rain.

[Illustration: FIG. 16.--Spring Remontoire.]

In all these gravity remontoires, however, only the friction of the
heavy parts of the train and the dial-work is got rid of, and the
scape-wheel is still subject to the friction of the remontoire wheels,
which, though much less than the other, is still something
considerable. Accordingly, attempts have frequently been made to drive
the scape-wheel by a spiral spring, like the mainspring of a watch.
One of these was described in the 7th edition of this encyclopaedia;
and Sir G. Airy invented another on the same principle, of which one
specimen is still going well. One of the best forms of such a
remontoire is shown in fig. 16, in which A, B, D, E, e, f are the same
things as in fig. 15. But e, the scape-wheel pinion, is no longer
fixed to the arbor, nor does it ride on the arbor, as had been the
case in all the previous spring remontoires, thereby producing
probably more friction than was saved in other respects; but it rides
on a stud k, which is set in the clock frame. On the face of the
pinion is a plate, of which the only use is to carry a pin h (and
consequently its shape is immaterial), and in front of the plate is
set a bush b, with a hole through it, of which half is occupied by the
end of the stud k, to which the bush is fixed by a small pin, and the
other half is the pivot-hole for the scape-wheel arbor. On the arbor
is set the remontoire spring s (a moderate-sized musical-box spring is
generally used), of which the outer end is bent into a loop to take
hold of the pin h. In fact, there are two pins at h, one a little
behind the other, to keep the coils of the spring from touching each
other. Now, it is evident that the spring may be wound up half or a
quarter of a turn at the proper intervals without taking the force off
the scape-wheel, and also without affecting it by any friction
whatever. When the scape-wheel turns in a minute, the letting-off
would be done as before described, by a couple of notches in the
scape-wheel arbor, through which the spikes A, B, as in fig. 15, would
pass alternately. During the half-minute that the spring is running
down the impulse on the pendulum constantly diminishes; but this error
is small if the spring be properly shaped, and besides, being
periodic, does not affect the _average_ time-keeping of the clock. It
would be inadmissible in astronomical c