Short answer
Lux measures how bright light looks to a person, not how much damage it does, so two lamps at the same lux can fade the same object at different rates. Which wavelengths matter depends on the colourant: for fugitive reds and yellows damage falls steadily towards the red end, but for blues it does not follow that rule. Conservation science has added three tools to the lux budget — choosing the spectrum, measuring the object's own sensitivity by microfading, and removing oxygen — and each comes with documented exceptions.
Illuminance weights every wavelength by the eye's sensitivity, which peaks in the green. A colourant responds to what it absorbs. David Saunders and Jo Kirby at the National Gallery exposed fugitive pigments and dyed textiles to separate bands of visible light and found that for red and yellow pigments damage decreased as wavelength increased, as photon energy would suggest; for blue pigments and for the first three ISO blue wool standards the relationship was less straightforward, and blue light was not necessarily the most damaging. Their conclusion was that damage depends both on the energy of the radiation and on the spectral absorbance of the colourant. It also means the blue wool standards, which are themselves blue dyes, do not report equally well on every colour in a collection.
White LEDs emit almost no ultraviolet or infrared in the beam, which removes two sources of damage at once. The concern has been the blue emission peak that drives the phosphor. The Canadian Conservation Institute's technical bulletin on LED lighting, written with the Getty's Jim Druzik, concludes that an LED chosen for high colour quality — a colour rendering index above 90 — causes no more light damage than halogen or ultraviolet-filtered daylight, while lamps with a large blue spike can cause up to twice the damage of a filtered 3000 K halogen. It adds that fading is modestly faster at higher colour temperature, up to about half as fast again at 6500 K as at 3000 K for most colours other than blues. The spectrum, in other words, is something a museum now selects rather than accepts.
The Getty's lighting research also investigated three-band filtered sources intended to cut the total energy delivered to the object.
Category tables assign an object a sensitivity from what it is thought to be made of. Microfading measures it. In the method Paul Whitmore and colleagues published in 1999, an intense beam fades a spot 0.4 millimetres across while a spectrometer follows its colour, and the test stops once a small, definite change has registered — enough to rank the material against the blue wool standards and, in particular, to pick out anything more fugitive than Blue Wool 2. The results compared well with conventional accelerated tests. The authors were open about the limits: the risk of leaving a visible mark was judged very small, but the spot can heat to about 50 °C, so waxes and other low-melting materials are at risk. The Getty's lighting programme went on to refine and standardise the technique for use in risk assessment.
Most fading is photo-oxidation, so taking oxygen away should slow it. The Getty Conservation Institute tested the idea on 125 samples — pigments, dyed textiles, watercolours, inks and natural-history specimens — given about 17.5 million lux-hours in air and in near-anoxic conditions. In 113 of them (90%) colour changed less without oxygen, and nearly half of those improved by a factor of four or more. Six changed more: three Prussian blue watercolours, Antwerp blue, verdigris as dry pigment and a fluorescent yellow gouache. Prussian blue is the important one because it is so common, and separate work on dyed textiles found that three weeks of anoxic insect fumigation faded it significantly, with only partial recovery. The dyes of early autochrome photographs, by contrast, all faded more slowly without oxygen.
The Getty's summary of its lighting programme describes the shift: guidelines once applied a few rules of thumb to everything, and it was 'inconceivable to openly speculate on acceptable levels of light-induced damage'; managing damage over time is now a familiar idea. The CIE's technical report 157:2004 set out a procedure for establishing and monitoring a display on that basis. The standard controls remain those the Getty lists — lower intensity, shorter exposure, removal of wavelengths the eye does not use, and making sure the lighting does not magnify other risks such as heat. What has changed is that each can be tuned to the object: a spectrum chosen for its colourants, a dose set from a microfading result, and an atmosphere chosen with the known exceptions in mind. The lux budget pages in this section give the arithmetic those refinements adjust.
| Outcome in near-anoxia | Samples | Share of 125 |
|---|---|---|
| Changed less than in air (all) | 113 | 90% |
| — of which four or more times less | 53 | 42% |
| — of which two to four times less | 44 | 35% |
| — of which less than twice | 16 | 13% |
| Changed more than in air | 6 | 5% |
| Neither group (not described in the abstract) | 6 | 5% |
| Colourant and change | Verdict | Why | What is documented as done |
|---|---|---|---|
| Verdigris (copper acetate) and copper greens in oil–resin media: Transparent green turns brown. | Irreversible | No reversal is reported. | None reported; laboratory studies of mechanism. |
| Prussian blue and Prussian green watercolour: Blue fades; partial return of colour in air in the dark. | Partly reversible | Recovery on re-exposure to air is incomplete and depends on the sample. | Tate's researchers propose about 5% oxygen, rather than none, for sealed 'anoxic' frames that may contain Prussian blue. |
| Prussian blue dyed on textile: Blue fades; recovers wholly or partly in air. | Partly reversible | After two weeks back in air some samples had fully recovered and others had not; aged historic dye may recover less. | The fumigation study recommends that anoxic treatment not be used on historic textiles dyed with Prussian blue. |
| Prussian blue image of a cyanotype: Blue image pales in light and regains colour in the dark. | Partly reversible | Short-term bleaching recovers; for the pigment in paint, long exposure produced loss that did not. | None reported in the source read. |
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
For fugitive red and yellow pigments, damage from visible light decreases as wavelength increases; for blue pigments and ISO blue wool standards 1–3 the relationship is less straightforward and blue radiation is not necessarily the most damaging.
Caveat: From the abstract; a limited set of fugitive pigments and dyed textiles.
Source: Wavelength-dependent fading of artists' pigments (Studies in Conservation, 1994)
FactStrong evidence
Microfading performs an accelerated light-fading test on an area about 0.4 mm in diameter while monitoring colour change, can identify materials more light-sensitive than Blue Wool 2, and can heat the test area to as much as 50 °C.
FactStrong evidence
Of 125 samples given about 17.5 Mlux·h, 113 showed less colour change in a near-anoxic environment than in air, and six showed more: three Prussian blue watercolours, Antwerp blue watercolour, verdigris dry pigment and a fluorescent yellow gouache.
Measured: 125 samples of pigments, dyed textiles, dyes, gouaches, watercolours, inks and natural-history specimens.
FactModerate evidence
A white LED of high colour quality (CRI above 90) causes no more light damage than halogen or UV-filtered daylight, whereas LEDs with a large blue spike can cause up to twice the damage of a UV-filtered 3000 K halogen benchmark.
Caveat: Taken from a summary of the CCI bulletin, not from the full text; figures should be checked against the bulletin before reuse.
Source: LED Lighting in Museums and Art Galleries — Technical Bulletin 36
StandardModerate evidence
CIE 157:2004 is a technical report that reviews the principles of radiation damage to museum objects and recommends a practical procedure for setting up and monitoring display lighting.
Caveat: Described from the publisher's abstract; the report itself was not read and no limit values are quoted.
Source: CIE 157:2004 Control of damage to museum objects by optical radiation
Colourwise analysisModerate evidence
In the Getty anoxia test, 90% of samples benefited, 5% were harmed and the remaining 6 samples fall in neither group; about 53 samples improved fourfold or more.
Based on: Colourwise arithmetic on the published counts: 113 and 6 of 125 samples, the remainder by subtraction, and 47% of the 113 that benefited.
Caveat: The abstract does not describe the remaining samples; they are a remainder, not a reported finding.
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.