ON TIMEKEEPERS.

195

action through an equal space in the middle of the vibration. Yet a goodclock on this construction may keep time without an error ot the tenthousandth part of the whole, and a watch within a two thousandth. In thecommon watch scapement, there is little friction, for the force acts almostperpendicularly on the pallet; it appears to have been the oldest scapement,and was employed before the application of springs to balances: it requires aconsiderable extent of motion in the balance, and cannot therefore well beapplied to clocks with such pendulums as vibrate in small ares. The crutchscapement, on the contrary, cannot be applied immediately to a vibration ina very large arc; but by the interposition of a lever with a roller, or of a part°f a wheel with a pinion, it may be adapted to the balance of a watch; andsome watches thus constructed by Emery, Letherland, and others, appear tohave succeeded very well.

To avoid the inconveniences of the recoiling scapements, Mr. Graham in-v ented or introduced the dead beat for the clock, and the cylinder for the^ateh. In both of these, the tootlji of the scape wheel rests, during thegreater part of the vibration, on a cylindrical surface, and acts on the in-clined plane for a short time only, in the middle of each vibration; so that achange of the sustaining power scarcely produces a sensible derangement of*he isochronism; for which ever way we turn the key of a horizontal watch,long as it continues to go, the frequency of its vibrations is scarcely‘ejected. A good horizontal watch will keep time within about a tenthousandth part, especially if a little oil be frequently applied to it, or if thecylinder be made of a ruby: and the timekeeper in the observatory at Greeu-' v *ch, with a dead beat scapement, made by Graham, varies from true time0l % two parts in a million. (Plate XVI. Fig. 203, 204.)

Still, however, the friction of the teeth of the scape wheel on the cylinder0r pallet, and the tenacity of the oil, where it is employed, may interfere ma »Hght degree with the time of vibration, especially by the irregulant.es to' v hich they are liable. If the friction were perfectly uniform, it wouldscarcely disturb the isochronism, but friction is always increased by an in-case 0 f pressure; hence, therefore, the effect of any addition to the sus-sing force must tend in some degree to retard the vibrations; and to ob-vi:il c this, the surfaces, on which the teeth rest, have sometimes been so