Regulation.—The reader will have noticed that in describing the various forms of seconds pendulums we have specified either eighteen or thirty-six threads to the inch; this is because a revolution of the nut with such a thread gives us a definite proportion of the length of the rod, so that it means an even number of seconds in twenty-four hours.

Moving the bob up or down ¹/18 inch makes the clock having a seconds pendulum gain or lose in twenty-four hours one minute, hence the selecting definite numbers of threads has for its reason a philosophical standpoint, and is not a matter of convenience and chance, as seems to be the practice with many clockmakers. With a screw of eighteen threads, we shall get one minute change of the clock’s rate in twenty-four hours for every turn of the nut, and if the nut is divided into sixty parts at its edge, each of these divisions will make a change of the clock’s rate of one second in twenty-four hours. Thus by using a thread having a definite relation to the length of the rod regulating is made comparatively easy, and a clock can be brought to time without delay. Suppose, after comparing your clock for three or four days with some standard, you find it gains twelve seconds per day, then, turning the nut down twelve divisions will bring the rate down to within one second a day in one operation, if the screw is eighteen threads. With the screw thirty-six threads the nut will require moving just the same number of divisions, only the divisions are twice as long as those with the screw of eighteen threads.

The next thing is the size and weight of the nut. If it is to be placed in the middle of the bob as in Figs. 10, 12 and 15, it should project slightly beyond the surface and its diameter will be governed by the thickness of the bob. If it is an internal nut, worked by means of a sleeve and disc, as in Fig. 9, the disc should be of sufficient diameter to make the divisions long enough to be easily read. If the nut is of the class shown in Fig. 5, 6, 7, a nut is most convenient, 1 inch in diameter, and cut on its edge into thirty equal divisions, each of which is equal to one second in change of rate in twenty-four hours, if the screw has thirty-six threads to the inch. This gives 3.1416 inches of circumference for the thirty divisions, which makes them long enough to be subdivided if we choose, each division being a little over one-tenth of an inch in length, so that quarter-seconds may be measured or estimated.

With some pendulums, Fig. 13, the bob rotates on the rod, and is in the form of a cylinder, say 8½ inches long by 2½ inches in diameter, and the bob then acts on its rod as the nut does, and moves up and down when turned, and in this form of bob the divisions are cut on the outside edge of the cover of the bob, and are so long that each one is subdivided into five or ten smaller divisions, each altering the clock .2 or .1 second per day.

On the top of the bob turn two deep lines, close to the edge, about ?-inch apart, and divide the whole diameter into thirty equal divisions, and subdivide each of the thirty into five, and this will give seconds and fifths of seconds for twenty-four hours. Each even seconds division should be marked heavier than the fraction, and should be marked from one to thirty with figures. Just above the cover on the rod should slide a short tube, friction-tight, and to this a light index or hand should be fastened, the point of which just reaches the seconds circle on the bob cover, and thus indicates the division, its number and fraction. The tube slides on the rod because the exact place of the hand cannot be settled until it has been settled by experiment. After this it can be fastened permanently, if thought best, though as described it will be all sufficient. While the bob is being raised or lowered to bring the clock to its rate, the bob might get too far away or too near to the index and necessitate its being shifted, and if friction-tight this can be readily accomplished, and the hand be brought to just coincide with the divisions and look well and be a means of accomplishing very accurate minute adjustments.

Suspensions.—Suspensions are of four kinds, cord, wire loop, knife edges and springs. Cords are generally of loosely twisted silk and are seldom found except in the older clocks of French or Swiss construction. They have been entirely displaced in the later makes of European manufactures by a double wire loop, in which the pendulum swings from a central eye in the loop, while the loop rocks upon a round stud by means of an eye at each end of the loop. The eyes should all be in planes parallel to the plane of oscillation of the pendulum, otherwise the bob will take an elliptical path instead of oscillating in a plane. They should also be large enough to roll without friction upon the stud and center of the loop, as any slipping or sliding of either will cause them to soon wear out, besides affecting the rate of the pendulum. Properly constructed loops will give practically no friction and make a very free suspension that will last as long as the clock is capable of keeping time, although it seems to be a very weak and flimsy method of construction at first sight. Care should be taken in such cases to keep the bob from turning when regulating the clock, or the effect upon the pendulum will be the same as if the eyes were not parallel.

