was made up of bodies so vastly distant and so completely
isolated that it was difficult to conceive of them as standing in any
definable relation to one another. It is true that they all emitted
light, else we could not see them, and the theory of gravitation, if
extended to such distances, a fact not then proved, showed that they
acted on one another by their mutual gravitation. But this was all.
Leaving out light and gravitation, the universe was still, in the time
of Herschel, composed of bodies which, for the most part, could not
stand in any known relation one to the other.

When, forty years ago, the spectroscope was applied to analyze the
light coming from the stars, a field was opened not less fruitful than
that which the telescope made known to Galileo. The first conclusion
reached was that the sun was composed almost entirely of the same
elements that existed upon the earth. Yet, as the bodies of our solar
system were evidently closely related, this was not remarkable. But
very soon the same conclusion was, to a limited extent, extended to the
fixed stars in general. Such elements as iron, hydrogen, and calcium
were found not to belong merely to our earth, but to form important
constituents of the whole universe. We can conceive of no reason why,
out of the infinite number of combinations which might make up a
spectrum, there should not be a separate kind of matter for each
combination. So far as we know, the elements might merge into one
another by insensible gradations. It is, therefore, a remarkable and
suggestive fact when we find that the elements which make up bodies so
widely separate that we can hardly imagine them having anything in
common, should be so much the same.

In recent times what we may regard as a new branch of astronomical
science is being developed, showing a tendency towards unity of
structure throughout the whole domain of the stars. This is what we now
call the science of stellar statistics. The very conception of such a
science might almost appall us by its immensity. The widest statistical
field in other branches of research is that occupied by sociology.
Every country has its census, in which the individual inhabitants are
classified on the largest scale and the combination of these statistics
for different countries may be said to include all the interest of the
human race within its scope. Yet this field is necessarily confined to
the surface of our planet. In the field of stel