the phenomenon of two colours That matchunder one set of conditions, but fail to match under a different set.
The two lights must be physically different.With coloured lights,a white light could be a mixture of approximatelyequal amounts of all wavelengthsand could be matched by amixture of two complementary wavelengths.
With surface colours,a sample dyed with a given set of dyes could bematched,under certain conditions,with a sample dyed with different dyes. In each case the match would be physicallydifferent, and the match would fail to hold when theconditions changed.For the white light a match for one observer wouldprobably not be accepted as a match by a secondobserver.
For the dyed samples, the match wouldprobably not hold if the light source were to bechanged; the match might be satisfactory in daylight,for example,but very poor under fluorescent light.Again, a good match forone observer might be perceivedas unsatisfactory by a second.
Suppose we have two objects whosetristimulus values (for a specified standard illuminant andstandard observer) are identical.The two objects illuminated by the specified sourceand viewed by the specified standard observerwould look identical, yettheir reflectance curves might differsignificantly.
For example,one object could have constant R values of, say, 30% at all wavelengths(and would look grey under all normal conditions),while the other object could be a piece of nylon dyed to match thefirst using a mixture of yellow, red and blue dyes.We can generally produce any particular grey colour usingsuch a mixture of dyes,but it is likely that the R values will vary with wavelength. (The lowest R values would probably be found at the wavelengthsof maximum absorption for each dye, say 450, 520 and 620 nmrespectively.)
We can match most coloured objects(under specified conditions) usinga mixture of three suitable dyes,but the reflectance curve of the dyed fabric will not be the same asthat of the object to be matched.In these cases the dyeing will normally not match the target colour if theconditions are changed, and the two samples are said to be metameric,or to form a metameric pair.In terms of colour measurements, the two sampleshave the same tristimulus valuesfor a specific combination of illuminant and observer, but the reflectance curves are different.Usually such pairs of reflectance curves cross at least three times.
In calculating the tristimulus valuesthe effects of higher R valuesfor the dyeing at some wavelengths must be balancedby the effects of lower R values at other wavelengths. If the illuminant used in the calculation is changed,not surprisingly The tristimulus values will usually changeas well.This is also true if the observer is changed.