The rule under draping was found in a dye house

In 1824 Michel-Eugène Chevreul, by then a well-known chemist, was made director of the dyeing department at the Gobelins manufactory in Paris. Within a few years the weavers were complaining about his dyers. The complaint was specific: the black wool used in the shading of the blue and violet draperies looked weak. Chevreul did what a chemist does and set his blacks against the best wool he could get from London and Vienna. The dye was fine.

What was wrong was not in the wool. Black laid beside blue was picking up a cast of orange; black laid beside violet was going yellowish. The material had not changed — its neighbours had. He read a memoir on it to the Académie des sciences on 7 April 1828, then spent eleven more years on the book, held up in large part by the difficulty of getting colour plates he could trust. It came out in 1839 under a title that runs on through tapestry, painting, carpets, mosaic, stained glass, printing, clothing and horticulture, because he thought the law reached all of them.

The law is one sentence, and it is worth having in the shape he gave it. When the eye sees two colours touching, it sees them as unlike each other as it can — in hue and in strength both. The word "simultaneous" is his, chosen to mark this case off from the successive one, where two colours land on the same patch of retina one after the other.

The lightness case is the one you can check without equipment, and the picture above is it: same paper, two grounds, and they will not sit still as the same grey. Chevreul was once called as an expert in a suit between a wallpaper maker and the client who had ordered a grey pattern on a green ground and then refused to pay, because the grey had come out reddish. He testified that both of them were right. To show it he cut the pattern's outline out of a white sheet and laid it over the paper so the green was hidden, and the grey went grey. His remedy was to mix a little of the ground's own colour into the pattern, cancelling what the eye was adding.

The explanation is where it gets contested. Chevreul held that the eye adds to each colour a little of its neighbour's complementary, and his definition of a complementary ran the mixing of lights together with the mixing of pigments — a confusion common in his day, and still a confusion. He later conceded that the demonstration of his own law was not satisfying. And a different account had been available since 1789: Gaspard Monge had argued that contrast falls out of the visual system's estimate of the illuminant. Chevreul never engaged with it, and Georges Roque, whose account of all this is the one to read, reports that the reason was probably political rather than scientific.

Two centuries on, the phenomenon is not in doubt and the mechanism still is. Soranzo's survey of it sets the argument up as Hering against Helmholtz — interaction between retinal neurons, against a judgement about the whole configuration — and describes the balance as having moved towards the higher-level accounts over recent decades without the low-level ones being abandoned. Lateral inhibition on its own cannot carry it: White's 1979 pattern is the standing counter-example — two identical greys set into a striped ground, of which the one lying along the white stripes is the one that looks lighter, the reverse of what a surround that inhibits would predict. Shevell and Kingdom's review states the working conclusion for anything with more than one thing in the frame: the colour of an object is not settled by the light coming from that object.

Faces are not exempt, and this has been measured on them. Chang, Cheang and So asked 46 observers — 23 ethnically Chinese, 23 White — to match the lightness of one photographed face to another. Matching within a group, the errors were small and even. Matching across groups, they were systematic, and their direction depended on which face was the standard. The two observer groups made the same errors as each other: a lifetime of looking at one kind of face protected nobody from the effect, which the authors describe as simultaneous contrast.

All of which is the argument for a dull procedure. Clear the frame — hair back, face bare, nothing with a hue in it behind you or on you. That is not fastidiousness; it is the only way to stop the surround writing part of the answer. Change one thing at a time, because two changes at once and you cannot say which one the face answered. Watch the face rather than the cloth: the shadow under the eye, the line beside the mouth, whether the skin reads even or blotchy. And compare side by side rather than one after the other — Chevreul coined a word to separate those two cases because they are not the same comparison.

A colour that suits you does not make itself look better. It makes the face look rested. That is an observation rather than a measurement, and the reason it has to be made carefully is a law from 1839 about black wool.