574 PROCEEDINGS OP THE AMERICAN ACADEMY. 



thus obtained the resultant of the combined radiation and gas forces could 

 be determined for the time of every turning point. Every value was 

 divided by the deflection at standard sensitiveness of the galvanometer 

 G 2 read at the same time and was thus reduced to a standard lamp. 

 Results thus obtained, together with the ballistic measurements, showed 

 the direction and extent of the gas action as well as its variation with 

 length of exposure. 



The reasons for reversing the suspension follow : The beam from the 

 lamp, before reaching the balance, passed through three thick glass 

 lenses and two glass plates. All wave-lengths destructively absorbed by 

 the glass were thus sifted out of the beam by the time it reached the 

 balance vanes. The silver coatings on the vanes absorbed therefore 

 more than the glass. The radiation pressure was always away from the 

 source irrespective of the way the vanes were turned, while the gas 

 action would be exerted mainly on the silvered sides of the vanes. 



At the close of the pressure and energy measurements, when the re- 

 flecting power of the silver faces of the vanes was compared with that of 

 the glass-silver faces, the reflection from the silver faces was found very 

 much higher than that for the glass faces backed by silver. This result 

 was the more surprising because the absorption of the unsilvered vanes 

 was found by measurement to be negligibly small.* This unexpected 

 difference in reflecting power of the two faces of the mirrors prevented 

 the elimination of the gas action, by the method described, from being as 

 complete as had been hoped for. But by choosing a gas pressure where 

 the gas action after long exposure is small, the whole gas effect during 

 the time of a ballistic exposure may be so reduced as to be of little con- 

 sequence in any case. 



By exposing each of the vanes in turn and by reversing the suspension 

 and averaging results, nearly all errors due to lack of symmetry in the 

 balance or in the position of the light images with reference to the rota- 

 tion axis, or errors due to lack of uniformity in the distribution of 

 intensity in different parts of the image, could be eliminated. 



The changing character of the gas action, both with time of exposure 

 and gas pressure surrounding the balance vanes, is well illustrated in 

 eight series of static observations in which the glass faces of both vanes 

 were exposed.f The results obtained on the two vanes were averaged 



* Lord Ra3'leigh records a similar difference between the reflection from air- 

 silver and glass-silver surfaces. Scientific Papers, Cambridge, II 538-539 (1900). 



t Observations were alsr) made on the silver faces, but the gas action when the 

 glass faces were exposed was nearly double that for the silver faces, so the least 

 favorable case is shown. 



