REPORT OF THE SECRETARY 119 



COMPAlilSON OF RESULTS 



With the coinpletion of the reduction of all the solar-constant 

 observations from the three field stations results of much interest are 

 found by comparing them. Figure 1 shows the monthly mean solar- 

 constant values derived from Table Mountain, Montezuma, and 

 Mount Brukkaros since 1926. The probable error of the weighted 

 mean curve shown as a heavy line in Figure 1 is less than 0.1 per cent. 

 In short, it is adequately accurate to show all that needs be known 

 of the general march of solar variation. 



Figure 2 shows the preferred monthly mean solar-constant values 

 from 1920 to 1930, inclusive. The extreme range of it is 2.8 per cent. 

 Although apparently so irregular, Figure 3 shows that the march of 

 solar variation may be expressed with surprising fidelity as the sum 

 of five regular periodicities, of 68, 45, 25, 11, and 8 months' intervals. 

 It is interesting to note that, though derived with no regard to it, 

 all of these intervals turn out to be nearly related to the lli4-year 

 sun-spot period. Thus 68 months is its half, 45 months its third, 

 and so on. Other periods are found which are not so long-lived as 

 these. Thus, curve H in Figure 3 shows periods of 45 and 5.6 days, 

 respectively, which lasted throughout the year 1924. The excellent 

 representation of the original curve A by the sum of the five peri- 

 odicities, as shown at B, encourages me to give in curve I the 

 expected march of solar variation in 1931 and 1932. 



Figure 4 gives the results of an attempt to represent the tempera- 

 tures of Washington, D. C, and Williston, N. Dak., as made up of 

 periodicities having these same five intervals, 68, 45, 25, 11, and 8 

 months. It proved necessary to add a period of 18 months in each 

 case. The original temperature curves A and C are found by taking 

 consecutive means of 5-month departures from normal. Thus, 1/5 

 (Jan. + Feb. + Mar.+Apr.+May) : 1/5 (Feb. + Mar. + Apr. + May + 

 June), and so on. This eliminates the shorter irregularities and 

 brings out prominently the principal departures from normal tem- 

 perature that have occurred since 1918. 



Curves B and D are 5-month consecutive means of curves repre- 

 senting the observed march of temperature as the sum of the six 

 periodicities above described. I do not insist that this method of 

 treatment gives certainty as yet, but I look forward for five more 

 years to 1936, when it can be subjected to a more rigorous test. Time 

 will show whether or not it is the germ of the method of forecasting 

 weather for future years, to which Langley looked forward when he 

 founded the Astrophysical Observatory. 



The comparison of stations shows that the daily solar-constant 

 values are not as accurate as are needed. Montezuma results are by 

 far the best. Yet they lack many days of completeness and many 



