200 LECTURE XVir. 



Stances. He employed, for a pendulum, a tube partly filled with mercury ; 

 when the tube expanded by the effect of heat, the mercury expanded much 

 more; so that its surface rose ai. little more than the end of the pendulum was 

 depressed, and the centre of oscillation remained stationary. This mode of 

 compensation is still sometimes practised with success; but the gridiron pen- 

 dulum is more commonly used: it was the invention of Harrison, who com- 

 bined seven bars, of iron or steel, and of brass, in such a manner, that tlie bars of 

 brass raised the weight as much as the bars of iron depressed it. At present 

 five bars only are usually employed, two of them being of a mixture of zinc 

 and silver, and three of steel. Mr. Ellicott suspended a pendulum at the ex- 

 tremity of a lever, which was supported by a pillar of brass, much nearer 

 to the fulcrum ; as the pendulum expanded, the end of the lever was raised 

 in the same degree, and the weight remained at its original distance from 

 the point of suspension, which was determined by a fixed plate, transmitting 

 the slender spring, as usual, between two opposite edges. The same efl'ect is 

 produced more simply by suspending the pendulum from the summit of a bar 

 nearly parallel to it, and of the same substance with itself, resting on a fix- 

 ed support, and either of the same length with the pendulum, or a little 

 longer, accordingly as the distance of the fixed plate from the point of sup- 

 port of the bar, is determined by materials which may be considered as 

 nearly of an invariable length, or as liable to a certain degree of expansion. 

 (Plate XVI. Fig. 210.) 



All these methods of compensation are peculiar to clocks ; for watches, it 

 is usual to unite together two metals which differ in. expansibility, so, as to 

 form a compound plate; one side of the plate is commonly of steel, the other 

 of brass, and it is obvious that any increase of temperature, by causing the 

 brass to expand more than the steel, must bend the Avhole plate. Such a 

 plate is variously applied ; the most accurate method, which is employed by 

 Arnold and other modern artists, is to make it a part of the balance itselfj 

 fixing a weight on its extremity, which is brought nearer to the centre, by 

 the increase of curvature of the plate, whenever the expansion of the arms of 

 the balance tends to remove it further off. The best way of making the 

 plate appears to be to turn a ring of steel, and to immerse it in melted brass, 

 and then to turn away what is superfluous of the brass. The magnitude of the 

 weight, and the length of the plate, may easily be so regulated, as to com- 



