Ex Libris 
 C. K. OGDEN
 
 ON THK 
 
 Springing 
 
 AND 
 
 fid lasting Watches 
 
 Beini;' a description of the Balance Spring and the Compensation 
 Balance with Directions for applying the Spring and Adjustii' 
 Isochronism and Temperature. 
 
 J . BRITTEN, 
 
 Author of the Watch & Clock Makers' Handbook, Former Clock 
 and Watch .Makers, &c. 
 
 fLonUon : 
 
 K. & F. N. SroN, 125, STRAND, W.C. 
 
 fjorfc : 
 
 Srox & CiiAMiiKKi.Aix, 12, Cortlandt Street. 
 1898. 
 
 \All rights rr served. \
 
 PREFATORY NOTE. 
 
 To meet many enquiries for a book of moderate price 
 explanatory of the springing and adjusting of watches, this 
 little work is submitted. 
 
 It is intended for those tolerably conversant with watch- 
 making generally, yet who desire guidance in this particular 
 branch, rather than for beginners, and therefore a knowledge 
 of many elementary facts is assumed. 
 
 The examination of modern watch escapements is dealt 
 with because it belongs particularly to the art of the adjuster. 
 For drawings of the escapements, particulars of their action, 
 and other details, the student is referred to the " Watch and 
 Clock Makers' Handbook," to which this volume may be 
 regarded as supplementary. 
 
 A brief historical notice of the balance spring and 
 compensation balance is included here ; a comprehensive 
 account of the early craftsmen may be found in " Former 
 Clock and Watch Makers and their Work." 
 
 If springs and balances of nickel-steel alloy answer all 
 expectations it may be that, in the future, adjustment for 
 varying temperatures will be unnecessary. But in such 
 an event so many springs and balances of other material will 
 remain as to justify the inclusion of those pages devoted 
 to compensation, even apart from their historical interest. 
 
 F.J. B. 
 35, Northampton Square, 
 
 London, E.C.
 
 CONTENTS 
 
 CHAPTER I. 
 Introduction and Effect of Various Springs. 
 
 CHAPTER II. 
 
 Theoretically Correct Terminal Curves. Effect of Disturbing 
 Influences. 
 
 CHAPTER III. 
 Compensation for Varying Temperature. 
 
 CHAPTER IV. 
 Method of Procedure in Springing and Adjusting. 
 
 CHAPTER V. 
 
 The Manufacture of Balance Springs. 
 
 CHAPTER VI. 
 
 Non-magnetic Material and Material insensible to Changes of 
 Temperature. 
 
 CHAPTER VII. 
 Gauges. 
 
 CHAPTER VIII. 
 
 Observatory Tests, and Note on Timing Repeating Carriage Clocks. 
 
 CHAPTER IX. 
 Examination of Escapements. Revolving Escapements. 
 
 1077183
 
 ON THK 
 
 SPRINGING AND ADJUSTING 
 OF WATCHES, 
 
 CHAPTER I. 
 
 The vibrating wheel of a watch or chronometer which, 
 in conjunction with the balance spring, regulates the 
 progress of the hands is called the balance. The time in 
 which a balance will vibrate cannot be predicated from its 
 dimensions alone. A pendulum of a given length always 
 vibrates in the same time as long as it is kept at the same 
 distance from the centre of the earth, because gravity, the 
 force that impels it, is always the same ; but the want of 
 constancy in the force of the balance spring, that in watches 
 and chronometers takes the place of 
 gravity, and governs the vibrations of 
 the balance is one of the chief diffi- 
 culties of the timer. There is another 
 point of difference between the pendu- 
 lum and the balance. The time of 
 vibration of the former is unaffected 
 by its mass, because every increment 
 of mass carries with it a proportional 
 addition to the influence of gravity : but by adding to the 
 mass of a balance, the strength of the balance spring is 
 not increased at all, and therefore the vibrations of the 
 balance become slower.
 
 There are three factors upon which the time of the 
 vibration of the balance depends : 
 
 (1) The weight, or rather the mass of the balance.* 
 
 (2) The distance of its centre of gyration from the centre 
 of motion, or, to speak roughly, the diameter of the balance. 
 From these two factors the moment of inertia may be 
 deduced. f 
 
 (3) The strength of the balance spring, or, more strictly, 
 its power to resist change of form. 
 
 I append the usual formula for ascertaining the time of 
 vibration of a balance, though it is difficult of -application 
 
 in actual practice : 
 
 T = A /~AL 
 
 M 
 
 A being the moment of inertia of the balance, M the 
 moment of elasticity of the spring, L the length of the 
 spring, and TT 3-14159. 
 
 The Moment of Elasticity of a spring is its power 
 of resistance. It varies directly as the modulus of elasticity 
 of the material, and as the breadth and cube of the thick- 
 ness of the spring when its section is rectangular. Mo = E 
 is a usual formula, E representing the modulus of elasticity, 
 b the breadth, and / the thickness of the spring. 
 
 The moment of elasticity must not be confounded with 
 the bending moment. The bending moment is a measure 
 of the resistance a spring offers to bending, and of the 
 amount of bending which has been produced, which varies 
 directly as the angle wound through, and inversely as the 
 length of the spring. 
 
 M ' L -~ is a formula for ascertaining the bending 
 moment, E being the modulus of elasticity, b the breadth, t 
 
 * The mass of a body is the amount of matter contained in that body, and is the same 
 irrespective of the distance of the body from the centre of the earth. But its weight, 
 which is mass multiplied by gravity, varies in different latitudes. 
 
 t The centre of gyration is that point in a rotating body in which the whole of its 
 energy may be concentrated. A circle drawn at seven-tenths of its radius on a circular 
 rotating plate ot unilortn thickness would represent its centre oi gyration. The moment 
 of inertia or the controlling power of balances varit-s as their mass and as the square of 
 the distance of their centre of gyration from their centre of motion. Although not Strictly 
 accurate, it is practically quite near enough in the comparison of plain balances to take 
 their weight and the square of their diameter measured t the middle of the rim.
 
 the thickness, and L the length of the spring, and A the 
 angle through which it is wound. 
 
 This formula also determines the value of the force 
 which has produced the bending, for if the forces are in 
 equilibrium, the moment of the resisting force must be 
 exactly equal to the moment of the bending force. 
 
 The Modulus of Elasticity is a constant, represented 
 by E, whfch is used for ascertaining what proportion of its 
 length material is strained when subjected to stress. If the 
 body is stretched, the strain is a lengthening, and if it is 
 compressed, a shortening of its original dimensions. In 
 Young's formula, which is usually accepted, the stress in 
 pounds per square inch of section, divided by E, gives the 
 strain ; E being the force in Ibs. that would stretch a rod 
 one square inch in section to twice its original length, 
 supposing its elasticity to remain perfect all the time. 
 Young gives 29,000,000 as E for steel, but Mr. Robert 
 Gardner considers this too high for the average quality of 
 steel used in balance springs, and places it at 23,000,000. 
 
 One end of the balance spring is fixed to a collet fitted 
 friction tight on the balance staff, and the other to a stud 
 attached to the balance cock or to the watch plate. The 
 most ordinary form of balance spring is the volute or flat 
 spiral, like Fig. 2. An overcoil or Broguet spring is a 
 volute with its outer end bent up above the plane of the 
 body of the spring, and carried in a long curve towards the 
 centre, near which it is fixed. (Fig. 3.) For marine 
 chronometers helical springs, in which both ends curve 
 
 inwards, are universally 
 
 used. Either helical or 
 
 Breguet springs are as a 
 
 rule applied to pocket 
 
 chronometers, although a 
 
 form of spring called " duo 
 
 in unfl."' invented, I believe, 
 
 by Mr. Hammersley, is 
 sometimes preferred. The bottom of this spring is in the 
 form of a volute, from the outer coil of which the spring 
 
 Fig. 2. -Ordinary 
 balance spring. 
 
 Fig. 3. Flat 
 spring with 
 overcoil.
 
 is continued in the form of a helix ; the upper end is curved 
 in towards the centre as in the ordinary helical spiing. 
 
 Elevation. Plan. 
 
 Fig. 4. Helical spring. 
 
 Among fancy shapes, which may be dismissed in a few 
 words, are spherical springs, introduced by Frederic 
 Houriet ; they present no superiority, and having to be 
 prepared on a solid spherical block are difficult to harden 
 except by exposure to a high temperature. " Bird cage 
 springs," having a helical body with top and bottom 
 terminals formed into volutes, were another short lived con- 
 ceit, with no advantage except the difficulty of making them. 
 
 The introduction of the balance spring which marks 
 such an important epoch in the manufacture of watches is 
 due to the investigations of Dr. Robert Hooke. He 
 discovered that the potential energy of a spring is pro- 
 portional to the angle through which it has been wound, 
 and propounded the whole theory in the sentence, " Ut 
 tensio sic vis," meaning that the force is proportionate 
 to the tension. He proposed to patent his discovery in 
 1660, and to quote his own words, " Sir Robert Moray drew 
 me up the form of a patent, the principal part whereof, viz., 
 the description of the watch, is his own handwriting, which 
 I have yet by me ; the discouragement I met with in the 
 progress of this affair made me desist for that time." 
 
 Derham describes the earliest of Hooke's essa) 7 s in this 
 direction as " a tender straight .spring, one end whereof 
 played backward and forward with the ballance." It is 
 stated that several watches were made under Hooke's 
 supervision at this period, and one of the first to which the 
 balance spring was applied he is said to have presented to 
 Dr. Wilkins, afterwards Bishop of Chester, about 1661.
 
 It appears that Hooke then conceived it to be an advan- 
 tage to have two balances coupled together, and had two 
 double balance watches constructed. In the first, which 
 had no balance spring, the escape wheel was placed in 
 the centre of the movement with its teeth in a horizontal 
 plane. There were two verges standing vertically on 
 opposite sides of the wheel and connected with each other 
 by means of toothed wheels of equal size ; each verge had 
 one pallet and carried a balance at its upper end, one 
 balance overlapping the other. 
 
 In the second watch the verge escapement was arranged 
 in the ordinary way, the balance being mounted on a verge 
 with two pallets ; on the verge was also a toothed wheel 
 which engaged with another of the same size mounted on a 
 stud, and the pipe of this wheel carried the second balance ; 
 the toothed wheels being of small size one balance was 
 placed a little higher than the other and overlapped it. 
 Each balance was controlled by a balance spring. 
 
 However, Hooke turned his attention to other matters, 
 and in January, 1673, Huygens addressed a letter to Henry 
 Oldenburg, secretary of the Royal Society, in which he 
 described as his invention the application of a spring to 
 control the balance in watches. This aroused the wrath of 
 Hooke, who accused Oldenburg of having divulged the 
 discovery in his correspondence with Huygens. Hooke 
 enlisted the interest of Charles II., and in a lecture, entitled 
 " Potentia Restitutiva," &c., said, " His Majesty was pleased 
 to see the experiment that made out this theory tried at 
 Whitehall, as also my spring watch." 
 
 A watch with two balances and springs, subsequently 
 made for the King, was inscribed, " Robt. Hooke, inven ; 
 t'>sN. T. Tompion, fecit, 1675." 
 
