Provenance Is Part of the Object

Here is a fact that offends most people when they first meet it. Two paintings can be visually indistinguishable (same panel, same pigments, same hand-level quality) and one is worth four hundred million dollars and the other eight thousand. The difference is paperwork.

The intuition that this must be a market pathology is understandable and wrong, and it is worth seeing why.

A painting is not only an image. It is a physical object with a history, and part of what a buyer is acquiring is a claim: this object was made by that person, in that place, at that time. Provenance, the documented chain of ownership, is the evidence for the claim. Remove the evidence and the claim becomes unsupported, whatever the surface looks like.

There is also a legal dimension. Provenance establishes that the seller can pass good title, which matters enormously for anything that was in Europe between 1933 and 1945. A gap in a picture's ownership across those years is now the first thing a serious museum investigates, and restitution research has turned provenance from a dealer's courtesy into a discipline with archives, databases and specialists.

The awkward part is that provenance is easier to fake than painting. John Drewe, working with the forger John Myatt in the 1980s and 1990s, got into the archives of the Tate and the Victoria and Albert Museum and inserted fabricated documents. Catalogue entries, letters, photographs of works in fictitious collections. Then the paintings had a history, and the history was in the institutions everyone checks. Some of those works have never been recovered, and the archives themselves were damaged in a way that cannot be fully undone.

Science helps, and it mostly helps in one direction.

Analysis can identify pigments, date supports by dendrochronology or radiocarbon, image an underdrawing and map changes in composition. A pigment first manufactured in 1834 rules out a seventeenth-century date, and that is a hard, publishable result. Ruling in is different: materials consistent with a period are consistent with everyone working in it. So the laboratory is very good at saying no, and rarely able to say yes.

Which leaves attribution as a judgement. Connoisseurship (deciding from visual evidence, trained by looking at very large numbers of pictures) has a poor reputation from a century of confident errors, and it remains indispensable, because no instrument reads a hand. Giovanni Morelli's method is the nearest thing to a system: compare the details a painter executes automatically, ears, fingernails, folds of cloth, on the grounds that a copyist attends to the parts that matter and reproduces the rest by habit (their own habit).

Forgeries age badly, and this is the most reliable protection anyone has. A forger works inside the taste of their own decade, and cannot see it from within.

Han van Meegeren's Christ at Emmaus was accepted in 1937 as an early Vermeer, praised by the leading authority on Dutch painting, and bought for the Boijmans museum for a vast sum. It is now almost painful to look at: the faces belong to 1930s cinema, and the picture is obviously of its decade to anyone who has seen a few films from it.

Van Meegeren was arrested in 1945, not for forgery but for selling a Dutch national treasure to Hermann Göring, which was collaboration and carried the death penalty. His defence was that the treasure was his own work: that he had swindled the Reich rather than served it. Nobody believed him, so he painted another Vermeer in custody, under guard, and was convicted of forgery instead. One year, rather than a rope. He died of a heart attack before serving it, briefly a national hero for having cheated the occupiers.

The case is the neatest possible demonstration of the argument. The object did not change at any point. What changed was the claim about who made it, and everything followed from that: the price, the prestige, the criminal charge and the verdict.