f course, in accordance with the principles
already explained, have a tendency to make their vibrations in shorter
periods; and those which are furthest, in longer periods. But all
these particles are bound together firmly by the power of cohesion,
and must move connectedly. They, therefore, come to an agreement to
move at a mean rate--that is, between the two extremes. The top
particles hurry on the middle ones; the bottom particles retard them
in a like degree. Consequently, the whole of the weight moves as if
its entire mass were concentered in the position of those middle
particles; and the exact place of this central position in relation to
the point of suspension, becomes the important condition which
determines the time in which the instrument swings.

In pendulums of ordinary construction, this relation is by no means an
unvarying one--changes of temperature alter the bulk of all kinds of
bodies. A metal rod runs up and down under increase and diminution of
heat, as certainly as the thread of mercury in the tube of the
thermometer does. A hot day, therefore, lengthens the metallic
suspending-rod of a pendulum, and carries the centre of its weight to
a greater distance from the point of suspension. By this means, the
period of each vibration is of necessity lengthened. An increase of
temperature to the extent of ten of Fahrenheit's degrees, will make a
second's pendulum with a brass rod lose five vibrations in a day. All
substances do not, however, suffer the same amount of expansion under
like increments of heat. If the rod of the pendulum be made of
varnished or black-leaded wood, an addition of ten degrees of heat
will not cause it to lose more than one vibration in a day. But even
this small irregularity is too vast for the purposes of precise
science, and accordingly ingenuity has been taxed to the utmost to
find some means of removing the source of inaccuracy, to invent some
plan whereby the pendulum may be made sensitive enough to discover and
correct its own varying dimensions as different temperatures are
brought to bear upon its material.

The first successful attempt to accomplish this useful purpose was
made by George Graham in 1715. He replaced the solid weight at the
bottom of the rod by a glass jar containing mercury. The rod he
formed of steel of the usual length; and because mercury expands five
times more than steel, he fixed the height of the column of mercury in
the jar at only 6-1/2 inches.