# The Spectrum Is a Line. The Wheel Is Us.

*Newton’s light runs from red to violet and stops. Painters have always drawn it as a circle. The gap between the two—the purples, the three cones, the quarrel with Goethe, the thin science of colour and mood—is where colour theory actually lives, and a painter who has checked the physics uses the wheel better for knowing which part of it is physics.*

Art · September 2026 · 18 min · by Epimystic
Canonical: https://epimystic.com/essays/the-spectrum-is-a-line-the-wheel-is-us/
Topics: perception, beauty, form, optics, colour

The board is tilted about ten degrees and the paint is moving. A pour is mostly waiting: you lay the colours down, tip the board, and for a few minutes physics does the painting and you watch. Where the cadmium red meets the phthalo blue there is a seam, and if I have been careless there will be a band of mud there by morning. If I have been careful there is a stripe of titanium white between them, and the red and the blue reach across it without touching, and the eye, which cannot help itself, supplies a purple that is not on the board.

That purple is the subject. Every colour wheel on every studio wall has a purple sector between red and violet, closing the circle. But the spectrum a prism throws on a wall is not a circle. It is a line, and it stops: red at one end, violet at the other, and no purple anywhere, because no wavelength of light is purple. The wheel is round because of something in us, not something in the light. I painted for years before I checked this, and checking it changed how I use the wheel.

## The Line and the Circle

Newton made the line in 1666 with a prism and a hole in a shutter, and printed it in 1704 in the _Opticks_. White light is a mixture, and the prism sorts the mixture by how much each part bends. He counted seven colours in it, for reasons that had more to do with music than with looking,[^1] and then did something that has been copied ever since without much thought. He bent the line into a circle. The two ends, red and violet, when mixed, gave a colour that appeared nowhere on the line—a purple—so he joined the ends through it, put white at the centre, and the colour wheel was born as a diagram of mixing rather than of light.

> For the Rays to speak properly are not coloured.
>
> — —Isaac Newton, Opticks, Book I, Part II

He knew what he was doing. That sentence is the modern position entire: the light has wavelengths, and the colour is what a nervous system makes of them. Goethe, a century later, refused it, and the argument has been told as poet against scientist ever since, which does neither justice. Goethe looked through a prism at a white wall and saw no spectrum, only white; the colours appeared at edges, where light met dark. From this he built a theory in which colour is born from the meeting of light and darkness, published it in 1810, and was sure to the end that it was his most important work. As physics it is wrong. As a record of what the eye does with edges, afterimages and coloured shadows it is the first careful description of a system nobody would name for seventy years. His wheel has six hues, with a purple he called _Purpur_ at the top, and beside every sector a paragraph on its effect on the feeling: yellow serene and a little vulgar when soiled, blue a contradiction of excitement and repose. He called this the sensual-moral effect. It is the ancestor of every colour-and-mood chart in every design textbook.

The wheel then grew a third dimension. Albert Munsell, a Boston painter, published _A Color Notation_ in 1905 and separated three things the wheel had always run together: hue, where you are round the circle; value, how light or dark; chroma, how far from grey. Set them out in space and the result is not a wheel but a lopsided tree, because yellow is strongest high up near white and blue low down near black, and a saturated dark yellow does not exist. Johannes Itten, teaching at the Bauhaus, flattened this back into a diagram a student could draw in an afternoon: twelve hues, three painter’s primaries of red, yellow and blue, seven kinds of contrast. _The Art of Color_ came out in 1961, and its wheel is the one on the wall.

The colour scientists’ version says the same thing more honestly. In the diagram adopted in 1931, every pure wavelength lies on an open horseshoe curve, and what closes it is a straight line drawn across the bottom from the violet tip to the red: the line of purples. Every colour on it is a mixture of the two ends, and none of them is a wavelength. On a modern hue circle those purples take up something like a quarter of the way round—a quarter of the wheel made of colours that light alone cannot produce, and exactly the colours that let it be a wheel.

