For as long as anyone alive has looked at it, the picture has been a scene between two people. A man leans back with a clay pipe. A woman raises a tall glass. Behind them a brick wall, a doorway, a strip of Delft sky. It is one of those quiet, sunlit Dutch courtyards that the seventeenth century seems to have invented specifically so that later centuries could feel calm about domestic life.
It was never a scene between two people. There was a third figure sitting right in the middle of them — a soldier in armour, tankard in hand — and at some point between roughly 1660 and the early 1800s, somebody painted him out. Not the artist. Somebody else. This month the Mauritshuis in The Hague put him back.
What the picture actually shows
The painting is A Man Smoking and a Woman Drinking in a Courtyard, inventory number 835, oil on canvas, 78 by 65 centimetres, painted by Pieter de Hooch around 1658–60 — the Delft years, the good years, when he was working within walking distance of Vermeer and producing the courtyard pictures that made his reputation. It came to the museum in 1947 as a gift from Mr and Mrs Ten Cate-van Wulfften Palthe.
De Hooch painted the composition twice. The other version, A Dutch Courtyard, hangs in the National Gallery of Art in Washington, is slightly smaller at 69.5 by 60 centimetres, and has never lost anybody. In the Washington picture the soldier is unmistakable: seated, red-faced, armour over a brown shirt, watching the woman drink. A second man smokes. A child comes in from the side carrying embers to light the pipes.
Once you know he is there, the subject changes completely. Sabrina Meloni, who led the Mauritshuis restoration with Abbie Vandivere, put it flatly: the men are challenging the woman to a drinking game — that is the subject. The glass in her hand is the sort of tall, ringed vessel that Dutch painters used as shorthand for exactly that: a pass-glass, banded with horizontal trails of glass, where the game is to drink from one ring to the next in a single swallow and the penalty for overshooting is another swallow. It is a machine for getting people drunk, and it is sitting in the middle of a painting that generations have filed under domestic tranquillity.
Take the soldier out and a drinking game becomes a marriage. That is not a restoration decision. That is an edit.
The 1920s photograph nobody acted on
The soldier was not a surprise. The Dutch conservator Martin de Wild found him in the 1920s, uncovered him, photographed him in black and white — and then painted over him again.
That sentence is worth sitting with. De Wild had the figure in front of him, documented him properly, and put the overpaint back. By the standards of his moment this was tact rather than vandalism: the picture had been read as a two-figure composition for a century, with a market and a literature built on that reading, and the job of a restorer in 1925 was not to relitigate what a painting was about. He left the evidence and closed the door.
What has changed since is not the technology so much as the licence. The 2026 campaign used macro X-ray fluorescence scanning — macro-XRF, carried out with the Rijksmuseum — which maps the chemical elements in the paint across the whole surface, reading shapes under later layers by their pigments rather than their colour. It confirmed De Wild and gave the conservators what he never had: a full map of what survived underneath, and how far the surrounding paint had drifted in tone.
Why they repainted him — and why softly
Here is the part that makes this more than a nice museum story.
Conservation has run for most of the last sixty years on a principle of restraint: clean, stabilise, remove what is demonstrably foreign, and make good only what the eye needs to read the picture — visibly, reversibly, never inventively. Museums take away overpaint. Adding a whole figure back to a canvas is close to the opposite move, and rare enough that the Mauritshuis has been careful to call it a first for the institution.
Their argument is straightforward. The soldier is not a guess: the composition survives intact in Washington, the underlying paint is mapped, and the picture without him is a mutilation rather than a variant. Leaving him buried preserves a nineteenth-century censor's opinion, not De Hooch's.
The interesting decision is the second one. Meloni and Vandivere did not copy Washington. They painted the soldier less detailed, deliberately subdued, because roughly 360 years have shifted the pigments around him — the Mauritshuis surface has gone quieter and browner than its brighter Washington twin, and a crisply reconstructed figure would have sat on top of the picture like a sticker. So the returned soldier is a suggestion: present, legible, unmistakably part of the scene, and just soft enough that you can tell he is a modern reading of a lost passage rather than the thing itself. And all of it is applied in materials that come off again without touching the original. If the next generation disagrees, they can wipe him away and be back where they started.
"A nice balance has now been restored to the painting," Meloni said. It is a modest line for what is, technically, a museum choosing to paint on a Dutch Golden Age masterpiece.
