26 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



No explanation for this displacement is offered. The mean curve, I3 

 is computed by transposing the phase of as by 5 months and taking a 

 straight mean as of period 11 months o days. The result is as follows, 

 still in units of one-tenth degree Centigrade : 



0.7 0.4 I.I — 3.6 — 3.8 0.9 0.1 2.6 4.0 i.2 I.O 



The range is nearly o.°8 Centigrade. 



The skew relationship of period between the best 11 -month periodi- 

 cities as determined for the low, medium, and high sun-spot activities 

 is puzzling, but perhaps not impossible to account for. It will be re- 

 called that the periods found were 1 1 months minus 3 days, 1 1 months 

 minus i-| days, and 11 months o days, respectively. What this implies, 

 as far as the ii-month periodicity goes, is the advance of the tempera- 

 ture influence associated with high sun spots over that associated with 

 low sun spots by 14 months in 13O' periods. As the ii-month peri- 

 odicity is only one of many, and produces less than a tenth of the 

 total influence which, as we shall see, is exerted by those periodicities 

 which are nearly aliquot parts of 23 years, the eft'ect is not conspicuous. 



B. The 8-month Periodicity 



Figure 12 shows, in curves ai to ki, the mean 8-month periodicity 

 results derived from the intervals of low sun-spot activity. As shown 

 by the inclined lines there seems to be an advance of 5 months in no 

 years, corresponding to a corrected period of 8 months plus i day. 

 Taking account of this modification, but preserving the same phase 

 expected as of 1811-1815, the mean results are as follows: 

 — 0.6 0.4 1.6 2.2 2.3 0.1 —2.4 —2.5 



The range is almost o.°5 Centigrade, which owing to the modifying 

 influence of the 5-month smoothing, already referred to, must be less 

 than the real average range of this periodicity. The mean curve, U, 

 figure 12, is based on 75 lines covering the intervals of low sun-spot 

 numbers from 181 1 to 1925. Scanning the curves ai to ki on figure 12, 

 the pairing tendency, already referred to in discussing the 11 -month 

 analysis, is recognizable. The only marked inconsistency of the pairs, 

 as arranged with a beginning in 1819, occurs for curves dx and Ci. 

 It will be noted that for 8-month periodicities, as with the 11 -month 

 results, the pairs palpably begin with the second curve, about 181 9. 



Turning to the intervals when the Wolf sun-spot numbers lay 

 between 40 and 80, we again find the greatest amplitude by assuming 

 a period of 8 months plus i day. Choosing the phase to agree with 



