ter spring remontoires the pendulum, as it goes by, flips a
delicate spring and releases a small weight or spring which has been
wound up in readiness by the action of the scape-wheel and which by
leaping on to the pendulum gives it a push. One on this principle made
about the middle of the 19th century by Robert Houdin is to be seen at
the Conservatoire des Arts et Metiers. It is very complicated. The
following is more simple. In fig. 21 a scape-wheel AB has 30 pins and
360 teeth. It is engaged with a fly vane EP mounted on a pinion of 12
teeth. Each pin as it passes raises an impulse arm CD which is hooked
upon a detent K. A pall NM then engages the fly vane and prevents the
scape-wheel from moving farther. The impulse arm being now set, as the
plate F attached to the lower end of the pendulum flies past from left
to right a pall G knocks aside the detent K, and allows a pin O
projecting from the end of the impulse arm to fall upon an inclined
pallet h, which is thus urged forward. As soon as the pallet has left
the pin, the impulse arm in its further fall strikes N, which
disengages the pall at P and allows the scape-wheel to move on and
again wind up the impulse arm CD, which is then again locked by the
detent K. On the return journey of the pendulum the light pall G,
which acts the part of a chronometer spring, flips over the detent.
The pallet is double sided, h and h', so that if by chance the clock
runs down while the pendulum swings from left to right the impulse arm
will be simply raised and not smashed. It has a flat apex, on which
the pin falls before descending. The impulse given depends on the
weight of the impulse arm and may be varied at pleasure. The work done
in unlocking the detent is invariable, as it depends on the pressure
of the fly vane at P and is independent of the clock-train. The
duration of the impulse is very short--only about 1/10 of the arc of
swing. It is given exactly at the centre of the swing, and when not
under impulse the pendulum is detached.

_Clock Wheels._--Since, as we have seen, any increase in the arc of a
pendulum is accompanied by a change in its going rate, it is very
desirable to keep the force which acts on the pendulum uniform. This in
fact is the great object of the best escapements. Inasmuch as the
impulse on the pendulum, derived from the work done by a falling weight
or an unwinding spring, is transmitted thro