1455 
conclude from these observations, that at the central phase of the 
annular eclipse the solar radiation fell below ‘/,,,, of its ordinary 
value. 
This remainder must in part be due to the unscreened ring of 
the disk. Assuming the apparent surface of that photospheric ring to 
be '/.s, Of the surface of the disk (which certainly is a low estimate), 
and its apparent radiating power per unit of disk-surface to be */, 
of the average intrinsie radiating power of the disk, we may say 
that at the epoch of centrality the photosphere was still able to 
furnish us with at least */,,,.. of the ordinary amount of radiation. 
Consequently, less — and probably much less — than 1/,,5,, ot 
the sun’s total radiation toward the earth is left as proceeding from 
the annular part of the solar atmosphere visible round the moon’s 
edge. 
So far, the inference is pretty sure, because it depends on the 
outcome of direct observations only. 
What we want to deduce next, however, is an estimate of the 
radiation due to the entire solar atmosphere — or rather to the 
visible half of it. This we cannot do without making some simply- 
fying assumptions concerning the absolutely unknown conditions 
prevailing in the sun. 
Let ZZ (fig. 1) be the photosphere (with radius 7), 
ABE the direction toward the earth. If the radiating 
and scattering power of the solar atmosphere were distri- 
buted homogeneously through its whole depth d, the 
emission due to the hemispherical shell would bear 
approximately the same ratio to the atmospheric emission 
observed at mid-eclipse, that the volume of the hemi- 
spherical shell (217? .d) bears to the volume of the ring 
produced by the rotation of the segment ABC about 
the sun’s diameter, which is parallel to AB, (2ar-segm. 
A KE. 
That proportion is 
Fig. 1. 
rd 
p= — —, 
p segment ABC 
For small values of d the surface of the segment is nearly 
*/, @. AB, and the ratio becomes 
ie 
= 3 
IN ts AR: 
Suppose we may replace the actual heterogeneous atmosphere by 
