52 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



of 23 years as the assumed periodicities whose validity was to be 

 tested, or any particular time as better than another for departure, 

 or any preferable time interval for delimiting the tables. Such hasty 

 critics may suppose that any other periods or lengths of tables would 

 probably have been equally successful. 



To test this objection, computations were carried through to test 

 for the existence of periodicities of 7f , lo, 12, I2f , i5f, 19, and 29 



Fig. 22. — Trials of periodicities not related to 23 years. Compare with figures 

 19 and 23. 



months in the temperature of Berlin. These tabulations, like the 

 others, commenced with January 1819, but were arranged in tables of 

 10 lines. Thus they covered intervals of time having no particular 

 relation to the 23 years which previous computations proved to be 

 so important. The results are shown in figure 22. 



With regard to the 12-month periodicity, this analysis differs but 

 little from that shown in figure 15. The first and second 12-month 

 curves in figure 22 cover about the same intervals of time as in figure 

 15. Also other pairs in figure 22, as the sixth and seventh, the eighth 



