230 



J. P. Hall — Short Cycle in Weather. 



large rise or fall occurs a little too early or a little too late to 

 be regarded confidently as a repetition in the sense here meant. 

 And if one were to employ this phenomenon as a basis for 

 practical forecasting, he would experience frequent failure. 

 Yet in spite of non-recurrences and regardless of the doubts 

 which may arise as to the reality of the parallel in certain in- 

 stances, there is still so strong a suggestion of periodicity as to 

 command respect when attention is once directed to it ; and, 

 as will appear later, some of the failures and discrepancies are 

 susceptible of reasonable explanation ; so that investigation 

 eventually strengthens one's confidence in the genuineness of 

 the phenomenon. 



The first half of some of the 

 traces presented herewith re- 

 sembles the last half, and the 

 resemblance may prove to be 

 more than accidental ; so that, 

 judging from this alone, one 

 might believe that the true cycle, 

 if there be one, is 13 or 14 days 

 in length, rather thau 27 ; but 

 in Series III, as in others which 

 might be offered, the latter in- 

 terval seems to be requisite to 

 make out the full outline ; one 

 portion having oscillations either 

 of wider range or greater fre- 

 quency than the other. Reduc- 

 ing, then, the data embodied in 

 all three series to tabular form, 

 we find that while there is often 

 a variation of a day or two from 

 the exact period, only rarely does 

 the departure amount to three 

 days, one way or the other ; 

 and the mean length of the 109 

 cycles exhibited is 27"016 days, which must, however, be re- 

 garded as only an approximate result. A slight change would 

 be effected in this value by omitting from the computation 

 some of the instances which have been employed but which 

 may not properly belong therein. The temperature-curve in 

 our latitudes is a very complicated affair, apparently including 

 not only features that for convenience may be called accidents 

 but also several fairly regular undulations of different lengths, 

 which partially mask each other. But even after free elimina- 

 tion from the evidence here exhibited and from other traces 

 which might be presented, there still remains a residuum hint- 

 ing strongly at a period of not far from 27 days. 