*Figure: the-colour-wheel — Twelve hues in the order of the spectrum. Nine of them have a wavelength and a frequency; the three purples have neither, and the dashed arc marks where the line of light has been closed into a circle by the eye.* (drawn in the essay: https://epimystic.com/essays/the-spectrum-is-a-line-the-wheel-is-us/)

## The Numbers

It is worth having the physics in numbers, because the numbers are smaller and stranger than the mythology around them. Visible light runs from about 380 nanometres at the violet end to about 750 at the red. Frequency is the speed of light divided by wavelength, so the band runs from roughly 400 million million cycles a second at the red end to roughly 790 at the violet. Seven hundred and fifty over three hundred and eighty is not quite two: the whole of colour, everything a painting can do, fits inside a little less than one octave. Hearing gets ten. The sense we think of as rich is the narrow one, narrow in part because the molecule that catches the light, retinal, is tuned to flip its shape for photons in that range.

Each photon carries an energy equal to its frequency times Planck’s constant: between about 1.7 and 3.3 electron-volts across the band, about 2.3 for a green one. That is the order of a chemical bond, and it is no coincidence that vision sits here: a photon in this range can change the shape of a molecule and cannot break it. It is also why sunlight fades a painting from the blue end first.

What the numbers do not license is the other kind of claim: that red is a slow vibration and violet a fast one and the body feels the difference, that a wavelength carries an emotion. A red photon and a violet photon differ by less than a factor of two, and the retina does not measure either. It counts absorptions in three kinds of cell and throws the frequency away. [The rainbow chart on the meditation-room wall](https://epimystic.com/essays/the-wheels-that-were-never-coloured/) is a nineteenth-century graft onto a tradition that never mentioned colour, and between the frequencies of sound and of light lie thirty-four doublings with no ladder across them.

*Figure: the-visible-band — The visible band on one axis, read three ways beneath it—wavelength, frequency, energy per photon—with the sensitivity of the three cones above it, and the rods and melanopsin for company. The curves are schematic; the peaks are measured.* (drawn in the essay: https://epimystic.com/essays/the-spectrum-is-a-line-the-wheel-is-us/)

## Three Cones and a Quarrel

There are three kinds of cone in the retina and none of them is a colour. They are pigments with broad, overlapping appetites, absorbing best around 420, 534 and 564 nanometres. The names short, medium and long are more honest than blue, green and red, because the last two are nearly the same curve shifted a little, and the one called red peaks in the yellow-green. A single cone cannot tell a dim light at its favourite wavelength from a bright light at one it likes less; it just counts. Colour begins when the counts are compared. Thomas Young guessed this in 1802, Helmholtz worked it out in the 1850s, and it is why every screen gets away with three phosphors: the eye was only ever asking three questions.

But three questions do not make a circle. Ewald Hering pointed out in 1878 that some colours cannot coexist: there is a reddish yellow and a greenish yellow, a reddish blue and a greenish blue, but no reddish green and no yellowish blue. He proposed that beyond the receptors the signals are sorted into three opposed pairs—red against green, blue against yellow, light against dark—each a single channel that can swing one way or the other but not both. In the 1950s Leo Hurvich and Dorothea Jameson measured it, asking observers to cancel the redness in a light by adding green and the yellowness by adding blue, and the curves crossed zero where Hering said they should. Both theories were right, one after the other: three cones at the front, two opponent channels and a lightness channel behind.

Now the circle explains itself. The two colour channels are two axes, and two axes make a plane. Any hue is a direction in that plane, and directions in a plane close into a circle. Go round from red through yellow to green to blue and you arrive back at red having passed through the purples on the way, not because any light lives there but because the plane has no gap in it. The wheel is a map of the second stage of the visual system.

> **The wheel is round because opponent space is two-dimensional. It would be round even if the spectrum stopped dead, which it does.**

This is also the honest settlement of Newton and Goethe. Newton described the light, and about the light he was right. Goethe described the seeing, and everything he catalogued—the green shadow beside a red-lit one, the afterimage in the complement of what you stared at, the colours that bloom at edges—is the opponent system caught in the act, each channel rebounding from where it has been pushed. He mistook the rebound for the nature of light. But the painter has no access to light, only to the rebound. Goethe’s book is the wrong physics and the right studio manual, and a sentence in his preface is exactly true of the eye.