Who erased him, and the thing nobody can prove
The honest answer is that we do not know. The Mauritshuis says so plainly. The window is wide — sometime after about 1660, and before 1833, when the dealer John Smith catalogued the picture in London and described it without any soldier. By then the deletion was already old enough to be invisible.
What can be said is that the erasure fits a pattern. Dutch genre painting was resold for two centuries into markets that wanted it to mean something gentler than it did. An armoured soldier plying a woman with drink in a courtyard is not a domestic idyll; in seventeenth-century terms it sits close to the guardroom scene, with everything that implied about soldiers, drink and the women who served them. Remove him and the same canvas becomes a husband, a wife, a pipe, a glass, a well-kept yard — a picture a respectable collector could hang in a drawing room without explaining it.
That is a reading, not a finding, and it should be labelled as one. No document names the person who did it or the reason. But the shape of the intervention — surgical, competent, aimed precisely at the one figure that changes the subject — makes accident unlikely. Somebody knew what the painting meant and preferred it to mean something else.
Two restorations, two opposite answers
What makes this month's news genuinely instructive is that it lands beside a decision that went exactly the other way. In Florence, restorers of Fra Angelico's Fiesole altarpiece found the original marble throne under a 1501 repaint — and left it buried, because the overpaint had itself become four centuries of the object's history. In The Hague, conservators found a figure under a later hand and brought him back.
Both teams were right, which is the point. The difference is what the later layer is. Fra Angelico's sky was a Renaissance artist's considered revision, part of the panel's biography. De Hooch's missing soldier is a market edit, made anonymously, that reversed the meaning of the picture. One is history; the other is interference. Telling them apart is not a technical judgement — the scanners cannot make it for you — but a historical one, and it is why conservation remains an argument rather than a procedure.
It also belongs to a longer thread we keep returning to here: the paintings that turn out to be two paintings. The turban that surfaced under a Rembrandt, the Velázquez that spent decades under someone else's name — every one of them a reminder that the surface of an old picture is a negotiated document. Looking at it properly, as we argue in our guide to reading a masterpiece, means reading it as a stack of decisions, not a single image — a job that no authentication algorithm has yet learned to do. The Mauritshuis has just added another decision to the stack, and had the grace to make it removable.
Honest limits
Three cautions before this hardens into a headline.
First, the reconstructed soldier is an interpretation. It is guided by an autograph second version and by macro-XRF, which is about as good as the evidence gets — but the paint you now see in that part of the canvas is 2026 paint, and no visitor should mistake it for De Hooch's hand. Second, the motive for the original erasure is unknown, and my reading of it as a commercial cleanup is inference from the pattern, not documentation. Third, the Washington and Mauritshuis pictures are not identical twins; the Washington version is generally judged the stronger of the two, and using it as the template for the other is a reasonable choice rather than an automatic one.
What is not in question is the object. A painter in Delft set two men and a woman around a pass-glass in a brick courtyard and painted the light coming over the wall. Somebody later decided one of them was surplus. A restorer in the 1920s found him, took a photograph, and thought better of it. And in the summer of 2026 two conservators looked at a scan, mixed a deliberately quiet flesh tone, and gave a soldier his beer back — in paint that can be taken off again, which may be the most respectful thing about the whole business.
He is on the wall in Room 15 in The Hague until 24 August 2026, after which the Mauritshuis closes for renovation until 20 September and the picture travels to Madrid for Mauritshuis ♥ Thyssen-Bornemisza. Three figures now. That was always the number.
Frequently asked questions
Which de Hooch painting had a hidden soldier? A Man Smoking and a Woman Drinking in a Courtyard (inv. 835) at the Mauritshuis, painted around 1658–60. A seated soldier in armour holding a tankard originally sat between the two figures. He was painted out by another hand between about 1660 and the early nineteenth century, and returned to the picture in 2026.
Did Pieter de Hooch paint the soldier out himself? No. The overpaint is not his. The second autograph version, A Dutch Courtyard in Washington, still shows the soldier, and a London catalogue of 1833 already describes the Mauritshuis picture without him. Who removed him, and why, remains unknown.
Is the repainted soldier reversible? Yes. Sabrina Meloni and Abbie Vandivere applied the reconstruction in materials that can be removed without harming the original paint, and kept the figure softer and less detailed than the Washington version because the surrounding pigments have shifted over some 360 years.