A large number of psychophysical experiments have been conducted in which observers judged the quality of reproduction either in terms of accuracy or pleasantness. However, all that these experiments say is how well a set of approaches that was used for reproducing some originals performs. What would be more important is to understand in what way the various colour reproduction methods fail so as to attempt an improvement in those areas. The aim of this paper is therefore to try to understand what factors contribute to judgements made by observers in experiments where they judge the quality of colour reproduction. Having an understanding of these then provides a new kind of basis for developing better colour image reproduction solutions. To this end the present paper describes the experimental method, data analysis and initial results of a psychophysical experiment where observers were asked to identify what differences they saw between a range of reproductions and their corresponding originals.