Knife edge suspensions are also rare now, having been displaced by the spring, as it was found the vibrations were too free and any change in power introduced a circular error (See Fig. 4) by making the long swings in longer time. They are still to be found, however, and in repairing clocks containing them the following points should be observed: The upper surface of the stud on which the pendulum swings should carry the knife edge at its highest point, exactly central with the line of centers of the stud, so that when the pendulum hangs at rest the stud shall taper equally on both sides of the center, thus giving equal freedom to both sides of the swing. Care should be taken that the stud is firmly fixed, with the knife edge exactly at right angles to the movement, and also to the back of the case. The suspension stud and the block on the rod should be long enough to hold the pendulum firmly in line, as the angle in the top of the rod must be the sole means of keeping the pendulum swinging in plane. The student will also perceive the necessity of making the angle occupy the proper position on the rod, especially if the latter be flat. In repairing this suspension it is usual to make the plate, fasten it in place and then drill and file out the hole, as it is easier to get the angles exactly in this way than to complete the plate and then attempt to fasten it in the exact position in which it should be. After fastening the plates in position on the rod, two holes should be drilled, a small one at the apex of the angle (which must be exactly square and true with the rod), and a larger one below it large enough to pass the files easily. The larger hole can then be enlarged to the proper size, filing the angle at the top in such a way that the small hole first drilled forms the groove at the apex of the angle in which the knife edge of the stud shall work when it is completed. Knife edge suspensions are unfitted for heavy pendulums, as the weight causes the knife edge to work into the groove and cut it, even if the latter be jeweled. Both the edge and groove should be hardened and polished.

Pendulum Suspension Springs.—Next in importance to the pendulum is its suspension spring. This spring should be just stiff enough to make the pendulum swing in all its vibrations in the same time; that is, if the pendulum at one time swung at the bottom of the jar 1¼ inch each side of the center, and at another time it swung only 1 inch each side, that the two should be made in exactly one second. The suspension springs are a point in the construction of a fine pendulum, that there has been very much theorizing on, but the experiments have never thus far exactly corroborated the theories and there are no definite rules to go by, but every maker holds to that plan and construction that gives his particular works the best results. A spring of sufficient strength to materially influence the swing of the pendulum is of course bad, as it necessitates more power to give the pendulum its proper motion and hence there is unnecessary wear on the pallets and escape wheel teeth, and too weak a spring is also bad, as it would not correct any inequalities in the time of swing and would in time break from overloading, as its granular structure would finally change, and rupture of the spring would follow. The office of a spring is to sustain the weight without detriment to strength and elasticity, and if so proportioned to the weight as to be just right, it will make the long and short swings of the pendulum of equal duration. When a pendulum hung by a cord or knife edge instead of a spring is regulated to mean time and swings just two inches at the bottom, any change in the power that swings the pendulum will increase its movement or decrease it, and in either case the rate will change, but with a proper spring the rate will be constant under like conditions. The action of the spring is this: In the long swings the spring, as it bends, lifts the pendulum bob up a little more than the arc of the normal circle in which it swings, and consequently when the bob descends, in going to the center of its swing, it falls a little quicker than it does when held by a cord, and this extra quick drop can be made to neutralize the extra time taken by the bob in making extra long swings. See Fig. 4. This action is the isochronal action of the spring, the same that is attained in isochronal hair springs in watches.

As with the hairspring, it is quite necessary that the pendulum spring be accurately adjusted to isochronism and my advice to every jeweler is to thoroughly test his regulator, which can easily be done by changing the weight or motive power. If the test should prove the lack of isochronism he can adjust it by following these simple rules. Fig. 16 is the pendulum spring or leaf. If the short arcs should prove the slowest, make the spring a trifle thinner at B; if fastest, reduce the thickness of the spring at A. Continue the test until the long and short arcs are equal. In doing this care must be taken to thin each spring equally, if it is a double spring, and each edge equally, if a single spring, as if one side be left thicker than the other the pendulum will wabble.

The cause of a pendulum wobbling is that there must be something wrong with the suspension spring, or the bridge that holds the spring. If the suspension spring is bent or kinked, the pendulum will wabble; or if the spring should be of an unequal thickness it will have the same effect on the pendulum; but the main cause of the pendulum wobbling in American clocks is that the slot in the bridge that holds the spring, or the slot in the slide that works up and down on the spring (which is used to regulate the clock) is not parallel. When this slot is not parallel it pinches the spring, front or back, and allows it to vibrate more where it is the freest, causing the pendulum to wabble. We have found that by making these slots parallel the wobbling of the pendulum has ceased in most all cases. If the pallet staff is not at right angles to the crutch, wobbling may be caused by the oblique action of the crutch. This often happens when the movement is not set square in the case.

