The colour that was not measured

Two prints of the same photograph, made from the same file by two laboratories, differ visibly. Both are correct.

Colour as a physical quantity is a spectrum, and colour as something seen is a response of three types of cell in the eye to that spectrum. Reproducing a colour therefore does not require reproducing the spectrum; it requires producing any spectrum that triggers the same response, which is why a screen with three narrow primaries can show a colour that in nature arrives as a broad band. The practical consequence is that two different ink mixtures can match under one light and separate under another — a phenomenon called metamerism, familiar to anyone who has bought paint that looked grey in the shop and brown at home.

This makes 'the right colour' an incomplete specification. A printing profile defines what a file's numbers mean for one printer, one ink set, one paper and one viewing light. Change the paper and the numbers mean something else. The two laboratories were using different papers, and each had produced the closest available match on its own stock.

Professional practice handles this by specifying the conditions rather than the appearance. A proof is approved under a stated illumination; a brand colour is defined by a measurement in a standard colour space rather than by a sample that fades; and a photographer who cares supplies the profile along with the file.

The idea generalises beyond printing. A measurement without its conditions is not a measurement, and an appearance is a relationship between an object, a light and an observer. Asking which print is the true one is asking a question that has no object, which is why the answer is a specification and not a colour.