How colour became a science

Newton passed sunlight through a prism in the 1660s and showed that white light is a mixture rather than a purity — the decisive fact, and the one that made everything afterwards possible. He then did something less often mentioned: he bent the spectrum into a circle so that the ends could meet, and divided it into seven named colours.

Seven is not what the eye reports. Newton was working partly by analogy with the seven intervals of the musical scale, an idea he found congenial and which has no basis in the physics he had just established. Indigo survives in schoolbooks today as a fossil of that decision. The circle, meanwhile, went on to become the colour wheel, an instrument painters would use for two centuries — inherited from a man who was not interested in painting.

Goethe published his Theory of Colours in 1810 as a direct attack on Newton, and as physics it is wrong. As a description of what happens in a human being looking at colour, large parts of it are excellent. He documented afterimages, coloured shadows and the way a colour induces its opposite in the surrounding field — phenomena Newton's account has nothing to say about, because they happen in the observer rather than in the light.

The split he opened has never really closed. There is the colour that can be measured with an instrument and the colour that is experienced by a nervous system, and they routinely disagree. Every argument about whether two things "match" is a version of it.

The most consequential experiment of the nineteenth century began as a customer service problem. Michel Eugène Chevreul, a chemist, was appointed director of dyeing at the Gobelins tapestry works, where clients kept complaining that certain blacks came out weak. He analysed the dyes and found nothing wrong with them. The weakness was optical: the blacks looked weak because of the colours woven next to them.

His law of simultaneous contrast, published in 1839, said that no colour has a fixed appearance — every colour is altered by its neighbours. It went from the dye works to the studios within a generation. Delacroix filled notebooks with complementary pairs; the Impressionists painted shadows in the opposite colour of the light and were mocked for it; Cézanne built solid form out of patches whose colour depends entirely on what surrounds them.

Albert Munsell, an American painter and teacher, wanted colour to be teachable to children without recourse to poetry. In 1905 he split it into three independent dimensions — hue, value, chroma — and arranged physical samples into a three-dimensional solid with a neutral grey axis running up the middle.

The most instructive thing about his solid is that it is lopsided. It bulges where colours can be intense and pinches where they cannot: yellow reaches maximum chroma high on the value axis, blue and violet only down in the dark. That asymmetry is a fact about pigments and about human vision, and it is the reason every palette system built as a neat symmetrical grid eventually fails to render.

In 1931 the international lighting commission ran colour-matching experiments with a small number of observers and averaged the results into a mathematical description of human colour vision. Every device that reproduces colour today descends from that model, through sRGB in 1996 and everything since. The screen you are reading this on is calibrated against a statistical average of a handful of people in a laboratory ninety years ago.

The same lineage runs into accessibility. A WCAG contrast ratio is a luminance calculation with hue thrown away — the modern, legally enforceable version of squinting at a canvas until it goes grey. Which is roughly where painters started. The science did not discover that value carries structure and that colours change their neighbours; it formalised what workshops had known and handed it back as something you could specify, standardise and check.