NO. 4 



SUN S VARIATION AND TEMPERATURES — ABBOT 



21 



Mean values computed by shifting solar-constant values as tempera- 

 tures were shifted, as in tables 3 and 4, give less satisfactory curves, 

 and of lower amplitudes, than those computed with exact periodicity, 

 like figure 9. This result is shown gi^aphically in figure ii. The solar- 

 constant values for September are treated with strict periodicity to 



-s 



-2 



a 



10 



Fig. 10. — Average variation of the solar constant. Upper curve for 80 recur- 

 rences of the 6.6456-day period when large terrestrial temperature effects at 

 Washington occurred. Lower curve for 56 occasions of small terrestrial tem- 

 perature effects. 



derive the upper curve, and with phase shiftings to derive the lower 

 one. Hence the terrestrial shifting of phases is not caused by direct 

 shifting of phases in the solar emission. 



13. SPECIAL INVESTIGATION AS TO WHETHER INDIVIDUAL 



RECURRENCES OF THE 6.6456-DAY PERIOD IN SOLAR 



VARIATION ARE OF DIFFERENT AMPLITUDES 



There remains one other point which I have attempted to clear up. 

 As stated above, the average amplitude of the 6.6456-day period in 

 solar constants is but 0.13 percent. Yet in my two papers of 1936, 



