6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 95 



fulfilled at Washington. Certain months, as January, March, April, 

 October, seem to be nearly unexceptionable as regards oppositeness of 

 features. Yet for several reasons it must not be counted as a great 

 blemish that some of the months are less perfect examples than these; 

 First, it is clear from figure 2 that the temperature effects, if at- 

 tributable to solar changes, continue for many days after the assumed 

 solar causes have died away. Hence, each individual case of the many 

 whose averages are plotted in figure 2 must probably have differed 

 decidedly from the average during its individual course because the 

 effects of previous or immediately succeeding solar changes were still 

 incomplete. Only in the mean of a very large number of cases, many 

 more than the 10 to 20 of each group going to form the averages, 

 could these vestiges of effects of previous or succeeding solar changes 

 be totally eliminated. Second, the arrangement of results by months 

 is arbitrary. The proper arrangement would be according to the 

 general meteorological condition prevailing in the Northern Hemi- 

 sphere. Since there are wide differences between the curves in dif- 

 ferent months, it is highly probable that the cases which are forced 

 together by the monthly classification should not. in fact, all go to- 

 gether under a proper classification. Third, owing to errors in solar- 

 constant determinations, there is no doubt that some of the solar 

 sequences referred to in table i are spurious. Hence, on these three 

 accounts it ought not to be expected that every pair of curves in 

 figure 2 should be precisely in opposition at all points. 



Figure 3 gives the mean of April results at Washington for the years 

 1924 to 1930. Heavy lines give the general averages corresponding 

 to rising and falling solar sequences respectively. Light lines give the 

 averages of the two or three cases of each sort when rising and fall- 

 ing solar sequences well above the average in amplitude were observed. 

 It will be seen that the ranges of these thin-lined curves are much 

 greater than those of the thick-lined curves representing the general 

 averages. Thus criterion no. 2 is supported. To add to its support, a 

 tabulation was made of 150 cases in total to show the average ranges 

 compared to the ranges of the 46 cases among them which presented 

 exceptionally large solar sequences. The comparison showed average 

 amplitudes in the ratio i to 1.95. 



As for criterion no. 3, figure 4 gives the results for rising and 

 falling sequences in the years 1924 to 1930 as compared to the years 

 1 93 1 to 1935. The reader must again be cautioned as above not to 

 forget the influences of continuing effects of preceding and imme- 

 diately succeeding solar changes, of unsuitable classification, and of 



