MAGNETIC CURVES FOR APRIL 1841-1860]

[Sidenote: Illustration of spurious connection.]

But I may perhaps repeat the word of caution already uttered against
extending without sufficient evidence this notion of the influence of
sun-spots to other phenomena, such as weather. A simple illustration will
perhaps serve better than a long argument to show both the way in which
mistakes have been made and the way in which they can be seen to be
mistakes. There is at the Royal Observatory at Greenwich an instrument for
noting the direction of the wind, the essential part being an ordinary
wind-vane, the movements of which are automatically recorded on a sheet of
paper. As the wind shifts from north to east the pencil moves in one
direction, and when it shifts back again towards the north the pencil
moves in the reverse way. But sometimes the wind shifts continuously from
north to east, south, west, and back to north again, the vane making a
complete revolution; and this causes the pencil to move continuously in
one direction, until when the vane has come to north again, the pencil is
far away from the convenient place of record; on such occasions it is
often necessary to replace it by hand. Then again, the vane may turn in
the opposite direction, sending the pencil inconveniently to the other
side of the record. During the year it is easy to count the number of
complete changes of wind in either direction, and subtracting one number
from the other, we get the excess of complete revolutions of the vane in
one direction over that in the other. Now if these rather arbitrary
numbers are set down year by year, or plotted in the shape of a diagram,
we get a curve which may be compared with the sun-spot curve, and during a
period of no less than sixteen years--from 1858 to 1874--there was a
remarkable similarity between the two diagrams. From this evidence _alone_
it might fairly be inferred that the sun-spots had some curious effect
upon the weather at Greenwich, traceable in this extraordinary way in the
changes of the wind. But the particular way in which these changes are
recorded is so arbitrary that we should naturally feel surprise if there
was a real connection between the two phenomena; and fortunately there
were other records preceding these years and following them which enabled
us to test the connection further, and it was found, as we might naturally
expect, that it was not confirmed. On looking at diagrams (Plate XV.) for