2. _The colour-blind name colours in accordance with their colour-perception, and thus show definitely to which class they belong._ I have not come across a man who has guessed correctly when examined with my test. A man who did guess would know that he was incompetent. As the colour-blind are often not aware of their defect they answer as they see, only guessing when they feel uncertain as to the nature of the colour shown. There is probably more misapprehension on this point than on any other in the practical testing of colour-blindness. 3. _Colours may be changed to the colour-blind, whilst leaving them unaltered to the normal-sighted._ 4. _The phenomena of simultaneous and successive contrast are much more marked for the colour-blind than for the normal-sighted._ Two colours, which have not changed in the slightest degree to the normal-sighted on being contrasted, have apparently altered very considerably to the colour-blind. As an example of this, let us take a pure deep yellow, a bright red, and a bright green. To the normal-sighted the yellow will be altered very little by comparison with the red or the green, but a trichromic would say that the colour was green when contrasted with the red, red when contrasted with the green. This principle of exaggerated contrast must be borne in mind when examining a candidate. Thus if a trichromic be doubtful about a yellow, but seems inclined to call it green, he should be given a pure green to compare with it. In the same way, in showing the coloured lights, the same colour produced in a different way should often be shown. Thus an orange-red may be shown immediately after a pure red. This will not alter the colour to the normal-sighted, but greatly facilitate the examination of the colour-blind. 5. _Many colour-blind match correctly, but name the principal colours wrongly._ Therefore the test must be a naming test, the examinee being rejected if he confuse the colours which it is essential he should distinguish between in his occupation. 6. _Many colour-blind recognise colours easily when they are close to them, or the surface is large, but fail to distinguish between them when they are at a distance or the image on the retina is small._ The test must be constructed in conformity with these facts. 7. _The colour-blind are more dependent upon luminosity than the normal-sighted, and are liable to mistake a change in luminosity for a change of colour._ The test should have a means of rapidly changing the luminosity of a colour. 8. _The colour-blind find special difficulty with faint and dim colours._ The test should have colours of this kind. 9. _The colour-blind who have shortening of the red end of the spectrum cannot see reds reflecting or transmitting only rays corresponding to the shortened portion._ It is essential that reds of this kind should form part of the test.