> Colours are the deeds of light, its deeds and sufferings.
>
> — —Goethe, Theory of Colours, preface (own rendering)

## Blue Light and the Clock

There is one route by which a wavelength reaches the mind that physics can trace end to end, and it does not go through seeing. In the late 1990s Ignacio Provencio found a fifth photopigment, melanopsin, and by 2002 David Berson had shown that it sits in a small population of retinal ganglion cells that respond to light on their own, without the rods or cones. They resolve no images and do not report to the parts of the brain that make pictures; they report to the clock. Their pigment absorbs best near 480 nanometres, in the blue-cyan, and the melatonin-suppression curves two groups measured in 2001 peak close by, around 460. This is why a day under an open sky sets the body’s rhythm and a warm lamp does not, and why screen light, rich in that band, has a measured effect on sleep.

The measurement worth knowing is a small controlled study from 2015 in which people read for four hours before bed, on five nights from a light-emitting tablet and on five from a printed book. On the tablet nights their evening melatonin was suppressed by about half, their clocks shifted later, they took about ten minutes longer to fall asleep, and they were less alert the next morning. That is the effect: real, modest, reproducible, and about timing and dose rather than beauty. A blue painting on a wall is not a screen held to the face at eleven at night. Short-wavelength light at the wrong hour moves the clock, and the clock moves the mood. That is not the claim that blue is calming, which is a different and much weaker one.

## What Colour Does to Mood

The largest survey ever done of what colours mean asked 4,598 people in thirty nations, speaking twenty-two languages, which emotions each of twelve colour terms carried. The pattern that came back was mostly shared. Red was love and anger nearly everywhere; yellow was joy; black was sadness and fear. Fourteen pairings turned up in almost every country. The edges were local: Nigerians alone associated red with fear, Egyptians alone did not associate yellow with joy, and the closer two nations were in language and geography the closer their patterns ran. Associations are real and widely held; part of that is plainly biology—blood, fire, sun, night—and part is learned, and the study cannot say which is which.

Association is not effect. The question that matters for the poster is whether a colour in the room changes what the people in it do or feel, and here the record is thin and getting thinner. Baker-Miller pink is the famous case: a 1979 report that a particular bubble-gum shade sapped the strength of anyone who looked at it, taken up by police cells across the West on a handful of small trials. Tested properly, with fifty-nine inmates of a Swiss prison randomly assigned to pink or plain cells for three days, it made no difference to aggression at all. The red-and-attraction effect, in which a woman photographed against red is rated more attractive, was the best-known finding of the field’s best-known laboratory; a decade of replication has shrunk it to something small and unstable, and the 2018 meta-analysis its originator wrote with two sceptics found an effect so modest that the co-authors could not agree whether it was there at all. His review of the field is unusually frank: small samples, colours named rather than measured, lightness and chroma left uncontrolled so that nobody can say whether an effect belonged to the hue at all.

That last point is the one a painter should keep, because it is the one that survives. When effects on feeling do turn up in careful work, they mostly track lightness and saturation rather than hue: dark and saturated reads heavy whatever the colour, pale and desaturated reads light. The rest is context. Red on a stop sign, a robe, a wound and a rose is one wavelength doing four jobs, and the job is assigned by everything around it. So here is what can honestly be said. Colours carry associations that are widely shared and partly learned. Colour can nudge arousal a little, in the direction the context already points. Blue light at night moves the clock. What cannot be said is that there is a table with wavelengths down one side and emotions down the other; every attempt to build one has come apart in someone else’s laboratory. Goethe’s sensual-moral effects are true the way proverbs are true.

