in a gallery the metal is the first and often the only thing a visitor registers; the face, lit obliquely and painted thinly, retreats. That imbalance is the picture's whole rhetorical strategy and it is also, as the twentieth century eventually noticed, unlike Rembrandt.

A helmet nobody can place

The Berlin museum's own object record describes the helmet as an ornamental Spanish or Italian war helmet of the late sixteenth century, and calls it an item of great rarity. Other accounts identify it as Augsburg work, which matters because Johann Ulrich Mayr, one of the painters proposed as the author, was an Augsburg painter who worked in Rembrandt's studio around 1648 to 1649.

Those two identifications are not compatible, and this record does not choose between them. An armour identification is a dating and localising instrument of exactly the kind that settles arguments about paintings, and here it points in two directions at once, each conveniently supporting a different reading. It should be treated as unresolved and worth resolving.

Who is he, and why the old answer is impossible

Nineteenth-century tradition identified the sitter as Adriaen van Rijn, Rembrandt's brother. Cornelis Hofstede de Groot, cataloguing the picture in 1915 as number 261, took the identification apart on plain biographical grounds. Adriaen was a shoemaker and lived in Leiden while Rembrandt was in Amsterdam. He died in 1652. Models with the same features appear in works dated 1654.

Neither objection is decisive on its own and together they are close to it. What is worth noticing is the shape of the reasoning: an unnamed head in a studio genre acquired a name because a famous painter needed a plausible relation to have painted, and the name survived a century after the evidence for it had gone.

How it reached Berlin

The picture is said to have come from the personal estate of Clemens August, the Wittelsbach archbishop-elector of Cologne, who died in 1761. That is a claim rather than a documented link, and it is the sort of claim that attaches itself to good pictures.

What is recorded is that it was in the collection of the De Boccard family at Fribourg in Switzerland and emerged from there in the early 1890s through British dealers. Hofstede de Groot's 1915 catalogue names the London firm of P. and D. Colnaghi and Company as the intermediary. In 1897 the Kaiser-Friedrich-Museums-Verein, the private association that buys pictures for the Berlin galleries, acquired it for the collection; it appears as number 811A in the Kaiser Friedrich Museum's catalogue of 1911, and it carries that number still. The Verein, not the state, owns it.

Ernst A. Busche published new findings on the provenance in the Jahrbuch der Berliner Museen in 2015, which indicates that this chain is still being worked on more than a century after the purchase.

Doubted from the beginning, and endorsed anyway

The two most formidable Rembrandt scholars of the day, Hofstede de Groot and Abraham Bredius, questioned the attribution from the moment the picture arrived. They were overruled, in effect, by Wilhelm von Bode, the general director of the Prussian collections, whose restoration campaign and personal endorsement persuaded the field.

The pattern is worth stating without moralising, because it recurs. A picture is bought by an institution on the authority of its director. Specialists object. The director's authority, plus the institution's need to be right about a major purchase, converts a contested attribution into a settled one, and the objections drop out of the literature. Fifty years later the picture is a fixture, taught and reproduced, and dislodging it costs far more than declining it would have done in 1897.

Bode had done the same thing eleven years later with the wax bust of Flora bought as a Leonardo in 1909, over similar objections. The Berlin museums have carried both cases.

The most reproduced face in Germany

Between the purchase and the war the picture became extraordinarily popular. Prints of it were mass-produced and hung in ordinary German homes; it was worked in embroidery and reproduced in other media. It functioned less as a museum object than as a domestic icon, and its title, not its author, is what most people knew.

That popularity is part of the record for two reasons. It is the measure of what was at stake in 1985, when a curator had to tell a country that its favourite Rembrandt was not one. And it explains the aftermath: since the attribution was withdrawn the picture's fame has fallen away, although nothing about the object has changed. The quality that made it worth mass-producing is still on the canvas. What was being consumed was the signature that was never there.

1957 to 1985: the doubt comes back

Werner Sumowski raised the question again in 1957, in a set of addenda to the Rembrandt year of 1956. In 1968 a group of Dutch scholars founded the Rembrandt Research Project, with the explicit aim of separating the autograph works from the very large body of pictures that had accumulated around them; over the following decades it removed a great many paintings from the corpus, and the Berlin picture was an obvious candidate.

