Having occasion to examine a large number of persons who have abnormal colour vision, the author has found that the so-called anomalous trichromates are clearly divisible into two groups, which groups can be sharply differentiated by quite simple experimental means. By anomalous trichromate is meant a person who, while not a dichromate,
i. e
., a person who can match all colours of the spectrum by a mixture of two suitably chosen monochromatic lights, has not got normal colour vision. Such cases were first investigated by Lord Rayleigh, and have since been investigated by many observers. It has already been noticed by several investigators that there are two classes of anomalous trichromates. These are distinguished by the fact that when a monochromatic orange (D-light) is matched by a mixture of green and red light one class, the protanopes, take more red in the mixture than the normal, while the other class, the deuteranopes, take more green than the normal. The following paper is intended to prove that both of these classes require to be subdivided, or rather that anomalous trichromates ought preferably to be divided into two other classes, and then each of these classes subdivided into protanopes and deuteranopes. As has been pointed out by Sir W. Abney, if white light is transmitted through a thickness of about 0·8 cm. of a saturated solution of chromate of potash the colour of the resulting light can be matched by a single spectral colour, and, further, in the case of persons having abnormal colour vision the wave-length of the spectral colour differs from the one which matches the chromate light for the normal. As a result of applying this test to a large number of persons who have not got normal colour vision, the author noticed that, while for some observers the monochromatic colour which matches the chromate light is of quite definite wave-length, for others a band of the spectrum matches the chromate light; further, that one edge of this band always coincides with the position which matches the chromate light to the normal eye. Thus if a slit is traversed through the spectrum and the light which passes is received on a screen alongside the light which has traversed the chromate solution and the position noted when the two lights match, the slit being moved alternately first from the red end of the spectrum and then from the green end, in the case of persons having normal colour vision the settings will lie between 578 and 580
μμ
when the source of the white light is the electric arc. With one class of abnormals, which will be called class A, the setting coming in one direction will agree with the normal setting, but the setting coming in the other direction will differ markedly from the normal setting. This class is divided into two subdivisions, one, A
r
, in which the abnormal setting, and hence the whole of the band of the spectrum which matches the chromate light, lies on the green side of the normal match, and the other, A
g
, where the band lies entirely on the red side of the normal. It must be remembered that the match with the chromate appears to be equally good throughout the band included between the extreme settings, and, further, that one edge of the band always coincides with the normal setting.