SMITHSONIAN MISCELLANEOUS COLLECTIONS , VOL. 68 



days this mterval was extended to six days. The observed departures 

 and the smoothed means are given in table 4.' The results are plotted 

 in figures i and 2 in whicli the observed departures for 1913 and 1914 

 are connected by continuous lines and the means for each successive 

 five days are joined by broken lines. Under these in each case are 

 plotted the successive five-day means of temperature at Pilar for the 

 same intervals. It is evident to the eye that these temperature changes 

 followed the same general course as the changes in solar radiation 

 shown by the five-day means. 



Calculations were now made to determine the correlation values 

 for the five-day means of solar radiation observations and the five-day 

 means of the daily maxima of temperature observed at Pilar, In 

 these calculations account was taken of the tenths of degrees instead 

 of using departures to whole degrees as in the first computation. 

 (See table i.) 



The results obtained from this computation were as follows : 



Table 3. — Correlation of Solar Radiation and Temperature at Pilar from 



5-Day Means 



By comparing tables 2 and 3 it is seen that the correlation factor 

 is much higher for the mean of five days than for the individual 

 observations being nine times the probable error in 1913, and nearly 

 eight times in 1914. This increased agreement may be due to the 

 fact that the individual solar measurements are approximate, varying 

 on each side of the true result, or, else, that the changes of short 

 period are too rapid to produce any appreciable change on the earth. 



To determine whether a better result would be obtained by using the 

 mean temperature of each day instead of the maximum, the correla- 



* Note. Table 4 is omitted in the English text. 



