longer equally successful when for the salt we substitute alum or any other diaphanous substance. But as recourse might be had to supposed differences between the conducting, the absorptive, or the emissive powers of these bodies, it seems advisable first to prove the refraction of the nonluminous rays without using lenses.
With this view I place, at a certain distance from the thermoelectric pile and out of the direction of its axis, a plate of copper heated to 390° by an alcoholic lamp, or, what is still better, a vessel filled with water in a state of ebullition. The pile being lodged at the bottom of a metallic tube blackened inside, the rays of nonluminous heat emitted from the vessel in a direction oblique to the axis cannot reach the thermoscopic body, and the index of the galvanometer remains perfectly at rest. Matters being now in this state, I take a prism of rock salt and fix it at the mouth of the tube with its axis placed vertically and its refractive angle turned towards the angle formed by a line drawn from the source to the extremity of the tube. (See Plate I. fig. 2.) A considerable deviation is immediately perceived in the galvanometer. The rays of heat are therefore conveyed into the tube by the action of the prism.
To show that the effect is really due to the refraction and not to the heat of the salt it will be sufficient to turn the angle of refraction in a contrary direction; for as soon as this is done the needle falls again to zero, notwithstanding the presence of the prism. The experiment is no less successful with the heat of the lamp, or that of the incandescent platina. Calorific rays of every kind are therefore, like luminous rays, susceptible of refraction.
But on the principle of analogy, as each species of light, so will each species of heat possess a different refrangibility. Hence it is evident that if the prism be left in its position and the radiant source changed it would become necessary at the same time to change the angle formed by the axis of the pile with the direction of the rays, in order to obtain the desired effect on the galvanometer. If however we attempt to verify this conjecture we obtain no decisive result. This is easily conceived when we reflect that the aperture of the tube has a certain diameter and that it is placed quite close to the prism, so that the rays refracted at angles differing but very little from each other can alvays reach the pile though no change should be made in the inclination of the axis of the tube.
But there is another process by means of which, if we cannot exactly measure the refrangibility of each, species of calorific rays, we prove at least that the angle of refraction varies with the measure of the radiating source. I took a graduated circle (Plate I. fig. 3.) 22 inches in diameter carrying a ruler as a moveable radius. At the extremity of this ruler I placed a thermoelectric pile composed of fifteen pairs disposed in one line perpendicular to the plane of the circle.
