108 



ANNUAL EEPOKT SMITHSONIAN INSTITUTION, 19 3 



change was less than 0.4 per cent in the solar constant value, and also 

 have omitted cases dependini^ on isolated or unsatisfactory values. 



Corresponding to each of these 189 cases I have tabulated the mean 

 temperature of Washington, Williston. and Yuma, for a 9-day period, 

 of which the day of culmination of the solar change is the fifth or 

 central day. Taking each month of the year by itself, I have com- 

 puted the average march of temperature over such 9-day intervals. 

 In illustration, I gave Tables 1 and 2, showing the Washington tem- 

 perature results of March, and Figure 4, which shows, at A, B, C, 

 and D, the mean values for March, May, July, and October. There 

 is given at E the average changes of solar radiation values corre- 

 sponding thereto. 



There are several reversals of sign of the average temperature 

 effects during the year. In Figure 5, 1 give a study of these changes 



JANUAWr FCBmM^Y 



Figure ,5. — Inversions of temperature dependence on solar variations 



of sign at Washington. All the cases have been arranged in con- 

 secutive order of days throughout the year, irrespective of what 

 year they occurred. The 98 cases corresponding to ascending se- 

 quences are given in Diagram F, and those corresponding to descend- 

 ing sequences in Diagram G. The quantity which is plotted is the 

 difference of temperature between that of the day of culmination of 

 solar change and that of four days previous. To guide the eye as 

 to the prevailing trend of the results, a zigzag line connects the 



