THE KANSAS UNIVERSITY 

 SCIENCE BULLETIN. 



Vol. XIIL] MAY, 1920. [No. 9. 



Application of Marvin's Periodocrite to Rainfall 



Periodicity.* 



BY DINSMORE ALTER. 



(Plates YII and VIII.) 



PROFESSOR MARVIN has recently ^ published a criterion 

 for discrimination between real periodicities and fortui- 

 tous ones. This criterion, called by him the periodocrite, 

 seems to me to fill a real need, and I hope that it, or a slight 

 modification of it, may be adopted generally for such purposes. 

 If the data covers q of the suspe3ted cycles they are arranged 

 in q rows and p columns. The total number oi^ observations is 



N. ^o= =^ V - ^n" = ^ "^n is then formed. Let n be any 



A^ .T979 N 



number of the rows or cycles. The mean is taken of the n ob- 

 servations, in each column, and 't,, = ± \j - "^' = ^ is 



n .1919 n 



formed. The ratios — are plotted as ordinates and i . — as 



To 1/ n 



abscissae. "When y is substantially and consistently greater 

 than X a real periodicity is indicated of greater or less amplitude." 

 In the first of these two papers published here I have given 

 two tables continuing the work of the previous paper on a 

 rainfall period equalling one-ninth the principal sun-spot 

 period. The first of these tables shows the percentages of 

 normal for each phase of each of twenty-four consecutive 



* Received for publication August 5, 1921. 



1. Monthly Weather Review, March, 1921, pp. 115-124. 



(109) 



