You have to hold the painting at the right angle to see it at all. The signature sits low on the canvas, under the emperor’s elbow, and the contrast against the dark ground behind it is so faint that from straight on there is nothing there. Turn it, let the light come in from the side, and eight letters surface out of the varnish. L daVinci. After them, another word, worn down but almost certainly pinxit, the Latin for painted this.
The family who own the picture have had it since the late 1930s. They are not a museum and they have never wanted their name attached to it. In the 1920s, before it came to them, Christie’s had catalogued the painting as a Titian, which is a serious attribution and by itself would make it a significant object. The signature was not visible then, or at least nobody recorded it. For decades the canvas hung with generations of soot and darkened varnish sitting on top of it, and whatever was written under the elbow stayed under the elbow.
Then the siblings who inherited it in the mid-1990s started asking the question that anyone would ask, and kept asking it for the better part of thirty years. If the name is really there, is the name true?
It helps to know why anybody would bother. A painting attributed to Leonardo is not just a nice picture with a good name on it; there are fewer than twenty in the world that scholars broadly accept, and the arithmetic that follows from that scarcity is brutal. A panel called Salvator Mundi, heavily damaged and heavily restored and argued over by specialists who could not agree how much of it Leonardo actually touched, sold at auction some years ago for a reported 450 million dollars, the highest price ever paid for a painting. That single number sits behind every claim of this kind. It is why families dig out old canvases, why laboratories get asked to look at them, and why art historians treat every new candidate with a wariness that can come across as unkind.
That question eventually reached a materials science department at Uppsala University in Sweden, where a group ran the painting through Raman spectroscopy. The result, published online in the journal Materials Today Advances earlier this year and picked up widely by the press some months later, is one of the more honest pieces of art science to come out in a while, because it refuses to give anybody the answer they want.
Raman spectroscopy is a beautifully simple idea dressed up in expensive equipment. You shine a laser at a spot on a surface. Almost all of the light bounces back at exactly the energy it arrived with, which tells you nothing. But a tiny fraction of it comes back shifted, having given up or gained a sliver of energy to the vibration of the molecules it hit. Every compound has its own set of vibrations, so every compound shifts the light in its own particular pattern. Collect the scattered light, read the pattern, and you have a molecular fingerprint of whatever is sitting at that spot. Jakob Thyr, who ran the analysis, describes it in roughly those terms: the scattering gives you a fingerprint for every pigment present.
What makes it useful for paintings is that pigments have birthdays. A painter can only use what exists. Nobody in 1500 could reach for a colour that would not be made for another two hundred years, and no amount of skill or intention gets around that. So if you can name the compounds on a surface, you can put a floor under the date of the surface. You cannot prove a painting is old. You can absolutely prove that part of it is not.
The team sampled thirteen locations across the canvas and came back with thirteen pigments. Eleven of them are exactly what you would expect from a painter working in the Italian Renaissance, and reading the list is like reading a period supply order. Lead white through the flesh and the textile. Cinnabar, the brilliant mercury red, in the hairband and the armour and the lit arm. Red ochre and yellow ochre and plain earth pigments. Gypsum and chalk. Massicot, a lead yellow, in the ear. Lead-tin yellow type one, which is genuinely diagnostic, because it fell out of use in the eighteenth century and was then forgotten so completely that it had to be rediscovered by chemists in the twentieth. Azurite in the armour. Terre verte, the green earth that Italian painters traditionally laid under skin tones. Natural ultramarine, ground from lapis lazuli hauled overland from Afghanistan, which in this period cost more by weight than gold and which nobody wasted on a fake.
Then there is location B, a dark laurel leaf in the wreath around the emperor’s head, and location I, the jewel set into the pommel of his sword. Both came back with Prussian blue.
Prussian blue was not around when either Leonardo da Vinci or Titian was alive. It was not around when anybody was alive who might plausibly have painted a Renaissance portrait. It was made for the first time in Berlin in the early 1700s, and the paper puts the date at 1704. It was also an accident. A colour maker was trying to produce a red lake pigment and used potash that had been contaminated with animal residue, and instead of red he got a deep, strong, startlingly cheap blue. Until that moment, if you wanted a serious blue you either bought ultramarine at a price that could bankrupt a commission or you settled for something duller. Prussian blue undercut all of it, and within a couple of decades it was everywhere in European painting.
For anyone who studies pictures for a living, that accident is one of the most useful clocks in existence. Leonardo died in 1519. Titian died in the 1560s. Prussian blue simply cannot be in an untouched fifteenth or sixteenth century paint surface, and its presence is not a matter of interpretation or connoisseurship or taste. It is a compound with a manufacturing start date.
Alongside it at the same laurel-leaf spot, the analysis found what is most likely titanium white, a pigment that exists in trace form in nature but which only became a commercial paint in the twentieth century. So the wreath has two anachronisms in one place.
The obvious reading is forger, and the obvious reading is probably wrong, and the reason is the pattern. A forger painting a whole picture in the wrong century contaminates the whole picture. The modern pigments show up everywhere, in the flesh, in the background, in the drapery, because the forger was working with one palette across one surface. That is not what this canvas looks like. Eleven period pigments are spread across the arms, the armour, the textile, the ear, the hairband, the background. The two modern ones are concentrated in a wreath and a jewel. That is the fingerprint of somebody going back into a finished painting and working on specific areas, which is what restorers and retouchers have done to almost every old picture that has survived this long.
