Memoirs presented to the Cambridge philosophical society on the occasion of the jubilee of Sir George Gabriel Stokes, bart., Hon. LL. D., Hon. SC. D., Lucasian professor.

VI. The Periodogram of Mlagnetic Declination as obtained from the records
of the Greenwich Observatory during the years 1871-1895. By ARTHUR
SCHUSTER, F.R.S., Professoer of Physics at the Owens College, Manchester.
[Received 1899, Aug. 1.]
I. INTRODUCTION.
THE science of Meteorology deals with variable quantities which are subject to
continuous and apparently irregular changes. Irregularities in the strict sense of the
word do not however exist in nature; there is never absence of law, though often an
appearance of lawlessness caused by the effects of several interacting causes. Our efforts
rnust be directed to disentangle these causes, and to discover for that purpose the
hidden regularities of the phenomena.
If we look for instance at the curve which represents the barometric changes, we
see at once that though irregular, there is a tendency towards an average position,
large deviations from that position being less frequent than small ones. Prof. Karl
Pearson has investigated statistically the laws of deviation from the mean, and obtained
valuable and interesting results. But enquiries of this kind necessarily leave out of
account one of the most essential points in the phenomena they deal with, which is
the regularity which may exist in the succession of events. In taking the average daily
values of barometric pressure and studying their deviations from the mean, the same
importance is attached to an exceptionally high barometer when it follows another day
of high barometer, as when it follows one of low pressure. But a high pressure is more
likely to be followed by a high pressure than by a low one, and the regularity which
this succession implies seems to me to be of greater importance than the laws of
distribution based on the assumption that successive days are quite independent of each
other.
I intend in this paper to describe a method, applying it to a particular case, which
seems to me to yield some valuable information concerning the hidden regularities of
fluctuating changes, though it does not pretend to give a complete representation of ail
that it is important to know.
14-2