Jennifer B. Barnes sent us this photograph, and a question with it: why do certain pigments seem to glow, and why does putting two particular colors side by side make the glow stronger?
Here is the surprise, and it is most of the answer in one sentence: every color on that wall is duller than it will be. Those jars hold dry powder, and dry is the weakest a pigment ever looks. Nothing on those shelves has its real color yet.
Try this first, it takes ten seconds
Find a stone. Any stone, but one from a beach or a river is best. Look at it dry. Now wet it, or breathe hard on it, or drop it into a glass of water.
It goes darker, and the color goes deeper. Everybody has seen this. Almost nobody has been told why, and the why is the same physics that makes a Rembrandt look lit from inside.
Air is the problem
Light bends when it crosses from one material into another, and some of it bounces back instead of crossing at all. How much of it bounces depends on how different the two materials are. Physicists put a number on each material for this, the refractive index. Air is about 1.0. Water is about 1.33. Linseed oil and the resins in varnish are about 1.5. The minerals that pigments are ground from run from about 1.5 to well over 2.
Dry pigment is a heap of those mineral grains with air in every gap. So every grain surface is a hard boundary, 1.0 on one side and 1.6 on the other, and light hitting that heap bounces off the first few grains and comes straight back out. It never gets in far enough to be colored much. What returns is a pale, chalky version of the hue with a white haze of surface bounce laid over the top of it.
Now wet it. The air is gone, and the gaps are full of water, or oil, or gum. The difference across each grain boundary collapses from 1.0-against-1.6 down to 1.33-against-1.6, and the bouncing largely stops. Light goes in.
Two things follow, and together they are what people mean by glow. The light now takes a long, wandering path through the material before it finds its way out, so it has far more chances to be absorbed. That is the darkening. And every photon that does escape has spent that whole path inside the color. That is the saturation.
In watercolor, the paper is the lamp
Watercolor is the purest case of this. It is pigment, a little gum arabic to hold it down, and nothing else. No white filler, no opaque body. You lay it thin.
So light goes down through the wash, strikes the brilliant white cellulose of the paper, and comes back up through the wash to your eye. The paper is the light source. The paint is a colored window you are looking through, twice.
That is why the brightest thing in a good watercolor is nearly always a place the brush never went. You cannot add light to a watercolor. You can only choose where not to take it away.
One piece of arithmetic, and it changes how you paint
Say a wash is thin enough to let 70% of the light through.
Light goes down through it, so 70% survives. It reflects off the paper. It comes back up through the same wash, so 70% of that survives.
0.7 × 0.7 = 0.49About half. One thin wash has already halved your paper.
Now put a second coat on. The light crosses two layers going down and two coming back up:
0.49 × 0.49 = 0.24Here is the part worth sitting with. The second coat did not take you from 70% down to 40%. It took you from 49% down to 24%. It halved what was left, again. Paint counts twice on every trip, and coats multiply rather than add.
That is the real reason watercolor is unforgiving, and it has nothing to do with skill. Every wash is a permanent subtraction from the only lamp you have.
Oils play the same trick with a better lamp
The Dutch painters got there two centuries earlier by another route. Rembrandt would lay in a monochrome underpainting first, a pale tonal layer sometimes called a dead layer, let it dry, and then float thin transparent glazes over the top of it.
A glaze works because the oil binder and the transparent pigments suspended in it have refractive indices close to one another, both near 1.5. With so little difference between them there is almost nothing to scatter off, so light travels down through the stack of dry layers, bounces off that pale underpainting, and climbs back out saturated.
The mechanism is the watercolor mechanism exactly. What differs is the reflector. A watercolorist’s lamp is the paper, and it is flat, white and fixed. An oil painter builds the lamp first, in paint, and can put the brightness precisely where the picture needs it.
Some pigments are simply cleaner than others
A color looks radiant when it absorbs a narrow slice of the spectrum and hands back everything else. A color looks muddy when it takes a little bit out of everywhere.
Cadmium pigments are crystals that behave like semiconductors. They have a sharp cutoff: light above a certain energy is absorbed almost completely, light below it is returned almost completely, and the changeover is closer to a cliff than a slope. That near-vertical edge is what makes cadmium yellow read as a pulse rather than a shade.
The modern transparent organics do it with chemistry instead. Quinacridone, phthalocyanine and dioxazine are built from long chains of alternating single and double bonds, and their electrons spread across the whole chain rather than sitting still in one place. That lets the molecule remove one narrow band of wavelengths with real precision and pass the rest straight through, which is also why so little of it goes such a long way.
And then the part that is not physics at all
Everything so far only gets saturated light to your eye. None of it makes anything glow, because your brain does not measure light. It measures ratios.
Put a high-chroma warm, an Indian yellow or a new gamboge, hard against something dark, dull and roughly opposite in hue, and your visual system exaggerates the boundary between them before you are aware of it. Michel Eugène Chevreul wrote the law down in 1839, while he was running the dye works at the Gobelins tapestry factory and trying to work out why perfectly good yarns kept looking wrong beside each other.
