ON SOLAR RADIATION. 



133 



The mean values of solar radiation given above in the 

 first Table are deduced from observations made on every 

 day in the year, and therefore in every possible state of 

 the atmosphere — clear, cloudy, rainy, foggy, calm, stormy, 

 &c. ; but it was evidently desirable to determine the calo- 

 rific intensity of the sun^s rays on days when the sky was 

 cloudless at the time of maximum temperature. The 

 printed observations, however, do not always show when this 

 was the case, and it became necessary to adopt some arbi- 

 trary principle of selection. As the one which appeared 

 to me to be least open to objection, I selected, in the 

 first instance, the highest value in each month of the six 

 years, and taking the means obtained the following num- 

 bers, a projection of which is shown in curve No. i of 

 diagram C : — 



January 16*4 



February 20*5 



March i9'3 



April zz'6 



May 2o*9 



June 20'i 



July 2i'i 



August 2o*3 



September 21*8 



October 19*6 



November 17*2 



December 15*3 



C 



This curve exhibits two principal maxima in April and 

 September, and a low minimum in December. 



Taking next the means derived 

 from the three highest values in 

 each month, and then those of the 

 five highest, we have curves No. 2 

 and No. 3. The slight irregu- 

 larities in the first curve have now 

 disappeared, and curve 3 has two 

 well-defined maxima in April and 

 September, a principal minimum 

 in December, and secondary mi- 

 nimum in June. 

 Proceeding, now, a step further, 

 and taking the ten highest values in each month, we have 



