A wash glows while it’s wet and goes duller as it dries. Painters often say watercolor “dries lighter”, but careful measurements say the bigger change is chroma: the dry color is grayer, and some colors even dry darker. Nothing has left the paper. Where does the color go?
The short answer
The pigment is all still there, and it absorbs exactly the same light wet or dry. What changes is how much light comes back before the pigment has colored it. That light is close to white, and white light mixed into a color makes it grayer.
The surface goes matte. A moist wash has a smooth, glossy surface. Its reflection goes off in one direction, like a mirror’s, and you look past it. A dry wash is rough, and its surface sends a few percent of white light in every direction: a thin white veil over the color.
Air replaces water between the particles. Dry watercolor is porous. A particle touching air bounces far more light than one touching water, because air’s refractive index (1.0) is much further from the pigment’s than water’s (1.33).
The paper whitens. Wet paper fibres scatter less light, so wet paper is darker and more see-through. As it dries it brightens, which lifts the whole wash.
The first two add white light and cost chroma. The third mostly adds lightness. Pigments whose refractive index is close to water’s change the most in the pores, because wet, they hardly scatter at all.
The same wash, wet and dry
Computed wavelength by wavelength, then turned into the color you’d see in daylight.
The same wash while it’s still wet and once it’s dry. Drying mostly makes the color grayer. The table shows how much each change does on its own.
MoistDry
What each change does on its own
Each row starts from the wet wash and changes one thing.
Change
Wet · after
ΔL*
ΔC*
ΔE
1What the measurements say
Bruce MacEvoy measured the drying shift of 75 common watercolor pigments and mixtures with a spectrophotometer, on Arches cold-press paper. He took the first reading 30 to 60 seconds after painting, while the swatch was “still glistening moist”, and the second at least four hours later [9]. His conclusion: loss of chroma is consistently the main change in appearance, many colors actually darken as they dry, and the common claim that “watercolors lighten as they dry” is inaccurate [9]. He also notes that the shift is largest for fine pigments: “many watercolors become lighter and less saturated as they dry, especially those with particle sizes below 1.0µm” [8].
Painters feel it too. Stephen Berry describes purples that go from “richly colored and dark in value” wet to “much paler and grayed out” dry, and sums up the practical lesson: “If it looks right wet, then it’s wrong” [2].
So the thing to explain is mainly a loss of chroma. The figure above follows the three changes one at a time. For hansa yellow at the starting settings, the matte surface takes away 17 units of chroma while barely changing the lightness. The paper drying adds 5 units of lightness. Together, the dry wash is 15 units less chromatic and 5 units lighter, and its purity falls from 92% to 83%.
2The surface: from glossy to matte
About 4% of light is reflected at a gum arabic–air surface (refractive index 1.48 against 1.0). What that light does depends on the surface. A wet wash is smooth: the reflection is a mirror image of the window, off in one direction, and from most viewpoints you look straight past it into the color. A dry wash is rough at the scale of the paper and the pigment, and it scatters that 4% evenly in every direction. It reaches your eye from every angle, a faint white layer over the color.
Saunderson’s correction for paint surfaces handles exactly this [11]. A matte surface adds a flat k1 to the reflectance. A flat addition is unfiltered white light, and adding white light slides the color toward gray. The Glaze vs. mix page shows the same effect for a glaze, section 5.6. A dark color suffers most, because 4% is a larger share of the little light it returns.
3The pores: air instead of water
Watercolor is a lean paint, with only a little gum arabic between the pigment particles. Coatings scientists call a film like this one above its critical pigment volume concentration: there isn’t enough binder to fill the space between particles, so the dry film has air pockets [1]. Those extra air–pigment interfaces scatter more light and make the film lighter [4].
While the paint is wet, water fills the pockets. A particle’s scattering depends on how far its refractive index is from its surroundings (see Transparent vs. opaque). Water (1.33) sits between air (1.0) and most pigments, so swapping water for air raises the contrast at every particle. The bars below use Mie theory [10][3] to compare each pigment’s scattering with its gaps full of water and full of air.
How much more each pigment scatters when dry
Mie theory at 550 nm, with the gum arabic share set above.
Once the water is gone, air fills the gaps around the pigment, and the grains bounce more white light back. Pigments that almost disappear in water change the most.
Refractive indices are from MacEvoy’s table [8] (burnt sienna is taken as 2.8, like the iron oxide venetian red). Particle sizes are typical values, assumed here. The space around a particle is treated as a simple mix of gum arabic (1.48) and water or air; the bar is the ratio of scattering per unit volume, dry over wet. The scale is logarithmic and stops at 100×.
Phthalo blue and hansa yellow (index about 1.4) almost vanish optically when wet: half gum arabic and half water averages to 1.41, nearly their own index, so they barely scatter at all. Dry, they scatter hundreds of times more. That extra scattering is unfiltered light, so it costs chroma: for hansa yellow, 6 units on its own. High-index pigments such as cadmium yellow contrast strongly with water as well as air, so drying changes their scattering only by about a third.