 Hooke, Huygens and other investigators experimented 
 with various materials to find the most satisfactory 
 controller. 
 
 The engraving (Fig. 5) represents a watch of German 
 origin from the collection of Mr. Evan Roberts. It 
 has a day of the month ring, and is generally of the construc- 
 tion usual soon after the middle of the i7th century. But the
 
 peculiar feature of the movement lies in the application of 
 a straight hog's bristle to regulate the balance. There is no 
 
 Fig. 5. Hog's Bristle as a Balance Controller, 
 sign of any other spring having been attached, and the 
 accessories of the bristle are quite in character with the rest 
 of the work. There are two arms which embrace the 
 bristle and practically determine its acting length, and by 
 means of a screw these may be shifted to act over a 
 considerable range. 
 
 Steel springs were however found to be the most suitable'. 
 The primitive straight ones would, of course, allow but a 
 very small vibration of the balance, while the to and fro 
 motion between pins where it made contact with the balance 
 involved considerable friction. Of others curved somewhat 
 to the shape of a pothook there are still examples, but 
 eventually the more convenient and correct form was found 
 to be a volute which at first had but one or two coils. The 
 coils were increased to four or five as the advantage of a 
 larger spring was understood, but the very long springs 
 with which we are now familiar were not applied till the 
 advent of the lever and other detached escapements which 
 allowed the balance to have a larger arc of vibration. 
 
 To lengthen or shorten the acting length of the spring 
 Toinpion appears to have used the circular slide with an 
 index from the first. This arrangement, which remained 
 in favour for a long period, is shown in Fig. 6. Below, and 
 attached to a silver disc, graduated and figured as a guide 
 to regulation is a pinion which gears with teeth on the
 
 10 
 
 outer edge of the circular slide; from the inner edge 
 projects an arm carrying two upright pins which embrace 
 the spring. The projecting end of the pinion is square so 
 that it could be turned by means of a watch key. 
 
 Fig. 6. Tompion's 
 Regulator. 
 
 rig. 7. Barrow's Regulator. 
 
 Mr. Albert Schloss has a clock-watch by Nathaniel 
 Barrow, dating from about 1675, in which the outer end of 
 the spring is continued in a straight line to the stud at the 
 edge of the plate, and the regulation accomplished very 
 much in the same way as the hog's bristle watch already 
 delineated. Fig. 7 is a plan of this watch movement. The 
 curved stud on the left is continued in a sort of zig-zag 
 shape to hold one end of the regulating screw. The upper 
 end of the nut points to an index engraved on the plate, and 
 the lower extremity is notched to receive the spring. 
 
 The chief drawback to Tompion's regulator is that owing 
 to the backlash or freedom between the teeth of the pinion 
 and slide, a slight reversal of the index has no effect on the 
 curb pins. The simple regulator now generally employed 
 consists of a lever, fitting friction tight over a boss on the 
 balance cock ; the shorter end of the lever carries the curb 
 pins which embrace the balance spring, while the longer 
 end through which it is moved serves also as an indicator 
 of alterations in the position of the curb pins. This device 
 was patented by Bosley in 1755. 
 
 There is one point about the stud used in those of 
 Tompion's watches I have seen which might well be 
 revived. The hole in the stud for the reception of the
 
 II 
 
 spring was square. The modern system of pinning by 
 squeezing the flat side of a spring against the surface of a 
 round hole is altogether unmechanical and must distort the 
 spring. 
 
 The action of different forms of springs with 
 opinions of various experts. -- A very generally 
 accepted rule is that the diameter of a steel balance spring 
 for a watch should be half the diameter of the balance 
 (rather under than over). 
 
 The dimensions of the spring, its form at the attach- 
 ments, the position of the attachments with relation to each 
 other, are all factors affecting its controlling power. 
 
 The length is important, especially in flat springs with- 
 out overcoils By varying the thickness of the wire two 
 flat springs may be produced, each of half the diameter of 
 the balance, but of very unequal lengths, either of which 
 would yield the same number of vibrations as long as the 
 extent of the vibration remained constant ; yet if the spring 
 is of an improper length, although it may bring the watch 
 to time in one position, it will fail to keep the long and 
 hort vibrations isochronous. Then, again, a good length 
 of spring for a watch with a cylinder escapement vibrating 
 barely a full turn would clearly be insufficient for a lever 
 vibrating a turn and a half. 
 
 The great advantage of an overcoil spring is that it 
 distends in action on each side of the centre, and the 
 balance pivots are thereby relieved of the side pressure 
 given with the ordinary flat spring. An overcoil spring, in 
 common with the helical and all other forms in which the 
 outer coil returns towards the centre, offers opportunities 
 of obtaining isochronism by slightly varying the character 
 of the curve described by the outer coil and thereby 
 altering its power of resistance. 
 
 The position of the points of attachment of the inner and 
 outer turns of a flat spring without overcoil in relation to 
 each other has an effect on the long and short vibrations 
 quite apart from its length. For instance, a different per- 
 formance may be obtained with two springs of precisely 
 the same length and character in other respects, but pinned
 
 12 
 
 iii so that one has exactly complete turns, and the other a 
 little under or a little over complete turns. This property, 
 which is more marked in short than in long springs, is 
 depended upon by many for obtaining isochronism. A short 
 spring as a rule requires to be pinned in short of complete 
 turns, and a long one beyond the complete turns. In 
 duplex and other watches with frictional escapements, 
 small arcs of vibration and short springs, it will be found 
 that the spring requires to be pinned in nearly half a 
 turn short of complete turns. In watches the point of 
 attachment to the collet with relation to the pendant has 
 also to be considered with plain flat springs and with 
 flat spring springs having an outer terminal curve. It is 
 not easy to see why a helical or other spring with two 
 theoretically correct terminal curves should be affected 
 by varying the positions of the points of attachment with 
 relation to each other, yet marine chronometer springs are 
 found to isochronize better and act truer when pinned 
 in at about a quarter of a turn short of complete turns. 
 
 If a spring is too long the short vibrations will be slow. 
 It may be taken as a very good rule that a steel balance 
 spring should be half the diameter of the balance, and have 
 twelve turns if it is a flat spring or eighteen turns if a 
 Bn'-guet for a lever watch of ordinary size. Small watches 
 are more difficult to time than those of a reasonable size, 
 and are generally slow 7 in the short arcs, being relatively 
 more affected by the retarding action of the escapement. 
 For small lever watches a spring shorter by a turn or two 
 will therefore be desirable. These lengths, it will be under- 
 stood, only apply where the work is good ; with coarse work 
 a shorter spring is usually required in order to get the short 
 arcs fast enough. Springs for cylinder watches should have 
 from eight to twelve turns. 
 
 Watch springs of thick and narrow w r ire are apt to 
 cockle with large vibrations, while springs of wide and thin 
 wire keep their shape and are more rigid. It is of even 
 greater importance that the springs of marine chronometers 
 subjected to the tremor of steamships should be of wide 
 and thin wire.
 
 13 
 
 Helical springs for ordinary two-day marine chrono- 
 meters are made from '4 to -54 of an inch in diameter, 
 and about a quarter of a turn short of either eleven, twelve, 
 or thirteen turns. 
 
 It is remarkable that while in watches the difficulty is 
 generally to get the short arcs sufficiently fast, precisely the 
 reverse is the case with the marine chronometer, in which 
 the trouble is usually to get the short arcs slow enough. 
 The escapement is not entirely responsible for the difference, 
 because pocket chronometers follow the same rule as watches 
 with lever escapements. 
 
 The relative slowness of the long arcs in marine chrono- 
 meters is much greater if the rini of the balance is narrow 
 and thin, and the weights large, than if a wider and stouter 
 rim with proportionately smaller weights is used ; for in the 
 former case the greater enlargement of the rim in the long 
 arcs from centrifugal tendency will be more marked. Mr. 
 Kullberg, by substituting a chronometer balance of ordinary 
 proportions for an uncut one, has demonstrated that the 
 effect of centrifugal tendency by increasing the arc of 
 vibration from three-quarters of a turn to a turn and a 
 quarter amounts to twelve or fourteen seconds in twenty- 
 four hours. Mr. B. Dennison some years ago advocated a 
 method of making the long arcs slower in watches having 
 compensation balances, by drilling a small hole in each half 
 of the balance rim, close to the arm, and broaching these 
 holes out as much as should be found necessary to produce 
 isochronism. The larger the holes, the more the balance 
 would expand from centrifugal tendency, which, of course, 
 would have more effect in the long than in the short arcs. 
 
 The size of the pivots in proportion to the size of the 
 balance is partially the cause, for in very small watches, 
 where of course the pivots are relatively large, the slowness 
 of the hanging position is proverbial, and a shorter spring 
 by a turn or two has often to be substituted. Very quick 
 trains should be avoided on this account. Watches have 
 occasionally been made with 19,800 vibrations, carrying of 
 course corresponding light balances, and the great trouble 
 has always been to get them fast enough in the short arcs.
 
 14 
 
 A balance staff pivot slightly too large for the hole is 
 occasionally the cause of undue slowness in the hanging 
 position of a watch ; a very small reduction in the diameter 
 will in such cases reduce friction and quicken the short arcs. 
 
 Mr. Robert Gardner insists that the relative proportional 
 inertia of the fourth wheels in marine chronometers and 
 watches accounts for much of the difference observed in the 
 long and short arcs, and there is no doubt that watches with 
 small fourth and escape wheels are comparatively easy to 
 time. J. F. Cole suggested a stronger mainspring to quicken 
 the short arcs in going-barrel watches. 
 
 With single beat escapements, such as the chronometer 
 and duplex, the short arcs are quickened if the drop of the 
 escape wheel tooth on the pallet is decreased ; if the drop 
 is increased the long arcs are quickened. The theory of 
 this appears to be that with more drop in the long arcs 
 (when the balance is travelling faster) the pallet gets further 
 away before being overtaken by the wheel than it does in 
 the short ones, and therefore the amount of impulse given 
 before the line of centres is proportionately less ; and Mr. 
 Robert Gardner asserts that the same effect may be produced 
 by putting the piece out of beat, e.g. if it were desired to 
 quicken the short arcs, the balance spring collet would be 
 shifted so as to bring the pallet more away from the unlock- 
 ing when the balance spring was quiescent ; but it must not 
 be forgotten that if the piece is out of beat it is more likely 
 to set. 
 
 Fig. 8 is a careful reproduction ot the curves of a marine 
 chronometer spring, a is the lower coil. The two short 
 lines crossing the spring denote the face of the collet and 
 the face of the stud respectively, and the dotted lines the 
 direction of the ends of the spring, which form nearly a 
 right angle. Occasionally, if the short arcs are fast, the 
 upper turn is slightly bent just as it enters the stud so as to 
 throw the end outward. The pin is then placed on the 
 inside. 
 
