SOLAR CONSTANT OF EADIATION ABBOT. 321 



heat produced by each spectral ray. At each of the Fraunhofer 

 lines the thermometer would fall slightly. The great A band of 

 oxygen would produce a large decrease of temperature, but beyond 

 the red you would think several times you had reached the end of 

 the spectrum if you did not know better, and that you were exam- 

 ining great water-vapor bands. Suppose now that several hours 

 later you repeated the experiment. You Avould find that, excepting 

 in these great water-vapor bands, practically every part of the 

 spectrum was hotter than before, and that the change .had been 

 greatest in the violet end. At any selected wave length you could 

 then apply the method of Pouillet, and find what your instrument 

 would have indicated if you could have read its rise of temperature 

 due to the heat of the solar spectral ray outside of the air altogether. 



It would be natural to plot upon a convenient scale the spectral 

 distribution at the earth's surface, and outside the atmosphere, using 

 intensities of the spectrum as ordinates, and wave lengths, or pris- 

 matic deviations, as abscissae. The total area included between such 

 a curve and the axis of abscissae (or zero intensity) is proportional 

 to the total radiation of all colors combined. Hence the ratio be- 

 tween the computed area outside the atmosphere and that measured 

 at the earth's surface is the ratio which would be found between the 

 readings of the pyrheliometer if one could read it outside the atmos- 

 phere and again at the given hour at the earth's surface. So we 

 should determine the " solar constant " by multiplying the pyrhelio- 

 meter reading at the earth's surface for the given hour by the ratio 

 just mentioned, and then reducing the result to mean solar distance. 

 One thing, however, is to be considered. The energy in the great 

 atmospheric absorption bands of the infra-red spectrum does not 

 increase fast enough, as the path of the beam diminishes, to fully 

 obliterate the bands in the energy curve computed for outside the 

 atmosphere. But we know that there is no absorption in these bands 

 due to the sun itself. They are solely atmospheric. Hence in draw- 

 ing our extra-atmospheric computed curve we draw it smoothly so 

 as to eliminate all atmospheric bands. The remaining solar Fraun- 

 hofer lines may be blurred over by using a wide slit of the spectro- 

 scope, or, better still, a smooth energy curve representing average 

 intensities may be drawn to allow for them, both within and without 

 the atmosphere. As for the ultra-violet and infra-red regions be- 

 yond what is convenient to observe, corrections of a few per cent 

 are added for them. 



Such, in brief, is the method of Langley for determining the solar 

 constant of radiation. Unfortunately in this pioneering Avork he 

 came to distrust the application of the exponential formula of 

 Bouguer to the atmosphere, even when applied as he did it to homo- 



97578°— SM 1910 21 



