Vol.. 6, 1920 
ASTRONOMY: C. G. ABBOT 
5 
method employed hitherto has been the method of Langley which consists 
of two essential parts. First, measurements with the pyrheliometer of 
the intensity of solar radiation at the earth's surface. Second, measure- 
ment by the spectro-bolometer, covering several hours' time, designed to 
determine the transparency of the atmosphere for all colors in the spec- 
trum. This latter part of the work demands a uniform transparency of 
the atmosphere during the several hours required for determining it. 
This condition is never perfectly met by terrestrial conditions. The 
changes in the atmospheric transparency not infrequently introduce an 
error of 1 and sometimes 2 or 3% in the results of the observations at 
the best stations, although on many days of observation the condition of ^ 
uniform transparency is so well fulfilled that the results appear to be 
accurate to a fraction of 1%. If we were interested only in the absolute 
value of the solar constant of radiation it would be possible to nearly elim- 
inate this source of error by a sufficient number of observations. But 
it has been shown that it is highly probable, I may say indeed certain, 
that the solar radiation is variable from day to day through a range of 
several per cent. The source of error which has been described is a 
principal difficulty against determining the real variations of the sun on 
which variations of the weather seem to depend. 
My colleagues and myself have long sought to discover a method of 
determining the solar radiation outside the atmosphere which would be 
independent of these variations of transparency. We sought to do this 
through the observations of the brightness of the sky. It is well known 
that the sky owes its brightness to the scattering of the solar rays by the 
molecules and small particles which compose our atmosphere. The more 
hazy the sky, the greater its brightness and the less the atmospheric trans- 
parency. Accordingly, it seemed to us hopeful to determine the atmos- 
pheric transmission coefficients in terms of the brightness of the sky. 
We employed in the investigation a new instrument which we call the 
pyranometer designed to measure the brightness of the whole sky or of 
any fraction of it, either on a horizontal surface or on a surface at right 
angles to the average direction of such a portion of the sky as might be 
under observation. While at Calama, Chile, in June, 1919, I took up 
the matter with Mr. Moore, the director of the Smithsonian observing 
station there. He placed at my disposal determinations of the trans- 
parency of the atmosphere for nearly forty wave-lengths, pyranometer 
values representing the intensity of the radiation of the sky in a zone 
15° wide surrounding but not including the sun, and measurements of the 
intensity of the solar radiation by the pyrheliometer and by the spectro- 
bolometer. All of these values were tabulated with the solar constant 
values for 60 days of observation. The pyranometer values and values 
of a function which we call p/psc were taken on each day at the periods 
