rches, but
also to the labor involved. Instead of a single measure for each star,
in the case of the so-called spectroscopic binaries, we must make enough
measures to determine the dimensions of the orbit, its form and the
period of revolution.

What has been said of the motions of the stars applies also, in general,
to the determination of their distances. A vast amount of labor has been
expended on this problem. When at length the distance of a single star
was finally determined, the quantity to be measured was so small as to
be nearly concealed by the unavoidable errors of measurement. The
parallax, or one half of the change in the apparent position of the
stars as the earth moves around the sun, has its largest value for the
nearest stars. No case has yet been found in which this quantity is as
large as a foot rule seen at a distance of fifty miles, and for
comparatively few stars is it certainly appreciable. An extraordinary
degree of precision has been attained in recent measures of this
quantity, but for a really satisfactory solution of this problem, we
must probably devise some new method, like the use of the spectroscope
for determining motions. Two or three illustrations of the kind of
methods which might be used to solve this problem may be of interest.
There are certain indications of the presence of a selective absorbing
medium in space. That is, a medium like red glass, for instance, which
would cut off the blue light more than the red light. Such a medium
would render the blue end of the spectrum of a distant star much
fainter, as compared with the red end, than in the case of a near star.
A measure of the relative intensity of the two rays would servo to
measure the distance, or thickness of the absorbing medium. The effect
would be the same for all stars of the same class of spectrum. It could
be tested by the stars forming a cluster, like the Pleiades, which are
doubtless all at nearly the same distance from us. The spectra of stars
of the tenth magnitude, or fainter, can be photographed well enough to
be measured in this way, so that the relative distances of nearly a
million stars could be thus determined.

Another method which would have a more limited application, would depend
on the velocity of light. It has been maintained that the velocity of
light in space is not the same for different colors. Certain stars,
called Algol stars, vary in light at regular intervals when partially
eclipsed by th