tion to be made to learn its causes, and what can be done to arrest it and restore the trade.
Perhaps we ought yet to pause, before taking up the balance itself, to take note of a form of escapement employed in chronometers (using the word in its strict sense as applied to clocks which may be moved, but must not be jarred or tossed about, and are to remain in nearly the same position). In this escapement the tooth of the scape-wheel acts directly upon the balance, and after so acting is caught upon a shoulder, from which it is only released by the return of the balance. Better results have been obtained with this than with the detached lever for chronometers, but it is not so good for watches, as it will not bear violent and sudden tossing about into all positions.
To measure time accurately with a balance-wheel, three sources of variation must be overcome—that is, the balance must be adjusted to heat and cold, to position, and to "isochronism," as it is termed—that is, it must in some way counteract its own variation, due to temperature as well as that of the hair-spring; its bearings must be so cut or shaped that changes of position will not occasion unequal friction; it must be made to beat uniformly, in spite of the variations in power which result from doing away with the fusee.
The method of compensating a balance is to-day everywhere the same. A compensated balance consists of a bar of steel, to the opposite ends of which are attached semicircular bows of brass and steel soldered together, the brass outside, and into these bows are screwed what are termed "set-screws." It will be seen that, in case of a change for the warmer, the brass outer rim will expand more than the inner steel one, in each of the arms, and that this will throw the extremities of the arms in toward the center, thus compensating, if the proportions are right, for the general expansion of the balance; and vice versa in case of a change for the colder. Of course, it is a very nice piece of experimentation which ascertains these proportions. After the approximate proportions are secured, the exact ones are obtained by means of the screws in the rim. Compensation in a balance-wheel is of far more importance than in a pendulum, for the variation of the rate of the time-piece, if not compensated, is far greater. And this is due not only to the expansion and contraction of the balance, but also to the variation in the power of the hair-spring under various temperatures, as above remarked.
I may pause to note that here is one of the tests of a watch that any one may resort to. There are many imitation compensating balances which look very like the genuine, save that they have not the cut, or have only a notch at the extremity of the bow, so that the bow is not free, and, of course, there is no compensation.
After a watch has been adjusted to heat and cold, it must be adjusted to "position"—that is, so that its rate will not be altered by changes of position. This is a nice piece of work. It is accomplished
