by comparing the position of the neutral point with that usually found by the two types. We have, however, an unerring criterion by the luminosity method. The red is placed beside the white, and he is asked to say which he considers the darker; he will give an answer of some kind, and probably protest that the two colours are not alike. A soothing answer will disarm his objection, and he will quickly see what you mean. If he be red-blind he will match in brightness a brilliant red and a feeble white; if he be green-blind he will make a match very similar to normal vision. In the case of the red-blind the slit is then moved into the extreme red, when he will say he sees but one patch of light, whilst the green-blind will see it as a person of normal vision would do. If time permits, the whole luminosity curve may be taken and registered. This is not essential, but interesting for reference. Where complete colour blindness exists, it should be possible to cause him to match a green with a red. To do this a second instrument, as described in page 18, may be used, but it is quite sufficient if a piece of red glass, such as is used for railway signals, or of bottle-green glass, be placed in the white beam. There is then a red or green patch alongside the patch of spectrum colour. The red will stimulate the red sensation of the green-blind, but not being spectrum red it contains a certain amount of yellow, which stimulates the green sensation if the observer be red-blind. The green is of such a colour that it will stimulate both the red and the green sensations. In the path of the reflected beam between G and the prisms (Fig. 3) a sheet of plain glass is inserted, which reflects a proportion of white on to the red patch. The sectors are placed in this beam. If the red glass is being used, the slit is moved into the green near E, and the colour blind will say that both are the same colour, but one darker than the other. By opening or closing the slit in the spectrum, he will possibly say that both colours are alike and of the same darkness, but he may say one is paler than the other, in which case the white light must be increased or diminished by means of the sectors till equality of tone is established. This applies to the red-blind and the green-blind. The former will require a very bright red to match a feeble green, whilst with the latter the red will require a fairly light green. When the green glass is used the spectrum colour patch should be red, and the match be made as before. With the violet-blind the neutral point will be in the yellow, and with monochromatic vision matches can be made throughout the spectrum. So far it will be seen that no mention of any colour is required. It may next be advisable to ask him the names of colours. This is best done by placing the white patch of light over the spectrum colour patch, and opening and closing, as may be required, the sectors. If the sectors are closed it is very probable that