The greater regularity of this form of escapement is due to a smaller oscillation of the pendulum being required than with the form first described; for it is found that the motion of a pendulum when vibrating through not more than six degrees is practically cycloidal, and it is only with larger arcs that the circle materially differs from the theoretical curve required. The pendulum is kept in vibration by the escape wheel, or rather by its teeth pressing against the inclined surfaces of the pallets, and forcing them outwards, and so giving the pendulum an impulse prior to each tick. [Illustration: FIG. 92.—The Anchor Escapement. ] This anchor escapement, which was invented by “Clement, of London, clockmaker,” forms, as it were, the basis of our modern clocks, and, with the exception of the dead beat, which was due to Graham some years afterwards, is in almost exclusive use at the present date. We see that as soon as a tooth has escaped on one side, a tooth on the other begins immediately to retard the action of the pendulum by pressing against the inclined surface of the other pallet, and as the pendulum swings on, the tooth gives way, and the motion of the wheel is reversed; then when the pendulum begins to return, it is assisted again by the tooth, so that the pendulum is always under the influence of the escape wheel, some times accelerated, and sometimes retarded. The principle of Graham’s dead beat is to get rid of the retarding action of the escape wheel, so that there should be no necessity for so much accelerating power, and the pendulum should be out of the influence of the escape wheel during a large portion of its vibration. This he accomplished by doing away with a large portion of the inclined surface of the pallet (Fig. 93), so that the teeth have no accelerating action on the pendulum until just as they leave the ends of the pallets where they are inclined; the greater portion of both the pallets on which the escape wheel works being at right angles to its direction of motion, the teeth have no tendency to force the pallet outwards. In Fig. 93 the tooth V has fallen on the pallet D, the tooth T having just been released, and as the pendulum still swings on in the direction of the arrow, the pallet D will be pushed further under the tooth C but without pressing the wheel backwards, and without retardation other than that of friction. When the pendulum returns and the pallet just gets past the position shown, it gets an impulse, and this is given as nearly as possible as much before the pendulum reaches its vertical position as after it passes it, its action is therefore neither to increase nor diminish the rate. In this escapement not only is the arc of oscillation considerably lessened and the motion of the pendulum brought near to the cycloidal form, but in addition to this there is this important point, that the weight is acting upon the pendulum for the least possible time. [Illustration: