Metamerism

Two colours match under one light and separate under another. Nobody made a mistake; the two surfaces return light differently, and the first lamp could not tell. This is the most expensive thing in colour, and it is also the easiest to predict — if you kept the spectrum.

Why two different things can look identical

The eye has three kinds of cone. Whatever arrives at it — a whole spectrum, hundreds of wavelengths — is reduced to three numbers before it reaches the brain. Three numbers cannot describe a spectrum, so an enormous number of different spectra collapse onto the same three.

Two surfaces whose spectra differ but whose three numbers agree are a metameric pair. They match. They genuinely match — there is no sense in which the observer is being fooled.

Change the light, and the three numbers are computed from a different part of each spectrum. The two answers stop agreeing, and the match falls apart.

Where it turns up in production

Everywhere two colours were made by different means and matched by their Lab values.

The library shows you one of these already

Open any sample and the CMYK column carries a ΔE₀₀. That figure is the round trip: the colour converted into process ink through an ICC profile and converted straight back. It says how far the process build lands from the spot colour under the profile’s own illuminant.

Read that number as a metamerism warning, not just as an accuracy score. Where it is small, the build is close and the two will drift together. Where it is large, the build is already a different colour under the reference light — and it will move further under any other. A colour you can see is hard to build in CMYK is a colour to think twice about before putting it on a product that will also be printed in process.

And the observer varies too

Everything above assumes one pair of eyes. Real people differ, and a metameric pair is exactly where that difference shows — two careful people can honestly disagree about whether it matches. See Observer Metamerism.

Metamerism is not colour inconstancy

They are constantly muddled, and the distinction decides who has to fix it.

What to do about it

Specify spectrally. A ChromaSpot code carries the whole curve. An ink kitchen given the spectrum can aim at the shape, not only at three numbers — and a recipe that follows the shape will hold under lights nobody tested.

Name the lights that matter. If the product will live under shop fluorescent, say so before the ink is made, not after the reprint. The Illuminant control lets you watch any colour in the library move between eleven of them.

Check the pair, not each half. Two colours that each measure within tolerance of the target can still be a poor match to each other under a second light. The only way to know is to compute both under both.