The accompanying diagram (Fig. 6) will give an idea of the apparatus we shall employ. On the usual slit S₁ of a collimator C is thrown, by means of a condensing lens L₁, a beam of light, which emanates from the intensely white-hot carbon positive pole of the electric light. The focus is so adjusted that an image of the crater is formed on the slit. The collimating lens L₂ is filled by this beam, and the rays issue parallel to one another and fall on the prisms P₁ and P₂, which disperse them. The dispersed beam falls on a corrected photographic lens L₃, attached to a camera in the ordinary way. It is of slightly larger diameter than the height of the prisms, and a spectrum is formed on the focusing-screen D, which is slewed at a slight angle with the perpendicular to the axis of the lens L₃. This is necessary, because the focus of the least refrangible or red rays is longer than that of the more refrangible or blue rays. By slewing the focusing-screen as shown, a very good general focus for every ray may be obtained. When the focusing-screen is removed, the rays form a confused patch of parti-coloured light on a white screen F, placed some four feet off the camera. The rays, however, can be collected by a lens L₄, of about two feet focus, placed near the position of the focusing-screen, and slightly askew. This forms an image on the screen of the near surface of the last prism P₂; and if correctly adjusted, the rectangular patch of light should be pure and without any fringes of colour. The card D slides into the grooves which ordinarily take the dark slide. In it will be seen a slit S₂, the utility of which will be explained later on. We shall usually require a second patch of white light, with which to compare the first patch. Now, although the light from the positive pole of the carbons is uniform in quality, it sometimes varies in quantity, as it is difficult to keep its image always in exactly the centre of the slit. If we can take one part of the light coming through the slit to form the spectrum, and another part to form the second patch of white light, then the brightness of the two will vary together. At first sight this might appear difficult to attain; but advantage is taken of the fact that from the first surface of the first prism P₁ a certain amount of light is reflected. Placing a lens L₅, and a mirror G, in the path of this reflected beam, another square patch of light can be thrown on the same screen as that on which the first is thrown, and this second patch may be made of the same size as the first patch, if the lens L₅ be of suitable focus, and it can be superposed over the first patch if required; or, as is useful in some cases, the two patches may be placed side by side, just touching each other.