ounts for the sun's gravitational
power. Indeed, its attraction cannot be felt by the most delicate
instruments at the distance of the earth, and would still be unknown
were it not for the influence of magnetism on light.

Auroras, magnetic storms, and such electric currents as those that
recently deranged several Atlantic cables are due, not to the magnetism
of the sun or its spots, but probably to streams of electrons, shot
out from highly disturbed areas of the solar surface surrounding
great sun-spots, traversing ninety-three million miles of the ether
of space, and penetrating deep into the earth's atmosphere. These
striking phenomena lead us into another chapter of physics, which
limitations of space forbid us to pursue.

STELLAR CHEMISTRY

Let us turn again to chemistry, and see where experiments performed
in cosmic laboratories can serve as a guide to the investigator.
A spinning solar tornado, incomparably greater in scale than the
devastating whirlwinds that so often cut narrow paths of destruction
through town and country in the Middle West, gradually gives rise
to a sun-spot. The expansion produced by the centrifugal force at
the centre of the storm cools the intensely hot gases of the solar
atmosphere to a point where chemical union can occur. Titanium
and oxygen, too hot to combine in most regions of the sun, join
to form the vapor of titanium oxide, characterized in the sunspot
spectrum by fluted bands, made up of hundreds of regularly spaced
lines. Similarly magnesium and hydrogen combine as magnesium hydride
and calcium and hydrogen form calcium hydride. None of these compounds,
stable at the high temperatures of sun-spots, has been much studied
in the laboratory. The regions in which they exist, though cooler
than the general atmosphere of the sun, are at temperatures of
several thousand degrees, attained in our laboratories only with
the aid of such devices as powerful electric furnaces.

[Illustration: Fig. 34. Splitting of spectrum lines by a magnetic
field (Babcock).

The upper and lower strips show lines in the spectrum of chromium,
observed without a magnetic field. When subjected to the influence
of magnetism, these single lines are split into several components.
Thus the first line on the right is resolved by the field into
three components, one of which (plane polarized) appears in the
second strip, while the other two, which are polarized in a plane
at right angles to that of the middl