There is a second move that goes with it, borrowed from photography, where it is called halation. When a light really is too bright for a camera or an eye to resolve, it bleeds a little past its own edges. Painters imitate that deliberately: soften the edge on the bright side of a highlight while keeping the outer silhouette crisp. The brain reads that soft edge as this was too bright to focus on, and fills in the rest by itself.
So the answer to Jennifer’s second question, why particular colors next to each other glow more, is that the glow was never in either color. It is in the border between them.
Where this stops being true
Paint does not emit light. A painted surface cannot send back more light than falls on it, so a canvas in a dark room is just a dark canvas. Everything above is about steering light that was already in the room.
There are two honest exceptions to that, and they pull in opposite directions.
The first is real. Daylight fluorescent pigments genuinely do return more visible light than lands on them, because they take in ultraviolet you cannot see and give it back at wavelengths you can. That is actual emission rather than an effect of contrast, and it is why a fluorescent orange looks like it has been plugged into something. It is also why those pigments never sit comfortably in a painting built the other way.
The second is where the evidence runs out. There is a point at which a patch of color stops reading as a bright surface and starts reading as a source, a light rather than a thing lit. That point has been measured. In the 1990s two vision scientists, Frederick Bonato and Alan Gilchrist, showed people gray patches under controlled light and asked one question: does this look lit, or does it look like a light? Across many different setups the answer flipped at about the same place, when a patch was roughly 1.7 times brighter than whatever read as white in the same scene.
Painters had been working that line for centuries without the number. Joseph Wright of Derby’s An Experiment on a Bird in the Air Pump (1768) is lit by a single flame, and he hides it behind the glass bowl on the table. Paint cannot be brighter than the white priming it sits on, so no painter can put down a flame that is 1.7 times whiter than white. What Wright could do was push everything else in the room down, further and further, until the lit faces sat far enough above their surroundings to cross the line. The glow is manufactured out of darkness.
And then Monet breaks the rule. In Impression, Sunrise (1872) the orange sun looks like the brightest thing on the canvas, and it is not. Margaret Livingstone, a neuroscientist at Harvard, measured it with a photometer and found the sun has the same luminance as the gray sky around it. Make a black-and-white copy of the painting and the sun all but disappears. It reads as a source with no brightness advantage at all, on color alone.
So the threshold is real, and it moves. Bonato and Gilchrist found it climbs as the patch gets bigger, and it shifts with the surround and from one person to the next. No formula will tell a painter in advance where it sits for the picture in front of them. Painters find it by trying, comparing, and trying again. The physics explains why the effect is available. It does not tell you when you have arrived at it.
Try it, three ways
- The stone. Wet one half of a flat beach stone and leave the other half dry. Look at the line between the halves. That line is the whole first half of this article.
- The same yellow twice. Cut two identical squares from one sheet of yellow paper, one sheet so you know they match. Glue one onto black paper and one onto white. Then argue with somebody about whether they are the same color.
- One wash against two. This is the arithmetic above, made visible. Paint a stripe of any watercolor across white paper and let it dry completely. Paint a second stripe of the same color beside it, let that dry, and give the second stripe one more coat. Now compare the two stripes in two ways. First hold the sheet up to a window with the light coming through from behind. The window is the lamp, and light crosses each stripe once on its way to your eye. Then lay the sheet flat on the table. Now the paper is the lamp, and light crosses each stripe twice, down and back. What to watch is the gap between the stripes. Against the window the two-coat stripe is darker. On the table it should be darker by more, because on the table every coat is counted twice. Same paint, two different jobs, and the table is the one you paint for.
Where this comes from
- Reflectance and albedo differences between wet and dry surfaces — Twomey, Bohren and Mergenthaler, Applied Optics 25(3), 1986, on why wet things go dark
- Why does wet sand look darker than dry sand? — Hong Kong Observatory, the same physics in plain language
- Why do wet materials become darker? — Revista Mètode, University of Valencia
- How watercolor paints are made — handprint, on particle size, transparency, and letting the paper show through
- Contrast effect — on simultaneous contrast, and on Chevreul’s law of 1839
- The perception of luminosity on different backgrounds and in different illuminations — Bonato and Gilchrist, Perception 23, 1994, the 1.7-times-white threshold
- Perceived area and the luminosity threshold — Bonato and Gilchrist, Perception & Psychophysics 61, 1999, on why a bigger patch needs to be brighter before it reads as a light
- An Experiment on a Bird in the Air Pump — Joseph Wright of Derby, 1768, National Gallery, London
- Impression, Sunrise — Monet, 1872, with Margaret Livingstone’s photometer measurement of the sun against the sky
The two diagrams above are original drawings made for this post. The photograph of the pigment wall is by Jennifer B. Barnes and appears with her permission.