NO. 3 SOLAR RADIATION AND THE WEATHER CLAYTON 39 



If the period is associated with solar rotation, then the outbreaks 

 would be in about latitude 40 from 1911 to 1914 and in latitudes 

 l>etween 50 and 60 in 191 5 to 191 8. These changes agree fairly- 

 well with the changes of latitude of solar prominences, which accord- 

 ing to the Lockyers are found in about latitude 40° or near the time 

 of sun-spot minimum and move to higher latitudes at the time of 

 sun-spot maxima. (Proceedings of the Royal Society, Vol. 71, 

 pp. 446-452.) 



The approximate regular advance of the for long intervals, as, 

 for example, from 1911 to 19 14, is surprising. It apparently indi- 

 cates that outbreaks on the sun continue in approximately the same 

 regions for long intervals, or else there is some other cause of the 

 period than solar rotation. 



An analysis of the 22-day period for several years did not give 

 any clear indication of continuity, but an analysis of a period of about 

 nine days gave a remarkably steady progression of the azimuths, 

 indicating that the true period lay between 8.78 and 8.80 days, and 

 hence about one-third of a synodical rotation of the sun at the 

 equator. In this case there was no evidence of change of length with 

 the sun-spot period. 



Study of Periodic Changes in Solar Radiation by Means of 

 Periodograms Derived from Fourier Series 



From the evidence found in the case of the 30-day period and in 

 the 8.8-day period it seems probable that, in the average of nine or 

 more periods, the periodic or semi-periodic changes in solar radiation 

 continue in the same phase for sufficient intervals to be studied by the 

 method of the periodogram described by Schuster. 



For the first inquiry of this kind there were used the means* for 

 the 75 days following and five days preceding the values of 2.00 gram- 

 calories per square centimeter in solar radiation for the years of sun- 

 spot maxima (see table V, column 2). 



Daily averages were obtained for successive periods between 23 

 and 44 days and a Fourier series was made to the sixth subharmonic 

 and in a few cases to the eighth and tenth. The amplitudes computed 

 by the harmonic formulas are given in table XV, part I, and plotted 

 in curve A^ figure 16. The plot is made on a logarithmic scale, so 

 that the distance between the periods is proportional to their length. 

 Turning to the plot and looking first at the periods above 16 days, it 

 is seen that there is a maximum at 20 to 21 days and a prolonged 

 maximum between 28 and 31 days. This latter is of much interest 



