t if we make it to occupy about half more angular extent on the
circle _y_ than will coincide with the angular extent of the jewel pin,
it will be perfectly safe and effectual. We previously stated that the
jewel pin should occupy about twelve degrees of angular extent on the
circle _c_, and if we make the passing hollow occupy eighteen degrees
(which is one and a half the angular extent of the jewel pin) it will do
nicely. But if we should extend the width of the passing hollow to
twenty-four degrees it would do no harm, as the jewel pin would be well
inside the horn of the fork before the guard pin could enter the passing
hollow.

[Illustration: Fig. 63]

We show in Fig. 61 the fork as separated from the roller, but in Fig.
62, which is a side view, we show the fork and jewel pin as engaged.
When drawing a fork and roller action it is safe to show the guard pin
as if in actual contact with the roller. Then in actual construction, if
the parts are made to measure and agree with the drawing in the gray,
that is, before polishing, the process of polishing will reduce the
convex edge of the roller enough to free it.

It is evident if thought is given to the matter, that if the guard pin
is entirely free and does not touch the roller in any position, a
condition and relation of parts exist which is all we can desire. We are
aware that it is usual to give a considerable latitude in this respect
even by makers, and allow a good bit of side shake to the lever, but our
judgment would condemn the practice, especially in high-grade watches.

RESTRICT THE FRICTIONAL SURFACES.

Grossmann, in his essay on the detached lever escapement, adopts one and
a half degrees lock. Now, we think that one degree is ample; and we are
sure that every workman experienced in the construction of the finer
watches will agree with us in the assertion that we should in all
instances seek to reduce the extent of all frictional surfaces, no
matter how well jeweled. Acting under such advice, if we can reduce the
surface friction on the lock from one and a half degrees to one degree
or, better, to three-fourths of a degree, it is surely wise policy to do
so. And as regards the extent of angular motion of the lever, if we
reduce this to six degrees, exclusive of the lock, we would undoubtedly
obtain better results in timing.

We shall next consider the effects of opening the bankings too wide, and
follow with various conditions which are sure to co