opposite. The lever is knocked downwards, and the tooth, which had been arrested by the locking-face a of pallet P, now presses on the impulse face b, forcing the left end of the lever up. The impulse pin I receives a blow, assisting the unwinding of the spring, and C again locks the lever. The same thing is repeated in alternate directions over and over again.
COMPENSATING BALANCE-WHEELS.
The watchmaker has had to overcome the same difficulty as the clockmaker with regard to the expansion of the metal in the controlling agent. When a metal wheel is heated its spokes lengthen, and the rim recedes from the centre. Now, let us suppose that we have two rods of equal weight, one three feet long, the other six feet long. To an end of each we fasten a 2-lb. weight. We shall find it much easier to wave the shorter rod backwards and forwards quickly than the other. Why? Because the weight of the longer rod has more leverage over the hand than has that of the shorter rod. Similarly, if, while the mass of the rim of a wheel remains constant, the length of the spokes varies, the effort needed to rotate the wheel to and fro at a constant rate must vary also. Graham got over the difficulty with a rod by means of the compensating pendulum. Thomas Earnshaw mastered it in wheels by means of the compensating balance, using the same principle—namely, the unequal expansion of different metals. Any one who owns a compensated watch will see, on stopping the tiny fly-wheel, that it has two spokes (Fig. 206), each carrying an almost complete semicircle of rim attached to it. A close examination shows that the rim is compounded of an outer