82 Prof. 8S. P. Langley on a Possible 
As nearly all the clouds, water-vapour, and other variable 
constituents of the atmosphere are contained in the layers 
below a two-mile elevation, it is immediately seen that the 
region over which approximately constant conditions must, 
for good observations, continue for three hours, is not many 
square miles, and this is a demand which can ordinarily be 
approximately met. 
Nevertheless it is obvious that if there was no criterion 
as to whether constant conditions had prevailed during the 
measurements, the mere smallness of the period of time and of 
the area concerned, would be of little weight for inference of 
the value of the resulting solar constant. 
There are various kinds of evidence to be derived from 
the measures themselves, which aid in forming a judgment 
of their value for determining the constant, and enable us to 
reject certain days’ observations which the impressions of 
the eye alone as to the clearness of the air might have 
inclined us to retain. These evidences are principally as 
follows :— 
First, the bolographic measurements of a single day on 
homogeneous rays, when plotted logarithmically as already 
described, should be represented closely by straight lines, if 
the air and instrumental conditions are nearly invariable. 
Examples of good results of this kind are shown in 
Plate I., which gives for March 25 and 26, 1903, the 
observations and representative lines for wave-lengths 1°027 y, 
0°656 pw, 0°468 w, and 0°395 w*. 
Second, the area of the (calculated) spectrum energy- 
curve outside the atmosphere ought to be nearly the same, 
whether estimated from one or another bolograph of a given 
date. As examples of this and the two following criteria, 
the reader is invited to consult the table of solar constant 
values at a later page. 
Third, the several solar constant values computed for a 
given date ought to agree within the limits of experimental 
error, though depending on independent bolographs and 
pyrheliometer measures. 
Fourth, if no variation of solar radiation occurs over 
several months, all the values of the solar constant computed 
during that interval should agree within close limits, regard- 
less of alterations in the transparency of the atmosphere and 
the altitude of the sun which may have intervened. In this 
* The dotted portions of the lines of March 26 are for morning observa- 
tions. IJtis seldom found that the morning observations bear this test of 
uniformity of conditions of the atmosphere, and they are now omitted, 
and the transmission ccefficients based on the afternoon bolographs alone. 
