30 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I07 



scale of ordinates in figure 12 starts from zero, although the zero of ob- 

 served temperatures is at +3°. Hence in the comparison between pre- 

 diction and event, given at the bottom of table 9, a scale-correction of 

 3° F. has to be made, as indicated in column i of table 9. 



Readers will observe that maxima and minima agree well. The 

 average deviation in the curves, as plotted, between prediction and 

 event for 36 days is 2.6 degrees F. But the factor 4/10 having been 

 applied to the actual temperature departures, the true average devia- 

 tion is 6? 5 F. It is largely made up by the sporadic discordances of 

 October 24-25, and by the excessive temperature departures of No- 

 vember 13-18. This latter discrepancy comes, of course, from the 

 assumption made that within the interval of 36 days there would 

 be no change in the amplitude of the 6.6456-day periodicity. It failed 

 after 22 days. If one omits 7 days of large discrepancy, the average 

 deviation for the remaining 29 days is i?7 on the graph, or 4?2 in 

 reality. I presume meteorologists would hardly expect to make much 

 smaller averages of deviations in forecasting for i or 2 days in advance. 



21. SUMMARY 



Extension of tabulations published in 1936 having shown the prob- 

 able existence of a period of about 6f days in solar variation, this 

 assumed period was traced in Washington temperatures from 19 10 to 

 1945, and found to prevail with the exact average period of 6.6456 

 days. 



Shiftings of phase rarely occur in the recurrence of temperature 

 effects of this cycle within a single month, but in different months and 

 years phase shiftings of plus and minus i, 2, and 3 days occur. 



The amplitudes of the temperature effect change moderately in the 

 several recurrences of the period within a month, but from month 

 to month the changes of amplitude are sometimes large. The range 

 of amplitudes for months of the same name but in different years, is 

 from 2° F. to 20° F. The average amplitude, 1910 to 1945, for all 

 months, is about 5° F. For summer months less, for other months 

 more. 



Tabulations of values of the solar constant of radiation of the years 

 1924 to 1944 were made to fix the form, amplitude, and phase of the 

 curve of variation of solar radiation in the period of 6.6456 days. 



The form of the average solar curve shows an abrupt rise and 

 slower fall. The phase is such as to give maximum radiation on 

 January 3, 1924. The average amplitude, corrected by a factor of 

 10/7 for influence of supplying by interpolation of days missing in 

 the record, is 0.13 percent of the solar constant of radiation. This is 



