Joshua Reynolds moved Mai’s hand. He painted it in one place, looked at it, and painted it again somewhere else, right on top. He reworked the head, too. For two and a half centuries those changes sat under the surface. Then, this summer, conservators at the Getty Center in Los Angeles scanned the canvas and found the first tries still there, under the paint that hides them.
Here is the surprise. One of the two instruments that found them never looked at the painting’s colors at all. It read the painting’s atoms. And it reads them by the same rule that makes the northern lights green.
A portrait of the first Polynesian in Britain
Mai was born on Ra’iātea, an island near Tahiti in the Society Islands. He sailed to England on HMS Adventure, part of Captain Cook’s second voyage, and arrived in October 1774. London made him a celebrity. He was received by King George III.
Around 1776 Reynolds painted him full-length, on a canvas nearly eight feet tall. That format was usually kept for European aristocrats. Mai stands in a white robe and turban, and the tattoos on his hands are painted with care. He went home on Cook’s third voyage, reached Huahine in 1777, and died there around 1780, still in his twenties.
The National Portrait Gallery in London and the Getty bought the painting together in 2023. It is on view at the Getty Center now, through March 14, 2027. Davide Gasparotto, the Getty’s senior curator of paintings, put it plainly: “It’s not just a painting, it’s also an anthropological artifact.” It records a meeting of two cultures, including the difficult parts.
Before it went on the wall, the museum spent a few intense weeks looking into it. The studio desk flagged the story in Fresh Paint on September 17. This is the long version.
A painter’s repentance
Painters have a word for a change of mind left under the paint. It is Italian: pentimento, “repentance.” The plural is pentimenti.
They are everywhere once you know to look. In Jan van Eyck’s Arnolfini Portrait of 1434, the husband’s feet were painted in three different positions before van Eyck was satisfied. Some pentimenti were always faintly visible. Others have surfaced on their own, because oil paint grows more transparent as it ages. The oil and the white grains in it scatter less light than they once did. The top layer grows easier to see through, and a first thought drifts up from under it. That is the physics of The light is behind the paint, working very slowly.
Most pentimenti never surface at all. Finding those takes light the eye cannot use.
Knock out an electron, and the atom says its name
The first instrument the Getty used is called X-ray fluorescence, or XRF. It is easiest to understand through something you may have already seen: an aurora.
In an aurora, a fast particle from the Sun strikes an oxygen atom high in the air and bumps one of its electrons up to a higher energy level. The electron cannot stay there. When it drops back, the atom gives off one particle of light. The size of the drop sets the color, and for that drop in oxygen it is always the same green, at 557.7 nanometers.
XRF does the same thing deeper inside the atom. A thin beam of X-rays knocks an electron out of an atom’s innermost shell. An electron from farther out falls in to fill the hole, and the atom gives off an X-ray of its own. That drop is huge, so the X-ray is powerful. And it is a different size for every element. Iron always answers with the same energy. So does calcium. So does lead. The detector never learns what color anything is. It listens for which answers come back.
Physicists measure the energy of a single particle of light in electronvolts. The aurora’s green carries about 2.2 of them. The X-ray that iron sends back carries about 6,400, nearly three thousand times as much. That is why it can climb out through layers of paint and still be counted.
Move the beam across the canvas one small spot at a time, and you get a set of maps: one for lead, one for mercury, one for iron. None of them is a picture of the painting as you see it. Each one is a picture of a single element.
The arithmetic that put the elements in order
Why is every element’s answer different? In 1913 a young English physicist, Henry Moseley, fired electrons at one metal after another and measured the X-rays that came back. He found a rule so simple you can check it with a pencil.
Take the element’s atomic number, the count of protons in its center. Subtract one. Square it. Multiply by 10.2. That is the energy of its strongest X-ray, in electronvolts.
Calcium, 20: 19 × 19 = 361 361 × 10.2 = 3,682 (measured: 3,692)
Iron, 26: 25 × 25 = 625 625 × 10.2 = 6,375 (measured: 6,404)
Copper, 29: 28 × 28 = 784 784 × 10.2 = 7,997 (measured: 8,048)Each one lands within one percent. Before Moseley, the periodic table was sorted by how heavy each element was, and a few pairs had to be swapped by hand because their weights put them in the wrong order. His rule showed that the atomic number is a real count and not just a place in a list. It also left gaps in the sequence, where elements belonged that nobody had found yet. Moseley was killed at Gallipoli in 1915. He was twenty-seven.
Now run the rule backward, the way a conservator does. A peak near 6,400 on a painting means iron is there, and in an eighteenth-century painting iron usually means the earth colors: ochres, siennas, umbers. A mercury answer means vermilion, the red made from mercury and sulfur. And the lead answers mean lead white, the white of European oil painting for centuries.
The second camera looks past red
The Getty’s second instrument was reflectance imaging spectroscopy. It measures how much light each spot on the painting sends back, at hundreds of wavelengths. It starts at violet, runs through red, and carries on into the near infrared, which the eye cannot see at all.
Every pigment makes its own curve across those wavelengths. So this is a second set of fingerprints, this time of how a material treats light rather than what atoms it is made of. The two instruments need each other. XRF cannot see carbon at all, because carbon’s answer is too weak to escape the paint and cross the air to the detector. So a black made of soot is invisible to it. The spectrometer finds carbon black easily, because carbon black swallows every wavelength, visible and infrared alike.