Scattering also changes direction when the contrast drops. Twomey, Bohren and Mergenthaler showed that particles in a liquid throw more of their light forward [12]. It travels deeper, meets more pigment, and is more likely to be absorbed before it escapes. The bars include this, since they count the light turned back, not just the light scattered.
4The paper
Paper is a mat of cellulose fibres (index about 1.55) with air between them. That is why it’s white: each fibre–air boundary reflects 4.7% of the light that hits it straight on, and there are thousands of boundaries. Wet the paper and water fills the gaps. A fibre–water boundary reflects only 0.6%, so the sheet scatters less. Treated as 20 µm cellulose spheres, the fibres scatter about three times less wet than dry. Wet paper is darker, and more see-through.
Wet paper turns see-through
The same sheet of paper, dry and wet, on a white board and over black ink.
Wet paper bounces less light, so it looks darker and you can see through it. As it dries it gets brighter again.
A sheet with Kubelka–Munk scattering S·X = 6 when dry, a little thinner than heavy watercolor paper. Wetting cuts the scattering to a third and leaves the absorption alone. The wet sheet shows the ink through it.
In a wash, the drying paper works against the other two effects. Brighter paper sends more light up through the pigment, which raises lightness and even adds a little chroma. In the table above, the paper is where most of the lightening comes from.
5A film of water on top
A wash that is still puddled has a continuous layer of water on top, and that adds a further effect. Light coming up out of the paint meets the water’s surface from below. Diffuse light inside the water meets that surface at every angle, and beyond about 49° all of it is reflected back down (total internal reflection). Averaged over angles, the surface sends 47% of the upcoming light back into the paint for another pass through the pigment. Each extra pass absorbs more, so the wet wash looks darker and richer. Lekner and Dorf used this to explain why some things look darker when wet [7].
Light trapped under a smooth surface
Fresnel reflection of diffuse light, averaged over every angle.
Solid line: diffuse light reaching the surface from inside a layer of the given index, reflected back down. Dashed line: diffuse light arriving from the air, reflected away. The two are linked by ri = 1 − (1 − re)/n². At n = 1.5 this gives 0.596, Judd’s classic value [5].
Switch the figure at the top to “Water film on top” to add it. The film makes the wet wash much darker, so the dry wash comes out far lighter: hansa yellow then gets 15 units lighter on drying instead of 5. MacEvoy’s measurements, taken while swatches were glistening but not puddled, show nothing like that. So in an ordinary wash, the film effect is much weaker than in this ideal flat layer of water. Measurements on wet particulate layers found that neither trapped light nor forward scattering explains wet darkening on its own [13].
6What painters do about it
Mix a little past the color you want. Painters learn to judge a wet wash as stronger and richer than it will dry [2]. Glazing over the dry wash to restore it adds pigment, so it tends to push the value darker too.
Fill the pores. A varnish or fixative soaks into the dry film and replaces the air with material of index about 1.5, which undoes much of the pore scattering. Its smooth surface also turns the matte veil back into a mirror reflection. That is why varnished watercolor looks deeper.
More binder. In this model, more gum arabic in the gaps means less air when dry and a smaller shift; try the slider. Treat that as the model’s prediction, not a measurement.
7What this model leaves out
The colors that darken. MacEvoy found many colors darken as they dry. Nothing in this model darkens a wash on drying, so it leaves out whatever does. One candidate is pigment moving as the water leaves: settling into the paper and its fibres, as staining pigments do [8], changes how much pigment sits in the light’s path. Read the model’s lightness changes as rough, and its chroma changes as the main result.
A simple picture of the gaps. The space around each particle is treated as an average of gum arabic and water or air, and each particle as an isolated sphere of an assumed size. Real films have particles touching, clumped and partly coated.
Illustrative pigments. The paint curves have the right kind of absorption and roughly the right opacity, but they aren’t measurements of real tubes.
Paper sizing. Watercolor paper is sized with gelatin or starch, which slows how fast water gets into the fibres. Here the paper is either fully wet or fully dry.
The math
The wash is one Kubelka–Munk layer on the paper [6]. Absorption K is the same wet and dry. Scattering is the dry value times f = σ(nwet) / σ(ndry), where σ is the Mie transport scattering per unit volume [3][10], averaged over a log-normal spread of sizes, and the surrounding index is n = φ·1.48 + (1 − φ)·nfill, with φ the gum arabic share and nfill = 1.333 (water) or 1.0 (air).
The paper keeps its absorption; its K/S, from R∞ via K/S = (1 − R∞)²/2R∞, grows by the Mie ratio for 20 µm cellulose spheres (1.55) in water versus air.
The surface follows Saunderson [11]. Dry and matte: Robs = k1 + (1 − k1)R, with k1 = ((1.48 − 1)/(1.48 + 1))² = 3.75%. Moist and glossy: Robs = R, since the mirror reflection misses the eye. With a water film: Robs = (1 − re)(1 − ri)R / (1 − riR), with re = 6.6% and ri = 47% for water, from Fresnel’s equations averaged over diffuse light [5].
Purity is the CIE excitation purity, 0 for white and 1 for light of a single wavelength. Chroma C* and ΔE are CIELAB and CIEDE2000.
justpaint color theory