It occasionally happens in mantel clocks that the pendulum when brought to time is just too long for the case when too thick a spring is used. In such a case thinning the spring will require the bob to be raised a little and also give a better motion. If compelled to make a spring use a piece of mainspring about .007 thick and ? wide for small pendulums and the same spring doubled for heavier pendulums, making the acting part of the spring about 1.5 inches long.

The suspension spring for a rather heavy pendulum is better divided, that is, two springs, held by two sets of clamps, and jointly acting as one spring. The length will be the same as to the acting part, and that part held at each end by the clamps may be ¾ inch long; total length, 1.5 inches with ? inch at each end held in the clamps. These clamps are best soldered on to the spring with very low flowing solder so as not to draw the temper of the spring, and then two rivets put through the whole, near the lower edge of the clamps. The object of securing the clamps so firmly is so that the spring may not bend beyond the edge of the clamps, as if this should take place the clock will be thrown off of its rate. After a time the rate would settle and become steady, but it only causes an extra period of regulating that does not occur when the clamps hold the spring immovable at this point. About in the center of each of the clamps, when soldered and riveted, is to be a hole bored for a pin, which pins the clamp into the bracket and holds the weight of the pendulum.

The width of this compound spring for a seconds’ pendulum of average weight may be .60 inch, from outside to outside, each spring .15 inch wide. This will separate the springs .30 inch in the center. With this form of spring, the lower end of each spring being held in a pair of clamps, the clamps will have to be let into the top of the rod, and held in by a stout pin, or the pendulum finished with a hook which will fit the clamp. In letting the clamp into the rod, the clamp should just go into the mortise and be without side shake, but tilt each way from the center a little on the pin, so that when the pendulum is hung it may hang perpendicular, directly in the center of both springs. Also, the top pair of clamps should fit into a bracket without shake, and tilt a little on a pin, the same as the lower clamps. These two points, each moving a little, helps to take any side twist away, and allows the whole mechanism to swing in line with the center of gravity of the mass from end to end. With the parts well made, as described, the bob will swing in a straight line from side to side, and its path will be without any other motion except the one of slight curvature, due to being suspended by a fixed point at the upper clamp.

Pendulum Supports.—Stability in the movement and in the suspension of the pendulum is very necessary in all forms of clocks for accurate timekeeping. The pendulum should be hung on a bracket attached to the back of the case (see Fig. 6), and not be subject to disturbance when the movement is cleaned. Also the movement should rest on two brackets attached to the bracket holding the pendulum and the whole be very firmly secured to the back board of the case. Screws should go through the foot pieces of the brackets and into a stone or brick wall and be very firmly held against the wall just back of the brackets. Any instability in this part of a clock is very productive of poor rates. The bracket, to be in its best form, is made of cast iron, with a large foot carrying all three separate brackets, well screwed to a strong back board and the whole secured to the masonry by bolts. Too much firmness cannot be attained, as a lack of it is a very great fault, and many a good clock is a very poor timekeeper, due to a lack of firmness in its supports and fastenings. The late Edward Howard used to make his astronomical clocks with a heavy cast iron back, to which the rest of the case was screwed, so that the pendulum should not swing the case. Any external influence that vibrates a wall or foundation on which a clock is placed, is a disturbing influence, but an instability in a clock’s attachment to such supports is a greater one. Many pendulums swing the case in which they hang (from unstable setting up) and never get down to or maintain a satisfactory rate from that cause. This is also aggravated by the habit of placing grandfather clocks on stair landings or other places subject to jarring. The writer knows of several clocks which, after being cleaned, kept stopping until raised off the floor and bolted to the wall, when they at once took an excellent rate. The appearance of resting on the floor may be preserved, if desirable, by raising the clock only half an inch or so, just enough to free it from the floor.

Crutches.—The impulse is transmitted to the pendulum from the pallet staff by means of a wire, or slender rod, fastened at its upper end to the pallet staff and having its lower end terminating in a fork (crutch), loop, or bent at right angles so as to work freely in a slot in the rod. It is also called the verge wire, owing to the fact that older writers and many of the older workmen called the pallet fork the verge, thus continuing the older nomenclature, although of necessity the verge disappeared when the crown wheel was discarded.

