436 Proceedings of the Royal Society of Edinburgh. 



gradually i*eappears in a plane perpendicular to the former, at greater 

 angles than this. (4.) That the polarization is more intense in the 

 neighbourhood of the horizon than of the zenith. (5.) M. Babinet 

 has recently remarked that, under certain circumstances, there is a 

 second neutral point in the neighbourhood of the sun. 



Professor Forbes has verified these facts in nearly every particular, 

 by the aid of a modification of Savart's polariscope, constructed of 

 two plates of quartz, peculiarly cut and combined, together with Mr 

 Nicol's single-image calc-spar prism, which the author has substi- 

 tuted with great advantage for the tourmaline commonly used in 

 France. 



With this instrument he finds, (1.) That a uniformly cloudy sky 

 exhibits distinct traces of polarization. (2.) That rain-clouds gene- 

 rally polarize light ; but not, so far as he has observed, those charged 

 with snow. (3.) That the common rainbow entirely vanishes in one 

 position of Nicol's prism (the fact of its polarization was discovered 

 by Biot). (4.) That the polarization of moonlight reflected by the 

 sky is very sensible, and likewise the diffuse light, or burr, which 

 surrounds the moon in cloudy weather. (5.) That the light reflected 

 from dry clear air, between the observer and objects a mile distant, 

 is sensibly polarized. 



With respect to the planes of polarization of skylight, he considers 

 that they may be represented by a fiction of this kind : That there is 

 a certain amount of polarization due to the regular reflection of sun- 

 light from the sky in meridional planes passing through the sun and 

 the obsei'ver ; the polarization being most intense towards azimuth 

 90°, and vanishing at 0° and 180°. Combined with this polarization 

 is another, distinct from it, and represented by a more intense effect 

 due to reflection parallel to the ■plane of the horizon in all azimuths, 

 which unites with, modifies, and even overpowers the regular polari- 

 zation in meridional planes just referred to. 



The result will be the composition of the effects of the reflection 

 of lio-ht at a concave spherical surface having the sun for one pole, 

 with that due to reflection at a cylindrical surface perpendicular to 

 the horizon. 



If the latter be tolerably uniform in all azimuths, it will evidently 

 overpower and replace the former in points nearly opposite to the 

 sun, and which become visible when the sun is low. 



The author stated his conception of the physical cause for such an 

 arrangement of the planes of polarization to be, that whilst, at con- 

 siderable elevations, the number of reflecting particles is not so great 

 as near the horizon, the effect due to a single reflection will be the 



