The Light of the Stars 43
equal energy, making but a relatively weak impression) so
that our visual magnitudes are substantially measurements
of the stars with yellow-green light. The ordinary plate on
the other hand is influenced almost entirely by violet light.
It is, therefore, not surprising that the stellar magnitudes
determined in the two ways often disagree. Red stars pho-
tograph relatively faint, and bluish ones bright. This has
been known since the first stellar photographs were made,
but its full importance was recognized much more recently.
In this difference between the visual and photographic
magnitudes we have a way of measuring the color of a
star’s light—for which reason it is called the color-index.
The results are in excellent agreement with the direct esti-
mates of color by visual observation, but have two great
advantages—first, that we can assign to them a definite
physical meaning, and second, that they are more trust-
worthy, for the eye is a rather poor judge of color in
objects as faint as the stars. Indeed, by using an isochro-
matic plate, with a “color screen” to cut off the violet light
and let through the green and yellow, we may get “photo-
visual” magnitudes which are substantially equivalent to
the usual determinations, and in some cases better—notably
for very faint stars, where the behavior of the eye has
annoying peculiarities. By either method we find that sun-
light is near the middle of the range of starlight colors.
Stars like Vega, or, even more, those in the belt of Orion,
are much bluer, while those like Arcturus, and especially
Antares, are much redder. But the colors of the stars lie
between definite limits, which may be described as cor-
responding to the emission of three times as much violet
light, in proportion to the yellow, as in the Sun’s case, and
again one-third as much violet light. Between these limits
all shades of color are found in nearly equal abundance;