and the deduced value of μ would therefore be higher for slow oscillations, the longer waves being thus the more refrangible. The order of refrangibilities would in such a case appear to be somewhat analogous to that in an anomalously dispersive medium like iodine vapour.
With exceedingly quick ethereal vibrations which give rise to light, there is an inversion of the above state of things, i.e., the shorter waves are generally found to be the more refrangible. It would thus appear that there is a neutral vibration region for each substance at which this inversion takes place, and where a transparent medium produces no dispersion.
It would be interesting to be able to determine the indices of refraction corresponding to different wave lengths, chosen as widely apart as possible, and plot a curve of refrangibilities. A curve could thus be obtained for rock salt, which is very transparent to luminous and obscure radiations, and fairly so to electric radiation. Carbon bisulphide, which is very transparent to all but the ultraviolet radiation, would also be a good substance for experiment.
For the construction of a curve of refrangibility for electric rays, having different vibration frequencies, the indices could be determined by the method of total reflection referred to above. The determination of the corresponding wave-lengths, however, offers great difficulties. Hertz used for this purpose the method of interference, the positions of nodes and loops of stationary undulation produced by perpendicular reflection being determined by means of tuned circular resonators.
Sarasin and De la Rive subsequently repeated these experiments with different sized vibrators and resonators. They found that the apparent wave-length depended solely on the size of the resonators. The wave-length found was approximately equal to eight times the diameter of the circular resonator. From these experiments it was supposed that the radiator emitted a continuous spectrum consisting of waves of different lengths, and that the different receivers simply resonated to vibrations with which they happened to be in tune. If this supposition be true the emitted radiation should, by the action of a prism, or better still, a diffraction grating, spread out in the form of a continuous spectrum. If, on the contrary, the radiation is monochromatic, the spectrum should be linear. The experiments to be described below may throw some light on this question.
Professor J. J. Thomson, referring to the above case, is of opinion that the hypothesis of a continuous spectrum is highly improbable. It is more likely that, owing to the oscillation being of a dead-beat character, the resonator is set in vibration by the impact of incident electric waves. Each resonator vibrating at its particular free period,
