240 ABBOT— SOLAR CONSTANT OF RADL\TION. [April 21, 



curve, extending f roni about wave-length 0.30 ^i in the ultra-violet to 

 about wave-length 3.0/1, in the infra-red. Its ordinates are deflec- 

 tions of the galvanometer, proportional to energ>' in the spectrum, 

 and its abscissae are proportional to difit'erences of prismatic devia- 

 tion. The Fraunhofer lines, and great oxygen and water-vapor 

 bands, show as depressions in the curve. In order to eliminate dis- 

 tortions which are due to differences, for differing wave-lengths, 

 in the reflecting power and transmission of the mirrors and prism 

 used in the optical train, special investigations of the transmission of 

 the apparatus are made from time to time, and the curves corrected 

 accordingly. 



In our ordinary practice, from six to eight holographs are taken 

 in a single forenoon, between the times when the sun's zenith dis- 

 tance is 75° and (say) 30°. The curves are measured at about 

 thirty positions, uniformly spaced in the prismatic spectrum. Each 

 group of six to eight measurements, at a single spectrum place, fur- 

 nishes means of computing from Bouguer's formula the transmis- 

 sion of the atmosphere for that wave-length, and also the ordinate 

 which would have been found there if the observations had been 

 made outside the atmosphere. The sum of the ordinates measured 

 on any holograph is approximately proportional to the total energy 

 of all wave-lengths observed. Similarly the sum of the ordinates 

 computed for outside the atmosphere is proportional to the total 

 energ}' there.^ 



In order to reduce the total energy, as determined bolometrically, 

 to calories per square centimeter per minute, the pyrheliometer is 

 read, while the spectro-bolometric work is in progress, on each day 

 of observation. Thus a factor is obtained for deducing from the 

 areas of the l3olometric curves the true heat units corresponding.^*' 

 A complete determination of the solar constant of radiation requires 



" In the regions of great water-vapor and oxygen absorption the extra- 

 atmospheric curve is determined by interpolation between adjacent com- 

 paratively unaffected wave-lengths on either side> for we know that there is 

 no oxygen or water-vapor absorption of these bands produced in the sun, so 

 that they ought not to show in the extra-atmospheric curve. Small allowances 

 are also made for the energy of lesser and greater wave-lengths than any 

 observed. 



^'' For further details consult Annals, Vol. IL 



