The AI Journal

In the summer of 1945, a prisoner in an Amsterdam cell was given canvas, pigments ground to an old recipe, and fine badger-hair brushes, and told to paint a Vermeer. He was not being humiliated. He was being tested. Dutch investigators needed to know whether Henricus Antonius van Meegeren, under arrest for selling Johannes Vermeer’s Christ and the Adulteress to Hermann Göring, could possibly have painted it himself, because that was now his own defence, and his life depended on whether anyone believed it.

Van Meegeren had been arrested weeks earlier as a collaborator, a charge that in liberated Holland could mean the firing squad: he had sold a national treasure, Dutch cultural patrimony, to the Reich’s second-most-powerful man. Facing that charge, he made a claim that sounds, on its face, like a worse confession rather than a better one. The painting was not a Vermeer. He had painted it himself, in his own studio, within living memory, using materials no seventeenth-century workshop possessed. He was not a traitor who had sold his country’s heritage. He was a forger who had sold a worthless fake to a man who deserved to be swindled. It was, transparently, the story a desperate man facing death would have every reason to invent. No chemist had said a word.

What investigators did next was demand proof, and the proof they asked for was theatrical rather than analytical: paint one again, here, under guard, while we watch. Van Meegeren produced Jesus Among the Doctors in custody, in a style close enough to his earlier fakes that the demonstration was credited, by the officials who watched it made, as persuasive. But a man’s talent for mimicking his own earlier fraud is not, by itself, proof of anything beyond talent. What his claim actually required, if it was going to substitute for a capital charge, was a case that did not depend on anyone’s word, including his own. That case had not yet been built. It was only now commissioned.

The commission went to Paul Coremans, a Belgian chemist, and the examination he assembled over the following years is the part of the story usually remembered as the discovery: X-radiographs showing the canvas had been scraped and reused; a craquelure pattern that, examined closely, had been induced rather than grown, the cracks running through the ground layer in ways three centuries of flexing never produces; a binding medium baked hard at temperatures no seventeenth-century oven reached, built from a synthetic resin, phenol formaldehyde, chemically related to Bakelite, unavailable before the twentieth century; and traces of cobalt blue, a pigment first synthesised in the nineteenth. None of this was faked or exaggerated. It was, and remains, a genuinely rigorous piece of conservation science, still cited as a model of how to authenticate a painting by what it is made of rather than by who claims to recognise a hand.

It is also, on the calendar, an afterthought. Van Meegeren was tried in November 1947, not for treason but for forgery and fraud, a charge that only makes sense if the court had already accepted, before the scientific record was in any publishable form, that the paintings were his. He was convicted, sentenced to a single year, and dead of a heart attack within weeks, before he served a day of it. Coremans’s full report, Van Meegeren’s Faked Vermeers and De Hooghs, the document now cited as the forensic case against him, was not published until 1949. The trial that settled his legal fate closed two years before the book that is supposed to have proved it.

The usual shorthand for this case is that science unmasked a forger. Unmasking implies a disguise still in place when the mask is removed. Van Meegeren took his own mask off, in public, under a threat he had every reason to invent an exit from, and the state’s problem was never identifying him. It was deciding whether to believe him. Chemistry was not the detective here. It was the instrument a government reached for because a confession, however convenient, is not the same thing as a defence that will hold, and a forger who had already lied to sell the paintings in the first place was not a source anyone could simply take at his word, even when his word happened also to be the only thing standing between him and execution.

Forensic science can do either of two jobs, and they are easy to confuse because the apparatus looks the same from the outside: a laboratory, a report, a verdict that follows. One job is detection, finding what nobody has yet said and nobody is motivated to volunteer. The other is corroboration, testing a claim that someone has already made, usually because they need it believed and the people who must decide whether to believe it cannot simply take their word. A fingerprint lifted from an unexplained scene is doing the first job. A paternity test ordered after a man already says the child is or is not his is doing the second. Both produce the same kind of confident sentence afterward, “science showed that,” and the retelling of a case routinely drops the distinction, because a confession is a messy, human, legally awkward fact to build a story around, while a laboratory result sounds like the story ending on its own. Van Meegeren’s case is retold as the first kind of forensic triumph. The documentary record, the order in which the confession, the demonstration painting, the trial and the published report actually occurred, shows the second.

The strongest objection to this reading is that it understates what the chemistry achieved, and the objection has a specific, decisive answer on the other side of the ledger. Coremans’s case, built around a man already in the dock, was not the end of the forensic record. A Belgian dealer named Jean Decoen spent years publicly disputing Coremans’s conclusions, and the question of exactly which paintings van Meegeren had and had not forged stayed genuinely open for two decades after his death. It was closed not by any further testimony but by Bernard Keisch and his colleagues, who in 1967 published a method for dating white lead pigment by measuring its content of polonium-210 against its parent isotope, radium-226, a ratio that shifts measurably over a human lifetime and barely at all over three centuries. Applied to the disputed paint, it returned an age of decades, not centuries, settling the Decoen dispute without reference to any courtroom, confession or forger’s biography at all. That is real, independent detection, exactly the kind the popular story credits the case with from the start, and denying it would be dishonest.

What the objection does not restore is the sequence in van Meegeren’s own trial. The lead-210 method is confession-independent because it was built and deployed years after the man who would have needed to confess to anything was already dead. In 1945, no such instrument existed, and the chemistry that did exist was called in only once a liar’s claim about his own paintings had become a legal problem requiring an answer that did not rest on his say-so. The forger spoke first. The laboratory, assembled afterward to make that speech usable, has simply kept the credit that was going when the courtroom closed.