X-ray imaging supported exactly that. The laurel wreath is not simply painted; it has been structurally altered. The wreath in the painting as it now stands is more elaborate than what lies underneath it. Somebody, at some point after 1704, decided the emperor’s crown of leaves needed more leaves, and gave him more, in a blue that had not existed when the rest of the canvas was made. Thyr’s own framing is that the Prussian blue was most likely added later, during a restoration or retouching, rather than being evidence of a fake.
Which brings the whole thing back to the eight letters under the elbow, and here the study quietly declines to hand anybody a verdict.
The signature was sampled too. It is carbon black, which is lamp black or ivory black, one of the oldest pigments humans have, made essentially by collecting soot. No other pigment was found in it. Carbon black is entirely consistent with the Renaissance, and it is also entirely consistent with every other century since, because soot never stopped being soot. The inscription is reported to sit behind aged varnish rather than on top of it, which argues against somebody having scratched a name onto the surface last week, but varnish can be removed and reapplied, and a signature added in the 1800s would sit behind varnish applied in the 1800s just as comfortably.
So the chemistry of the signature proves that it could have been written by Leonardo. It equally proves that it could have been written by a dealer in 1780 with a pot of lamp black and a strong commercial motive. The single most loaded object on the canvas turns out to be the one the laser has the least to say about.
The art historians who have looked at the painting are, to put it gently, not persuaded. Charles Hope of the Warburg Institute has said flatly that the picture looks nothing like any work by Leonardo, and has suggested that the signature is the work of an unscrupulous dealer, which was a thriving trade for centuries and remains one. Frances Coulter has described the handling as far too robust and too sharply defined for Leonardo or for anything from his workshop, which is a specific criticism rather than a vague one: Leonardo’s whole reputation as a painter rests on edges that dissolve, on faces that emerge from shadow without ever quite committing to an outline. A hard, confident, well-lit Roman emperor is close to the opposite of that.
The sitter is worth a moment too, because he is not a flattering choice. Domitian ruled Rome at the end of the first century and was killed in a palace conspiracy. The historians who wrote about him afterwards, working under the regime that replaced him, left a portrait of a suspicious and isolated man, and the Roman senate went so far as to order his name scrubbed from inscriptions and his images destroyed. Some of the cruelty in that account is almost certainly political score-settling by the winners, and modern historians treat it carefully. But it meant that for centuries he was a byword rather than a hero, and an artist who chose to paint him was not painting a crowd-pleaser. He is an emperor in armour with a laurel wreath and a sword, and the wreath is the part somebody later decided to make grander.
There is also a straightforward problem of chronology that has nothing to do with chemistry. Domitian was not a fashionable subject in Leonardo’s lifetime. He was a Roman emperor whose reputation, as written by the historians who survived him, was thoroughly unflattering, and painted portraits of him seem not to appear at all until the 1560s, decades after Leonardo was dead. Painters of the later sixteenth century did produce sets of the Caesars, and a Domitian belongs comfortably in that tradition. In Leonardo’s, it is an orphan.
Reported carbon dating of the canvas itself is said to place the support somewhere around 1470, which would put the material firmly in Leonardo’s era, though it dates the cloth rather than the paint on it and dating a support has never been the same thing as dating a picture. Old canvas and old panel have been available to anyone with the money to buy them for as long as there have been people willing to pass off new work as old.
The authors of the study are clear-eyed about all of this, and their caution is the most valuable thing in the paper. They state that establishing authenticity requires stratigraphic, material and contextual analysis taken together, rather than any single test, and they point out the uncomfortable fact that a sufficiently informed forger can deliberately work in historically appropriate materials precisely because they know somebody will eventually point a laser at the result. Pigment analysis can exclude. It cannot crown.
What the work does deliver is a real and specific history of an object, which is more than most paintings in private hands ever get. There is a canvas whose paint is overwhelmingly consistent with the Italian Renaissance, made with materials including one, natural ultramarine, that nobody wasted casually. There is a laurel wreath and a sword jewel that somebody went back into at least two centuries later and repainted, making the wreath fuller than it had been. There is an inscription in soot, sitting under old varnish, naming the most famous painter who ever lived, which the science can neither confirm nor kill. And there is a family who have lived with the thing for the better part of a century and now know considerably more about it than they did, without knowing the one thing they actually wanted to know.
That is a more interesting outcome than a clean answer would have been. An attribution either way would close the file. What the laser found instead is the record of the painting’s own life: a picture made a long time ago, valued enough by somebody to be improved, damaged enough to need fixing, signed by a hand nobody can name, catalogued as one master, claimed for another, and passed down through a family that still will not say who they are.
It is still hanging in a private home. Nobody is being asked to believe anything. The painting simply carries more documented truth than it did before, and the last piece of it, the eight faint letters that started the whole enquiry, is still exactly as open as it was on the day somebody first tilted the canvas toward the light and saw them there.