lO 



SMITIISOMAN MISC'KLI.ANKOUS COLLKCI IONS VOL. 66 



results of 1914. A slightly clifft>rent wci^^htin^ of two discordant 

 points on each of the curves for wave lenj.;ths 0.4265*, 0.5062, 0.6702, 

 and 1.008 /x in the curves for U)oy would hriui; them all (excepting 

 that at wave length o.HsHo/i) into harmony with the conclusions 

 ahout to be stated. C'ontining- attention to the more accurate observa- 

 tions of 1 91 4, the change of contrast is greater for short wave lengths 

 than for longer ones, in the proportions indicated by the following 

 table (7) : 



Table 7. — Deviations of contrast of mean results of 1914 front stcindoni form 

 of Hjij. {^Derived from mean curi'es) 



Figure 5 shows graphically the dependence of the i(>i4 departures 

 of contrast on the wave length. 'J'he evidence of future years will be 

 re(|uiitd to show whether the irregularities of the curves of tigure 5 



Wave- lcn^ths 



Fig, S. — Dependence of brightness variability on wave length 



are accidental or solar, but of the reality of the increase of departures 

 attending decrease of wave length there seems to be little (piestion. 



CHANGES OF DISTRIBUTION FROM DAY TO DAY 



We now pass to a consideration of changes of contrast from day 

 to day. Still using the mean distribution of Kji^ as a standard, we 

 took logarithms of the values and subtracted these from the corre- 



