42 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. Jl 



because it corresponds to solar rotation in the region of the promi- 

 nences; but the maximum near 21 days cannot thus be explained, 

 unless it be a half-rotation at very high latitudes. The most striking 

 maxima, however, are those at 3.5 and 11. 3 days. 



These maxima were derived from an inspection of the curves S5 

 in figure 3. It is now seen from figure 16 that these maxima are not 

 the only ones, but that there are maxima 5.7, 9, and 13 days almost 

 as marked as at 11. 3 days, while there are less marked maxima at 

 5.25 and 6.8 days. This analysis proves that by dividing the solar 

 radiation values into separate classes we do not eliminate any of the 

 complex periods. The dififerences shown by figures i and 2 are due 

 to relative differences in the strengths of the different periods follow- 

 ing different solar values. 



It is further seen that the maxima bear some kind of relation to 

 each other. The period of 3.5 days is one-third of a period of 10.5 

 days and the later period is one-half of 21 days and one-third of 

 31.5 days. Again 5.25 days is nearly one-fifth of 26.3 days, 9 days is 

 nearly one-third, and 13 days is nearly one-half of this number. 

 The maximum at 6.8 days is between one-fourth of 26.3 and one- 

 third of 21 days, while 5.7, 6.8, and 11. 3 are submultiples of 34. 

 Owing to the difficulty of dividing the periods into fractions there 

 is some uncertainty about the amplitudes of the uneven divisions of 

 periods. 



It now seemed worth while to make a periodogram of a series of 

 observed values of solar radiation without separating them into 

 grades. The longest available series of observations is that made in 

 Calama from July 27, 1918, to May 16, 1919. These values were 

 averaged in successive periods running from 23 to 44 days and were 

 further averaged in submultiples of one-half, one-third, one-fourth, 

 one-fifth, one-sixth, and one-eighth of these, and the amplitudes of 

 the resulting periods were computed by means of the harmonic 

 formulas given above. The results are shown in table XV, part II, 

 and plotted in curve B, figure 16. In this curve the maxima are 

 remarkably near those of curve A for the periods below 14 days 

 and unlike those of curve A for the longer periods. Both curves 

 show maxima at 3.5 days, 6.8 days, 9 days, 11. 3 days, and 13.2 days ; 

 but instead of the maxima at 21 days as in curve A, there are now 

 found maxima at 18 and 24 days. 



There are still found maxima at 28.32 and 42 days, as in the upper 

 curve, but with different intensities. This difference may be owing 

 to the few periods embraced in the lower curve, which was made up 



