380 Ferret — Measures of Intensity of Solar Radiation. 



of nine hours, and then decreases regularly until the setting of 

 the sun. 



The smallest clouds and the least atmospheric disturbances 

 are followed by oscillations of the curve. There is a great con- 

 trast between the apparent constancy of the light of the sun, 

 and the continued oscillations of the curve, especially during a 

 fair sky and calm weather." 



Again the following deductions are made from the curves 

 subsequently obtained after his first notice of the apparatus 

 (Compt. Eend., cii, p. 962, April, 1886) : 



" 1. The oscillations during the days of summer (with a very 

 pure sky and without apparent clouds) are the more marked 

 as the atmosphere is most calm and its temperature the high- 

 est ; two maxima, the one before and the other after mid-day, 

 are sufficiently distinct, the' one from the other. 



2. During the days of autumn the oscillations diminish in 

 amplitude, and the maxima approach mid-day. 



8. During the days of winter the oscillations persist, but 

 their amplitudes still diminish more ; the two maxima tend 

 more and more to be confounded with each other. 



4. In fine, during winter days, when the temperature is the 

 lowest, and the mass of vapor contained in the air the least, 

 the two maxima unite into one, which occurs at midday : with 

 these conditions, especially if the atmosphere is strongly braced 

 by violent winds, hourly curves are obtained almost entirely 

 symmetrical with reference to the ordinate of mid day." 



Similar curves were subsequently obtained at places a long 

 distance from the ocean, where the air was supposed to be 

 much dryer than at Montpellier, where the first ones were 

 obtained. 



In this apparatus it is assumed that, for all temperatures, the 

 intensities of the thermo-electric currents, indicated by the 

 deflecting of the galvanometer needle, are proportional to the 

 thermo-electric forces, that these are proportional to the differ- 

 ences of temperature between the two ends of the thermopile, 

 and that these again, are proportional to the intensities of the 

 solar radiation by which these differences are maintained, and 

 so that the deflections of the galvanometer needle are pro- 

 portional to the intensities of the radiation, and are conse- 

 quently true relative measures of them. But none of these 

 assumptions, it seems to the writer, is strictly correct. 



By Ohm's law we have 



*=l (•) 



in which I is the intensity of the thermo-electric current, E is 

 the thermo-electric force and R is the resistance. In this ex- 



