There is one class of colours, it must be remembered, which can never be mistaken for simple colours. I refer to the purples--mixtures of red and blue--for there are no spectrum colours which unmixed can possibly match them. All other colours, as no doubt will soon be apparent, can be referred to some one spectrum colour, either in its pure state or else mixed with some variable quantity of white light. We are all familiar with the fact that there are three primary colours, and we are naturally led to consider these in the light of the experiments just made. As good a definition as any other of a primary colour is that it is a colour which cannot be formed by the mixture of any two or more colours. The original investigators in colour phenomena were the artists, and they found that neither red, nor yellow, nor blue could be formed by any mixture of pigments on their palette, but that all other colours could be made by a mixture of two or more of these three. Hence to these three were given the name of primary colours. When, however, the physicist began to work with the simple colours of the spectrum, it was speedily found that, at all events, the yellow was not a primary colour, as it could be formed by a mixture of green and red, whilst a green could not be formed by a mixture of any other two colours. This we can prove with our apparatus. [Illustration: FIG. 7.] Three apertures, all of which can be opened or closed as required (see Fig. 7), are placed in the spectrum, one in the red, one in the green, and one in the violet. The last we shall not require at present, so it is entirely closed; but we vary the width of the other two. We find that with a little red added to a bright green, a yellow green is produced; with more red added we have yellow; with still more red, an orange. The relative brightness of the two colours mixed together can be shown by removing the lens which recombines the spectrum to form the patch of light. Each colour issues through its slit and forms its own patch on a white screen which, for the purpose, we make rather larger than usual. The two patches overlap in the middle (Fig. 8), and the pure colours are seen one on each side of the mixed colours. [Illustration: FIG. 8.] Now, placing one slit in the yellow and another in the blue of the spectrum, we find that whatever width of slit we take, no green is produced, but that, in fact, a yellowish or a bluish white results, and that when the two slits are properly adjusted, a pure white is produced. Evidently since none of the intermediate spectrum colours between the blue and the yellow can be made by their mixture, certainly green cannot. Hence, with pure colours a green and not a yellow is one of the primaries.