as though simultaneously a small amount of violet light had been added to the mixture. This fact naturally suggests an experiment like the following: Suppose we combine a purple and a green disk as indicated in Fig. 84, employing equal parts, and thus obtain a pure grey. Let

us now replace 10 parts of purple by white, also 10 parts of the green by white (Fig. 85): will we then still obtain a pure grey, or will the grey be tinged with violet? Several experiments of this kind have been accurately made by the author, but in every case the result was the production of a grey identical with that given by mixing by rotation black and white. The explanation would seem to be that the green and purple instantly combine to produce the sensation of grey, and then of course adding white to this grey can only make it paler, but can not at all alter its tint. It would seem from this that, when a colour is altered in the manner above described by admixture with white, time comes in as a necessary element in the process; the mixture of white and coloured light must be allowed to act on the undisturbed during an interval of time which is not too short, otherwise these peculiar effects will not be produced.

eye

One might suppose that the same result would be produced by spreading thin washes of coloured pigments on

white paper that is obtained by mixing white with coloured pigments by the method of revolving disks. The tint in these two cases, however, is usually somewhat different. A pale wash of carmine, for example, was allowed to dry on white paper, and an effort was made to imitate it by combining a deep-coloured carmine disk with one of white cardboard by the method of rotation. It was soon ascertained that the hue of the water-colour wash was considerably more saturated or intense than a tint of equal luminosity produced by the rotating disks; it was also found to be more of a purplish hue. When the luminosities in the two cases were made equal, the water-colour wash showed far more colour than did the simple mixture of the red and white light. Treated in the same way, a thin wash of vermilion was more orange in hue than a mixture of vermilioncoloured light with white light; a thin wash of gamboge looked yellow, while the mixture by rotation had more of an orange-yellow appearance. The reason of these changes is quite evident, and lies in the well-known fact that thin layers of coloured substances have in general a different absorptive action on white light from thicker layers of the same substances. A thin layer of vermilion allows, for example, more of the orange rays to pass; hence in very thin layers this pigment is sometimes used by artists to represent very pale tints of orange, or even of orange-yellow. The other fact above mentioned, viz., that thin layers are often relatively more saturated than those that are thick, is to be explained in a different way. It was shown in Chapter X. that, when a pigment is mixed with one of a different colour, not only is the hue changed, but an effect is produced as though at the same time some black had been added to the mixture. It now appears that, even when a pigment is made darker by mixing with it a larger quantity of itself, a similar change is to some extent produced, the darker wash of the pure pigment acting as though some black were mingled with it. In the experiment with the pale wash of

carmine it was actually found necessary to combine black by rotation with the water-colour wash, so as to reduce it, before it could be matched by a disk composed of white and a deep tint of carmine.

The fact now under consideration can perhaps be rendered more intelligible by a different statement. Carmine, as we know, absorbs powerfully nearly all the coloured rays of light except the red; these latter it reflects in considerable quantity, and to this circumstance its red colour is due. But the experiments just mentioned indicate that it absorbs also to a considerable extent even the red rays, so that a deep wash of carmine sends to the eye less red light than we should expect. The author found that yellow glass presented a parallel case. Yellow glass transmits the orangeyellow, yellow, and greenish-yellow rays abundantly, and to this power mainly its yellow colour is due. But it does not transmit even these rays at all as perfectly as ordinary window-glass. In one experiment it was found that a plate of yellow glass absorbed about 25 per cent. even of these rays. Most coloured substances, pigments, and glasses probably act in a similar way.

CHAPTER XIII.

ON THE DURATION OF THE IMPRESSION ON THE RETINA.

AMONG different forms of fireworks none excite more admiration than revolving wheels of fire with their brilliant colours, ruby, emerald, or sapphire, and their wonderfully blended surfaces, which so often suggest fanciful resemblances to roses, carnations, and other flowers. It is quite possible to arrange matters so as to obtain an instantaneous view of one of these fiery objects, without at all interfering with its rapid movement; and when this is done, it is seen that much of its beauty depends upon an illusion: the broad, variegated, shaded surface vanishes, and we have before us simply a few jets of coloured fire, in no wise particularly remarkable. The appearance of these brilliant objects depends, then, upon an illusion, and this has for its foundation the fact that the sensation of sight is always prolonged after the light producing it has ceased to act on

the eye.

The most familiar illustration of this fact we find in an old experiment, which no doubt was the parent of our revolving fireworks: If a lighted coal on the end of a stick is caused to revolve rapidly, it describes a ring of fire which is plainly seen at night to be quite unbroken. The light from the moving coal falls upon the retina of the eye, and an image of it is produced, let us say at the point 1, Fig. 86; an instant afterward, owing to its having moved into a new position, the image will be found at 2 and then at 3,

and so on all the way around the circle. Now, if the sensation due to the first image lasts while the circle is being traversed, then it will be renewed before it has a chance to fade out, and consequently will be present continuously; the same will be true of all other points on the circle, which consequently will produce on the beholder the appearance of an unbroken ring of light. In order to produce this condition of things, it is necessary of course that the coal of fire should move with a certain velocity; according to

the observations of D'Arcy, it is essential that it should traverse its circular path completely in thirteen hundredths of a second.

It is very easy to experiment on matters like these with the aid of revolving disks. If we take a circular disk which is painted black, and has on it a white spot like that represented in Fig. 87, and set it in rotation, as soon as the motion is quick enough we shall see a ring of white, just as in the previous case we obtained a ring of fire. Fig. 88 shows the appearance of the disk when in rapid rotation.