of colour on the surfaces of waves is made up largely of these two elements; and in a more subdued way we find them also producing the less marked translucency of foliage or of flesh.

One of the resources just mentioned the painter never employs the light which is more or less regularly reflected from the outermost surface, he endeavours to prevent from reaching the eye of the beholder, except in minute quantity, his reliance being always on the light which is reflected in an irregular and diffused way, and which has for the most part penetrated, first, some little distance into his pigments.

The glass-stainer and glass-painter make use of the principle of the direct transmission of light for the display of their designs. Now, as painted or stained glass transmits enormously more light than pigments reflect in a properly lighted room, it follows that the worker on glass has at his disposal a much more extensive scale of light and shade than the painter in oils or water-colours. Owing to this fact it is possible to produce on glass, paintings which, in range of illumination, almost rival Nature. The intensity and purity of the tints which can thus be produced by direct transmission are far in advance of what can be obtained by the method of reflection, and enable the designer on glass successfully to employ combinations of colour which, robbed of their brightness and intensity by being executed in oils or fresco, would no longer be tolerable.

CHAPTER II.

PRODUCTION OF COLOUR BY DISPERSION.

In the previous chapter we have seen that the sensation of sight is produced by the action of very minute waves on the nervous substance of the retina; that is to say, by the aid of purely mechanical movements of a definite character. When these waves have a length of about 3ʊ of an inch, they produce the sensation which we call red-we see red light; if they are shortened to 10 of an inch, their action on us changes, they call up in us a different sensation

-we say the light is coloured orange; and as the lengths of the waves are continually shortened, the sensation passes into yellow, green, blue, and violet. From this it is evident that colour is something which has no existence outside and apart from ourselves; outside of ourselves there are merely mechanical movements, and we can easily imagine beings so constructed that the waves of light would never produce in them the sensation of colour at all, but that of heat.

The colour-sensations just mentioned are not the only ones capable of being produced by the gradual diminution of the wave-length between the red and orange we find every variety of orange-red and red-orange hue; the orange, again, changes by a vast number of insensible steps into yellow, and so of all the other tints. Types of all colours possible, except the purples, could be produced by this method. The colours generated in this way would not only pass by the gentlest gradations into each other, forming a long series of blending hues, but they would also be

perfectly pure, and, if the light was bright, very intense. The advantage of providing, in the beginning of our colour studies, a set of tints possessing these precious qualities, is evident without much argument.

Now, white light consists of a mixture of waves possessing every desirable degree of length, and it is only necessary to select some instrument which is able to sort out for us the different kinds of light, and neatly arrange them side by side in an orderly series. Fortunately for us, we find in the glass prism a simple and inexpensive apparatus which is able to effect the desired analysis. We may, if we are willing to take a little trouble, arrange matters so as to

repeat the famous experiment made by Newton many years ago: viz., admit a small beam of sunlight into a darkened room, and allow it to fall on the prism, as indicated in Fig. 1. We shall notice, by observing the illuminated path of the sunbeam, that the prism bends it considerably out of its course; and, on tracing up this deflected portion, we shall find it no longer white, but changed into a long streak of pure and beautiful colours, which blend into each other by gentle gradations. If this streak of coloured light be received on a white wall, or, better, on a large sheet of white cardboard, the following changes in the colours can

be noticed: It commences at one end with a dark-crimson hue, which gradually brightens as we advance along its length, changing at the same time into scarlet; this runs into orange, the orange becomes more yellowish, and contrives to convert itself into a yellowish-green without passing noticeably into yellow, so that at first sight yellow does not seem to be present. The orange-yellow and greenishyellow spaces are brighter than any of the others, but the rise in luminosity is so gradual that the difference is not striking, unless we compare these two colours with those at a considerable distance from them. As we pass on, the tendency to green becomes more decided, until finally a full green hue is reached. This colour is still pretty bright,

FIG. 2.-Mode of isolating a Single Colour of the Prismatic Spectrum.

and not inferior to the red in intensity; by degrees it changes into a greenish-blue, which will not at first attract the attention; next follows a full blue, not nearly so bright as the green, nor so striking; this blue changes slowly into a violet of but little brightness, which completes the series. If we wish to isolate and examine these tints separately, we can again follow the example of Newton, by making a small, narrow aperture in our cardboard, and use it then as a screen to intercept all except the desired tint, as is indicated in Fig. 2. In this manner we can examine separate

portions of our spectrum more independently, and escape from the overpowering influence of some of the more intense tints. Under these circumstances the greenish-blue becomes quite marked, and the blue is able to assert itself to a greater degree; but the yellow will not be greatly helped, for in fact it is confined to a very narrow region, and it is only by greatly magnifying the spectrum that we can obtain a satisfactory demonstration of its existence.

These experiments, though very beautiful, are quite rough; every two minutes the beam of sunlight strays away from the prism and needs again to be directed toward it; and besides that, the colours blend into each other in such a subtile, puzzling way, that, without a scale or landmark of some kind to separate them, it seems hopeless to undertake any exact experiments. In this difficulty it is to the spectroscope that we must turn for aid; it was certainly not originally contrived for such purposes as these, but nevertheless is just what we need. It is not necessary to stop to describe the instrument, as this has been done by Professor Lommel in another volume of this series; it is enough for us that it is a convenient instrument for sorting out the different kinds of light which fall on it, according to their wave-length, and that it performs this work far more accurately than a prism used according to Newton's plan. Just at this point we can take advantage of a singular discovery made by Fraunhofer, and independently to some extent by Dr. Wollaston, early in the present century. These physicists found that when the coloured band of light just described is produced by a spectroscope, or by apparatus equivalent to one, the band is really not continuous, but is cut up crosswise into a great many small spaces. The dividing lines are called the fixed lines of the solar spectrum. Almost their sole interest for us is in the fact that they serve as admirable landmarks to guide us through the vague tracts of ill-defined colour. Fig. 3 shows the positions of some of the more important fixed lines of