centres EC, of the wheel and pallets; or rather C is the top of the pendulum spring to which the pallets CS, CS converge, though their actual action are a little below C. It is not worth while to crank them as Mr Bloxam did, in order to make them coincide exactly with the top of the pendulum, as the friction of the beat pins on the pendulum at P is in significant, and even then would not be quite destroyed. The pallets are not in the same plane, but one is behind and the other in front of the wheel, with one stop pointing backwards and the other forwards to receive the teeth alternately, it does not matter which; in this figure the stop S is behind and the stop S forward. The pendulum is now going to the right, and just beginning to lift the right pallet and free the stop S; then the wheel will begin to turn and lift the other pallet by one of the pins which is now lowest, and which moves through 45 across the line of centres, and therefore lifts with very little friction. It goes on till the tooth now below S reaches S and is stopped there. Meanwhile the pallet CS goes on with the pendulum as far as it may go, to the end of the arc which we have through out called a, starting from 7; but it falls with the pendulum again, not only to 7 but to-7 on the other side of 0, so that the impulse is due to the weight of each pallet alternately falling through 2y; and the magnitude of the impulse also depends on the obliqueness of the pallet on the whole, i.e., on the distance of its centre of gravity from the vertical through C. The defect of the original three-legged escapement was that the pallets were too nearly vertical.
Fig. 12. Four-Legged Gravity Escapement.
Another most material element of these escapements with very few teeth is that they admit of a fly KK on the scape- wheel arbor to moderate its velocity, which both obviates all risk of tripping, wholly or partially, and also prevents the bang which goes all through the clock where there is no ny. The fly is set on with a friction spring like the common striking-part fly, and should be as long as there is room for, length being much more effective than width. For this purpose the second wheel arbor is shortened and set in a cock fixed on the front plate of the clock, which leaves room for a fly with vanes 2 inches long. The back pivot of the scape-wheel is carried by a long cock behind the back plate, so that the escapement is entirely behind it, close to the pendulum. The pallet arbors are short, as they come just behind the centre wheel, which is here also necessarily above the escapement, and the great wheel arbor on a level with it, and at the left hand (from the front) or the string would be in the way of the fly. No beat screws are required, as the pallets end in mere wires which are easily bent. It is found better to make the tails of the pallets long, rather than short as Mr Bloxam did. It is essential, too, that the angle CSE formed by the tooth and the pallet which is struck upwards should not the least fall short of a right angle, nor the other angle CS E be the least obtuse, or the escapement may very likely trip. Practically, therefore, it is safer to let CSE be just greater than 90 and CS E a little less, so that there may not be the least tendency in the blow on the stops to drive the pallets outwards. For the purpose of calculation, however, we must make them both 90 and then it follows that, calling the length of the teeth r, and the distance of centres d, and the length of the pallets from C down to the stops p, r must d sin. 22^ J and _p - d cos. 22 J D . Therefore if r is made 2 inches CE or d will bo 5 22, say 5j inches, and p ? 4 "82. The distance of the lifting pins from the centre will be
Gravity escapements require more weight than a direct impulse escapement with an equally fine train; and they try the accuracy of the wheelcutting more severely. If there is a weak place in the train of a common clock the scape-wheel only follows the pendulum more weakly; but in a gravity escapement it always has to raise the pallets, and ought to raise them quickly, and especially in clocks for astronomical purposes where you take its exact time from the sound of the beats, and so the lifting must not lag and sound uneven. Therefore although a fine train of high numbers is not requisite it must be perfectly well cut. And as the force of the weight does not reach the pendulum its increase is of no consequence, within reasonable limits. It is worth while to put large friction wheels under the arbor of the great wheel in all astronomical clocks, and it makes a material difference in the friction on account of the necessary thickness of the winding arbor. A variation of arc in dead escapement clocks is sometimes visible between the beginning and the end of the week according as the string is nearest to the thick or the thin end of the great arbor, when there are no friction wheels.
Fig. 13.—Double Three-legged Escapement.
The other form of the gravity escapement, which is now adopted for large clocks by all the best makers, having been first used in the great Westminster clock, is the double three-legged which is shown in fig. 13. The principle of it is the same as of the four-legs; but instead of the pallets being one behind and the other in front of the wheel, with two sets of lifting pins, there are two wheels ABC, abc, with the three lifting pins and the two pallets between them like a lantern pinion. One stop B points forward and the other A backward. The two wheels have their teeth set inter mediately or 60 apart, though that is not essential, and the angle of 120 3 may be divided between them in any other proportions, as 70 and 50, and in that way the pallets may be still more oblique than 30 from the vertical, which however is found enough to prevent tripping even if the fly gets loose, which is more likely to happen from carelessness in large clocks than in astronomical ones. The Westminster one was once found to have been left with the spring loose for several days, and it had not gained a second, and therefore had never tripped. The two wheels legged must be both squared on the arbor, or on a collar common to them both, and must not depend upon the three pins or they will shake loose. If the wheels are set with the teeth equidistant, their centre is evidently twice the length of the teeth below C, the theoretical centre of the pallets. The pins should not be farther from the centre than one-24th of the radius of the wheel; and they should be so placed that the one which is going to lift next may be vertically over the one which has just lifted, and is then holding up the other pallet. The third will then be level with the centre; i.e., they will stand on the radii which form the acting faces of the teeth of one of the wheels, and half way between those of the other. Of course the fly for those escapements in large clocks, with weights heavy enough to drive the hands in all weather, must be much larger than in small ones. For average church clocks with 1 sec. pendulum the legs of the scape-wheels are generally made 4 inches long and the fly from 6 to 7 inches long in each vane by lj orl^ wide. For 1^ sec. pendulums the scape-wheels are generally made 4^ radius. At Westminster they are 6 inches.
Sir E. Beckett has come to the conclusion that these escapements act better, especially in regulators, if the pallets do not fall quite on the lifting pins, but on a banking, or stops at any convenient place, so as to leave the wheel free at the moment of starting; just as the striking of a common house clock will sometimes fail to start unless the wheel with the pins has a little run before a pin begins to lift the hammer. The best way to manage the banking is to make the beat-pins long enough to reach a little way behind the pendulum, and let the banking be a thin plate of any metal screwed adjustably to the back of the case. This plate cannot well be shown in the drawings together with the pendulum, which, it may be added, should take up one pallet just when it leaves the other. It is no longer doubtful that these two escapements are far the best of all for large clocks, the three-legs for very large ones, while the four-legs does very well for smaller turret clocks. And they cost no more to make, though rather more is charged for them by some makers under the pretence that they do. It is absolutely impossible for any large clock exposed to the variations of weather and dust to keep as good time as an ordinary good house clock unless it has either a gravity escapement, or a train remontoire, which last is much more expensive, to intercept the variations of force before they reach the pendulum. And though a detached escapement clock while kept clean and the oil in good condition is as good as a gravity one and perhaps better, the gravity one is less affected by variations of the oil, and its rate is altogether more constant. They seem also to have a smaller barometric error.
Going Barrels.
A clock which is capable of going accurately must have some contrivance to keep it going while you are winding it up. In the old-fashioned house clocks, which were wound up by merely pulling one of the strings, and in which one such winding served for both the going and striking parts, this was done by what is called the endless chain of Huyghens, which consists of a string or chain with the ends joined together, and passing over two pulleys on the arbors of the great wheels, with deep grooves and spikes in them, to prevent the chain from slipping. In one of the two loops or festoons which hang from the upper pulleys is a loose pulley without spikes,