## The Painter’s Wheel

None of this weakens the wheel in the studio. It tells you what the wheel is for. It is not a chart of light and not a chart of feeling; it is a chart of the opponent system, and every classical technique of colour is a way of playing that system on purpose. Complementaries sit across from each other because they push the same channel opposite ways, and placing them side by side makes each push harder, which is why a red beside a green looks redder than a red alone. Michel-Eugène Chevreul, running the dye works at the Gobelins tapestry manufactory, was sent complaints that his blacks were faulty; he found the blacks were fine and were being made to look brown or bluish by the colours woven beside them, and in 1839 wrote the law of simultaneous contrast to explain it. Delacroix read him. So did the Impressionists, who laid unmixed strokes side by side and let the eye mix them at a distance—broken colour, which makes a Monet haystack shimmer at ten paces and dissolve at one. Seurat took it to its limit with dots, on the promise that mixing in the eye would be brighter than mixing on the palette; it is not, quite, because the eye averages the dots rather than adding them, but the shimmer was real.

*Figure: the-grey-on-two-grounds — Simultaneous contrast and value. One grey on two grounds looks like two greys. Three hues of the same luminance sit above the single grey a photocopy would make of them. Beneath, the nine steps of value a painting is read in before any hue.* (drawn in the essay: https://epimystic.com/essays/the-spectrum-is-a-line-the-wheel-is-us/)

Warm and cool are the wheel’s other axis, and here the physics is on the painter’s side. Leonardo noticed that distant things go blue and pale and lose their edges, and told painters to render distance that way; he called it the perspective of colour. The reason is the air. Molecules scatter short wavelengths far more strongly than long ones, roughly six times more for blue than for red,[^2] so the light from a far hill has had its blue partly scattered out, and the air between you and the hill has scattered its own blue in. The hill loses warmth, the veil gains coolness, contrast falls, and the whole thing lightens toward the sky. Warm forward, cool back is not a convention. It is atmospheric optics, and a painter who pushes the foreground toward orange and the distance toward violet-grey is painting Rayleigh scattering without needing the name.

The limited palette is the same principle applied to harmony. The one attributed to Anders Zorn—yellow ochre, vermilion, ivory black and lead white—is a legend with a real palette behind it; the palettes preserved at his museum in Mora show he leaned on those four and reached for cobalt and viridian when a sky needed them. With four sources every mixture on the canvas shares ancestors, so the painting cannot help agreeing with itself, and the black stands in for blue by being cooler than everything around it: simultaneous contrast doing the work of a pigment. The same logic runs through every serious painter’s advice to put value before hue. A picture is read first as a pattern of light and dark, by the channel that carries almost all the detail, and only afterwards as colour.[^3] Photocopy a painting and if it still works, it works. Most of what goes wrong in a beginner’s colour is value, and no better red will fix a red at the wrong lightness.

Then there are the two ways paint can sit on a surface, which are two different pieces of optics. A glaze is a thin transparent layer over a dry one: light goes through it, hits the layer beneath, and comes back through it, filtered twice. The colour of a glazed passage is the colour of the light that survives that double passage, which is why it glows the way stained glass glows and why Van Eyck’s reds have a depth no opaque mixture can reach. Impasto is the other thing entirely: paint thick enough to have a surface, so that light scatters off the top of it and the paint casts its own small shadows. One works by transmission, the other by reflection, which is why a glaze must be transparent and why an impasto stroke changes colour when you move the lamp. [Cézanne’s mountain, painted in the wrong colours](https://epimystic.com/essays/what-cezanne-knew-that-the-camera-could-not/), is right for a related reason: he was recording the seeing, not the light.

*Figure: the-two-mixings — Two things called mixing. Three lights overlapping add their wavelengths and reach white; three pigments overlapping each subtract part of the light and reach something near black. The primaries differ, the complements differ, and the painter lives entirely on the right.* (drawn in the essay: https://epimystic.com/essays/the-spectrum-is-a-line-the-wheel-is-us/)

Under all of it sits the distinction that took physics until the 1850s to state cleanly. Lights add: red and green on one patch of wall make yellow, and with blue make white. Pigments subtract: a yellow pigment eats the blue, a blue pigment eats the red, and mixed they leave only the narrow band both let through, which is green. That is why the painter’s primaries are red, yellow and blue and the screen’s are red, green and blue, and why the painter’s complement of red is green and the physicist’s is cyan. The pour on my bench is subtractive from end to end, and the mud at a careless seam is what is left when two filters have between them eaten nearly everything.

