4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. IO7 



continuity of a supposed periodicity, I found it seemed indicated that 

 some changes should be made from the schedule of periodicities given 

 at page i8i of Annals, volume 6. For comparison I repeat here in 

 table 2, table 31 from that source, with two additional lines to show 

 the modifications found to be desirable. The amplitudes given in the 

 lowest line of table 2 relate to the second of two analyses which I made 

 of the data, as about to be described. 



Two complete analyses of the data were computed, of which I give 

 here only the second. The first started with August 1920. In making 

 it, two curves were drawn at the end. In one the sums of the effects 

 of 14 periodicities were used. In the other curve the best representa- 

 tion that I could make of Aldrich's determination ^ of the effect of sun- 

 spots on the solar constant was added to them. Neither of these 

 curves fitted the solar-constant results of 1945. The great depression 

 showed by the synthetic curves did not occur in the observations. 



It may very well be that the great depression in the observed values, 

 shown in figure 14, Annals, volume 6, in the years 1922 and 1923, 

 is real, and is a fortuitous phenomenon, not included in the sun's 

 ordinary course of variation as represented by the periodicities related 

 to the 273-month master period ; or it may be due to a long periodicity 

 like 45^ or 91 years. Possibly, on the other hand, the defective ob- 

 servations at Montezuma, and unsatisfactory sky conditions at Harqua 

 Hala may have been the cause of the great observed depression. In 

 short, possibly it was erroneous, though it is difficult to accept this 

 view as appears from the Annals, volume 6, page 176. 



I made a second analysis. Omitting the questionable interval, I 

 started in the middle of the year 1923, and ended with December 

 1945. Before giving the results of this second analysis I shall give 

 illustrative examples. First I show my method. Then I shall show 

 that individual periodicities continue throughout the entire period. 

 Finally I shall show why certain periodicities found in the earlier 1939 

 analysis are now omitted, and others substituted. 



Figure i gives a facsimile reproduction of the computation of the 

 periodicity of ii^j months for the interval May 1923 to September 

 1945. Although not quite identical, the three sections of the computa- 

 tion agree in supporting the continuing reality of this period of ii-J 

 months. 



The reason for introducing the periodicity of 16 months is that I 

 had found it in the residual curve C of figure 14, Annals, volume 6.^ 

 It is not an important periodicity, but I think a real one. 



1 Smithsonian Misc. Coll., vol. 104, No. 12, 1945. 



2 See Science, May 11, 1945, p. 483. 



