Often, an ink must be formulated to match a CIELAB value without any required spectral match. In this case, the ink formulator (whether in the ink company, or in the printer's ink kitchen) has a number of possible combinations of pigments from which to choose. Without a spectrum to match against, it is not possible to select one formulation that minimizes the metamerism. Measuring a metameric index requires a target spectrum.
If two objects match under one illuminant, then the metameric index is the color difference between them under a second illuminant. It is a measure of how one object changes color with respect to another object. Instead of print buyers specifying an upper limit for the metameric index under specified illuminants, they may make the deceptively simple request that the color of the ink formulation change as little as possible under different illuminants.
Naively, one might set the rule that the best formulation will be the one where the color difference between the ink under D50 and under D65 will be minimized. One of the conclusions of this paper is that this would be ineffective. If an object changes CIELAB values under a second illuminant, the majority of the false color inconstancy is likely attributable to a flaw in CIELAB.
A color space called ConeLab was recently introduced [Seymour, 2020] which corrects this flaw. ConeLab mimics CIELAB, but it is based on the spectral response of the cones rather than the xyz tristimulus functions. This paper demonstrates that ConeLab shows considerably less change in color values with an illuminant change. This suggests a different way of answering the root question: Which ink formulation has the smallest change in apparent color? Which formulation is least likely to cause metameric issues?