The infrared half does something else too. How much a grain scatters light depends on how big the grain is next to the wave, a cousin of the rule in Why the far mountain is blue. Infrared waves are longer, so pigment grains look smaller to them and scatter them less. Past red, many paints turn partly see-through, and the camera looks down into layers the eye stops at.
Why a map of lead can show a hidden hand
Here is how the two instruments together catch a painter changing his mind.
For centuries, lead white was the white every oil painter reached for. It went into the light parts of a picture: the highlights on skin, the bright folds of a robe, the sky. So a map of lead is roughly a map of where the painter put light.
Paint a hand, with lead white in its highlights. Decide it is wrong. Paint the background over it, and paint the hand again a little farther along. On the surface there is one hand. In the lead map there are two, because the first hand’s lead is still in the layer underneath. X-ray photographs of paintings, which are far older than either scanner, work by the same fact: lead stops X-rays, so a lead-white hand shows even under a dark coat.
The maps themselves have not been published yet, as far as we can find, so the shape of Mai’s first hand is not public. What the team has said is that the scans revealed multiple layers of paint and reworked areas, including Mai’s hand and head. They also named Reynolds’s palette. The paint held lead, vermilion, carbon black, and “one or more iron-earth pigments.”
Where the evidence runs out
That last phrase is the most honest one in the report, and it is worth reading slowly.
XRF names elements, not paints. Iron is in yellow ochre, red ochre, raw sienna and burnt umber alike. It is even in Prussian blue, first made in Berlin around 1706. The iron map cannot say which of them Reynolds used, or how many. So the scientists wrote “one or more.” Mercury points to vermilion only because almost nothing else on an eighteenth-century palette was made of mercury. That is a fact about history, not about the atom.
A map adds up every layer at once. It shows lead, and it cannot say on its own which lead went on first. Telling the order of layers takes other evidence. The team says the next phase of the study includes examining the paint layer structure.
The scans show that he changed his mind, not why. Sally Higgs, a conservator from the National Portrait Gallery, has a view: “[Reynolds] is notoriously someone that never is satisfied with what he’s done.” That is a judgment from someone who knows his work well. It is not something an instrument measured.
And the binder is still a question. Reynolds was known for stirring waxes and resins into his oil paint. Nobody yet knows whether he did it here. It matters, because those experiments, with a tar called bitumen and a red that fades in light, are why so many of his canvases are cracked or faded today. This one, the Getty’s Kari Rayner said, is in very good condition, “especially for a Reynolds.”
Watercolor keeps no secrets
A watercolor is the opposite of an oil painting in this one way. Watercolor is transparent, so you cannot bury a first try under it. Every pencil line and every earlier wash shows through what comes after. The paper is always part of the picture.
That is why a resist feels like magic. In a resist painting you draw with wax, then flood the page with color. The color slides off the wax, and the drawing you could barely see appears. You made a hidden layer on purpose, and the wash revealed it. It is a small, happy pentimento, and it works for the same reason the lead map does: two layers that treat the same paint differently.
Try it, three ways
- Find the hidden hand. Trace your own hand in white crayon on white paper, pressing hard. You will hardly see it. Brush a dark watercolor wash over the whole sheet. The hand comes up white, the way the first try came up in the lead map.
- Scan a drawing with a window. On thin printer paper, draw a hand in pencil. Paint over it with opaque white tempera until it disappears. Let it dry, then hold the sheet flat against a sunny window. The pencil hand shows through. Light passes through the thin paint and is blocked by the graphite, the same way X-rays pass through the oil and stop at the lead.
- Check a white with Moseley’s rule. Titanium is element 22. Work it out: 21 × 21 × 10.2. You should get about 4,500 electronvolts. Titanium white was first sold to painters about a century ago. So if a painting said to be from 1776 answers at 4,500 wherever it is white, somebody painted over it, or painted all of it, long after Reynolds.
Where this comes from
- Anthony Solorzano, Los Angeles Times, August 18, 2026 — the XRF and reflectance scans, the reworked hand and head, the pigments, and the quotes from Higgs, Rayner and Gasparotto (also carried by the Associated Press)
- Joshua Reynolds’s Portrait of Mai and 18th-Century Portraiture — Getty Museum, the display dates and the 2023 joint acquisition with the National Portrait Gallery
- Mai (Omai) — born on Ra’iātea, arrived on HMS Adventure in October 1774, home to Huahine in 1777
- Pentimento — the word, the Arnolfini Portrait’s three positions for the feet, and paint growing transparent with age
- Moseley’s law — the energy of the strongest X-ray line is about ¾ × 13.6 eV × (Z − 1)², which is the 10.2 used above
- X-ray fluorescence — how knocking out an inner electron produces an element’s own X-ray
- Visible and infrared reflectance imaging spectroscopy of paintings — Delaney and colleagues, National Gallery of Art, on pigment mapping and why paint grows transparent in the infrared
- The Materials and Techniques — the Wallace Collection on Reynolds’s wax, bitumen and fading reds, and why his canvases are so often cracked or faded
- Prussian blue — an iron pigment first made in Berlin around 1706
- Titanium dioxide — titanium white as a twentieth-century pigment
The two diagrams above are original drawings made for this post. The measured X-ray energies are standard tabulated values.