 Fig. 9 is a lower coil, and Fig. 10 the upper coil of a 
 pocket chronometer spring. It will be observed that the 
 sweep is longer than in the marine chronometer, and that
 
 the lower curve is carried rather farther back into the spring 
 than the upper. The spring makes complete turns, but the 
 upper turn just as it enters the stud has a slight sharp bend 
 
 Fig. 8. Fig. 9. Fig. 10. 
 
 throwing the end inwards. This little bend is of the utmost 
 importance, for it has the effect of quickening the short 
 arcs. Pocket chronometer springs are made half the 
 diameter of the balance, and from seven to ten turns. 
 
 Fig. 10 will also serve very well to represent the overcoil 
 of a Breguet spring, although the overcoil is sometimes 
 carried much farther into the spring. In some springs it is 
 not unusual to see the overcoil more than a complete turn 
 in extent. Breguet springs are now often used for pocket 
 chronometers, instead of the helical form. Just as good a 
 result can be got with the Breguet as with the helical, and 
 the latter takes up height, and in consequence is often made 
 too short. There is, though, one advantage in using helical 
 springs for pocket chronometers. The escapement may be 
 banked through the spring, and this is done more readily in 
 the helical form. Mr. Kullberg's method is to place two 
 nearly upright pins on the balance arm, so close to the 
 spring as to prevent it expanding more than is required for 
 a sufficient vibration. These pins should be slightly 
 inclined to the centre so as to touch the top of the spring 
 first, and thereby stop the balance more gradually. 
 
 Two pairs of pliers with curved noses lined with brass are 
 generally used for forming the overcoil of marine chrono- 
 meter springs. The coil of the spring where the overcoil 
 is to start is grasped by one pair curved exactly to corres- 
 pond with the spring, and the other pair used to bend the
 
 16 
 
 overcoil. The operation looks easy enough, but it really 
 requires great skill to get at once an overcoil of the desired 
 shape. The overcoils of watch springs are turned with 
 curved nosed steel tweezers. 
 
 Acceleration. It is noticed that new chronometers 
 and watches, instead of steadily gaining or losing a certain 
 number of seconds each day, go faster day after day. There 
 is no certainty as to the amount or ratio of this acceleration, 
 nor as to the period which must elapse before the rate 
 becomes steady, but an increase of a second a month for a 
 twelvemonth may be taken as the average extent in marine 
 chronometers. 
 
 It is pretty generally agreed among chronometer-makers 
 that the cause of acceleration is seated in the balance spring, 
 though some assert that centrifugal action slightly enlarges 
 the balance if the arc of vibration is large, as it would be 
 when the oil is fresh, and that as the vibration falls off 
 centrifugal action is lessened and acceleration ensues from 
 the smaller diameter of the balance. Though thin balances 
 do undoubtedly increase slightly in size in the long vibra- 
 tions from centrifugal action, this theory is disposed of by 
 the fact that old chronometers do not accelerate after 
 re-oiling. Others aver that the unnatural connection of the 
 metals composing the compensation balance is responsible 
 for the mischief, and that after being subjected to heat the 
 balance hardly returns to its original position again. If 
 true, this may be a reason for exposing new chronometers 
 before they are rated to a somewhat higher temperature than 
 they are likely to meet with in use, as is the practice of 
 some makers ; but then chronometers accelerate on their 
 rates when they are kept in a constant temperature, and 
 also if a new spring is put to an old balance, or even if a 
 plain uncut balance is used. 
 
 It is noticed that when the overcoil of a balance spring 
 has been much bent or "manipulated" in timing, the 
 acceleration is almost sure to be excessive. This is just 
 what might be expected, for a spring unduly bent so as to 
 be weakened but not absolutely crippled, recovers in time 
 some of its lost elasticity. But however carefully a spring
 
 17 
 
 is bent the acceleration is not entirely got rid of, even 
 though the spring is heated to redness and again hardened 
 after its form is complete. There is little doubt that the 
 tendency of springs is to increase slightly in strength for 
 some time after they are subjected to continuous action, 
 just as bells are found to alter a little in tone after use. As 
 a proof that the acceleration is due to the bending of the 
 overcoil, Mr. Hammersley asserts that, if the spring of an 
 old chronometer is distorted and then restored to its original 
 form, the chronometer will accelerate as though it were new. 
 Helical springs of small diameter have been advocated by 
 some as a means of lessening acceleration, on the ground 
 that the curves are less liable to distortion in action than 
 when the springs are larger. Mr. T. Hewitt tells me it was 
 noticed that if the balance springs were tempered on a 
 larger tube than the one they were hardened on, as was 
 formerly the practice of some makers, the acceleration was 
 sure to be excessive. Springs elongate in hardening, and 
 it has been suggested that they afterwards gradually shorten 
 to their original length and so cause acceleration, but there 
 does not seem to be much warrant for this assumption. 
 Unhardened springs do not accelerate, but then they rapidl)' 
 lose their strength, and are therefore not used. Flat springs 
 do not accelerate so much as springs with overcoils. Palla- 
 dium springs accelerate very much less than hardened steel 
 springs. 
 
 Having summarized the various opinions of practical 
 experts, let us see how far they are founded on scientific 
 bases. 
 
 CHAPTER II. 
 
 Theoretically correct terminal curves. Effect 
 of disturbing influences. Of late years science has 
 lifted the veil which formerly obscured the reasons for 
 various operations in connection with springing which were 
 accepted as the correct procedure solely from the result 
 obtained. In many instances these empirical dicta were 
 
 B
 
 i8 
 
 founded on sound bases, but there is always the danger in 
 accepting rules given as "practical experience" that the 
 result observed has been merely the accidental presence of 
 unsuspected factors. 
 
 M. Phillips, a distinguished French mathematician, in- 
 vestigated the laws governing the action of the balance 
 spring and in " Memoire sur le spiral teglant" &c., which 
 was printed in 1861, published the result of his researches ; 
 giving a number of examples of theoretically correct 
 terminal curves. 
 
 Jules Grossmann, professor at the Locle School of 
 Horology, in continuation of Phillips' labours^ examined 
 various factors connected with the balance spring as in 
 operation in a timekeeper, and brought, as it were, the 
 subject into the domain of the watchmaker. 
 
 L,. Dossier reduced to greater simplicity some of Jules 
 Grossmann's results, and produced a most excellent manual 
 of the Theory of Timing. His work, ably translated from 
 the "Journal Suisse d 1 Horlogeric" by Mr. George Walker 
 and Mr. W. N. Barber, appeared a short time ago in the 
 " Horological Journal." 
 
 Several other writers have contributed to the sources of 
 theoretical and practical information on the subject as will 
 appear from acknowledgments made in due course. 
 
 The first requirement for a spring to be isochronous is 
 that the centre of gravity of the spring shall be coincident 
 with the centre of gravity of the balance. 
 
 In a cylindrical spring, furnished with terminal curves 
 we have to consider two portions ; first, the whole of the 
 coils forming a certain number of complete turns commen- 
 cing and ending at the same point C (Fig. n), and whose 
 centre of gravity falls exactly on the centre of the figure O, 
 then the part composed of the two curves C A and C' A', 
 and the added part of the spring C D C' of any length 
 whatever. 
 
 In order that the whole spring may have its centre of 
 gravity on O, it is only necessary to form the curves in such 
 a manner that the three portions of the spring A C, C C',
 
 19 
 
 and C' A', are in equilibrium, and have a common centre of 
 gravity in O. 
 
 To do this we shall have to find the positions of the 
 centres of gravity of G and G' which satisfy these conditions. 
 
 HLJ. ir. 
 
 I do not propose to follow the demonstration of this, 
 which may be obtained from various text books on 
 Graphical Statics, but it leads to Phillips' formula : 
 
 This may be stated as but involving the following two 
 conditions : 
 
 i st. That the centre of gravity of the curve be on the 
 line of O G, making a right angle with O C. 
 
 2nd. That the distance of the centre of gravity of the 
 curve from the centre of the spring be equal to the square 
 of the radius of the spring divided by the length of the 
 curve. 
 
 Phillips demonstrated that when the centre of gravity 
 of the spring has been thus established by construction, on 
 the axis of the balance, it will remain there whatever may be 
 the deformation of the spring, so that this latter, either in 
 contracting or expanding, always remains concentric with 
 itself. 
 
 He also gave a method of graphically constructing the 
 theoretic curves, and Lossier has devised a practical rule of 
 
 B2
 
 20 
 
 correction, which may be with advantage adopted, and 
 which I venture to reproduce. 
 
 The two terminal curves of the same spring need not be 
 identical in form and dimensions. If, in fact, one of the 
 curves is, for example, longer than the other, it will be 
 necessary, in order to satisfy the formula of Phillips, that 
 its centre of gravity be nearer to the centre of the spring in 
 order that its moment with relation to the centre may remain 
 the same. 
 
 L,et us suppose that one of the curves is twice as long as 
 the other, so that its length will be 2 /; in order that it may 
 comply with the conditions of Phillips' formula, since R 
 does not change, the distance of its centre of gravity from 
 the centre of the spring should be but one half that of the 
 
 O C~ 
 other curve, that is ' The moment of this curve will 
 
 then be 2 / X ' or / x O G, which is the same as the 
 
 2 
 
 moment of the other curve, therefore the equilibrium of the 
 whole spring will not be modified. 
 
 We see, consequently, that we may change the length, 
 the form, and the position of the curves as we like, provided 
 that the condition holds for each of them, that O G be equal 
 to the square of the radius, divided b)' the length of the 
 curve. 
 
 The theory of Phillips adapts itself equally well to the 
 flat spring whether it be for the exterior curve, which does 
 not sensibly differ in form from that of a cylindrical spring, 
 or for the interior curve. This latter ought to be considered 
 as composed of that part of the spring joining the largest 
 and the smallest coils and of a connecting curve with the 
 collet. 
 
 It results from this that the small connecting curve can- 
 not of itself alone comply with Phillips' conditions, since it 
 has also to agree with an archimedian spiral. However it 
 need not differ much from it, but the difficulty of its exact 
 reproduction from a drawing increases considerably with 
 its smallness. Fortunately as we shall see this inner curve 
 may in most instances be dispensed with, but if it is desired,
 
 21 
 
 Dossier suggests forming a curve something approaching 
 the theoretical shape and then correcting it as the action 
 of the spring when the watch is going may indicate. 
 
 It should be noticed that when a spring is furnished 
 with Phillips' curves, not only is its centre of gravity not 
 displaced, but the spring itself expanding and contracting 
 in a uniform manner on all sides, does not exert any 
 pressure or any effect of torsion on the axis of the balance. 
 
 It will be understood that no particular form of curve is 
 prescribed by Phillips. So long as the curve conforms to 
 the conditions recited the character of the curve is un- 
 important. Appended are four of several examples given 
 by Lossier. In all a spring theoretically imperfect distend- 
 ing towards A r , is represented by dotted lines and a suggested 
 alteration of the outer terminal renders it theoretically 
 perfect. 
 
 These springs are all right handed, that is the volute 
 developes, starting from the left of the centre O, upwards 
 to the right. And with such springs the isochronal correc- 
 tion is made by drawing that part of the curve to the right 
 of the centre line M O A r , towards the centre, or the part 
 
 Fig. 12. 
 to the left of the centre away from the centre, or by a
 
 22 
 
 combination of these operations. With left handed springs 
 the correction would of course be in the reverse direction. 
 
