Effect of the Spectrum on the Haloid Salts of Silver. 251 



cast by the beam from G, and coming through the negative, just 

 touch, and the two are equalised in brightness by means of opening 

 more or less the rotating sectors K. 



The negative to be measured was marked with a scale of 

 ~ inches, and, in cases of sudden change of density, to less. It was 

 then placed in the measuring apparatus and measurements com- 

 menced. When the square patches were measured, a thickish rod 

 was employed, but for the photographed spectrum a knitting need]e, 

 ^ of an inch thick, was substituted. The opacities of the different 

 parts of the film were calculated from where the negative showed 

 "no deposit," and the opening of the sectors when the direct and 

 reflected light balanced was taken as the standard. The required 

 opening of the sectors showed but. little variation for any of the 

 photographs. 



The scale of density corresponding to different times of exposure 

 was plotted from the readings of squares, and the readings of the 

 different parts of the photographed spectra were applied to. the curve 

 so derived, and the density corresponding to the times of exposure 

 tabulated. From the photographed spectrum of the arc the positions 

 of the different measured densities were known and the curve with 

 the reference Fraunhofer lines plotted in the usual manner. From 

 these curves the curves for the normal or wave-length spectrum were 

 calculated, and it is these curves which are shown in the accompany- 

 ing figures. 



Reverting to the paper to which we have referred (' Roy. Soc. Proc./ 

 No. 217, 1881), it will be seen that the figures therein shown differ 

 from those here given. This is caused by the fact that the former 

 curves were only eye estimations of density, whilst the latter are the 

 comparative sensitivenesses derived from measured densities. 



