56 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I46 



Fig. 47. — Temperature, Washington, D.C. Opposed effects follow rise and fall 

 of solar-constant and ionospheric observations. 



of observed solar variation, but 46 cases of unusually great observed solar 

 changes were followed on the average by 1.95 times as large temperature changes 

 in the same phases as the mean of 150 cases of all amplitudes. Again, the 

 average trends of temperature following solar changes, as observed in the 

 years 1924 to 1930, were nearly identical in phase, magnitude and form with 

 those observed in the years 1931 to 1935. 



But now I offer a new evidence which I think is even more convincing. 

 If, in reality, the observed variations of the sun were real, and influenced 

 temperatures greatlj-^ for 16 days after their incidence, there still seems no 

 reason to think there should have been any unusual temperature effects imme- 

 diately before their occurrence. I have therefore computed for each of the 

 320 dates the march of temperature departures from normal for 16 days 

 preceding the dates in question. I have then computed correlation coefficients 

 for Washington as between the average marches of temperature attending 

 rising and falling solar sequences, both after and before the beginnings of the 

 sequences of solar change. 



To fix ideas, I recall that in each division of this test there are 24 lines 

 comprising 17 values each, two lines for each month of the year, selected from 

 the 12 years, 1924 to 1935. These pairs of 24 lines of the divisions are 

 separated into two types, one type containing 17 values for days following, 