 In Fig. 12 an overcoil is formed of two quarter circles 
 having a radius equal to half the radius of the spring O P 
 joined by a straight line C. It will be observed that in this 
 case the termination of the overcoil is not carried in towards 
 the centre so that the stud remains in the same position as 
 for a flat spring. Mr. Kullberg adopted a somewhat similar 
 overcoil but with a short outer concentric termination for 
 use with curb pins and he spoke well of watches so fitted. 
 
 In Fig. 13 the original curve which was not isochronal 
 was formed o r hilf-a-circle as dotted. The full lines show 
 the alteration to an isochronal overcoil without disturbing 
 the point of attachment to the stud. The part to the right 
 of the centre line lias been carried towards O along the line 
 R O, while the part to the left has been drawn out from O 
 towards S. In this compound correction the lines O S and 
 R O forming the centres of the respective corrections should 
 form a right angle. 
 
 In Fig. 14 the curve lias been carried in from the right, 
 and in Fig. 15 carried out to the left of the centre. 
 
 Dossier gives a curve, easy to construct, fulfilling the
 
 Fig. 14. 
 
 condition necessitated by the regulator, and which may 
 serve as a basis for the exterior curve of a flat or a cylindri- 
 cal spring, subject to slight modification afterwards, accord- 
 ing to the performance of the watch. 
 
 Fig. 15- 
 
 From the centre of the spring (Fig. 16), trace an arc of a 
 circle with a radius of -67 7? (R being the radius of the
 
 exterior coil), the arc ought to comprise beyond the curb 
 pins an angle of of 83, and connect it with the exterior coil 
 by a semi-circle of a radius 
 
 This curve may serve for the internal curve of either a 
 cylindrical or flat spring. For this latter, a space of about 
 four coils is left between the collet and the interior coil, 
 connection being made by a coil constructed like that in 
 Fig. 16, taking for R the radius of the internal coil. The 
 
 Fig. 17.
 
 25 
 
 part b c (Fig. 17), of the coil being concentric with the 
 collet would touch this latter, which would completely 
 destroy the isochronism. To avoid this inconvenience, the 
 extremity is curved in such a manner that there remains a 
 space, equal to the distance between two coils, between the 
 collet and the beginning of the curve. 
 
 These curves only give a concentric development when 
 they are reproduced with exactitude ; but as it is extremely 
 difficult to obtain the desired precision in the execution, 
 and besides there may arise circumstances obliging one to 
 deviate from the rigorously theoretical form, it is practically 
 more simple to reproduce these curves as near as possible 
 by the eye, and then correct them according to the action of 
 the spring. 
 
 Method of Forming and Testing Terminal 
 Curves. Messrs. George Walker and W. N. Barber have 
 devised a way of forming the curve, together with a sort of 
 scale for testing it. According to Phillips the conditions to 
 be satisfied by a terminal curve are : 
 
 ist. That the centre of gravity of the curve fall on the 
 line O B making a right angle with O N (N being the 
 starting point of the curve). 
 
 2nd. That the distance from the centre of gravity of the 
 curve to the centre of the spring be equal to the radius of 
 the spring squared, divided by the length / of the curve. 
 
 Then O G = Jf- 
 
 Take the radius of the spring O N as a constant, and / 
 of any convenient length determined by experience ; from 
 these data calculate the distance of the point G from the 
 centre of the spring. From the centre O draw a circle 
 having a radius equal to the distance of the inner side of 
 stud hole from the axis of the balance (the same scale being 
 used as for the rest of the construction.) From centre G 
 draw a circle of such a radius as will include the point N ; 
 cut this circle out of uniformly thick material ; this circle 
 will then balance on the point G. 
 
 Take a rod of uniform section of the pre-determined
 
 26 
 
 length of the curve and place one end on the point N, the' 
 other end on the circle drawn for the stud hole ; this rod may 
 then be bent to any form so that the whole system balances 
 on the point G, when the conditions for isochronism will be 
 satisfied. 
 
 By means of Messrs. Walker & Barber's Tester the 
 moment of a terminal curve with reference to the axis of 
 the spring may be weighed against a mass so placed that its 
 moment is equal and opposite to that of the terminal curve. 
 
 A circular plate is accurately balanced on a centre which 
 has a small vertical movement in a tripod stand fitted with 
 levelling device, a stepped annular ring embraces the 
 balanced plate, so that when the centre is lowered the plate 
 
 FiK. is. 
 
 rests on the step of the annular ring, thus allowing manipu- 
 lation of the terminal curve without danger of damage to 
 the centre. The coincidence of the top surface of the ring 
 and the balanced plate shows when the whole is in 
 equilibrium. 
 
 The terminal curve is arranged on the surface of the 
 balanced plate, the centre of the plate representing the axis 
 of the spring. Having determined the two circles on which 
 the ends of the curve will lie, and calculated the position of 
 the centre of gravity of the curve as before, the moment of 
 the curve may be found by multiplying the weight of the 
 curve by the distance of its centre of gravity from the axis. 
 By placing a mass, say equal to the weight of the cuvvj, on
 
 27 
 
 the line BO produced, and at the same distance from O as 
 the centre of gravity of the curve, the theoretical conditions 
 will be satisfied when the whole system is in equilibrium. 
 
 Fig. i8.\, I'laii of Walker and Barber's Tester. 
 
 PLAN (the same objects are represented by the same letters in plan 
 and elevation) : 
 
 B V E=Terminal curve. 
 D=Dovetail slide. 
 
 P=Dovetail slip with vertical pin, to take weights, which 
 may be adjusted at any distance from the centre Oi 
 the plate. 
 C=Counterweight, to secure adjustment of the plate when 
 
 the slip P is moved to different positions. 
 ELEVATION R M=Stand with levelling screws. 
 I*. L. L.=Levelling screws. 
 
 N.=Balanced plate on which the terminal curve is 
 adjusted.
 
 Fig. i8u, Klevation of Walker and Barber's Tester. 
 
 S. S. S.=Studs, screwed into the stand, on which the 
 balanced plate rests while the terminal curve 
 is being manipulated. 
 E V.=Terminal curve resting on balanced plate. 
 
 W=Weight, to secure equilibrium when the curve 
 
 is correctly formed. 
 
 Q Double-ended bar, with boss, which is snapped 
 into recess in balanced plate, and carries 
 slides with pins to keep the end of the ter- 
 minal curve at right distance from the axis. 
 C=Connterweight. 
 
 F= Plumb-line, for levelling apparatus. 
 K=Pump centre, shown dotted. 
 H=Handle of eccentric for raising and lowering 
 
 pump centre. 
 
 U.U.=*Pins, to prevent circular motion of balanced 
 plate.
 
 To use the Tester. Let us suppose that the ter- 
 minal curve is resting on the balanced plate, and arranged 
 to fulfil the theoretical conditions. Its C of G will be at a 
 point G at a certain distance from the centre O and on a 
 line at right angles to a line joining the centre O to the 
 beginning of the curve, measured round the curve. If the 
 plate be free to move it will turn round the axis O in such a 
 direction that the line G O moves in a vertical plane with a 
 moment G O + the weight of the curve. 
 
 If in the same line, on the opposite side of the axis O, a 
 weight W be placed such that its moment is equal to that 
 of the terminal curve the plate will be in equilibrium. Now 
 if the terminal curve be changed in shape the equilibrium 
 will be destroyed, and the curve will no longer fulfil the 
 theoretical conditions. 
 
 To design any terminal curve, taking 100 mm. as a stan- 
 dard diameter, it is only necessary to know the length and 
 weight of curve from which the requisite moment to pro- 
 duce equilibrium may be found, and at the same time fulfil 
 the theoretical conditions. 
 
 The terminal curve should be made of some easily bent 
 and heavy metal to facilitate manipulation. 
 
 A spring with theoretically perfect terminal curves 
 would be correct if applied to a balance also perfect, and 
 not subjected to external influences. But such conditions 
 do not, of course, exist in watch work, and, therefore, the 
 consideration of all the factors with which the practical 
 horologist has to contend is the next step. 
 
 AN EXTERNAL FORCE APPLIED TO THE BALANCE AT THE 
 DEAD POINT, THAT IS WHEN THE SPRING IS QUIESCENT, 
 HAS NO EFFECT ON THE TIME OF OSCILLATION, AND THE 
 EFFECT OF A FORCE IS MUCH ACCENTUATED AS THE 
 DISTANCE BETWEEN THE DEAD POINT AND THE POINT OP 
 APPLICATION INCREASES. 
 
 AS A RULE THE DISTURBING EFFECT OF A GIVEN 
 EXTERNAL FORCE WILL VARY ACCORDING TO THE AMPLITUDE 
 OF THE OSCILLATIONS, AND WILL BE MUCH MORE MARKED 
 AS THE AMPLITUDE DIMINISHES.
 
 30 
 
 AN IMPULSE DELIVERED BEFORE THE BALANCE REACHES 
 
 ITS DEAD POINT QUICKENS THE TIME OF VIBRATION. 
 
 AN IMPULSE DELIVERED AFTER THE BALANCE HAS PASSED 
 ITS DEAD POINT EXTENDS THE TIME OF VIBRATION. 
 
 SIMILARLY, ANY ABSTRACTION OF FORCE IF TAKING 
 PLACE BEFORE THE DEAD POINT IS REACHED EXTENDS THE 
 TIME OF VIBRATION, WHILE ABSTRACTION OF FORCE AFTER 
 THE DEAD POINT IS PASSED QUICKENS THE VIBRATION. 
 
 IN THE CYLINDER ESCAPEMENT AND IN THE LEVER 
 ESCAPEMENT ALSO THE IMPULSE IS NOT EQUALLY DIVIDED 
 ON BOTH SIDES OF THE DEAD POINT, THEREFORE, SO FAR AS 
 THE INFLUENCE OF THE ESCAPEMENT GOES ALL CYLINDER 
 AND LEVER WATCHES MUST LOSE IN THE SHORT ARCS. 
 
 With these data let us examine what takes place in a 
 cylinder escapement, taking the angle of vibration as jj of a 
 turn. Let the angle of lift be 40, of which 5 are for lock- 
 ing, leaving 35 for impulse. The balance, starting at 120 
 from the dead point, returns to within 15 of this point 
 without being subject to any other influence than friction. 
 From 15 to o it will receive impulse in the same direction 
 as its movement, and consequently there will be an accelera- 
 tion. After the dead point is passed, the impulse will 
 continue for 20 ; the first 15 will produce a retardation 
 equal to the preceding acceleration, thus leaving 5 during 
 which the retardation produced will not be compensated. 
 
 The final effect of the lift will then be to increase the 
 time of the oscillations, because the angle of repose neces- 
 sitates a longer impulse on one side of the dead point than 
 on the other. The watch will thus lose more in the short 
 than in the long arcs, more in the hanging than in the lying 
 position. 
 
