NO. 10 SOLAR RADIATION AND WEATHER STUDIES ABBOT 



51 



culty would be found in making a satisfactory Fourier analysis in 

 14 terms of the temperature departures of Berlin from 1819 to 1929 

 or in discovering by that method the remarkable reversals of phase 

 which occur at intervals which are integral multiples of ii| years 

 after 1819. Furthermore, I hope statisticians will be fair enough to 

 weigh carefully the arguments I have presented, and having done so 

 will suspend adverse judgment until they have examined what is yet 

 to follow. 



°B^^ 



:v£r?^^^^E^^^^ 



w 



«=T^ 



C 



%t^¥ 



5 



a' 



^BSg^efe 



^-m 



^: 



^^^ 



^^^^^^^^^^§-^ 



^^^n 



^sf° 



Fig. 21. — Residuals after removing periodicities, Berlin temperatures. Heavy 

 curves, original data, light curves, residuals. 



D. ANALYSES BY INTERVALS NOT INTEGRALLY RELATED TO 23 YEARS 



The periodicities employed in the preceding discussion were selected 

 partly because they had been found in solar variation, partly because 

 they seemed to appear in Berlin and other temperatures and precipita- 

 tions. But it will be objected by some who ignore the fact that we re- 

 peatedly scanned the curves, and sought all periodicities existing 

 therein, that there was no reason for selecting integral submultiples 



