Bornstein, Kessen, and Weiskopf (1976) reported that pre-linguistic infants perceive colour categorically for primary boundaries: Following habituation, dishabituation only occurred if the test stimulus was from a different adult category to the original. Here, we replicated this important study and extended it to include secondary boundaries, with a crucial modification: The separations between habituated and novel stimuli were equated in a perceptually uniform metric (Munsell), rather than in wavelength. Experiment 1 found Categorical Perception and no within-category novelty preference for primary boundary blue-green and secondary boundary blue-purple. Experiment 2 replicated the categorical effect for blue-purple and found no within-category novelty preference with increased stimulus separation. Experiment 3 showed category effects for a lightness/saturation boundary, pink-red. Novelty preference requires a categorical difference between the habituated and novel stimulus. The implications for the origin of linguistic colour categories are discussed.