the magnetic observatories can be, and have been, appealed to for
information, and they indicate that the facts actually accord with the
theory. Behind the veil of sunlight in the middle of the afternoon,
there is good reason to believe, auroral exhibitions often take place
which would eclipse in magnificence those seen at night if we could
behold them. Observation shows, too, that auroræ are more frequent
before than after midnight, which is just what we should expect if
they originate in the way that Arrhenius supposes. Second, the theory
offers an explanation of the alleged fact that the formation of clouds
in the upper air is more frequent in years when auroræ are most
abundant, because clouds are the result of the condensation of
moisture upon floating particles in the atmosphere (in an absolutely
dustless atmosphere there would be no clouds), and it has been proved
that negative ions like those supposed to come from the sun play a
master part in the phenomena of cloud formation.

Yet another singular fact, almost mystical in its suggestions, may be
mentioned. It seems that the dance of the auroral lights occurs most
frequently during the absence of the moon from the hemisphere in which
they appear, and that they flee, in greater part, to the opposite
hemisphere when the moon's revolution in an orbit considerably
inclined to the earth's equator brings her into that where they have
been performing. Arrhenius himself discovered this curious relation of
auroral frequency to the position of the moon north or south of the
equator, and he explains it in this way. The moon, like the earth, is
exposed to the influx of the ions from the sun; but having no
atmosphere, or almost none, to interfere with them, they descend
directly upon her surface and charge her with an electric negative
potential to a very high degree. In consequence of this she affects
the electric state of the upper parts of the earth's atmosphere where