to an escape funnel placed below the arm. Hence at
each double vibration of the pendulum part of the work done by a drop of
water falling through a short distance is communicated to the pendulum,
which is thus kept in motion as long as the water lasts. At this rate a
gallon of water ought to drive the clock for 40 hours. Care of course
must be taken to keep the water in the reservoir at a constant level, so
that the drops formed shall be uniform.

If it were worth while, no doubt the oscillations of a pendulum working
in a vacuum could be maintained by the communication and discharge at
each oscillation of a slight charge of electricity; or again, heat might
at each oscillation be communicated to a thermo-electric junction, and
the resulting current used to drive the pendulum.

The expansions and contractions of metal rods under the influence of the
changes of temperature which take place in the course of each night and
day have also been employed to keep a clock wound up, and if there were
any need for it no doubt a small windmill rotating at the top of a tower
would easily keep a turret clock fully wound, by a simple arrangement
which would gear the going barrel of the clock to the wind vane motion,
whenever the weight had fallen too low, and release it when the winding
up was completed. Even a smoke jack would do the same office for a
kitchen clock.

The methods of driving astronomical telescopes by means of clockwork
will be found in the article TELESCOPE. Measurements of small intervals
of time are performed by means of chronographs which in principle depend
on the use of isochronous vibrating tuning-forks in place of pendulums.
In practice it is needful in most cases that an observer should
intervene in time measurements, although perhaps by means of a revolving
photographic film a transit of the sun might be timed with extraordinary
accuracy. But if the transit of a star across a wire is to be observed,
there is no mode at present in use of doing so except by the use of the
human eye, brain and hand. Hence in all such observations there is an
element of personal error. Unfortunately we cannot apply a microscope to
time as we can to space and make the cycle of events that takes place in
a second last say for five minutes so as to time them truly. By personal
observations the divisions of a second cannot in general be made more
accurately than to 1/10 or 1/15 of a second. The most rapid music player
does not stri