NO. 3 RADIATION OF THE ATMOSPHERE — ANGSTROM 1 55 



The methods just discussed lead to a numerical value for the solar constant. 

 But the measurements in a selected part of the spectrum lead also to a direct 

 test of solar variability, which seems likely to be especially valuable because 

 these observations are not affected by aqueous absorption. 



MEASUREMENTS WITH ABSORBING SCREEN 



We may put : 



I=I e-vm 



where I is the energy transmitted through the absorbing screen at the limit 

 of the atmosphere, / is its value after passing through the air mass m, and 

 Y is a constant dependent upon the scattering power of the atmosphere. If 

 now we plot log / against m, the points should lie on a straight line, whose 

 ordinate for m = o is log h. 



The values of h thus obtained from our observations are given under the 

 heading h in table 19. The straight lines were run by the method of least 

 squares, not so much because the presuppositions of this method seemed here 

 to be satisfied, as because thereby all personal bias was eliminated. The 

 " probable error " e of each value of h is appended as a rough indication of its 

 reliability, and the weighted mean h is given at the bottom of the table. A 

 comparison between the different values of h shows that they all differ by 

 less than 2 per cent; half of them by less than Y?. per cent from the mean 

 value. The deviation falls as a rule within the limits of the probable error. 



This result thus fails to support the variability of the sun inferred by Abbot 

 from simultaneous observations at Bassour and Mount Wilson. We cannot, 

 however, with entire safety draw any conclusions about the total radiation 

 from measurements in a limited part of the spectrum. All that can be said 

 with certainty is that a change of the energy in the green part of the solar 

 spectrum exceeding 2 per cent during the period of our observations is 

 improbable. 



If we, from this, are inclined to infer that the total solar radiation during the 

 same period was constant, this inclination rests upon a statement by Abbot 1 

 himself to this effect : " So far as the observations 2 may be trusted, then, 

 they show that a decrease of the sun's emission of radiation reduces the 

 intensity of all wave lengths ; but the fractional decrease is much more rapid 

 for short wave lengths than for long." 



Yet unpublished measurements by Mr. A. K. Angstrom, in Algeria at 1,160 m. 

 altitude, give a mean value for h equal to 0.0708, which is in close agreement 

 with the value 0.0702 given above. On the former occasion Mr. Abbot's 

 spectrobolometric observations gave a mean value for the solar constant of 

 1.945. If we assume the energy transmitted by our green glass on Mount 

 Whitney to bear the same ratio to the total energy, the Mount Whitney 

 observations give a value for the solar constant reduced to mean solar distance 

 equal to 1.929, which differs by less than 1 per cent from the former value. 



1 Annals of the Astrophysical Observatory, Smithsonian Institution, 3, 133. 



I9I3- 



2 Observations of Bassour and Mount Wilson, 1911-1912. 



