<p>Colour primitivism is the thesis that the colours are simple qualitative properties of items in the external world—just as they appear to be. This paper considers colour primitivism in relation to a version of Frank Jackson’s knowledge argument. On the one hand, this argument seems to support primitivism, by delivering the conclusion that the colours cannot be reduced to underlying physical features of a sort that Mary might have known about even in the room. On the other hand, however, the argument threatens to show that primitivism cannot be combined with even a fairly minimal (non-reductive) form of physicalism, on which the colours are conceived as supervenient upon or grounded in underlying physical properties. To resolve the tension, the paper recommends a novel primitivist take on the knowledge argument, which turns crucially on the classical empiricist thesis to the effect that certain facts about the colours can only be learned on the basis of sensory experience.</p>