The converse problem of getting the spectrum of the electrodes, without that of the surrounding gas, has been solved by Demarcay, Schuster and Hemsalech, who have proved that practically all the "air lines" of a spectrum may be eliminated by placing in the discharge circuit a proper amount of inductance. In this manner the discharge is prolonged and is made to re semble that of the arc.
The electrodes are connected in series with an inductance I, and in parallel with a capacity C. Concerning the temperature of the spark very little is known. It has been generally assumed to be very much higher than that of the electric arc: but there is no direct evidence for this view and there is much indirect evidence against it.
3. Separation of Let us sup pose that we now have before us one of the above-mentioned sources whose spectrum it is desired to examine. The next step will be to separate the rays of various wave-lengths so that we may examine them independently. The reader may here assume that each different substance introduced into the flame or arc gives a different spectrum, peculiar to itself ; for this is the experimental basis of spectrum analysis. This analysis is usually accomplished either by passing the light through a prism or by allow ing it to fall upon a diffraction grating.
The Prism Spectroscope.— If the source of light be small and if it is necessary to make only a hasty visual examination, the most convenient plan is simply to view the source through a prism, placed immediately in front of the eye, as indicated in Fig. 3.
This is an especially useful method in ob serving the discharge in the capillary portion of a vacuum tube. The chief difficulty in the naked prism is that the source, however small, is generally still so large that the successive colored images formed by the prism will over lap each other, leaving the separation incom plete. To avoid this difficulty a small tele scope — called a collimator — is placed between the prism and the source as shown in Fig. 4. In the principal focus of its objective is placed a narrow, straight slit with movable metal jaws. The source is now placed immediately back of this slit, or the image of the source is focused upon the slit by means of a lens, called the image lens.
When the slit is properly placed and illumi nated the collimator will emit from each point of the slit a beam of nearly parallel light whose cross-section is equal to the effective aperture of the objective. If now this emergent beam be examined through a prism and the naked eye as before, a series of colored images of the slit will be seen, each appearing to be at an infinite distance from the observer; and if the slit is narrow these images will be sharp and fine and easily separated. But they will not, in general, appear very bright because the aperture of the human eye is too small to ad mit more than a fraction of the emergent beam. Accordingly these colored images, each at an infinite distance, are viewed through a small astronomical telescope, called the view tele scope. This serves to condense the entire beam which emerges from the prism and reduce it to another parallel beam just large enough to fill the aperture of the average human eye.
When it is desired to photograph a spectrum the view telescope is removed and is replaced by a camera with a photographic objective. Such an instrument is known as a spectrograph and an excellent modern form of it is illus trated in Fig. 5.
When the typical spectroscope, represented in Fig. 4, is provided with a graduated circle, and so arranged that the view telescope and the prism can rotate about the axis of the divided circle, the instrument is known as a spectrom eter, and can be employed for the measure ment of prism-angles and wave-lengths as well as for the examination of spectra. In this type of spectroscope it is nearly always necessary for the sake of good definition to place the prism in such a position that it will produce a minimum deviation in the incident pencil of rays. For it is proved in geometrical optics that a homocentric pencil will remain homo centric after passing through a prism only pro vided the prism is placed in its position of minimum deviation. If the prism be set in any other position the image of a point-source will not be a point, but a line.