 In the lever escapement the minimum amount of lock- 
 ing is i, measured from the axis of the pallets ; and we 
 may take 2 as the maximum admissible. To determine the 
 same angle of locking, considered with relation to the 
 movement of the balance staff, we must replace the angle 
 measured at the pallet staff by the corresponding angle 
 moved through by the balance. Multiplying the former by 
 the ratio between the length of the lines representing the
 
 lever and roller, according as this ratio is 4^, 4, 3^, 3, the 
 angle moved by the balance, the pallets having i of lock- 
 ing, will be 4i, 4, 3^, or 3 degrees ; and for 2 of locking it 
 will be 9, 8, 7, or t> degrees. 
 
 If the locking at the pallets be 2, and the ratio between 
 the length of the lever and that of the roller is 4^, the effect 
 of the locking commences at 22^ from the dead point (for a 
 lift of 10 at the pallets), and ceases 22^ 9= 13^ from the 
 dead point. If this ratio be 3:1, the effect commences 
 about 15 and ends at 15 6 --= 9 from the dead point. 
 As the effect of a force acting on a balance depends on the 
 angular distance of the point of application from the dead 
 point, the effect of the angle of repose on the timing will be 
 less in the second case than in the first. 
 
 The advantage of escapements, giving a small arc of 
 connection with the balance, is apparent. An ideal escape- 
 ment would be one delivering the whole of the impulse by 
 a blow at the dead point. No such escapement exists, and 
 in our endeavour to approach this ideal the motion of the 
 pallets and arc of connection with the balance are some- 
 times too much diminished, leaving insufficient force to 
 maintain the vibrations. 
 
 There is a limit to variations in the proportion of the 
 lever escapement, though it is not precise. Small locking 
 angles are desirable, but the locking must be safe for all the 
 teeth of the wheel. For the same balance arc Swiss escape- 
 ments have usually a lesser lifting angle for the pallets than 
 English, and a longer lever. With too great a lifting angle 
 the tooth is likely to set on the impulse planes, and with too 
 little to set on the locking. 
 
 There is another variable disturbing factor connected 
 with the lever escapement, and that is the drop on to the 
 impulse pallet after the unlocking, which affords an excep- 
 tion to the general proposition that all disturbing influences 
 have less effect in the long than in the short vibrations. 
 Even with the lightest escape wheel there must be an interval 
 between the unlocking of the tooth and its contact with the 
 impulse plane, and owing to the draw on the locking face 
 the wheel receives a recoil at each unlocking which
 
 lengthens this interval. In the long vibrations the unlock- 
 ing is more energetic and the recoil greater, and, as the 
 balance is moving quicker, the pallet will evidently get 
 farther away before the tooth reaches it than it would in 
 the short vibrations. The loss of impulse before the dead 
 point from this cause will therefore be greater in the long 
 than in the short vibrations, and the effect will be much 
 more marked if the escape wheel is large and heavy, 
 because then its inertia will render its action more sluggish, 
 and the point where it impinges on the impulse pallet will 
 be farther along the plane than would be the case with a 
 smaller and lighter wheel. 
 
 With the chronometer and other single beat escape- 
 ments the variation of the time of oscillation of the 
 balance is not necessarily so pronounced. L,ossier 
 suggests shifting the dead point in the chionoiiieter 
 escapement, so that when the spring is at rest the 
 impulse pallet is nearer to the exit tooth. The impulse 
 may be equalized on each side of the dead point in this 
 way, and the short arcs quickened thereby, but a chro- 
 nometer out of beat is more likely to set. 
 
 Point for attachment of the spring to the 
 collet. The position of the point where the spring is 
 attached to the collet WITH RELATION TO THE PENDANT OF 
 THE WATCH is of prime importance, except in helical or 
 other springs provided with two isochronal curves. The 
 centre of gravity of a flat or other form of balance spring 
 without an internal terminal curve and pinned to the 
 collet in the usual way is continually shifted during the 
 action of the spring, and it will be evident that its poise, and 
 practically the poise of the balance, will be affected during 
 the going of the watch. If the spring were affected equally 
 in its coiling and uncoiling this displacement might be 
 ignored, but in L,ossier's treatise is a demonstration by Jules 
 Grossmann which shows that a spring distends more from its 
 normal radius in uncoiling than it contracts in coiling. 
 
 Advantage may be taken of this fact to neutralize in 
 many instances positional errors, and to quicken the short 
 arcs.
 
 The exact effect of varying the point of attachment 
 cannot be invariably predicated, but generally with a spring 
 developing to the right as in Fig. 19 the result was found to 
 be as follows : 
 
 
 Fij 
 
 j. 19. 
 
 
 
 Position of the 
 
 Pendant 
 
 point of 
 attachment. 
 
 left, 
 
 right, 
 
 up. 
 
 X 
 
 (loNVll. 
 
 Left 
 
 o 
 
 O 
 
 gain 
 
 loss 
 
 Right 
 
 o 
 
 
 
 loss 
 
 gain 
 
 Up 
 
 loss 
 
 gain 
 
 
 
 o 
 
 Down 
 
 gain 
 
 loss 
 
 o 
 
 o 
 
 Palladium springs, being heavier, would show a greater 
 result from the shifting of the centre of gravity than steel 
 springs. 
 
 THE PRACTICAL REsri/r is THAT A WATCH WITH A 
 
 BALANCE SPRING DEVELOPING UPWARDS TO THE RIGHT, AS IX 
 FIG. 19, HANGING (OR PENDANT VI 1 '), AND LOOKING AT THE 
 BACK THE POINT OK ATTACHMENT SHOULD HE TO THE LEFT 
 OF THE BALANCE CENTRE AND ON A HORIZONTAL LINE WITH 
 IT. 
 
 To facilitate obtaining this position Jules Grossmann 
 suggested drawing on a sheet of paper two lines at right 
 angles (A, B, and C, D, Fig. 20), and placing the movement 
 so that the centre of the balance coincides with the point of 
 intersection, and that a line, E, F, parallel to A, B, passes 
 through the centre of the dial. The line G touches the 
 front of the stud and may be continued to the centre of the 
 
 c
 
 34 
 
 balance O, as in Fig. 21. T is the point for the attachment 
 to the collet, and the point for the stud at P can be marked 
 on the spring. 
 
 Occasionally it may happen from some constructional 
 defect that with the spring so pinned in there is a loss 
 on turning the watch pendant left. In such a case the plan 
 would be to cut a piece say seven-eights of a turn out of the
 
 35 
 
 eye of the spring so as to bring the point of attachment 
 towards M. 
 
 There appears to be but little doubt that so far as flat 
 springs with an exterior overcoil are concerned the old and 
 contradictor}' data as to pinning in springs at complete 
 coils, or short or over complete coils really amounted to 
 nothing more than getting the point of attachment to the 
 collet into the most favourable position. Cutting and 
 repinning the spring, as here recommended, in order to 
 shift the point of attachment could be avoided if the stud 
 were capable of being moved round concentrically with the 
 balance. Messrs. Walker and Barber and Messrs. P. & A. 
 Guye have devised adjustable studs of this description, but 
 in the majority of watches cutting and repinning the spring 
 must be resorted to. 
 
 For a left-handed spring or one developing from the 
 centre upwards to the left, as in Fig. 22, the result was as 
 follows : 
 
 Pendant 
 
 Position of the 
 
 point of 
 attachment. 
 
 Left 
 Right 
 Up 
 Down 
 
 left, ri^ht, up, 
 
 o o loss 
 o o 
 
 gain loss o 
 
 loss gain o 
 
 down, 
 gain 
 
 less 
 o 
 o 
 
 With a flat spring having no overcoil there is another 
 factor which should not be passed unnoticed. Here the 
 
 C 2
 
 36 
 
 spring in coiling and uncoiling is subjected to a pushing 
 from and drawing towards the stud which is tantamount 
 to an additional winding or unwinding of the spring. 
 
 If the spring consists of complete turns the stud and 
 collet attachments are both on the same side of the centre, 
 and then there is no. tendency to twist the collet when the 
 spring starts from its point of rest as there would be if the 
 spring were, say, a quarter of a turn short, or a quarter of 
 a turn over complete coils. In action the relative position 
 of the attachments is continually altering, and the effect in 
 any particular case is difficult to calculate. 
 
 The general conclusion shown by Dossier is in favour of 
 pinning in at a quarter of a turn over complete coils as 
 giving the best result in vibrations of ordinary amplitude ; 
 a quarter of a turn short showing a more even performance 
 than either complete turns or half turns. 
 
 With an adjustable stud a flat spring without overcoil 
 may be applied so as to bring the collet attachment in a line 
 with the centre of the balance in accordance with the 
 prescription given for obtaining the best result in positions, 
 and also to make a quarter of a turn over complete coils. 
 But with the position of the stud fixed one of these desi- 
 derata inav have to be sacrificed. 
 
 CHAPTER III. 
 
 Compensation for varying- Temperature. John 
 Harrison, in his efforts to obtain the Government reward of 
 ^2o,coo offered for a timekeeper sufficiently accurate to 
 determine the longitude, was the first to apply a corrective 
 to the balance spring for the temperature variations. 
 
 One of the chief features of his celebrated watch, now 
 preserved at the Royal Observatory, Greenwich, is a 
 bimetallic arm fixed at one end, and carrying at its free 
 ends two pins, to embrace the balance spring near its 
 outer point of attachment. This is his description of the 
 device : " The thermometer kirb is composed of two thin
 
 37 
 
 plates of brass and steel riveted together in several places, 
 which, by the greater expansion of brass than steel by heat, 
 and contraction by cold, becomes convex on the brass side 
 in hot weather, and convex on the steel side in cold 
 weather ; whence, one end being fixed, the other end 
 obtains a motion corresponding with the changes of heat 
 and cold, and the two pins at this end, between which the 
 balance spring passes, and which it touches alternately as 
 the spring bends and unbends itself, will shorten or 
 lengthen the spring." 
 
 Harrison at first provided additional curb pins for mean 
 time adjustment, but had to abandon them ; for it is clear, 
 if they were placed behind the pins on the compensation 
 curb, they would not act, and, if placed in front, the move- 
 ment of the temperature pins would be ineffective. 
 
 Mudge, Gumming, Kmery, and other celebrated horolo- 
 gists of the period adapted curbs on the Harrison principle. 
 John Arnold, leaving the spring, decided to compensate for 
 the loss of its elasticity by means of a laminated balance 
 rim composed of brass and steel soldered together, and 
 shaped to a circular form by bending with plyers. 
 
 It remained for Thomas Earnshaw to improve the con- 
 struction. He turned the steel centre and fused the brass 
 around it, thus obtaining a sound junction of the metals 
 and a perfectly circular rim. 
 
 The investigations of the two eminent French artists, 
 Ferdinand Berthoud and Pierre Le Roy, were nearly con- 
 temporaneous with those of Mudge and Arnold. Berthoud 
 used a gridiron arrangement of brass and steel to com- 
 pensate for temperature errors, and fitted his timekeeper 
 with two balances geared together. Le Roy experimented 
 with a balance composed of two glass thermometer tubes 
 containing mercury and placed radially, the bulbs, which 
 were filled with alcohol, being furthest from the centre of 
 motion and the tubes pointing inwards. 
 
