It is more instructive if we experiment with pure colours, and so we must resort to our colour patch apparatus described in Fig. 6. If a small circular aperture about quarter of an inch in diameter be cut in a card, and placed in front of the prism nearest the camera lens (Fig. 32), the colour patch, instead of being an image of the face of the prism, will be an image of the circular hole, and when the slit is passed through the spectrum we shall have a coloured spot on the screen, on which we can superpose a patch of white light from the reflected beam. There are two ways in which we can reduce the intensity of the spot, by narrowing the slit through which the spectral ray passes or by placing the rotating sectors in front of the coloured beam. This last, perhaps, is the readiest plan, as it only involves the reading of the sector. We can then diminish the intensity of the coloured spot to such a degree that by its dilution with white light it will entirely disappear. It will be found that red disappears at a different aperture of sector to that required for the green, and the green to that for the blue. From our previous experiments in chapter VII. we know the luminosity of the spectrum to the eye, and it will be of interest to see what relation the luminosity at which the spots of different colour disappear, when they are so diluted with white light, bear to the total luminosity of these rays. In a set of measurements made it was found that the reduced angular apertures required for the colours indicated by the following were: B required 300°* of aperture. C " 56° " D " 14° " E " 22° " F " 150° " G " 2100°* " The large numbers marked with an asterisk were obtained by placing the rotating sectors in front of the white reflected beam. The light of D had to be reduced to 14° before it was extinguished; therefore to extinguish the original light of this colour in the spectrum would require 180/14, or 12·9 times the intensity of the white light of the reflected beam. With the E light it would take 180/22, or 8·2 times the white light to extinguish it, and so on. If we tabulate the results in this manner, and take the white light necessary to extinguish the D light empirically as 98·5, which is its percentage luminosity in the spectrum of the electric light, we can then compare the extinguishing factor with the luminosity in each case.