Three colour systems, defined by the International Commission on Illumination (CIE), have been used to parametrise spectra from microspectrophotometry in the visible range for ten replicates of each of forty inks. The parametrised spectra were used to calculate a likelihood ratio (LR) for pairwise comparisons under the propositions implying that any ink from some suspect document came from the same pen, as that from a control document, versus, the converse proposition which implies the ink from the suspect document came from some other pen. Both univariate and bivariate likelihood ratios for each colour system were calculated. Empirical cross-entropy was selected as an appropriate measure of performance for each system. The bivariate combinations of the CIE-xyz colour system achieve the best results as well as a bivariate combination of a and b variables within the CIE-Lab colour system.