 Berthoud, in 1773, tabulated the effect of temperature 
 upon one of his marine watches. He reckoned that in 
 passing from 32 to 92 (Fahr.) it lost per diem by
 
 Expansion of the Balance . . . . . . 62 sees. 
 
 The loss of Spring's Elastic Force .. .. 312 
 
 Elongation of the Spring .. .. .. .. 19 ., 
 
 393 or 6111.33 s. 
 
 Doubtless Berthoud's observation was correct as far as the 
 total amount of the temperature error goes, but there 
 appears to be no warrant for assuming that a part of the 
 loss was due to elongation of the spring. The thickness 
 and the width of the spring would be increased in precisely 
 the same proportion as the length, and as the strength of a 
 spring varies as the cube of its thickness, the spring would 
 be absolutely stronger for a rise of temperature if the rela- 
 tive dimensions only were considered.* 
 
 Sir G. B. Airy, by experiment in 1859, showed that a chro- 
 nometer with a plain un compensated brass balance lost on 
 its rate 6'ii sees, in 24 hours for each degree of increase in 
 temperature. 
 
 To counteract this effect of change of temperature, 
 chronometers and fine watches are furnished with a balance 
 as constructed by Earnshaw, which expands and contracts 
 with heat and cold. The halves of the rim are free at one 
 end and fixed at the other to the central arm, as shown in 
 Fig. 23. The arm and inner part of the rim are of steel, and 
 the outer part of the rim of brass melted on to the steel. 
 Formerly the brass was preferred to be twice the thickness 
 of the steel. Yvon Villarceau, after investigating the action 
 of the compensation balance, decided that the relative 
 thickness of the metals should be in the inverse proportion 
 of the square roots of their elasticity, giving the thickness 
 of brass to steel as 17 to 12, and lately there has been a 
 disposition to make the proportion of brass to steel as 3 to 2, 
 thus following nearly Villarceau's prescription. Mr. Wright 
 tells me he has found the middle temperature error sensibly 
 less with the thickness of brass so reduced. As brass 
 expands more than steel, the effect of an increase of tem- 
 
 * It is curious that Berthoud's statement should have been accepted 
 without question by all authorities and writers till Mr. Wright, the 
 able theoretical teacher at the Horological Institute, pointed out its 
 fallacy in 1882.
 
 39 
 
 perature is that the brass in its struggle to expand bends 
 the rim inwards, thus practically reducing the size of the 
 balance. With a decrease of temperature the action is 
 reversed. The action, which is very small at the fixed ends 
 of the rim, increases towards the free ends, where it is 
 greatest. In a marine chronometer there is one large 
 weight at about the middle of each half rim, which is 
 shifted to or from the fixed end, according as the compen- 
 sation is found on trial to be less or more than is desired. 
 In pocket chronometers and watches a number of holes are 
 drilled and tapped in the rim, and the compensation is 
 varied by shifting screws with large heads from one hole to 
 another, or by substituting a heavier or a lighter screw. 
 The compensation screws should fit well in the thread and 
 be always screwed home till the heads touch just the rim. 
 If the thread of a screw is at all loose there is danger of 
 distorting the rim by turning the head too tight against it. 
 As a precaution, the shoulders of the screws are sometimes 
 rounded off, so that only the surfaces close to the threads 
 can be in contact with the rim. 
 
 In the marine balance there are two screws with heavy 
 nuts on opposite sides of the rim, close to the central arm, 
 for bringing the chronometer to time. These nuts are slit, 
 as shown in the drawing, to clasp the screw spring-tight and 
 so avoid blacklash. In watch balances there are four such 
 
 Fig. 23. 
 Marine Chronometer Balance, 
 
 Fig. 24. 
 Compensation Balance for Watches.
 
 screws placed at at equal distances round the rim. These, 
 of course, are not touched for temperature adjustment. 
 
 Fig. 23 shows the Marine Chronometer Balance. A 
 compensation watch balance is shown in Fig 24. In all 
 but the finest work the quarter mean-time screws are not 
 fitted with nuts, but made with heavy heads, and screwed 
 into the balance from the outside like the compensation 
 screws. It will be observed that the cuts in the rims of the 
 balances are not radial. The object of cutting them at an 
 angle, as shown, is that the free end of the rim may be 
 stopped from bending unduly towards the centre when the 
 balance is roughly handled. 
 
 Before using a compensation balance it is the practice of 
 good adjusters to spin it close to the flame of a lamp, so as 
 to subject it to a higher temperature than it is ever likely to 
 meet with in use. Mr. Arthur Webb raises the tempera- 
 ture till the ends of the rim butt at the notch. The balance 
 is then placed on a cold plate, and afterwards tested for 
 poise. If necessary the balance is trued, and the operation 
 repeated till the balance after heating is found to be in 
 poise, or is rejected. The late A. P. Walsh told me that to 
 get a low temperature after heating he immersed the balance 
 in ether. Mr. Arthur Webb suggests that compensation 
 balances would be more certain and permanent in their 
 action if they were hardened and tempered after the brass is 
 melted on to the steel. It appears probable that uncertainty 
 of action may sometimes be traced to careless hammering 
 of the brass, which is better compressed by rolling than by 
 hammering. 
 
 Middle Temperature Error. The element of 
 
 irregularity known under this name is one of the most 
 perplexing that troubles the maker of superior timekeepers. 
 The object of applying a compensation balance to a watch 
 or chronometer is that its vibrations shall be performed in 
 the same time notwithstanding that from changes of tem- 
 perature the energy of the balance spring is varied. In the 
 ordinary form of balance this is sought to be effected by 
 causing the balance to contract with a rise of temperature
 
 41 
 
 when the balance spring is weaker, and to expand with a 
 fall of temperature when the balance spring is stronger. 
 
 Airy demonstrated that the loss in heat from the 
 weakening of a steel balance spring is uniformly in propor- 
 tion to the increase of temperature. But the compensation 
 balance fails to meet the temperature error exactly : the 
 rims expand a little too much with decrease of temperature, 
 and with increase of temperature the contraction of the 
 rims is insufficient ; consequently a watch or chronometer 
 can be cor rectly adjusted for temperature at two points only. 
 
 While the energy of the balance spring varies equally 
 for equal increments and decrements of temperature, the 
 number of vibrations made by the balance in a given time 
 varies not inversely as the distance of its weights from the 
 centre, but inversely as the SQUARK of the distance of the 
 centre of gyration from the centre of motion. If a chro- 
 nometer keeping mean time at 60 Fahr. were tried first at 
 30 Fahr. and then at 90 Fahr., and the compensation 
 weights on the second trial approached the centre the same 
 distance as they receded from it in the first trial, the 
 chronometer might be expected to lose at 30 and gain at 
 90. But the difficulty is actually, as already stated, too 
 much expansion in cold, or too little contraction in heat. 
 The action of the balance is complex. The weights do not 
 move radially, and, with an increase of temperature, while 
 the weights are- moved inwards the central arm lengthens 
 and the parts of the rim adjacent thereto move outwards. 
 
 However as the adjustment can only be perfect at two 
 points, chronometer makers generally arrange so that the 
 compensation is right at the two extremes of temperature 
 which the chronometer is likely to encounter, leaving the 
 greater error at the middle temperature. The chronometer 
 will then be found to gain at all temperatures within the 
 extremes, and to lose if exposed to temperatures outside 
 the extremes. 
 
 A marine chronometer is usually adjusted at 45 and 90, 
 unless special adjustment is ordered to suit particularly hot 
 or cold climates; pocket watches at about 50 and 85. In a 
 range of 40 Fahr. there would be a middle temperature
 
 42 
 
 error of about 2 sees, in 24 hours with a steel balance spring. 
 The amount of the middle temperature error cannot be 
 absolutely predicated, for in low temperatures, when the 
 balance is larger and the oil thicker, the arc of vibration 
 is less than in high temperatures when the balance is 
 smaller and the oil thin ; consequently its time of vibration 
 is affected by the isochronism, or otherwise, of the balance 
 spring and the action of centrifugal force on the balance. 
 Advantage is sometimes taken of these circumstances to 
 lessen the middle temperature error by leaving the piece 
 fast in the short arcs. 
 
 To avoid middle temperature error in marine chro- 
 nometers, various forms of compensation balances have 
 been devised, and numberless additions or auxilaries have 
 been attached to the ordinary form of balance for the same 
 purpose. Of these some examples will doubtless be of 
 interest. 
 
 The late John Hartnup, director of the Liverpool 
 Observatory, advocated the acceptance of the ordinary 
 compensation balance, and proposed that navigators should 
 be instructed to allow for the middle temperature error, 
 which, after trial of 1,000 chronometers, with steel balance 
 springs, he found amounted to 1-5 sec. in 24 hours for a 
 change of 15 above or below the temperatures at which the 
 chronometers had been compensated. This can only be 
 accepted as an approximation, because, as we have seen, 
 the amount of the Middle Temperature Krror is to some 
 extent dependent on other disturbing influences. 
 
 With palladium balance springs the middle temperature 
 error is very sensibly less than with steel, which may be 
 accounted for in the following way : 
 
 There is a greater temperature error to be compensated 
 with palladium than with steel springs, and the compensa- 
 tion weights with palladium have consequently to be moved 
 more towards the free ends of the rim. With a rise of 
 temperature the superior expansibility of the brass or outer 
 metal of the rim not only carries the weights towards the 
 centre, BUT, BY KLOXC.ATIXG, CURLS THIS RIM ; AXD THK 
 
 PATH OK THK WKIGHTS IS THKRRKORR VARIKD ACCORDIXG
 
 43 
 
 TO THEIR POSITION ox THE RIM. It happens that the 
 movement of the weights at the point where the action of 
 the palladium spring causes them to be placed more nearly 
 conforms to what is theoretically required than if they were 
 farther back, as they would be with a steel spring. But the 
 middle temperature error is not yet eliminated or provided 
 for, and its amount, though reduced with a palladium 
 spring, is even less constant than with steel. 
 
 Checking the outward movement of the rim in low tem- 
 peratures, or causing a subsidiary weight to be carried 
 inwards in high temperatures may be taken to represent the 
 two principles on which most auxiliaries are constructed. 
 
 At the Guildhall Museum is a watch by John Leroux, 
 who was admitted as an honorary member of the Clock 
 Makers' Company in 1781, and the date mark in the case of 
 this watch (k) seems to indicate that it was made in 1785. 
 If so, it is of especial interest, and may be regarded as the 
 earliest attempt to modify the action of the balance. As 
 will be seen from Fig. 25, the rim is an entire circle of steel, 
 
 25. 
 
 with four very light arms. Midway between the arms are 
 left four projections above the upper surface, for receiving 
 the four mean-time screws, as the one marked i ; on the 
 upper portion of the balance axis is fitted, friction tight, a
 
 44 
 
 moveable steel bar, each end of which carries a circular- 
 lamina, similar to the rim of an ordinary compensation 
 balance, each piece of lamina being firmly attached at one 
 end to the bar by two screws; at the acting end of each 
 lamina is a circular flat weight fixed by two screws. 
 