In order to avoid friction at this very important point, the centers of both axes of oscillation, that of the pallet arbor and that of the pendulum spring, where it bends, should be in a straight horizontal line. If, for instance, the center of suspension of the pendulum be higher, then the fork and the pendulum describe two different arcs of circles; that of the pendulum will be greater than that of the fork at their meeting point. If, however, the center of suspension of the pendulum be lower than that of the fork, they will also describe two different arcs, and that of the pendulum will be smaller than that of the fork at their point of meeting. This, as can be readily understood, will cause friction in the fork, the pendulum going up and down in it. This is prevented when, as stated before, the center of suspension of the pendulum is in the prolonged straight imaginary line going through the center of the pivots of the fork, which will cause the arcs described by the fork and the pendulum to be the same. It will be well understood from the foregoing that the pendulum should neither be suspended higher nor lower, nor to the left, nor to the right of the fork.

If the centers of motion do not coincide, as is often the case with cheap clocks with recoil escapements, any roughness of the pendulum rod where it slides on the crutch will stop the clock, and repairers should always see to it that this point is made as smooth as possible and be very slightly oiled when setting up. If putting in a new verge wire, the workman can always tell where to bend it to form the loop by noticing where the rod is worn and forming the loop so that it will reach the center of that old crutch or loop mark on the pendulum rod. If the verge wire is too long, it will give too great an arc to the pendulum if the latter is hung below the pallet arbor, as is generally the case with recoil escapements of the cheap clocks, and if it is too short there will not be sufficient power applied to the pendulum when the clock gets dirty and the oil dries, in which case the clock will stop before the spring runs down.

An important thing to look after when repairing is in the verge wire and loop (the slot the pendulum rod goes through). After the clock is set up and oiled, put it on a level shelf; have a special adjusted shelf for this level adjusting, one that is absolutely correct. Have the dial off. If the beat is off on one side, so that it bangs up heavily on one side of the escape wheel, bend the verge wire the same way. That will reverse the action and put it in beat. So far so good—but don’t stop now. Just notice whether if that shelf were tipped forward or back, as perhaps your customer’s may, that the pendulum should still hang plumb and free. Now if the top of your clock tips forward, the pendulum ball inclines to hang out toward the front. We will suppose you put two small wedges under the back of the case. Now notice in its hanging out whether the pendulum rod pinches or bears in the throat of the verge; or if it tips back, see if the rod hits the other end of the slot. This verge slot should be long enough, with the rod hanging in the middle when adjusted to beat on a level, to admit of the clock pitching forward or back a little without creating a friction on the ends of the slot. This little loop should be open just enough to be nice and free; if open too much, you will notice the pallet fork will make a little jump when carrying the ball over by hand. This is lost motion. If this little bend of wire is not parallel it may be opened enough inside, but if pitched forward a little it will bind in the narrowest part of the V and then the clock will stop. The clock beat and the tipping out or in of the clock case, causing a binding or bearing of the pendulum rod in this verge throat, does more towards stopping clocks just repaired than all other causes.

Putting in Beat.—To put a clock in beat, hang the clock in such a position that when the pendulum is at rest one tooth of the escape wheel will rest on the center of a pallet stone. Screwed on the case of the clock at the bottom of the pendulum there is, or should be, an index marked with degrees. Now, while the escape wheel tooth is resting on the pallet, as explained above, the index of the pendulum should point to zero on the index. Move the pendulum until the tooth just escapes and note how many degrees beyond zero the pendulum point is. Say it escapes 2° to the left; now move the pendulum until the next tooth escapes—it should escape 2° to the right. But let us suppose it does not escape until the index of the pendulum registers 5° to the right of zero. In this case the rod attached to the pallets must be bent until the escape wheel teeth escape when the pendulum is moved an even number of degrees to the right and left of zero, when the clock will be in beat.

Close Rating with Shot.—Very close rating of a seconds’ pendulum, accompanied by records in the book, may be got with the nut alone, but there is the inconvenience of stopping the clock to make an alteration. This may be avoided by having a small cup the size of a thimble or small pill box on the pendulum top. This can be lifted off and put back without disturbing the motion of the pendulum. In using it a number of small shot, selected of equal size, are put in, say 60, and the clock brought as nearly as possible to time by the nut. After a few days the cup may be emptied and put back, when on further trial the value of the 60 shot in seconds a day will be found. This value divided by 60 will give the value of a single shot, by knowing which very small alterations of rate may be made with a definite approach towards accuracy, and in much less time than by putting in or taking out one or more shot at random.