IResearcb Corporation jfun^ 



THE PERIODOMETER : AN INSTRUMENT FOR FINDING 



AND EVALUATING PERIODICITIES IN LONG 



SERIES OF OBSERVATIONS 



By C. G. abbot, 



Secretary, Smithsonian Institution 



(With One Plate) 



In a paper entitled " Weather Dominated by Solar Changes," ' I 

 described and illustrated a method of finding and evaluating periodici- 

 ties by computation. My method consisted first in plotting the lengthy 

 series of observational data on a large scale and scanning them from 

 a distance in order to perceive tendencies, if any, toward a repetition 

 of minima at some nearly regular interval. 



Such an interval of 8 months seemed to appear in the plot of solar 

 variation in the years 1924 to 1930. To test it and evaluate it, I ar- 

 ranged the 1 0-day mean solar-constant values in a table of 24 columns. 

 The top line contained the values for the first 8 months, the second 

 line those for the second 8 months, and so on until the data were ex- 

 hausted. Mean values of the vertical columns were then taken. These 

 indicated plainly the reality of the 8-month periodicity in solar varia- 

 tion, and determined the distribution of it. The mean form found for 

 the curve of this periodicity did not approximate a sine curve, but 

 showed a short quick rise from the minimum and a long slow decline 

 from the maximum to the minimum. 



The second step in computation was to subtract from the original 

 data values representing the average march of the 8-month periodicity. 

 A new curve of partial solar variation resulted, from which the aver- 

 age 8-month periodicity had been cleared. This residual curve was 

 next scanned, and seemed to display an 1 1 -month periodicity. It was 

 evaluated and removed from the residual data in the same way that 

 the 8-month periodicity had been evaluated and removed from the 

 original data. The 11 -month periodicity showed a double maximum 

 and still less resembled a sine form than the 8-month periodicity. 



A 45-month periodicity and a 25-month periodicity were similarly 

 discovered, evaluated, and removed. In Figure i , the residual curve C 



* Smithsonian Misc. Coll., vol. 85, no. i, 1931. 



Smithsonian Miscellaneous Collections, Vol. 87, No. 4 