In 1984 a Dutch commission convened to examine doubtful Rembrandts looked at the painting in West Berlin and recorded its doubts. In November 1985 Jan Kelch, the Berlin curator responsible, announced that important features of the picture's handling did not match Rembrandt's, and that it had probably been painted about 1650 by someone from his circle. He was quoted at the time saying that it is not a fake, and that it remains a great and masterly work.

In 1986 the Gemaeldegalerie published the finding formally, in a documentation edited by Kelch and produced with the Rathgen research laboratory of the Prussian cultural foundation and the Hahn-Meitner-Institut in Berlin. The attribution to Rembrandt ended there.

What the surface turned out to be

The technical examination produced a finding that goes beyond the question of authorship, and it is the part of this case that ought to be better known.

The canvas had been cut down on all sides, so the composition anybody has ever seen is not the composition that was painted. Large areas had been filled in or painted over by restorers, and the overpainting was concentrated in the face. On top of that sat an ochre varnish, applied in an earlier restoration, which produced the warm brown so-called gallery tone that the nineteenth and twentieth centuries read as the authentic Rembrandt atmosphere.

Attribution by eye compares brushwork. If the brushwork under discussion is substantially a restorer's, and the colour balance is a varnish, then the comparison is being made against the wrong object. Everyone who defended or attacked the attribution before the mid-1980s, on both sides, was arguing about a surface that had been rebuilt. The picture was restored again and brought back towards its reconstructed original colouring in 2011.

Three candidates and no verdict

Nobody has been able to put a name on it, and the names that have been proposed are instructive about how narrow the evidence is.

Carel van den Pluym, a Leiden painter and a relation of Rembrandt's, has been suggested. So has Heyman Dullaert, a Rembrandt pupil; the argument here rests on Arnold Houbraken's report that a painting by Dullaert was sold in Amsterdam as an authentic Rembrandt, which establishes that such confusion happened in the seventeenth century rather than that it happened to this picture. Johann Ulrich Mayr, from Augsburg, who was in Rembrandt's workshop around 1648 to 1649, is proposed partly on the strength of the reading of the helmet as Augsburg work.

Each proposal is an argument from possibility. The Berlin record says only that the painter must have worked in Rembrandt's studio or at least have been in close contact with him, and the label says anonymous. On the evidence set out in the sources consulted here, that is the correct answer and the other three are hypotheses.

Withdrawing an attribution is not declaring a forgery

This distinction is the reason the record exists in this form, and it is routinely collapsed in retellings.

A forgery is an object made with intent to deceive about its authorship. Nothing about this painting suggests that. It is a seventeenth-century Dutch picture, painted within Rembrandt's orbit, in the period it appears to be from, by a highly accomplished hand. It acquired a false name in the nineteenth century, from dealers and scholars, some two hundred years after it was painted. The deception, if there was one, happened in the market and in the literature, not in the studio.

What the Gemaeldegalerie did in 1986 was remove a name it could no longer support and decline to substitute another. That is the harder and rarer institutional act. It is much easier to move an attribution sideways, to a named pupil, because a named pupil still gives a label something to say. Anonymous is an admission, and museums do not enjoy making it.

What the case leaves behind

Three durable points.

The first is that condition governs connoisseurship. An attribution argued over a cut-down canvas with an overpainted face and a tinted varnish is an attribution argued over a restorer's work. Any technical examination that dates the interventions before it judges the hand is doing the steps in the right order, and this one did.

The second is that the objections were on the record in 1897. Hofstede de Groot and Bredius said so at the time and were overridden by an institutional authority who had just bought the picture. The scholarly correction took eighty-nine years and the evidence that finally carried it was largely physical rather than stylistic.

The third is about the public. A picture that hung in a hundred thousand German homes lost its fame the moment it lost its name, and nothing about the canvas changed. What the market and the public were valuing was a proper noun. The object was always exactly as good as it is now.