a still lower illumination, the violet vanished, the red became red-brown, and the green was visible as a pale-green tint; then the red-brown disappeared, the green still remaining, though very feeble. With still less light, even this suggestion of colour vanished, and the light appeared simply grey.

The tendency in these experiments is evidently just the reverse of what was observed where the illumination was very bright. In that case the coloured light as it increased in brightness gradually set all three sets of nerves into action, and the result was white or yellowish-white; but here the action of the coloured light as it grows feebler is more and more confined to a single set of nerves. From this it results that those colour-sensations which are due to the joint action of two sets of nerves speedily diminish when the colour is darkened, and are replaced by the primary sensations, red, green, or violet. The sensation of orange is produced by those light-waves in the spectrum which have a length such as to enable them to stimulate the red nerves strongly and the green nerves to a lesser degree; hence, when orange-coloured light is made very weak, it fails to act on the green nerves while still feebly stimulating the red, and consequently the sensation of orange passes over into red. For similar reasons the sensations of yellow and greenish-yellow pass into green, as do also those of greenish-blue and cyan-blue; in the same way the sensations of blue, ultramarine-blue, and violet-blue pass into violet. It is quite evident that these changes furnish another argument in favour of Young's theory of colour, and also tend to approve the selection of red, green, and violet as the fundamental colour-sensations.

In the experiment of Von Bezold just mentioned, after the spectrum had been darkened to a certain degree only three colours remained-red, green, and violet; this dark red, however, as far as sensation goes, is somewhat changed in character, and, according to the unpublished experiments

of Charles Pierce, has become somewhat purplish; the same is true of the green, which is more bluish; the violet alone is unchanged. Now, just these same effects can be produced by mixing small quantities of violet with red or green; hence the final effect of darkening on all the colours of the spectrum is virtually to mix them with increasing quantities of violet light. The cause of these peculiar changes, according to the theory of Young and Helmholtz, resides in the fact that the violet nerves act more powerfully, relatively to the red and green nerves, when the light is feeble. For example, if we present to the eye pure green light, it will stimulate the green nerves strongly, the red and violet to a much less degree: we thus obtain a certain sensation, and call it green. If now we greatly diminish the intensity of the green light, it will of course affect the green nerves to a minor degree; but, besides this, it has now less action on the red than on the violet nerves, so that virtually we have a mixture of green and violet, which will cause the green to appear bluish-green. The same explanation holds good for the red, dull red light producing less effect on the green than on the violet nerves.

The change which colour undergoes when darkened is interesting from a practical point of view; and accordingly the author made a series of experiments on this subject, using for that purpose coloured disks and the method of rotation. In these experiments we do not deal with the pure colours of the spectrum, but with surfaces painted with brilliant pigments, which correspond more nearly to the cases that present themselves to the artist and decorator. A black disk was in each case combined with a coloured disk, as indicated in Fig. 75; a smaller disk of the same colour being either attached to the axis for comparison or held from time to time near the rotating disk. It was ascertained by previous experiments that the amount of white light reflected by the black disk was small; if we set the amount of light reflected by white cardboard as 100,

then the black disk which was employed on this occasion reflected two per cent. of white light, or The colour of the painted disks was in every case as intense, saturated,

FIG. 75.-Chrome-yellow and Black Disks FIG. 76.-The disk of Fig. 75 when in in combination. rotation becomes coloured olive-green.

and brilliant as possible. The results obtained by rotation —that is, by reducing the luminosity of the colours by mixing black with them-are briefly indicated below:

Ultramarine-blue (artificial). More violet, less blue.

Dark violet.

Violet....

Purple..

Carmine.

More violet, less red.

Not much changed.

It will be noticed that these results correspond more or less closely with those of Von Bezold, before given.

Some of the changes in the experiments just mentioned were so great as to be quite astonishing, and might well tempt the beholder to believe that the black disk exercised some peculiar influence on the result; this, however, was not the case, as the same results can be obtained without the black disk by simply reducing the illumination of the coloured disks by holding before the eye two Nicol's prisms, and turning them so as gradually to cut off the coloured light. On the other hand, if the tints that are obtained by using the black disk give the true appearances of surfaces painted with pure pigments, but viewed under a feeble illumination, then accurate copies of them ought, when powerfully illuminated, to appear once more brightly coloured, and of the original tints. This was found to be the case, for example, with gamboge, where the change in colour by darkening was from a slightly orange-yellow to a fine olivegreen. The olive-green colour was carefully copied with water-colours on a slip of paper, and afterward held in bright sunlight; this caused it to appear yellow, and made its colour resemble that of the gamboge disk placed near it, but in the shade.

The general result of these experiments is, that, if the illumination is feeble, the colours become weaker, and there is on the whole a general tendency toward a darkish blue; just as in the reverse case, where the colours are made very bright, there is a tendency toward a whitish-yellow. This average tint can best be studied by observing moonlight effects here the more luminous colour appears to be a somewhat greenish-blue, the darker shades more like an ultramarine-blue. With regard to this delicate point, the painters of moonlight landscapes are as good an authority as we have, and the best of them are very decided as to the prevalence of various shades of blue, greenish-blue, and violetblue. Similar effects, though smaller in degree, are ob

served on dull, cloudy days, when the prevailing tint is a bluish-grey. Indeed, as Helmholtz remarks, simply viewing a sunlit landscape through a pale-blue glass suggests the idea of a cloudy day; while reversing the process, and viewing a landscape on a dull, cloudy day through a paleyellow glass, gives the impression of sunshine. Corresponding to this, accidental streaks of yellow ochre or sawdust on the shaded pavement often suggest forcibly the idea of stray sunbeams; and other examples of this kind of illusion might be mentioned. If we mix lampblack directly with pigments on the palette, their colour will of course be darkened, but the effects produced are not identical with those obtained by the method of rotation. Paper was painted with a strong wash composed of carmine and lampblack, which imparted to it a dark-reddish, purplish hue. From this a disk was cut and an attempt made to match its tint by mixing, according to the method of rotation, carmine and lampblack. In order to accomplish this, it was

FIG. 77.-Small central Disk composed of Cardboard washed with a Mixture of Carmine and Lampblack. This is nearly matched by disks painted with carmine, black, and white, in the proportions indicated.

found necessary to introduce into this rotation-mixture a quantity of white; the best match being effected when the compound disk was arranged as indicated in Fig. 77. This shows that the saturation or intensity of a coloured pigment is greatly reduced by mixing lampblack with it on the palette, and is one reason why artists refuse to adopt this