The plates used for this are of a very special nature. In the first place, there is the basis of glass, but upon that there is laid what we might term the selective screen. This is a layer of starch grains, of exceeding smallness. The size of them is as little as a half a thousandth of an inch and there are about four millions of them on every square inch of plate. Next, upon the screen of starch grains is a layer of waterproof varnish, while over that is the ordinary sensitive emulsion such as forms the essential part of the usual non-colour plate. Now the starch grains which form the screen are, before they are laid on, stained in three colours. Some are blue, some red, and some a yellowish-green, which experience shows is preferable to pure yellow. The differently coloured grains are well mixed, and when the screen is held to the light and looked through the effect is almost that of clear glass. That is because red rays from the red grains, and green and blue rays from the grains of those colours, all proceed to the eye mingled together. This plate is placed in the camera differently from the usual way, since the glass side is turned towards the lens. The light, therefore, after entering the camera, passes through the glass, then through the screen, and finally falls upon the sensitive film. Suppose, then, that the camera were pointed to a red wall; red light would fall upon the plate and, passing through the red grains, would act upon the sensitive film behind them. The blue and green grains, on the other hand, would stop those rays which fell upon them, and so those parts of the sensitive film which they cover would remain unaffected by light. Then, if that plate were to be developed, a dark, opaque spot would be produced upon the film under each red grain, the film under the other grains remaining transparent. Hence, when held up to the light and looked through, the plate would appear a greenish-blue, for all the red grains would be covered up. In like manner, if the wall were blue instead of red, a greenish-red plate would result, while if it were green, the plate would be a purple, the result of the combination of red and blue. But this, it will be seen, is a topsy-turvy effect, the exact opposite of what we want, so that it is fortunate that by a simple chemical method we can set it right. After a first development in the ordinary way the plate is placed in another bath and exposed to strong daylight, with the result that those parts which were darkened by the first development become clear and the parts which were clear become opaque. Thus, after this twofold development of the photograph of the red wall, we find ourselves in possession of a red plate, in which only the red grains are visible, since all the others are covered up by opaque parts of the sensitive film. The photograph of the blue wall will also, after it has been subjected to the double development, show blue only, and the same with the green.