NO. 10 SOLAR RADIATION AND WEATHER STUDIES ABBOT 35 



3. The 23-year period, which is the approximate least common 

 multiple of the observed solar periodicities, and is also the approximate 

 length of Hale's solar magnetic cycle, is of dominating importance in 

 the terrestrial weather-responses to solar influences. 



4. While the intensity of sun-spot activity has some influence on the 

 phases of the temperature periodicities, it is by no means as important 

 as the arrival of integral multiples of the 23-year interval measured 

 from January 1819. These define large modifications both of phases 

 and amplitudes. 



5. The 23-year period governs not only periodicities which seem 

 to be of purely solar causation, but also the phases and amplitudes of 

 the 12-month periodicity in departures from normal temperature. This 

 periodicity might otherwise have been regarded as purely of accidental 

 terrestrial origin. 



6. It is not possible to arrive at definite conclusions as to the veridity 

 of periodicities of long duration from data restricted to 23-year in- 

 tervals, and further restricted to intervals of comparable sun-spot 

 activity. Another attack on this part of the subject follows. 



B. ANALYSES GROUPED IN PERIODS OF Il| YEARS AND 23 YEARS, BASED 



ON JANUARY 1819, AS DATE OF DEPARTURE, AND 



INCLUDING ALL DATA 



The preceding discussion of Berlin temperatures was restricted to 

 intervals of comparable sun-spot activity. But though this is desirable 

 it is not vital, and restrictions relating to 23-year intervals having 

 been proved to be more essential, it becomes necessary to merge all 

 data, whatever the prevailing sun-spot activity, in order to study fairly 

 the longer periodicities. It has been proved advantageous to base 

 our studies on the zero date January 18 19. 



A. The 1 1 -month Periodicity 



As before, we begin with the 11 -month period. As there is here 

 no intention of making a century-long comparison, no account need 

 be made of the correction (minus 3 days), nor when we deal with 

 the 8-month periodicity of its correction (plus i day). Table 7 

 gives, for illustration, a complete tabular determination of the mean 

 ii-month periodicity curves from January 1819 to October 1864. 

 The similarity of the two halves of each of the two 23-year periods 

 covered, and the complete opposition of these two 23-year periods, 

 each to each, are clearly shown in figure 16. It is instructive to note 

 how abruptly the transition occurs from one type to the other just 

 at the turn of 23 years after January 1819. The two types differ 



