One tooth of the wheel in the present position rests on the inclined surface of the inner part of the pallet c, upon which its disposition to slide tends to throw the point of the pallet further from the centre of the wheel, and consequently assists the vibration in that direction. While the pallet c moves outwards, and the wheel advances, the point of the pallet d, of course, approaches towards the centre, in the opening between the two nearest teeth; and when the acting tooth of the wheel slips off', or escapes from the pallet c, another tooth on the opposite side immediately falls on the exterior inclined face of d, and by a similar operation tends to push that pallet from the centre. The returning vibration is thus assisted by the wheel, while the pallet moves towards the centre, and receives the succeeding tooth of the wheel after the escape from the point of d. In this manner the alteration may be conceived to go on without limit. The celebrated George Graham improved this escapement very much, by taking off part of the slope furthest from the points of the pallets; instead of which part he formed a circular or cylindrical face, having its axis in the centre of motion.

Pallets of this kind are seen on the opposite side of the wheel at e and g, having h for their centre or axis. A tooth of the wheel is seen resting upon the circular inner surface of the pallet g, which is not therefore affected by the wheel, excepting so far as its motion, arising from any other cause, may be affected by the friction of the tooth. If the vibration of the pendulum be supposed to carry g outwards, the sloped surface will be brought to the point of the tooth, which will slide along it and urge the pallet outwards during this sliding action. When the tooth has fallen from the point of this pallet, an opposite tooth will be received on the circular surface of e, and will not affect the vibration, excepting when the slope surface of e is carried out so as to suffer the tooth to slide along it. In the two former escapements, there is always a certain portion of vibration takes place after the drop which drives the pallet back, and causes the index also to recede through a small arc: this has been distinguished by the name of a recoil. The escapement of Graham, and all such as have no recoil, have been called dead beat escapements, because the index for seconds falls directly through its arc, and remains motionless on the line of division till the next vibration.

It may be observed, that the maintaining power in Graham's escapement, may be applied during a small portion only of the vibration; and that an increase of the maintaining power tends to enlarge the arc of vibration, but scarcely interferes with its velocity. In the escapements just described, the escape wheel is in continual contact with the pallets belonging to the axis of the balance, and the friction arising from this circumstance may be considered as a principal cause of the irregularity in the going of watches. If we suppose a regulator to be made so perfect as to be perfectly isochronal, while vibrating in a free position, that advantage would be diminished, or lost, as soon as it was placed in connexion with a train of wheels; and the errors would be more or less, according to the nature and quantity of friction in the escapement. It would be, therefore, extremely useful to secure to the regulator a perfect liberty of vibration, except during the short intervals of time which may be necessary for the action of the escape wheel, to give it a new impulse. This ingenious idea was first started, and also carried into execution by P.

Leroy, who in 1748 presented to the Academy of Sciences in Paris a model of a detached escapement, the effect or action of which may be briefly described as follows: an escape wheel is kept in repose by a lever detent; the balance unlocks the detent, and receives an impulse or stroke on a pallet through a part of every second vibration, and during great part of its course it is free and detached. A great variety of escapements have been contrived by various ingenious men; those in which springs are used in the locking pieces instead of pivots are at present generally preferred; we shall, therefore, proceed to describe the escapement of Mr. Arnold, who we believe is the author of this improvement. The engraving on the next page is a representation of this escapement. The teeth of the escape wheel are of a cycloidal shape, in the face Dart, which is intended for action, the sections of which, with those of the two other sides, form a sort of mixed triangle, b b d represents the detent, which is formed of a .flexible piece or spring, bending between c and n; and in the part n b d, which is stronger than the other, is fixed the locking pallet a, opposite an adjusting screw f.

The pallet projecting below the spring detent locks upon the interior angle of the tooth, suspending the motion of the escape wheel, and leaving the balance to vibrate free, as pointed out in the preceding escapements. The action of the spring detent (for the joint of the detent is itself a spring) presses the locking pallet against the screw f, except at the time of unlocking the wheel. A very delicate spring n e, called the discharging, or unlocking spring, (and also the tender spring,) is attached by one end n to the spring detent c b n b a; and passing under the adjusting screw f, extends a little beyond the extremity d, by the detent itself; h h h is a circular piece attached to the axis of the balance, and o the discharging pallet.

Fig. 1.

Fig. 2.

This pallet when the balance is in motion from e to d presses against the end of the discharging spring n e; and carrying it together with the locking spring b b d disengages the locking piece a out of the internal angle of the tooth, with which it was in contact; and the escape wheel then communicates a new power to the balance by its impulse on a pallet m, which is fixed or set in the aperture of the circular piece. As soon as this is done, the spring detent, or locking spring, falls back to its position against the adjusting screw f; and the pallet, by receiving or intercepting the next tooth, stops the motion of the escape wheel. When the balance returns from d to e, the unlocking pallet acts again on the extremity of the discharging spring, but this being very delicate, gives way without disturbing the detent, or locking spring; and the balance, after suffering a trifling degree of resistance by that contact, continues its free vibrations. At the next vibration the unlocking takes place, and the action of the escapement proceeds successively, as explained before.