8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 87 



mean monthly temperatures 1918-1930, and have subtracted them 

 from the observed, thus giving monthly departures exactly suited to 

 the epoch studied. Lest the influences of shorter-period changes should 

 obscure the general march of events, we have smoothed the monthly 

 temperature departures by taking five-month consecutive means of 

 the form 



ai + a2 + ao + ai + 05 . a-i + Oz + ai + o-^ + C'^. ^^^ 

 5 5 



With the data thus prepared, we have sought and evaluated with 

 the periodometer all the periodicities which the curves disclosed. Our 

 procedure, as in the case of solar variation, is to subtract from the 

 data the effect of each periodicity as soon as determined, before pro- 

 ceeding to evaluate in the residual curve another periodicity. We 

 continued the search and evaluation until no more periodicities could 

 be perceived. 



The result obtained for Clanton is shown in Figure 4. Periodicities 

 of 8, 9-I, 18, 21, 25, 34, 39, 45, and 68 months were evaluated. These 

 periodicities and their partial determining curves are indicated by 

 letters C, D, E, F, G, H, I, J of Figure 4. The residual shown in 

 curve B plainly indicates the ii^-year period. We also note the large 

 positive departure shown in the residual curve B for the year 1930, a 

 year remarkable for the extraordinary drought and accompanying 

 cloudlessness. A similar extraordinary positive departure for 1930 is 

 shown in Figure 5, curve B, for Washington, and also in Figure 6 for 

 Williston. It will be noted that strong periodicities of 8, 21, 25, 45, and 

 68 months found in Clanton temperatures are found also in solar vari- 

 ation. The 1 1 -month solar period is indistinct in Clanton temperatures. 

 The 135-month period is doubtless of solar origin, although it does not 

 appear conspicuously in the solar variation between 1918 and 1930. 

 The other Clanton temperature periodicities of 9^, 18, 34, and 39 

 months were not found in the sun, but nevertheless 34 months is one- 

 half of 68 months, which is conspicuously found as a periodicity in 

 the sun. 



The results for Washington are shown in Figure 5. Periodicities of 

 8, 9^, 13^, 18, 25, 45, and 68 months are found as indicated at C, D, E, 

 F, G, H, I. The residual curve B shows clearly the 135-month peri- 

 odicity in practically the same phase, though lesser amplitude, than 

 Clanton. The extraordinary drought of 1930 produces its strong posi- 

 tive departure. Here again the strong periodicities of 8, 25, 45, and 

 68 months seem to reflect solar-radiation changes. The periodicities of 



