\} S. A. Hill — j-he Measurement of Solar Badiation. [No. 1, 



Table T. — Excess Temperature of the Sun thermometer on clear days at 

 Allahahad above the air temperature at noon. 



Year. 



Jan. 



Feb. 



Mar. 



Apl. 



May. 



June. 



Oct. 



Nov. 



Dec. 



Mean for 9 

 dry months. 



1876 



661° 



65-7° 



63-9° 



60-6° 



56-50 



57-3<' 



60 3° 



640° 



64-0° 



620° 



77 



63-0 



66-7 



63-6 



62-4 



600 



57-4 



59-8 



630 



63-4 



62-1 



78 



65-9 



64-3 



65-2 



64-5 



62-3 



57-8 



59-6 



64-7 



64-6 



632 



79 



64-2 



63-9 



63-7 



62-6 



61-8 



61-0 



61-1 



64-9 



63-5 



631 



80 



63-0 



63-8 



60 3 



59-9 



620 



61-5 



58-3 



62-5 



60-6 



613 



81 



623 



60-9 



61-9 



60-4 



60-3 



630 



548 



61-8 



59-9 



60-6 



82 



60-5 



61-5 



601 



61-1 



60-6 



591 



54-8 



62-2 



58-2 



59-8 



Mean. 



63-6 



638 



62-6 



61-6 



605 



59-6 



58-4 



63-3 



620 



61-7 



Variations in the fourth condition, above specified, cannot be allowed 

 for or corrected, unless by means of a long and troublesome experimental 

 investigation, but those of the first, second and third conditions may per- 

 haps be estimated by mathematical methods from observations already 

 made. Pouillet's formulse, it is true, has a rational basis only vs^hen it is 

 applied to radiation of one definite degree of refrangibility ; but, as an 

 empirical rule, it probably gives results not very wide of the truth when 

 the altitude of the sun above the horizon exceeds 40"*, as it does at noon 

 in Allahabad during every month of the year. If, then, we take a to re- 

 present the diathermancy coefficient of dry air, or the proportion of the 

 total radiation, transmitted vertically, through a layer of dry air which 

 produces a pressure of 1 inch ; /?, the diathermancy coefficient of vapour, the 

 tension of which is 1 inch ; and y, the proportion transmitted through an 

 atmosphere containing dust or haze to the extent of one unit on an arbi- 

 trary scale, we have — 



log r = log R + 5 sec 2; log a + /* sec 2; log p + d log y. 



The proportionate number for dust and haze, d, being somewhat uncer- 

 tain, there is no advantage in applying to it a correction for obliquity 

 especially as the vertical thickness of the dust layer is greatest in the 

 hot weather months, when the sun's rays fall almost perpendicularly. The 

 number for May, the dustiest month, being taken at 10, the proportionate 

 numbers I have assigned to the other months are : 



Jan. 



Feb. 



Mar. 



Apl. 



May 



June 



Oct. 



Nov. 



Dec. 



4 



7 



9 



10 



5 







1 



3 



The mean values of h and f barometric pressure and the tension of 

 vapour, are given in tables II and III. In strictness these should be 



