1 we need a movement-holder to hold the movement
as nearly a constant height as possible above the bench. The idea is,
when the clamp _A B_ is slipped on the pallet staff the index hand _A'_
will extend outward, as shown in Fig. 72, where the circle _C_ is
supposed to represent the top plate of a watch, and _A'_ the index hand.

HOW THE ANGULAR MOTION IS MEASURED.

[Illustration: Fig. 72]

Fig. 72 is supposed to be seen from above. It is evident that if we
remove the balance from the movement shown at _C_, leaving power on the
train, and with an oiling tool or hair broach move the lever back and
forth, the index hand _A'_ will show in a magnified manner the angular
motion of the lever. Now if we provide an index arc, as shown at _D_, we
can measure the extent of such motion from bank to bank.

[Illustration: Fig. 73]

[Illustration: Fig. 74]

To get up such an index arc we first make a stand as shown at _E F_,
Fig. 73. The arc _D_ is made to 11/2" radius, to agree with the index
hand _A'_, and is divided into twelve degree spaces, six each side of a
zero, as shown at Fig. 74, which is an enlarged view of the index _D_ in
Fig. 72. The index arc is attached to a short bit of wire extending down
into the support _E_, and made adjustable as to height by the set-screw
_l_. Let us suppose the index arc is adjusted to the index hand _A'_,
and we move the fork as suggested; you see the hand would show exactly
the arc passed through from bank to bank, and by moving the stand _E F_
we can arrange so the zero mark on the scale stands in the center of
such arc. This, of course, gives the angular motion from bank to bank.
As an experiment, let us close the bankings so they arrest the fork at
the instant the tooth drops from each pallet. If this arc is ten
degrees, the pallet action is as it should be with the majority of
modern watches.

TESTING LOCK AND DROP WITH OUR NEW DEVICE.

Let us try another experiment: We carefully move the fork away from the
bank, and if after the index hand has passed through one and a half
degrees the fork flies over, we know the lock is right. We repeat the
experiment from the opposite bank, and in the same manner determine if
the lock is right on the other pallets. You see we have now the means
of measuring not only the angular motion of the lever, but the angular
extent of the lock. At first glance one would say that if now we bring
the roller and fork action to coincide and act in unison