NO. 10 SOLAR RADIATION AND WEATHER STUDIES ABBOT 25 



But the amplitudes alter widely from time to time among the ii 

 curves shown. Not only do they thus vary, but the forms of the curves 

 differ widely also. When these features are carefully scanned, there 

 seems to be disclosed an interesting regularity. Beginning zvith the 

 y.ear i8ip, the forms and amplitudes may be arranged in pairs with 

 very good effect. The only deviation from noticeable similarity among 

 these pairs occurs for the pair covering the interval March 1886 to 

 July 1903. Of this pair of curves the first covers that period when 

 the sky was still filled with dust from the tremendous volcanic eruption 

 of Krakatoa. Dr. W. J. Humphreys has called attention to the dis- 

 turbance of weather which volcanic dust produces.^" We shall recur 

 frequently to the similarity of such pairs when considering other data. 



Curves a2 to ko, figure ii, similarly deduced, cover the intervals of 

 time in caption ii when the Wolf monthly mean sun-spot numbers 

 lay generally between 40 and 80. In part of the data the sun-spot 

 activity was increasing, and in the other part it was decreasing. But 

 no appreciable difl:'erence in the data seems to arise thereby. It appears 

 that neither 11 months o days nor 11 months minus 3 days gives 

 the maximum amplitude of the periodicity in this case. The best 

 period is 11 months minus i^ days. The following mean values show 

 this : 



The relative amplitudes as just given are 2.9, 2.6, and 4.8, respectively, 

 which show a decided preference for 11 months minus i^ days. In 

 each case the phase given is the same as that expected for the interval 

 1815-1819. Here, as before, it is noted that the curves show decided 

 similarity when grouped in pairs beginning with the second curve. 

 The only exception is the last pair which presents dissimilarity. Curve 

 I2 gives the mean result, assuming a period of 11 months minus i^ 

 days.'"" It depends on 58 lines of temperature departures, and shows 

 a range of o.°48C., and therefore, like the case already discussed, 

 may fairly be regarded as demonstrative. 



Turning now to the temperature data corresponding to Wolf sun- 

 spot numbers exceeding 80, these are graphically expressed in curves 

 as to 63 of figure 11. These curves rest on few data, only 4, 3, 4, 3, 

 and 2 lines, respectively. Excepting as, they are closely similar. The 

 curve as is in fact displaced 5 months in phase from all the others. 



'"Journ. Franklin Inst., vol. 176, pp. 131-172, 1913. 

 ^"" The mean for 1 1 m. o d. is given by curve hK 



