These experiments, taken as a whole, show that, in calculating for the effects produced by illuminating coloured surfaces by coloured light, we must be guided mainly by the laws which govern mixtures of coloured lights, rather than by those which can be deduced from experience with pigments; they are certainly useful in teaching us, when studying from nature, fearlessly to follow even the most evanescent indications of the eye, utterly regardless of the fact that they disobey laws which we have learned from the palette.

We pass on now to consider the changes in tint which take place when coloured surfaces are illuminated by lamplight or gas-light. If we undertake to make experiments in this direction simply by viewing coloured surfaces by lamplight in a room illuminated with it, correct results can not be obtained; for by this very method we have practically rendered ourselves colour-blind to a certain extent, and have become incapable of judging correctly of quite a series of hues. Gas-light is deficient in the violet, blue, and bluish-green rays; hence its resultant tint is not white, but orange-yellow. If we are immersed in this light, it will appear to us white, and our judgment of all colours will be more or less disturbed: yellow surfaces will appear white or whitish; blue surfaces, more greyish-blue, or, if pale,

even pure grey. The actual changes effected by artificial illumination may readily be studied by the following simple method, contrived by the author: A camera-obscura is placed in a room illuminated by ordinary daylight; in front of it, at a short distance, is placed a gas-flame or lampflame, in such a way that the lens of the camera is capable of forming an image of it of about half the natural size. (See Fig. 63.) This image is now allowed to fall on coloured stuffs or on coloured paper placed behind the lens of the camera at S; it can be viewed, as indicated, through

Paper.

the top of the camera, and the resultant tint noted. In the experiments made by the author a gas-flame was employed, along with a set of painted disks, representing the principal colours. The disks, fourteen in number, were the same as described in the following chapter, and constituted together seven pairs, the colours of which were complementary, two and two. Below are the results :

1. A carmine disk when illuminated by the gas-flame assumed an intense red hue, even more brilliant than by daylight; the complementary disk, painted blue-green, appeared of a yellowish-green, not saturated, but rather pale.

2. Vermilion appeared of an intense fiery red; its complement, green-blue, lost in strength, and became yellowishgreen and rather pale.

3. Orange appeared brilliant; cyan-blue, the complement, became greenish-yellow and lost in saturation.

4. Yellow became brilliant, showing a tendency toward orange; its complement, blue, appeared white, or rather pure grey. On the same occasion disks painted with chrome-yellow were examined: two of them were rendered somewhat orange-yellow; the third was brought to almost a full orange hue by the gas-light. A disk painted with gamboge-yellow acquired something of an orange tint under the gas-light.

5. Greenish-yellow was brought to a pure yellow; its complement, artificial ultramarine-blue, appeared violet.

6. Greenish-yellow became pure yellow; its complement, violet, was converted into a strong red-purple.

7. Full green passed into a bright, strong yellowishgreen; its complement, purple, assumed an intense purplish-red hue, displaying less blue than by daylight.

These are the actual changes produced by the artificial illumination as they appeared to an eye placed in ordinary daylight, and consequently able correctly to note the several tints. When the disks were examined at night by gaslight, in many cases a different result was reached. The carmine and vermilion disks still appeared very brilliant, the tint in the case of the former being a pure red, while the vermilion showed a tendency to red-orange. The orange disk seemed to be changed in tint to a redder orange hue; the yellow, on the contrary, appeared paler. The greenishyellow disks did not show much change. The full green was intense, appearing perhaps more bluish than by daylight. Blue-green was liable to be confused with blue, cyan-blue and blue with green; artificial ultramarine-blue appeared more purplish than by daylight; violet became purple, and purple a very red purple. Some other disks

were also examined on this occasion: gamboge and chromeyellow showed a loss in saturation, looking whitish; indigo appeared dull greenish-grey; Prussian-blue was confused with blue-green; genuine ultramarine-blue still was always blue with a slight tendency to purple; cobalt-blue exhibited this same tendency, which reached a maximum in French-blue. All the blues appeared much duller and

greyer than by daylight.

By comparing these two sets of experiments, it will be seen how greatly the judgment of colour was influenced by the circumstance that the prevailing illumination was yellow, and that hence a certain shade of yellow stood for white, and gave a false standard to which all the colours were referred. This was particularly noticeable in the case of the yellow disks; in point of fact, as the first set of observations showed, they reflected to the eye much yellow light, and, as far as the mere physical action went, ought to have produced the sensation of a strong, brilliant yellow hue; but, as all surfaces which professed to be white were really (owing to the gas-light) yellow, this competition caused the yellow disks to appear pale. Another case illustrates this disturbed judgment even better. In the first set of experiments it was found that the blue disk when illuminated by gas-light really assumed a pure grey hue without any trace of blue; but at night, although it must have sent to the eye this same pure grey light, it always appeared either blue, greenish-blue, or bluish-green; in other words, the blue disk, when held near the gas-flame, sent to the eye white light, which appeared blue, by contrast with the prevailing yellow illumination. It is hardly necessary to add that these causes affect our judgment of paintings and decorations at night to a very considerable degree, the blues being rendered less conspicuous, the blue-greys being mostly abolished, and the yellows losing in apparent intensity. Genuine ultramarine-blue is less affected than the other blues, cobalt and artificial ultramarine-blue becoming pur

plish, and Prussian-blue quite greenish. It hence follows that paintings in which the blue tones are rather overdone appear often better by gas-light; but this is hardly the case when the green hues are of somewhat too great strength, the evil seeming often to be exaggerated by artificial illumination, which must of course be due to an act of the judgment, as the greens really assume a more yellowish appearance by gas-light or lamp-light, as was proved by the first set of experiments. From this it follows that, if the chromatic composition of a picture is quite right for daylight, it will be more or less wrong when viewed by gas-light; hence it would be desirable to illuminate picture galleries at night with some kind of artificial white light, a problem which the future will no doubt solve.

All the appearances which have thus far been considered could be satisfactorily observed and studied by a person possessed of only a single eye. Let us now turn our attention for a moment to some very remarkable phenomena which occur when different colours are presented to the right and left eye. This is a case which happens occasionally, particularly when we look at the reflection from polished surfaces or from water. In order to simplify matters, let us take a case where, for instance, yellow light is presented to the right and blue light to the left eye. It is very easy to make an experiment of this kind with the aid of the stereoscope. Selecting one of the common paper slides, we colour it as indicated in Fig. 64, and then view it with the stereoscope. We have already seen that blue and yellow light when presented to the same eye undergo mixture on the retina and produce the sensation we call white. This would lead us very naturally to suppose that, if blue light were presented to the right eye and yellow to the left, the two sensations would be united in the brain and would call up that of white. The effect is, however, of a much more complicated character. Viewed in the