 From the outer rim as before described are two other 
 projections for receiving two small check screws, // ; these 
 are made to come in contact with the brass or outer surface 
 of the laminae. Adjustment for compensation is effected 
 by moving the bar forward or backward by pliers applied to 
 its flatted sides. This motion of the bar causes a different 
 part of the laminae to bear against the screws, //, thereby 
 lengthening or shortening the effective compensation 
 length. 
 
 Molyneux's auxiliary (Fig. 26) is attached by a spring to 
 each end of the central arm, and is acted on by the free ends 
 of the rim in high temperatures only. A screw in the end 
 of the rim and another in the auxiliary serve to adjust the 
 action as may be required. -Molyneux's patent also covered 
 the use of a short laminated arm instead of the spring by 
 which the auxiliary is attached to the central arm, and 1113113- 
 successful auxiliaries are now made in that way, of which 
 Fig. 27 may be taken as an example. The laminated 
 auxiliary arm is generally about one half as wide and one 
 
 Q 
 
 Kit,'. 27. 
 
 quarter as thick as the rim of the balance, and it can be 
 made to act either in high or in low temperatures according 
 as the brass is placed inside or outside of the arm. Some 
 chronometer makers consider there is an advantage in
 
 45 
 
 constructing the auxiliary for low temperatures, because 
 compensation balances, after exposure to heat, are found to 
 set slightly inward instead of returning exactly to the 
 normal position ; and. while an auxiliary acting in high 
 temperatures would add to the error when such setting 
 occurred, one acting in low temperatures would tend to 
 neutralize it. 
 
 John Poole's best-known auxiliary (Fig. 28) consists of 
 a piece of brass attached to the fixed ends of the rim, and 
 carrying a regulating screw, the point of which checks the 
 outward movement of the rim in low temperatures. Poole 
 also used a butting screw, fixed so as to stop the outward 
 movement at the free end of the rim ; and a more recent 
 inventor met with success by connecting the two iree ends 
 of the balance rim with a thin wire, and so checking their 
 outward movement in low temperatures. 
 
 William Hardy attempted to avoid the middle tempera- 
 ture error without using an auxiliary by altering the form of 
 the balance. Abandoning the cylindrical lamhue used by 
 Arnold and Karnshaw, he used a straight laminated bar 
 of brass and steel, the brass being underneath (Fig. 29). 
 A hole in the centre of the bar served to attach it to the 
 
 Fitf. 28. Fitf. 29. 
 
 staff, and at each end of the bar was a stalk carrying a 
 spherical weight. These weights could be made more or 
 less active as compensators by screwing them up or down 
 on the stalks which had threads cut on them. By slightly 
 curving the laminated bar upwards or downward, the 
 weights could be made to approach or recede from the centre 
 of the balance more or less as desired. At first sight it 
 appeared that the difficulty of the middle temperature error 
 had been overcome. But to obtain sufficient compensation 
 the laminated bar must be so thin and the stalks so long as 
 to leave the balance wanting in rigidity. Nevertheless,
 
 4 6 
 
 Hardy's attempt led to the invention of many other balances 
 on the same principle. 
 
 Massey, in 1814, patented (No. 3854) a balance very 
 much resembling Hardy's in appearance, but in which the 
 central arm was solid, and the upright stalks formed of 
 brass and steel. On a central arm, similar to Hardy's, 
 E. J. Dsnt mounted laminated pieces bent over like 
 staples thus C to increase the action. 
 
 In 1849 John Hartnup invented the balance shown in 
 Fig. 30. The rims are composed of brass and steel, as 
 usual, but they are neither upright nor flat, but bevelled, or 
 placed at an angle midway between these two positions. 
 The central arm a is also bimetallic, the brass being upper- 
 most, and connecting the arm with the sections of the rim 
 are two other bimetallic strips b c, the brass of these being 
 underneath, and the steel on top ; c c are the weights, and 
 at the ends of the rim the screws for timing and poising. 
 
 vSubsequently Victor Kullberg constructed a flat-rimmed 
 
 Fit;. 30. Hartnup' s balance. Fig. 31. Kullberg' s balance, 
 
 balance, as shown in Fig. 31. Here the central arm A, and 
 the rim B, composed of brass and steel, are in one piece, but 
 in the arm the brass is on top and in the rim underneath, 
 so that with a rise of temperature the ends of the arm bend 
 down and the free ends of the rim are lifted upwards and 
 inwards. The weights C are carried on stalks, which also 
 afford a support for the timing nuts I). 
 
 E. T. Loseby's Balance. Chronometers with this 
 balance were remarkably successful at the Greenwich trials 
 from 1846 to 1853. The rims are bimetallic of brass and
 
 47 
 
 steel, shorter than usual ; at the end of each segment of the 
 rim is a cup- joint in which is placed a glass vessel consist- 
 ing of a curved arm and a bulb, which contains mercury 
 (Fig. 32). The curved arm is sealed with a little air in it to 
 ensure the continuity of the thread of mercury when it 
 contracts. It is apparent that by bending the cup-joint the 
 direction of the glass arms may be altered, and in this way 
 a very exact temperature adjustment obtained. 
 
 Loseby's invention was admitted by the Greenwich 
 authorities to be an improvement, but his application for a 
 reward was refused, the Admiralty, as a sort of recompense, 
 buying a larger number of his chronometers than they 
 otherwise would have done. This rather shabby treatment 
 disgusted Loseby, who gave up chronometer- making and 
 
 I'ijr. 32. Loseby's balance. 
 
 returned to Warwickshire, where he died in 1890 ; but the 
 manufacture of balances on his principle has been occasion- 
 ally revive .I, with, I believe, encouraging results. 
 
 The late C. van Woerd, of Waltham, Mass., patented an 
 ingenious method of construction. Across the periphery of 
 each half of a steel rim, for about a third of its length, near 
 the fixed end, he formed V-shaped notches, Brass was 
 fused on to the steel and turned off level with the circular 
 part of the steel rim, leaving it bare except at the notches, 
 which were thus filled with prisms of brass. By varying 
 the number of notches he claimed to be able to obtain any 
 desired amount of movement of the free ends of the rim.
 
 CHAPTER IV. 
 
 METHOD OF PROCEDURE IN SPRINGING AND ADJUSTING. 
 
 Applying a flat spring. The number of vibrations 
 required depends, of course, on the train. For a modern 
 watch with a seconds train in which the fourth wheel turns 
 once in a minute, divide the number of the fourth wheel 
 teeth by the number of leaves in the escape pinion ; 
 multiply the quotient by 30 (double the number of escape 
 wheel teeth), the product will be the number of impulses the 
 balance receives in a minute. If it is an 18.000 train the 
 number will be 300, that is 5 beats a second. A 16,200 
 train gives 270 a minute ; 4-5 a second. 
 
 Should the \vatch not have a seconds train we must go 
 back to the centre wheel which rotates once in an hour. In 
 this case multiply together the numbers of centre wheel, 
 third wheel, fourth wheel teeth and 30. Also multiply 
 together the number of leaves of third pinion, fourth pinion, 
 escape pinion and 60. Divide the first sum by the last and 
 the quotient will be the vibrations per minute. 
 
 Having ascertained the number of vibrations required 
 proceed to select what appears to be a suitable spring. Lay 
 the spring with its centre coinciding with the cock jewel 
 and mark on the coil that is to be the outer one exactly 
 where it would enter the stud hole, bearing in mind that 
 the spring should be rather small in diameter than large ; 
 
 FOR A SPRING TOO LARGE IN RELATION TO THE INDEX 
 PINS AND STUD IS PRETTY WELL SURE TO HE SLOW IN THE 
 SHORT ARCS. 
 
 The next thing is to count the vibrations of the balance 
 when connected with this spring. 
 
 Put the eye of the spring over the balance staff down on 
 to the balance, and press the collet on to the spring so as to 
 confine the inner portion that will be broken away for the 
 eye. If it should happen, as in rare cases it may, that the 
 collet when pushed on the staff its proper way fails to hold
 
 49 
 
 the spring, the collet may be reversed and pushed on with 
 the smaller side of the hole first.* 4 
 
 Place in a convenient position on the bench a watch that 
 is known to be keeping correct time, then take hold of the 
 spring firmly with tweezers at about the distance of the 
 curb pins short of the spot marked for the coil to enter the 
 stud ; then lift the spring so that the balance hangs horizon- 
 tally and the lower balance pivot is just above the watch 
 glass. Give the balance about half a turn so that it will 
 vibrate for over a minute, Owing to the contraction and 
 expansion of the spring the balance will also acquire an up 
 and down motion. 
 
 With 3'our eyes on the watch dial count every alternate 
 vibration registered by the contact of the pivot with the 
 glass till a spring is obtained that gives 75 double vibrations, 
 if it is for an 18,000, and from 67 to 68 during half-a- minute, 
 if for a 16,200 train. The observation may extend over a 
 lesser or a greater period, always remembering that for an 
 18,000 train, there are to be 5 DOUBLE vibrations every 
 TWO SECONDS, and for a 16,200 train 4^ in the same time. It 
 is better for the spring to give say half a vibration less in a 
 minute rather than more, for it is sure to be a little faster 
 when the eye of the spring is broken out to suit the collet. 
 Extend the observation and counting to one minute if any 
 doubt exists respecting the watch which is taken as a 
 standard. It should not only be a good timekeeper but 
 also have the seconds dial correctly spaced, and the seconds 
 arbor concentric therewith if exact results are to be obtained 
 for shorter periods than a minute. Of course, if a chrono- 
 meter is available it may be used instead of a watch. Many 
 prefer to listen to the taps of the pivot on a watch glass and 
 to watch the seconds hand of a regulator. Others will 
 
 *In the absence of the collet a bit of beeswax or putty powder the 
 size of a small pin's head may be placed upon the end of the lower 
 balance staff pivot to catch the eye of the spring under, in order to count 
 the vibrations ; but the use of wax or powder, a portion of which may be 
 afterwards transferred to the watch, is not to be recommended. Besides, 
 clasping the spring between the balance and the collet places that part 
 of" the eye which will be broken off out of action, and the result will 
 be more exact than catching up the spring close to its centre. 
 
 D
 
 50 
 
 dispense altogether with the audible report on a watch glass 
 and catching up the spring will give a slight twist of the 
 hand holding the tweezers, and watch the number of vibra 
 tions given during 4 or 5 seconds by listening to the beats 
 of a regulator. 
 
 Yet another plan is to hold to the ear a watch having the 
 same number of vibrations as it is desired the balance under 
 trial should have, and note the discrepancy, if any. The 
 first method given is, I think, best suited to a tyro, and will, 
 I am sure, result satisfactorily if carefully conducted. 
 
 Although very accurate results may be obtained by 
 "vibrating" on the watch when continued for, say, two 
 minutes, the use of a "Vibrator" is recommended by Mr. 
 Geo. Walker. With it there is certainly the advantage that 
 the attention may be withdrawn from the balance under trial 
 for a few seconds without the observation being lost. 
 
