t that it tends to deflect from the ideal or true position and to
bind. This condition is aggravated by the fact that the ring gear was
made by cutting its teeth on an angle to the axis around which it is to
revolve, using only a saw of appropriate width. The teeth were then
rounded-up to form by hand in a separate operation which by its very
nature means that the teeth are not exactly alike. This lack of
uniformity of the ring gear coupled with an entirely inadequate bearing
for the barrel contributes to rather erratic transfer of power. These
irregular teeth would not, of course, be a factor in factory-made watches
where suitable machinery would be available for the work.

[Illustration: Figure 2.--PATENT DRAWING OF THE HOPKINS WATCH. The
mainspring barrel _E_, of a very large diameter in proportion to the
diameter of the watch, occupies nearly the full diameter of the movement.
The spring itself, narrower and much longer than usual, is made in the
patent model by riveting two ordinary springs together end to end. Over
this barrel and attached to the stationary frame of the watch is placed a
large thin ring A, cut on its inner diameter with 120 teeth. Near its
edge the barrel E carries a stud _g_ on which runs a pinion of 10 in mesh
with the ring gear _A_. On this pinion is a wheel of 80 driving a pinion
of 6 on the escape-wheel arbor. The 15-tooth escape wheel locks on a
spring detent and gives impulse to the balance in one direction only,
being a conventional chronometer escapement. The intermediate wheel and
pinion, balance wheel, and balance cock have been adapted from a Swiss
bar movement of the time.]

The second fault is in the ratio between the time of one revolution and
the number of revolutions necessary for a day's run. Three turns of the
spring are, of course, required to run the watch for an hour, since the
barrel and train revolve three times in that length of time. If we choose
to have the watch run for 30 hours on a winding, and this leaves but a
small safety factor, then we see that this will require 90 turns of the
main spring, a manifest impossibility in view of the space
available.[5]

[Illustration: Figure 3.--ORIGINAL PATENT MODEL OF THE HOPKINS WATCH,
U. S. Patent 161513, July 20, 1875, now in the U. S. National Museum
(_cat. no._ 309025).]

[Illustration: Figure 4.--DRAWING FROM U. S. PATENT 165831, showing
Hopkins' first design improvement, an arbor for the barrel and train to
turn on and