NO. 5 FORECASTS OF SOLAR VARIATION ABBOT 3 



ill my paper " Weather Duiiiinated by Solar Changes," rather than 

 the mechanical process of the periodometer. 



There were no months of observations available. This left unused 

 values of 5 months and 6 months, respectively, in evaluating periodici- 

 ties of 7 and 8 months period, and of 5, 10, and 20 months, respec- 

 tively, in evaluating those of 21, 25, and 45 months period. Only in the 

 case of the 11 -month periodicity were all available data used. It was 

 impossible without a longer interval to evaluate satisfactorily the 68- 

 month periodicity. I found that all the other periodicities agreed 

 closely in form and phase with those published in " Periodicities in 

 Solar \'ariation," except that the amplitudes now found average but 

 0.7 as great as those found in that paper by Mrs. Bond from the longer 

 series of earlier values. Hence to supply the place of the 68-month 



Table i. — Solar constants. Monthly mean values 



Year 

 A . ^ 



Month 19J4 19J5 19-6 19^7 19^8 1929 1930 1931 1932 1933 



January .. 1.938 1.944 1.940 1.941 1.938 1.933 1-941 1-944 1-943 1-94^^ 



February . r.936 1.943 1.936 1-94^ 1-942 1-933 1-940 1-943 1-959 1-947 



March ... 1.941 1-939 1-937 1-943 I-94S 1-932 1-939 1-944 1-952 .-.- 



April .... 1.942 1.947 I-93''J 1-944 1-942 1-934 i-94i 1-945 1-958 .... 



May 1.945 1-950 1.940 1-944 1-946 1-936 1-944 1-943 1-949 ---- 



June 1.952 1.945 1-943 1-947 1-948 1-933 1-943 1-945 1-942 •--- 



July 1.946 1.951 1-94-^ 1-945 1-943 1-933 1-943 1-948 1-944 ■--- 



August ... 1.940 1.944 1.943 1-943 1-940 I-932 1.941 1-947 1-944 



SeptemixM-. 1.946 i-9Si i-94i 1-944 1-939 1-929 I-935 1.946 1.937 



October .. 1.950 1.946 1-935 1-944 i-93i i-93i 1-935 1-949 i-940 



November. 1.947 1-945 1-93-' 1-944 I-929 1-938 1.938 1-945 i-940 



December. 1.943 1.945 1.938 1.941 1-928 1.942 1.941 1-954 1-945 ---- 



periodicity I used her curve for it, first reducing its amplitude to 0.7. 

 I found, too, that the periodicity assumed to be 7 months should 

 preferably be taken as 6-| months. Without re-evaluating it on that 

 basis I retarded the 7-month curve by one month at the end of each 

 21 months, in building up the curve of summation. 



As remarked above, none of the periodicities excepting that of 

 II -months interval were evaluated by using any data later than Sep- 

 tember 1932, and the intense periodicities of 25, 45, and 68 months 

 were determined solely by data that ended prior to April 1932. Hence 

 it is fair to regard at least all parts of the curve given in Figure 2 

 later than October 1932 as forecasts, and the plotted points of observa- 

 tion from October 1932 to February 1933 as parts of the verification 

 of the prediction. 



Table i gives the data used. Figure 2 shows the result. The average 

 deviation between observation and computation from January 1924 to 