## The Seam

So this is what a painter gets from checking the physics: not a rulebook, not the debunking of one, but a different sense of what the wheel is. It is not a picture of light, which is a line, and not a picture of feeling, which is a fog. It is a picture of the second stage of the visual system, and every technique that works—complementaries, broken colour, the cooled distance, the limited palette, the glaze—works by handling that system with intent. [A blue that once cost more than gold](https://epimystic.com/essays/on-ultramarine-and-the-price-of-blue/) carried its meaning by scarcity and contract, not by wavelength. Colour theory, done honestly, is the theory of a viewer.

Back at the bench the pour has stopped moving. At the base of the blue, where it ran thinnest over the white, the surface has broken into cells, a honeycomb of pale rings with blue walls, because the paints have different densities and the lighter has punched up through the heavier. That is fluid dynamics; I did not draw it. Above it the red and the blue still reach across their white stripe, and the purple between them is still there and still not on the board. I know now where it is: in me, at the second stage, in a plane that has no gap in it. That does not make it less. It makes it the one part of the painting that is not paint, the part [every eye brings, trained or not](https://epimystic.com/essays/seeing-is-a-skill-not-a-gift/), and the reason the wheel on the wall has been round, and useful, and wrong about light, for three hundred and twenty years.


---

[^1]: Newton’s seven—red, orange, yellow, green, blue, indigo, violet—were chosen to match the seven notes of the diatonic scale; he divided the circle in the proportions of the string lengths. Indigo has been quietly dropped by most people since, and nobody has missed it.
[^2]: Rayleigh scattering goes as the inverse fourth power of wavelength. Take 450 nanometres for the blue and 700 for the red: (700/450) to the fourth is about 5.9, so blue light is scattered out of a beam, and into the air between you and the hills, roughly six times as strongly as red.
[^3]: Munsell’s value steps were set by asking observers for equal-looking intervals between black and white, which is why they are not equal intervals of light. The modern CIE lightness scale, L*, is a cube root of relative luminance for the same reason: the eye compresses.

## Further reading

- **Opticks** — Isaac Newton. The 1704 book in which the prism experiments are written up and the first colour circle is drawn; Book I, Part II is where he says the rays are not themselves coloured, which is the whole modern position in one sentence three centuries early.
- **Theory of Colours** — Johann Wolfgang von Goethe, trans. Charles Eastlake. Wrong about light, right about looking. The sections on coloured shadows, afterimages and the ‘sensual-moral effect’ of colour are the first careful phenomenology of the opponent system, written nearly seventy years before anyone knew there was one.
- **The Art of Color** — Johannes Itten. The Bauhaus course that fixed the twelve-hue wheel and the seven contrasts on every studio wall. Its physics is loose and its psychology is asserted rather than shown, but as a manual for what complementaries and warm-cool do on a canvas it has not been bettered.
- **Interaction of Color** — Josef Albers. The book that replaces theory with exercises: the same grey on two grounds, one colour made to look like two, two made to look like one. Simultaneous contrast taught by the hand rather than the diagram.
- **Color Psychology: Effects of Perceiving Color on Psychological Functioning in Humans** — Andrew Elliot and Markus Maier. The 2014 Annual Review of Psychology survey, honest about how small and methodologically shaky the experimental record on colour and behaviour is, written by the people responsible for the best-known effect in it.
- **Universal Patterns in Color-Emotion Associations Are Further Shaped by Linguistic and Geographic Proximity** — Domicele Jonauskaite and colleagues. Psychological Science, 2020: 4,598 people in thirty nations asked which emotions each colour term carries. The largest survey of its kind, and the source for the claim that associations are widely shared while effects are not.