68 Recent Advances in Stellar Astronomy
the surface brightness, T the temperature, and A and B
are constants. This equation is not absolutely exact, but
the error only becomes important at temperatures exceed-
ing 10,000o, when the formula gives values which are a
little too low. Applying this method to the Sun, using
Abbot’s measures of the energy which it sends us in dif-
ferent wave-lengths, we find a temperature of 6100°. This
also should be a little too low if the Sun is not a perfect
radiator. The discordance between this and the previous
value shows that such is indeed the case.
For light of different wave-lengths the coefficient B in
the above equation varies inversely as the wave-length,
indicating that the change of violet light (and hence of
photographic magnitude) with the temperature is more
rapid than that of visual magnitude. It follows that the
variation of the color index with the temperature can
also be represented by an equation of the same form,
but with different values of the numerical coefficients A
and B, the latter being about one-quarter as great as in
the case of the visual surface brightness. Applying this
method again to the Sun (still using Abbot’s data), we
get a decidedly lower temperature, about 5200°. The
discrepancy can Ъе explained by the fact that the violet
part of the solar spectrum is full of dark lines, while
there are not nearly so many in the green. If we could
get rid of these, the violet rays would be considerably
more strengthened than the green, the Sun’s light would
appear blue, and the “color-temperature” would come out
higher. All things considered, we may adopt 6l)00o as
the surface temperature of the Sun, with reasonable as-
surance that we are within a few per cent, of the truth.
We must remember, however, that this is only a sort of
average temperature of the various layers in the sun’s