ON SOLAR RADIATION. 43 



lamp was kept within less than a tenth of a volt of 75*0 volts by the 

 automatic recording potentiometer, and the observations were taken 

 within a few days of each other. In case 1, with the clean metal surface, 

 the value of the coefficient of cooling g=-00512 is nearly that due to 

 convection and conduction alone, as the radiative power of clean metal 

 is very small at these low temperatures, although the absorptive power 

 for the lamp radiation is nearly 40 per cent. An extremely thin coating 

 of smoke-black (2) suffices to raise tho absorptive power for the lamp 

 radiation nearly to its maximum, although the radiative power for rays 

 of great wave length is still very low, as shown by the small value of 

 (y= -00595, and the high value of the final excess of temperature 

 )-/^=2-528° C. The thicker coating of smoke-black (3) lowei's the value 

 of the final excess to 2'373° C, because the coefficient of cooling is in- 

 creased in a much greater ratio than the absorptive power for the lamp 

 radiation. It appears from the great increase of q, in case (4), that the 

 dead-black varnish is a much more efficient radiator at low temperatures 

 than the smoke-black, although the absorptive power for the lamp radia- 

 tion is l)ut slightly increased. The back of the disc was not covered in 

 these experiments in order to obtain a greater rise of temperature. In 

 case (5), with a new thermo-couple, the diminution in the values of r and 

 I, as compared with case (3), may be due simply to unavoidable errors of 

 observation, or slight variations in the uniformity of the wires, or in the 

 quality of the smoke-film ; but it may also be caused by a variation in the 

 cooling of the junctions by conduction, due to slight differences in the 

 attachment of the wires to the disc. In any case it is satisfactory to find 

 that so large a change in the conditions produces a change of less than 

 one per cent, in the result. Similarly, in case (6), the effect of blacking 

 the back of the disc is to produce a very marked increase in the 

 coefficient of cooling ; but although the rate of cooling by radiation is 

 nearly twice as great as in case (5) — supposing that the conduction and 

 convection effects remain the same as in case (1) — the diminution in tho 

 result for I, as compared with (4), is not greater than might reasonably 

 be attributed to the thinness of the varnish, which possessed appreciable 

 reflecting power. 



It is clear from the above summary that the method is capable of 

 giving fairly consistent results in spite of wide variations in the experi- 

 mental conditions. But it is evidently necessary to investigate further 

 the absorptive powers of different coatings for radiations of different 

 qualities if it is desired to obtain an order of accuracy higher than one 

 per cent, in the absolute results. Another correction of some importance 

 is that for the cooling of the junction by conduction along the wires. 

 This correction depends on the size of the wires and on their mode of 

 attachment to the disc. Although enormously reduced by the adoption 

 of very fine wires for the couple, it remains distinctly appreciable and 

 requires further investigation. It is evidently possible to determine this 

 correction by employing wires of different sizes simultaneously, or the 

 whole correction may be included in the coefficient of cooling by a suit- 

 able arrangement of the junction. 



Measurement of Solar Radiation. — Owing to the great intensity and 

 incessant variations of solar radiation, it would not be possible to obtain 

 absolute measurements directly by exposure of the instrument above 

 described to direct sunshine, although such a course has been attempted 

 with instruments of the class of Pouillet's pyrheliometer. Even with the 



