WHICH REFLECT LIGHT. 305 



point is hereby gained. Each ray which encounters such a 

 mass will be partially reflected at the first and second surfaces 

 and partly permitted to pass through, and the latter will at its 

 entrance and exit be refracted if it does not by accident strike 

 the surfaces perpendicularly. In general we must assume that 

 the rays which have passed through such a mass do not retain 

 their original direction, but are more or less deflected from it, 

 and thus a dispersion of the total light passing through the 

 atmosphere must take place, which could not escape observation. 

 Now abstracting from certain particular phaenomena, such as 

 that of astronomical refraction, the cause of which is sufficiently 

 explained, it is known that the light is propagated in straight 

 lines, as by this alone are we enabled to see objects in their 

 correct position. We have now to examine whether and under 

 what circumstances the refraction which necessarily accom- 

 panies the reflexion is to be brought into harmony with this 

 known fact. 



Let us here distinguish the assumption before hinted at more 

 clearly, so that we may prove that by it the above harmony is com- 

 pletely established, and hence that it is admissible. The further 

 investigation must then be conducted with greater generality ; 

 and if it happen that all other assumptions are impossible, that 

 is, not to be reconciled with the propagation of light in straight 

 lines, the necessity of the former will follow as a matter of 

 course. 



Let us suppose the shape of the masses such that for every 

 ray which strikes them, the forward surface at the point of inci- 

 dence and the hinder surface at the point of exit are parallel, tlien 

 each ray refracted by the forward surface will be refracted back 

 again on leaving the mass to its original direction. While then 

 the reflexions sum themselves together, the refractions mutually 

 annul each other, and the transmitted light pursues its primitive 

 course unaltered. That this condition shall be fulfilled, we 

 must assume that all the reflecting bodies possess the form of 

 the thinnest possible plates, whose limiting surfaces are parallel. 

 At first sight such an assumption may appear somewhat forced, 

 but by a closer consideration of the circumstances, \ve find that 

 such bodies present themselves quite naturally in the atmo- 

 sphere. By the precipitation of vapour minute bladders of 



