HOUSTOUN1 has pointed out that a three-colour theory, such as that of Young and Helmholtz, inevitably leads to mathematical contradictions. It is well known in addition that a theory in which there are only three primary sensitivities, such as red, green and blue, becomes unsatisfactory when the many different forms of defective colour vision are studied in detail. Pieron2 has recently attempted to meet these difficulties, but only by a complex and speculative hypothesis. Chapanis3 has published results of experiments on variations of the saturations of spectrum colours, which are inconsistent with a three-colour theory.