*Project Gutenberg*. Public domain.

We erect from the point *a* a perpendicular to
the line *A a*, and, as previously explained,
establish the pallet center at *B*. Inasmuch
as we are to employ circular pallets, we lay off to
the left on the arc *m*, from the point *a*,
five degrees, said five degrees being half of the
angular motion of the escape wheel utilized in the
present drawing, and thus establish the point *c*,
and from *A* as a center draw through this point
the radial line *A c’*. To the right
of the point *a* we lay off five degrees and
establish the point *d*. To illustrate the
underlying principle of our circular pallets:
with one leg of the dividers set at *B* we sweep
through the points *c a d* the arcs *c’’
a’’ d’’*.

From *B* as a center, we continue the line *B
a* to *f*, and with the dividers set at five
inches, sweep the short arc *e e*. From the
intersection of this arc with the line *B f* we
lay off one and a half degrees and draw the line *B
g*, which establishes the extent of the lock on
the entrance pallet. It will be noticed the linear
extent of the locking face of the entrance pallet
is greater than that of the exit, although both represent
an angle of one and a half degrees. Really, in
practice, this discrepancy is of little importance,
as the same side-shake in banking would secure safety
in either case.

[Illustration: Fig. 19]

The fault we previously pointed out, of the generally
accepted method of delineating a detached lever escapement,
is not as conspicuous here as it is where the pallets
are drawn with equidistant locking faces; that is,
the inner angle of the entrance pallet (shown at *s*)
does not have to be carried down on the arc *d’*
as far to insure a continuous pallet action of ten
degrees, as with the pallets with equidistant locking
faces. Still, even here we have carried the angle
*s* down about half a degree on the arc *d’*,
to secure a safe lock on the exit pallet.

If we study the large drawing, where we delineate
the escape wheel ten inches in diameter, it will readily
be seen that although we claim one and a half degrees
lock, we really have only about one degree, inasmuch
as the curve of the peripheral line *m* diverges
from the line *B f*, and, as a consequence, the
absolute lock of the tooth *C* on the locking
face of the entrance pallet *E* is but about one
degree. Under these conditions, if we did not
extend the outer angle of the exit pallet at *t*
down to the peripheral line *m*, we would scarcely
secure one-half a degree of lock. This is true
of both pallets. We must carry the pallet angles
at *r s n t* down on the circles *c’’
d’* if we would secure the lock and impulse
we claim; that is, one and a half degrees lock and
eight and a half degrees impulse.