116 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1926 
As a result of all this painstaking work, the newly derived solar 
constant values show in their accordance, as well as in the various 
internal evidences which their computations afford, that they are 
of a new and higher order of accuracy than ever reached before. 
A new proof of solar variability—Many writers having expressed 
doubt as to the certainty of variations of the sun, either of short or 
long interval, a new and simple proof has been formed by Doctor 
Abbot, and will be published in the Monthly Weather Review 
for May, 1926. It rests on the basis that if the atmosphere 
had uniform temperature, transparency and humidity, and if the 
sun was observed by means of the pyrheliometer, always at the same 
altitude above the horizon, then the solar constancy or variation 
would exhibit itself directly, without recourse to the complex obser- 
Fic. 1.—Solar variation confirmed by results of selected pyrheliometry. 
Mount Wilson data of July, 1910-1920, excluding 1912, 1915, when volcanic dust from 
Mount Katmai made sky conditions not comparable. 
Thin full curve, pyrheliometry of selected days. 
Dotted curve, solar constant values hitherto published. 
Double curve, sun-spot numbers, 
vations and computations associated with the bolometer. In other 
words, at such times the atmosphere could be regarded as a screen 
of unchanging influence, and the readings of the pyrheliometer 
would be directly proportional to the intensity of solar rays. 
Testing this new idea on all the observations made in the months 
of July at Mount Wilson, Calif., between the years 1910 and 1920, 
Doctor Abbot found it necessary to exclude the years 1912 and 1913 
on account of the veiling effect of dust from the volcano, Mount 
Katmai. Many individual days were excluded also from each July, 
because the atmospheric conditions differed too much from the usual 
ones. 
From the remaining observations was plotted the full curve of 
Figure 1. Taking the identical days used in this study, the mean 
