28 Mr. R. H. M. Bosanquet on a new Form of Polariscope. 



We should expect such a polarization to be insensible ; and 

 that it is so is continued by the observation of the setting 

 sun with the polariscope which I have succeeded in making. 

 No trace of polarization is to be seen in the sun's light. 



Let us see what the positive polarization at the neutral 

 point will be if we compute it on the supposition that it arises 

 from the simplest case of diffraction, as is most probably the 

 case. 



Let A, B be the amplitudes which would produce the in- 

 tensity and polarization observed normally to the sun, so that 



A 2 — B 2 B 



the polarization there is -p — w; ; and put - = tan <f> n . Then 



the polarization of the same components looked at in a direc- 

 tion forming an angle a. with the line to the sun is 



A 2 (l-cos 2 a)-B 2 sm 2 a 

 A 2 (l + cos 2 a) + B 2 sin 2 a' 



Comparing this with the form 



cos 2 6 — sin 2 <b a , , . ,. 



9-7 . 9 , = cos 2 d> = polarization, 



cos- cj> + sm 2 <f> r r 



we have 



tan 2 (j> = cos 2 a + tan 2 (p n sin 2 a. 



In the sky cj) n is generally from 20° to 25° ; and putting 

 B=45° — (/>, we find for R values between 1° and 2°. 



These values are so much larger than that due to horizontal 

 astronomical refraction that it seems impossible that the two 

 can compensate each other at the neutral points. 



We are thus thrown back on the original speculation of 

 Arago ; that the horizontal polarization in the neighbourhood 

 of the horizon is due to the effect of light reflected from the 

 atmospheric regions surrounding the observed point. It is 

 clear that this is capable of furnishing an explanation ; for 

 the light diffused from sources in a line with the observer 

 would not be specially polarized by diffraction at the observed 

 point, while that from sources to the right and left would be 

 horizontally polarized, as the plane of source, point, and ob- 

 server would be in the mean horizontal. But it does not seem 

 practicable to establish any definite theory of this kind. The 

 principal interest of this horizontal polarization and its effects 

 would appear to be in connexion with the meteorological con- 

 dition of the atmosphere. 



