﻿298 M. H. Wild on the Absorption of Light by the Air. 



or 



a«-E=:i.C.tant> 2 ; 

 i 



and in this case the two unknown magnitudes C and - are de- 

 fined or eliminated by our placing both disks at the same distance 

 (that is, making E = e), on the assumption that the removal 



of the larger screen does not alter the ratio -. If the angle of 

 neutralization observed in this position is v u we have for this case 



lrr^C.tan^ 2 . 



This equation, divided by the above, gives 



a = 



On the above-mentioned days I succeeded in obtaining seven 

 complete observations uninterrupted by any disturbing influ- 

 ences. The unexpectedly large difference in the angles of neu- 

 tralization v and v } (about 2° when the screens were at the 

 greater distance 30 metres apart, while I had only expected 

 about J°) led me to fear that there might have been some dis- 

 turbing reflection from the house-walls on one of the screens. 

 Hence on the 10th of July the measurements were repeated on 

 the open road (one, that is to say, bounded on both sides by 

 meadows) ; these, however, only confirmed the previous results. 

 Introducing the values obtained for v and v v and the correspond- 

 ing ones for E — e, into the above formula, we get as the mean of 

 all observations for the coefficient of transparency of air, referred 

 to 1 metre as the unit of thickness, 



a= 0-9061, 



with a probable error of +0*0005. This number refers to white 

 light (that is, to the brightest rays in it, the colours that are be- 

 tween Fraunhofer's lines D and E of the solar spectrum), for a 

 mean temperature of the air =24° C, relative moisture =0*55, 

 and a mean pressure of 722 millims. The layer of air was about 

 1*2 metre above the ground. 



Before making any further observations on this result, I must 

 briefly mention the precautions which are indispensable for ac- 

 curate observations. It is first of all necessary that during an 



i 

 experiment the relative illumination j of the screens remain 



exactly equal. For this purpose the larger screen must always 

 be moved quite parallel, which is best effected by fixing a sight 



