Short answer
Water, warmth and reactive gases each change colour in the dark, and in the published record they do so about as often as light does: 28 of the 51 degradation records Colourwise has compiled name no light at all. Water vapour dissolves and moves salts and drives hydrolysis; heat speeds every thermal reaction and converts some minerals outright; ozone, nitrogen dioxide and sulfur gases attack dyes, arsenic pigments, lead white and silver directly. Which of them matters depends on the material, so a low light level protects only against one agent of several.
Moisture does two different jobs. As a reagent it takes part in hydrolysis: the Canadian Conservation Institute notes that acid hydrolysis proceeds at any humidity above zero and slows by more than half each time relative humidity is halved, which is the chemistry behind brittle, browning paper and failing cellulose plastics. As a carrier it moves soluble material to where it shows. In The Scream, moisture and chlorine compounds — not light — were found to drive the conversion of cadmium yellow to soluble sulfates that migrate and recrystallise. Modern oil paints kept twelve weeks at 75% humidity under strong light grew magnesium sulfate crystals on their surface without any polluted air. On walls, water passing through masonry leaves salts as a white veil. And in damp air mould takes over, discolouring organic materials: by the Canadian Conservation Institute's figures it appears in about a hundred days at 75% humidity and in a couple of days above 90%.
Warmth usually acts by speeding reactions that would happen anyway. The dark fading of chromogenic photographs and motion-picture film, the yellowing of PVC as it loses hydrogen chloride and the self-accelerating breakdown of cellulose acetate all run faster as temperature rises; the Canadian Conservation Institute's rule of thumb is that each five-degree reduction roughly doubles an object's chemical lifetime. Occasionally heat converts a colourant in one step. Yellow ochre is the mineral goethite, which dehydrates to red hematite when heated — the reaction potters and painters use to make red ochre, and the reason large yellow wall panels at Pompeii and Herculaneum are red today. Azurite heated or exposed to alkaline conditions turns to black copper oxide, readily enough that an analytical laser can blacken fine grains while measuring them.
In the 1980s Paul Whitmore and Glen Cass exposed artists' colourants to single pollutants in dark chambers. After twelve weeks in ozone at 0.40 parts per million, indigo and turmeric had reacted, orpiment was the only inorganic pigment to lose colour severely, and a nineteenth-century Japanese woodblock print had faded only in its indigo-containing blues and greens. Nitrogen dioxide at 0.50 parts per million changed ten natural organic colourants on paper by more than two ΔE units, the arsenic sulfide pigments and iron inks by more than five, and nine of twenty-three dyed silks by at least two. Reduced-sulfur gases such as hydrogen sulfide work differently, building dark products on the surface: they tarnish silver, including the metal threads of tapestries, and darken lead white.
The authors equated their nitrogen dioxide dose to about two years inside an unprotected museum in downtown Los Angeles at the time, or five to six years in many other cities.
Chlorides and sulfates rarely appear in lists of museum hazards, yet they sit behind several of the best-documented colour changes. Chloride turns vermilion grey-black on panel paintings and azurite green on plaster; sulfation of a lime ground laid a black gypsum film over Pompeian cinnabar; rainwater leaking through the Sistine Chapel vault left white efflorescence that earlier restorers disguised with glue, which then darkened. These reactions need a source of ions and usually moisture to move them, so they follow leaks, rising damp, sea air and accumulated dirt rather than the lighting plan. They are also the changes most likely to be made worse by well-meant wetting, which is one reason published guidance leaves the treatment of salt-affected surfaces to conservators.
30 of the 51 records name more than one driver, and several show one agent switching another on. Zinc yellow needed light and sulfur dioxide, and the humidity decided which product formed. Vermilion needs light and chloride. Sulfate crystals grew on modern oil paint only when high humidity and strong light coincided. Removing oxygen, which protects most colourants, makes Prussian blue and verdigris fade faster. This is the practical argument for the 'agents of deterioration' framework conservation bodies use: a store that is dark but damp or warm has controlled one agent and left the others. The table sets out the records in which light plays no documented part.
