NO. 10 SOLAR RADIATION AND WEATHER STUDIES ABBOT 1 7 



from normal temperature at Bismarck, N. Dak., from 1875 to 1925 

 as computed from " World Weather Records," computations similar 

 to those illustrated in connection with table 3 were made. It was soon 

 found that evidences of terrestrial counterparts of each of the seven 

 solar periodicities then known were apparent for short intervals, but 

 changes of phase occurred, showing that continuity is lacking. Further 

 studies seemed to show that these puzzling changes of phase were 

 absent if the computations were restricted to intervals when the sun- 

 spot activity as measured by Wolf's numbers is nearly constant. Later, 

 when longer series of weather records were studied, another phase 

 relationship of much more importance was disclosed. But of this we 

 shall write later. 



Figure 9 shows the results of analyses of Bismarck temperatures 

 aimed to disclose and evaluate the ii-month periodicity during the 

 interval 1875-1925. The data are segregated into four groups in which 

 low, medium low, medium high, and high sun-spot numbers prevailed. 

 The dates included in this classification are indicated on the curves. 

 It will be seen that a gradual shift of the maxima of the ii-month 

 periodicity amounting in total to fully half a period is disclosed by 

 the mean values. 



Here, as in what follows, the reader is reminded that owing to the 

 presence of other periodicities, and of accidental fluctuations besides, 

 it is not fair to expect perfect correspondence between periodic curves 

 of a given length of period, when these are determined from rather 

 brief intervals containing but a few repetitions of the periodicity in 

 question. Specifically, for instance, the curve D5 of low sun-spot 

 number statistics in figure 9 differs at months i, 2, 3 in its trend 

 from the other four. Also the four curves Bi to B4 corresponding to 

 medium high sun-spot numbers, show considerable disagreement, 

 although each of them has its maximum in the first half of the period. 

 But when it is recalled that curves D5, B2, and B4 in figure 9, which 

 are the most unsatisfactory of those shown, represent, respectively, 

 only two, two, and one recurrences of the 11 -month periodicity, it 

 does not seem surprising that they deviate as much as they do from 

 the better determined mean forms with which they are associated. 

 Naturally, the effects produced by the influences which determine all 

 other periodic and accidental changes of temperature departures can- 

 not be eliminated by taking the mean of only one or two recurrences 

 of the ii-month periodicity. 



9. Preparation of Weather Data 



When a large program of computation of periodicities in weather 

 departures was undertaken, it was soon found that the monthly flue- 



