ON TIMEKEEPERS. 197 



error, by a construction in which the scape wheel only assists the pendulum in 

 raising the lever; but it depends on the degree of force applied, to determine 

 what part of the weight the scape wheel shall sustain; this scapement cannot, 

 therefore, by any means be considered as detached. It is, however, easy to 

 remove the defect of Mr. Cumming's scapement, if it can be called a defect, 

 by a method similar to that which Mr. Haley has applied to watches; each 

 tooth of the wheel being unlocked by the descent of the lever on the opposite 

 side, at the moment that It ceases to act on the pendulum, and remaining in- 

 active until the pendulum meets it. (Plate XVI. Fig. 206, 207.) 



The detents of the scapements of Mudge and Gumming are parts of the 

 pallet, but in the timekeepers now commonly made by Arnold, Earnshaw, 

 and others, the tooth is detained by a pallet or pin projecting from a lever, 

 the point of which is forced back by the balance, at the moment that the 

 pallet presents itself to another of the teeth. Mr. Arnold employs an 

 epicycloidal tooth, acting on a single point of the pallet; Mr. Earnshaw 

 makes .ji flat surface of the tooth first act on the point of the pallet, and then 

 the; point of the tooth on a flat surface of the pallet. In other respects there 

 is littledifFerence in these scapements ; and both the artists have been judged 

 worthy of a public reward for their success. (Plate XVI. Fig. 208, 209.) 



The last of the three principal objects which require the attention of the 

 watchmaker, is to employ a pendulum or balance of which the vibrations 

 are in their nature perfectly isochronous. For this purpose the weight of 

 the pendulum ought to move in a cycloidal arc, but the difficulty of pro- 

 ducing svich a motion in practice is much greater than the advantage deriv- 

 ed from it, and a circular vibration, confined to a small arc, is sufficiently 

 isochronous for all practical purposes. The error of such a vibration is nearly 

 proportional to the square of the arc described by the pendulum, and 

 amounts to a second and a half, in a day of 24 hours, for a single degree on 

 each side the point of rest; so that a pendulum keeping true time in an Sic 

 of three degrees, would gain 13^ seconds if the arc were very much con- 

 tracted or made cycloidal, and would lose 104^ seconds by having the vibra- 

 tion extended to an arc of four degrees. In order to avoid the friction 

 which would be occasioned by the motion of the pendulum on an axis, it is 

 usually suspended by a flexible spring, which is wholly free from friction. 