| Colourant | Object class | Driver detail | What changed |
|---|---|---|---|
| Cadmium yellow (cadmium sulfide) in oil | Easel paintings | Moisture and mobile chlorine compounds are the key promoters; light plays a lesser role. | Alteration of the yellow paint to pale sulfate products. |
| Smalt (cobalt potash glass) in oil | Easel paintings | Reaction with the oil medium. The National Gallery authors suggest humidity swings and cleaning solvents could further alter the hygroscopic surface crust. | Blue turns grey, brownish or colourless, sometimes blanched. |
| Red lead and lead-tin yellow in oil | Easel paintings | Reaction between pigment and oil binder over time. | White, translucent lumps appear in and erupt through the paint, speckling the surface. |
| Zinc white (zinc oxide) in oil | Easel paintings | Pigment particle properties, fatty-acid profile, paint additives and environmental conditions are all named as factors. | Soap aggregates linked to deterioration of the paint layer. The review abstract read does not describe a colour change. |
| Emerald green (copper acetoarsenite) in oil | Easel paintings | Reaction with the binder; reproduced in the laboratory at room temperature and normal light. | Green turns brown; the picture darkens overall. |
| Verdigris (copper acetate) and copper greens in oil–resin media | Easel paintings | Browning tracks how easily copper is extracted — readily from verdigris, slowly from basic copper carbonate. Separately, verdigris dry pigment changed more under light without oxygen than in air. | Transparent green turns brown. |
| Cinnabar (mercury sulfide) on lime mortar | Wall paintings, sculpture and architectural surfaces | Chlorine attributed to sodium chloride from outside the painting; the gypsum layer is about 5 µm thick. | Deep red turns grey or black. |
| Lead white on plaster | Wall paintings, sculpture and architectural surfaces | Occurs where lead white was used in mural painting. A 2023 study says the oxidation is often attributed to the strong alkalinity of slaked lime or to oxidising agents, and links the darkened lead white it examined at Assisi to reactive chlorine compounds. | White passages turn brownish and dark. |
| Azurite (basic copper carbonate) on plaster | Wall paintings, sculpture and architectural surfaces | Needs a source of chloride near the painting. | Blue turns green. |
| Azurite (basic copper carbonate) on plaster | Wall paintings, sculpture and architectural surfaces | Thermal alteration, or chemical alteration under alkaline conditions; a measuring laser can itself blacken grains below about 25 µm. | Blue turns black. |
| Yellow ochre (goethite) | Wall paintings, sculpture and architectural surfaces | Heat from the AD 79 eruption — hot lapilli, gases or pyroclastic flow, depending on distance from the volcano. | Yellow panels turn red. |
| Fresco surface under soot, dust and restorers' animal glue | Wall paintings, sculpture and architectural surfaces | Centuries of candle and brazier smoke; glue applied repeatedly to revive colours. | Colours appear dark and brownish, 'as if through a smoked glass'. |
| Fresco surface with soluble salts | Wall paintings, sculpture and architectural surfaces | Rainwater infiltration through the vault. | White efflorescence over the painted surface. |
| Paint layers over an organic (lacquer) ground on terracotta | Wall paintings, sculpture and architectural surfaces | Long burial in waterlogged soil. | Painted colour is at risk of being lost with its ground. |
| Manganese-bearing medieval window glass | Wall paintings, sculpture and architectural surfaces | The abstract links the extent of browning to the manganese oxidation state and does not name an environmental trigger. | Brown spots and patches darken the glass. |
| Iron gall ink | Manuscripts and works on paper | Strongly influenced by temperature and humidity. In a chamber test, iron inks changed by more than ΔE 5 after twelve weeks at 0.5 ppm nitrogen dioxide. | The paper browns around the ink line; the brown spreads through the sheet and offsets onto facing pages; the line may finally drop out. |
| Lead white in watercolour, gouache and drawing highlights | Manuscripts and works on paper | Hydrogen sulfide and related gases in air. | White highlights turn grey, brown or black. |
| Indigo, turmeric and orpiment on paper | Manuscripts and works on paper | 0.40 ppm ozone at 22 °C and 50% RH for twelve weeks, without light. | Blues and greens containing indigo fade; orpiment loses its yellow. |
| Natural organic colourants, orpiment and realgar on paper | Manuscripts and works on paper | 0.50 ppm for twelve weeks — equated by the authors to about two years in an unprotected museum in downtown Los Angeles. | Ten natural organic colourants changed by more than ΔE 2; the arsenic sulfides by more than ΔE 5. |
| Silver and gilt metal threads wrapped on silk | Textiles | Sulfur-containing gases such as hydrogen sulfide. | Bright metallic threads turn dark and dull. |
