48 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



be ignored." Similarly the periodicity of 13.6 months was disclosed 

 by five waves in the preliminary curves computed for 68 months. 



But it is freely admitted that if only one station had been investigated, 

 some of the periodicities, especially those of 46 and 55 months might 

 have been regarded as doubtful. Much support for the veridity of all 

 of the longer periodicities is found by comparing results at several sta- 

 tions. In order to help the reader to appreciate the value of this sup- 

 port, I give in figure 19 some of the more convincing examples of 

 each of the 14 different periodicities which are accepted as caused in 

 terrestrial temperatures and precipitations by the solar influences 

 integrally related to 23 years. To save space in depicting so many 

 curves, certain special arrangements are made in figure 19, as follows : 

 In the periodicities of 21. 34, 46, and 68 months, certain curves are 

 inverted, as described in the legend. In the 25-month periodicity, odd- 

 numbered curves are given separately from even-numbered curves. 

 All of these arrangements emphasize the phase reversals already noted. 

 As amplitudes are large in the longer periodicities, the scales of ordi- 

 nates are diminished for them. 



It is believed that if the reader bears in mind the abundant evidence 

 already presented, which shows that periodicities of 8, 9!, 11, 12, and 

 21 months change phases and forms radically at the expiration of 

 integral multiples of 11^ years after January 1819, he will be prepared 

 to accept as veridical all of the periodicities shown in figure 19. 



Accepting this evidence as proving in general the veridity of all 

 of these periodicities because they are all so well marked at some 

 stations, and almost without exception in solar radiation, as shown in 

 figure 7, it seemed but a matter of course to compute them for each 

 and all stations, and for departures of both temperature and precipita- 

 tion from normal. All such computations gave more or less favorable 

 curves. Some curves covering short time intervals, had they stood 

 alone, might not have been regarded indeed as expressing a veridical 

 periodicity. But reinforced by the better curves representing that same 

 periodicity for the same station at other intervals within the century, 

 and by such evidence as is given in figure 19, even these less satis- 

 factory curves were acceptable. 



If I am so fortunate as to have carried the conviction of the reader 

 thus far, he will perhaps still ask, why I have stopped with 14 of the 

 23 periodicities which are integral submultiples of 2^ years, and 



' These two periodicities, 9I and 34 months, and also the 92-month periodicity, 

 re later discovered in solar radiation. (See fig. 7.) 



