I must now introduce to your notice a different method of experimenting with colour vision. If we throw the whole spectrum on the screen, and ask a person with normal vision to point out the brightest part, he will indicate the yellow, whilst a red-blind will say the green, and so on. This tells us that the various types of colour blind must see their spectrum colours with luminosity differing from that of the normal eye. The difference can be measured by causing both to express their sense of the brightness of the different parts of the spectrum in terms of white light, or of one another. Brightness and luminosity are here used synonymously. On the two small screens are a red and a green patch of monochromatic light--a look at the green shows that it is much brighter than the red. Rotating sectors, the apertures of which can be opened or closed at pleasure during rotation, are now placed in the path of the green ray. The apertures are made fairly small, and the green is now evidently dimmer than the red. When they are well open the green is once more brighter. Evidently at some time during the closing of the apertures there is one position in which the red and green must be of the same brightness, since the green passes through the stage of being too light to that of being too dark. By gradually diminishing the range of the “too open” to “too closed” apertures we arrive at the aperture where the two colours appear equally bright. The two patches will cease to wink at the operator, if we may use such an unscientific expression, when equality in brightness is established. This operation of equalising luminosities must be carried out quickly and without concentrated thought, for if an observer stops to _think_, a fancied equality of brightness may exist, which other properly carried out observations will show to be inexact. Now, instead of using two colours, we can throw on a white surface a white patch from the reflected beam, and a patch of the colour coming through the slit alongside and touching it. The white is evidently the brighter, and so the sectors are placed in this beam. The luminosity of (say) a red ray is first measured, and the white is found to require a certain sector aperture to secure a balance in brightness. We then place another spectrum colour in the place of the first, and measure off in degrees the brightness of this colour in terms of white light, and we proceed similarly for the others. Now how are we to prove that the measures for luminosity of the different colours are correct? Let us place three slits in the spectrum, and by altering the aperture of the slits make a mixture of the three rays so as to form white. The intensity of this white we can match with the white of the reflected beam. We can then measure the brightness (luminosity) of the three colours separately, and if our measures are correct there is _primâ facie_ reason to suppose that they will together make up the brightness of the white.