| Traditional Japanese plant dyes mordanted on silk | Textiles | After twelve weeks at 0.5 ppm nitrogen dioxide, nine of 23 dyed silks had changed by at least ΔE 2, one (from the pagoda tree) by more than ΔE 7. | Measurable fading or colour shift without any light. |
| Water-soluble dyes in textiles and watercolour | Textiles | Damp conditions or wetting. | Colour spreads into neighbouring areas. |
| Scheele's green and emerald green (copper–arsenic greens) | Historic interiors and wallpaper | Acidic conditions, on the common account. | Loss or alteration of the green. |
| Oil and alkyd house paint | Historic interiors and wallpaper | Worst behind pictures and inside cupboards, and near stoves, radiators and ducts. | Whites and pale tints turn cream to yellow. |
| Chromogenic cyan, magenta and yellow image dyes | Photographs and film | Proceeds at room temperature; rate rises with temperature and humidity. | A colour cast as the least stable dye goes, with yellowish staining of whites. |
| Chromogenic dyes in post-1950 motion picture film | Photographs and film | Significant fading in about forty years at room conditions. | The image drifts to a purplish-pink cast. |
| Poly(vinyl chloride), plasticised | Plastics | Lower temperature, lower relative humidity, higher molecular weight and higher plasticiser content all slow yellowing; activation energy 86 kJ/mol. | Clear or white PVC turns yellow (rising b*). |
| Cellulose acetate and cellulose nitrate | Plastics | Self-accelerating once acid is present. | Discolouration (yellowing or opacifying), warping, crazing and stickiness. |
| Driver | Records naming it | Share of all records | Of those, irreversible |
|---|---|---|---|
| Light and UV | 23 | 45% | 15 |
| Humidity and water | 17 | 33% | 12 |
| Heat | 9 | 18% | 8 |
| Gaseous pollutants | 10 | 20% | 5 |
| Chlorides and salts | 6 | 12% | 3 |
| Oxygen | 6 | 12% | 5 |
| Low oxygen (anoxia) | 4 | 8% | 1 |
| Acidity or alkalinity | 7 | 14% | 4 |
| Adjacent materials (binder, support, varnish) | 9 | 18% | 6 |
Why: Indigo-based colours react with ozone and nitrogen dioxide in ordinary city air; no light is needed.
Fix: The chamber studies point to air quality, not light; what suits a particular work is for a paper conservator to judge.
Why: Lead white darkened by sulfur-containing gases.
Fix: This is a recognised condition with a contested treatment history; do not attempt anything and ask a paper conservator.
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
FactStrong evidence
After twelve weeks at 0.50 ppm nitrogen dioxide in the dark, ten natural organic colourants on paper changed by more than ΔE 2, orpiment, realgar and iron inks by more than ΔE 5, and nine of 23 dyed Japanese silks by at least ΔE 2.
Method: Chamber exposure with colour difference calculated from reflectance spectra.
Caveat: Chamber concentration; the authors equate the dose to about two years in an unprotected 1980s Los Angeles museum.
FactStrong evidence
After twelve weeks at 0.40 ppm ozone at 22 °C and 50% RH without light, a nineteenth-century Japanese woodblock print faded significantly only in the blue and green areas containing indigo.
Caveat: One print; ozone well above ordinary indoor levels.
Source: The ozone fading of traditional Japanese colorants (Studies in Conservation, 1988)
FactModerate evidence
Cadmium yellow and French ultramarine oil paints aged for twelve weeks at 75% RH under elevated light formed hydrated magnesium sulfate crystals on their surfaces without elevated sulfur dioxide.
Caveat: Tube paints from one manufacturer.
FactModerate evidence
Heat from the AD 79 eruption converted yellow goethite to red hematite in wall paintings at Pompeii and Herculaneum, turning extensive yellow panels red.
Caveat: Cited from a conference abstract.
FactModerate evidence
Reduced-sulfur gases such as hydrogen sulfide tarnish silver and darken lead white, and acid hydrolysis slows by more than half each time relative humidity is halved.
Caveat: The two agents-of-deterioration pages were read in summary rather than in full.
Source: Agent of deterioration: pollutants; Agent of deterioration: incorrect relative humidity; Silver — Care and Tarnish Removal — CCI Notes 9/7
Colourwise analysisModerate evidence
28 of 51 degradation records (55%) name no light among their drivers, and 30 name more than one driver.
Based on: Colourwise count over its colour-degradation dataset: records whose driver list omits 'light', and records with two or more drivers. The counts describe the published cases selected, not their frequency in collections.
Source: The fading of artists' colorants by exposure to atmospheric nitrogen dioxide (Studies in Conservation, 1989); Probing the chemistry of CdS paints in The Scream by in situ noninvasive spectroscopies and synchrotron radiation x-ray techniques (Science Advances, 2020)
Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.