When the mechanism is in the state shown, the balance-spring should be free from strain; but the balance-wheel is moving with its maxi mum,velocity in the direction (let it be sup posed) of the arrow, so that in another instant the discharging7pallet, H, will trip the detent by momentarily withdrawing the locking-pallet, D. The escape-wheel, A, is then free to rotate until again arrested by the locking-pallet,.D; but as it does so it engages the impulse-pallet, C, and accelerates the balance-wheel. After the escape-wheel has been arrested by the lock ing-pallet, the balance-wheel continues its ex cursion until the energy of its motion has all been expended in winding up the balance spring. The balance-spring then preponderates, and the wheel returns to the opposite extreme of its swing without disturbing the locking pallet, D; the discharging-pallet, H, merely raising ex light gold spring, E, as it passes the detent. In most chronometers the balance spring is made in the form of a helix instead of a flat spiral.
The balance-wheel is also compensated for change of temperature, though this, feature can hardly be said to constitute a point of differ ence between a chronometer and a watch, since practically all of the better grades of watches are similarly compensated. The effect of an increase of temperature upon the rate of a chronometer (or watch) is of a dual nature. If the instrument keeps correct time at one temperature, then, unless some mode of com pensation is provided, it will lose time at all higher temperatures, because (1) the strength of the balance-spring is less at lugher tempera tures, and (2) the wheel itself is greater in diameter on account of its natural expansion, so that it has a greater moment of inertia, and hence •responds more sluggishly to the action of the balance-spring. To counteract these effects, as far as possible, certain weights are provided along the rim of the wheel and the wheel is so designed that as the temperature rises these weights are automatically thrown nearer to the axis of rotation. The moment of inertia of the wheel is thereby reduced, and the wheel is caused to respond to the weakened spring with precisely the same readiness as before. The essential parts of the balance wheel of a marine chronometer are shown in Fig. 2. The rim is divided by transverse cuts into two semicircular segments, each of which is supported near one end and it is built up of two concentric metal strips,. of which the inner is• steel, while the outer-(which is.twice as thick) is of brass.. As the temperature rises the brass expands more than the steel, so that the rim-segments are deflected inward, and the weights, AA, are caused to approach the axis (or "staff") of the wheel. The two masses AA must always be opposite each other, ,in order that the balance may not be thrown out of poise; but they are slotted,so that they may be slid along the rim. If a chronometer keeps
correct time at one temperature, but loses at a higher temperature, the compensation is not sufficient, and the masses AA must be shifted toward .the free ends of the rim-segments. If it gains at the higher temperature it is over compensated, and the masses AA must be shifted toward the fixed ends of the rim-seg ments. In practice it is found to be impossible to adjust the masses so that the instrument shall be perfectly compensated at all tempera tures. In fact, the theory of the balance shows that there are only two temperatures at which the chronometer can be expected to keep cor rect time, though these two may be selected arbitrarily, and the instrument adjusted accord ingly. In marine chronometers the tempera tures so selected are usually and 90 F., and at intermediate temperatures the instru ments will gain. Hartnup, the director of the Liverpool Observatory, concluded, as the result of experiments upon 1,000 chronometers, that the error in rate from imperfect compensation may amount to 1.5 seconds per 24 hours at temperatures F. above or below either of the points at which the balance is standardized, when the balance-spring is of steel. With pal ladium springs the error is smaller. A more delicate adjustment has been devised, in the form of short arcs of steel attached to the shaft of the wheel at its ends. They are set parallel to the rim and a short distance from it, and at •their free extremities carry an ad justable screw. Two are placed on the outside of the long ends of the,cut•riin, and,these check the outward thrust of the rim, under the in fluence of cold, thus automatically shortening the expanding arm; The two placed inside of the rim act in a similar way when the tempera ture is high, shortening the contracting arias. The practical effort is to bring the extremes of temperature for which the chronometer is regu lated much closer together and nearer the centre of the scale to be covered, the tempera tures above and below these points being taken care of automatically. When the balance has been compensated for temperature as accurately as possible, the moment of inertia of the wheel is adjusted so that the chronometer will keep correct time at the standard temperatures, by means of the slotted nuts BB. These turn upon screw-threads that are secured to the rim near the central arm, where their positions will not be sensibly affected by changes of tempera ture. If they are caused to approach the rim, the moment of inertia of the wheel is lessened, and the chronometer runs faster; and vice versa. The slots in BB are provided so that the nuts may g 1 a,p their :rews spring-tight, and so avoid backlash. Col suit Britten, F. J., Watch and ("1,,ckmakers' Handbook' (New York 1899) ; Fritts, C. E, The Watch Ad justers' Manual) (Philadelphia 1912).