 The "Vibrator" consists usually of a vertically pivoted 
 box, with glass top, containing a balance and spring accu- 
 rately adjusted to give a standard number of vibrations. If 
 adjusted to, say, 18,000 vibrations, another small box is 
 provided to give 16,200, and sometimes a third giving 
 14,400. Any of the two latter may be slipped over the 
 first when these numbers are required. Attached to the 
 side of the pivoted box there is a vertical pillar from which 
 projects a horizontal arm carrying tweezers to hold the 
 outer end of the spring. The construction of the tweezers 
 is such that the spring may be drawn out or in by simply 
 turning a small milled nut. 
 
 The balance and spring to be tested are adjusted over 
 the standard balance, the outer end of the spring being held 
 by the tweezers so that the bottom pivot remains in contact 
 with the glass throughout the entire vibration. The two 
 balances are then set in motion, exactly together, by a push 
 from a lever pivoted to the stand. With a little practice it 
 is easy to quickly adjust the length of spring so that the 
 balances are running accurately together. 
 
 For the drawing of the Vibrator (Fig. 32A) I am indebted 
 to M. Faure. 
 
 When a spring of the right strength is found it may be
 
 pinned to the collet. To break away sufficient of the eye 
 for it to pass over the collet place the spring on the board 
 
 Fig. 32A. Vibrator. 
 
 paper ; grasp it with tweezers at the point where it is to be 
 severed ; with a screwdriver, or other suitable implement, 
 bend the eye close to the tweezers, and it will break away. 
 
 Assuming the position of the stud to be fixed, and that 
 the spring should start from the collet on a level with the 
 centre of the cock jewel if the watch were on edge with the 
 pendant uppermost, place the spring on the cock, \vith the 
 mark already made for the outer attachment coincident with 
 the face of the stud, and mark on the eye of the spring the 
 point where it is to spring from the collet. The spring 
 should start away from the collet with an easy curve, and 
 must not hug the collet, or isochronism will be out of the 
 question. With tweezers, having a rounded grip rather 
 than a sharp edge, hold the eye a little farther up the spring 
 
 D 2
 
 52 
 
 than the mark just made, and with a peg gently bend round 
 to the requisite angle the piece which is to enter the collet. 
 Do not press with the peg close to the tweezers, or the 
 spring may break. A sketch of the collet showing the hole 
 for the spring may be made on the board paper as a guide 
 for the angle. The piece bent round, of course, is curved. 
 Hold it in the tweezers, and, with the assistance of the peg, 
 straighten it. 
 
 Occasionally it may be that the only spring of the required 
 strength available is too small with relation to the stud. If 
 unhardened a spring may be expanded by placing it on a 
 flat plate previously warmed, and heating it to a bluing 
 temperature. It must not be heated beyond a bluing tem- 
 perature, but the operation may be repeated more than 
 once, and the spring will expand at each heating. To 
 ensure the flatness of the spring remaining undisturbed it 
 may be covered by a small flat piece of brass or glass, which 
 must not be too heavy, or it ma)' prevent distention of the 
 spring. 
 
 The Collet. The depth of the collet should not be 
 less than twice the diameter of the central hole, if room 
 allows. The taper of the hole and the taper of the balance 
 staff where it fits should exactly correspond, so that the 
 collet does not rock, and that part of the staff should be left 
 grey to allow a tighter grip for the collet. The bottom of 
 the hole should be rounded, so that it readily enters on the 
 staff; the bottom outer edge of the collet should also be 
 rounded, to facilitate the entrance of the tool used for 
 raising it. The hole for the reception of the spring must be 
 exactly at right angles to the central hole, for if the spring 
 has to be forced or bent to suit the hole it is hopeless to 
 expect good results. Beginners are often in such a hurry 
 to pin the spring on that they proceed when the hole is 
 obviously not in the proper direction. The end of the 
 spring should be passed into the hole, just kept in position 
 with a temporary pin and examined. If any alteration is 
 needed, the spring is to be taken out and the hole broached 
 in the direction required till it is right. A round pin of 
 hard brass is then fitted to the hole. When this is done,
 
 53 
 
 take a waste piece of the spiing, insert it in the hole, and 
 flatten away the pin till it and the spring together fill the 
 hole properly. Then cut off the pin, leaving but little to 
 project, so that there is 110 danger of its touching the spring 
 in action. 
 
 The usual plan of pinning in is to put the collet on a 
 broach held between the thumb and finger of the left hand 
 while the pin is fitted and the spring pinned on, taking the 
 precaution to push a piece of paper on the broach before the 
 collet or the spring may touch the fingers, and in the case 
 of a damp hand the spring would be likely to be spoilt by 
 rusting. By previously fitting the pin as directed, there is 
 now very little fear of injuring the spring. The pin can be 
 pressed home with a small joint pusher. 
 
 Another mode of pinning on is to place the collet on the 
 board paper, adjust the spring to the collet, and with a 
 short piece of boxwood sloped away at the end press the 
 collet on the board ; the pin can then be fitted with comfort, 
 and without danger of shifting the collet. 
 
 Yet another way of proceeding to pin the spring to the 
 collet is to get a plate of brass about an inch square, and tap 
 a hole in the middle of it less in size than the hole in the 
 collet. A screw passing through the hole in the collet fixes 
 it to the plate. The advantage of this plan is that the 
 serface of the plate serves to show at once if the spring lies 
 in the proper plane. Care must be taken with the under- 
 side of the head of the screw, so that it does not scratch the 
 collet. 
 
 The collet may now be put on an arbor, and the arbor 
 rotated in the turns to ascertain if the spring is true. In 
 setting the spring it must only be touched close to the eye. 
 Steady timekeeping will be out of the question if the spring 
 is bent to and fro in reckless attempts to get it true. The 
 eye should be brought gradually round to get it in circle, 
 taking care not to overdo it. When this is right, and the 
 spring is also true on the face, many good timers heat the 
 spring and collet to a blue to set the eye. The outer coil 
 may now be pinned into the stud. 
 
 It will be prudent to first pin the spring in temporarily
 
 54 
 
 and notice if the eye is true with the cock jewel, or very 
 slightly towards the stud. If it is not, and the stud cannot 
 be shifted, the stud hole had better be broached in the 
 required direction. Or, of course, the spring can be slightly 
 bent close to the stud hole to bring the eye right with the 
 cock jewel. The largest part of the pin should be towards 
 the body of the spring, and therefore the hole should be 
 broached from that direction. First fit a round pin, and 
 then flatten it to suit with a waste piece of the spring as 
 directed when pinning to the collet. The pin at the stud 
 may be left fairly long. The waste outer part of the spring 
 should not be broken off close to the stud, but sufficient 
 should be allowed to project through the stud to allow for 
 any letting out of the spring that ma)- be required. 
 
 The curb pins should be of brass, only just free of the 
 spring, parallel inside and tapered back from the point on 
 the outside, without burr or roughness that would allow a 
 lodgment of the spring. Mr. Kullberg considered the curb 
 pins should not reach below the bottom of the spring, so 
 that there is no room left for the second coil to jump in. 
 The curb pins should not extend far beyond the stud, or the 
 outer coil may jump out. But apart from the jumping of 
 the spring it is objectionable to have the curb pin or pins 
 far from the stud. It is noticed that watches in which this 
 distance is great are very difficult to time. 
 
 The watch should be tried, lying and hanging, for 
 twenty-four hours in each position. If it gains in the short 
 vibrations the spring may be taken up, and if it loses in the 
 short vibrations let out a little. But if the alteration has 
 the opposite effect to that desired, as in some instances it 
 ma)', proceed in the contrary direction. 
 
 If the short arcs are slow, some recommend bending the 
 outer curb pin away from the spring and pressing the inner 
 one hard against it, so that the outer coil is really dragged 
 out of its position. This seems to be a most objectionable 
 nostrum, which doubtless originated through a spring too 
 large having been applied. 
 
 If the short arcs are fast the curb pins are sometimes 
 slightly opened as a remedy, but this is also open to objec-
 
 55 
 
 tion, for the pins then obviously cease to be parallel with 
 each other, and with undue play between the curb pins the 
 spring will get worn where they touch it, and the going of 
 the watch will not be satisfactory. 
 
 Unless theie is some grave constructional fault the 
 variation will not be so excessive but that the watch may be 
 brought to time with the index, dividing the error between 
 the long and short vibrations.*' 
 
 Where there is no seconds hand, the usual plan is to 
 make a mark on the fourth wheel ; but Mr. Bickley recom- 
 mends placing the watch to the ear and counting every 
 alternate beat for a minute or two, and comparing with a 
 regulator or watch going to time. However, a small dot 
 on the fourth wheel is an almost imperceptible disfigure- 
 ment, and will generally be found useful as the watch is 
 getting pretty close to time. 
 
 Macartney's Collet Adjuster. For twisting a 
 balance spring collet to position, the special tweezers 
 
 Fig- 33. 
 
 shown in the sketch (Fig. 33) are admirable. One jaw is 
 concave, and the other in the form of a knife edge 
 
 Plose's Collet Lifter. This is an excellent form of 
 pliers for raising a balance spring collet. The jaws are 
 curved and have knife edges. The leverage is much more 
 even and effectual than with the old style by one-sided 
 wedging. There is a set screw to prevent the edges closing 
 too far and so injuring the staff (Fig. 34). 
 
 * All watch balances should, I think, be furnished with screws for 
 read}' adjustment of weight. The extra cost, even for cheap watches, 
 is not worth consideration. There is no doubt that the proper way 
 of bringing the piece to time is by varying the balance instead of 
 lengthening or shortening the spring the the stud, which upsets any 
 prearranged position for the inner point of attachment.
 
 56 
 
 For pressing a collet home, a table with holes or slits 
 such as is depicted tinder the head of " Stud " will be found 
 useful. 
 
 Fig- 34- 
 
 Balance Spring Holder. This tool (Fig. 35) forms 
 a kind of self-acting tweezers, and is useful for holding the 
 outer coil of the spring, especially when it is desired to get 
 
 35- 
 
 the length accurately with the balance in its place in the 
 watch before pinning to the stud, as in the case of Breguet 
 spring prior to turning the overcoil.
 
 57 
 
 Breguet Spring 1 , or Flat Spring with overcoil. 
 
 The selection of a spring, counting its vibrations with the 
 balance and pinning to the collet may be conducted in the 
 same way as for a spring without overcoil, except that if it 
 is decided to keep the point of attachment to the collet in 
 a horizontal line with the cock jewel when the watch is 
 placed pendant up, allowance must be made for the bending 
 in of the spring to reach the stud, which, of course, is 
 nearer to the centre than a stud for a flat spring would be. 
 Experienced springers after selecting a suitable spring, by 
 vibrating, break awa) r the eye very accurately, to ensure the 
 desired result. The overcoil will not materially affect the 
 number of vibrations in a given time. It is, however, 
 difficult to frame a rule of any practical use to the beginner, 
 because theie are so many factors to be considered ; the 
 diameter of the spring, the length of the overcoil, the 
 distance of the stud from the centre, the thickness of the 
 wire from which the spring is made, all affect the amount to 
 be broken away at the eye. To be quite certain the spring 
 selected will bring the watch to time after the overcoil is 
 formed, it may be desirable to try it in the watch, and for 
 this purpose the holder (fig. 35) will be useful for gripping 
 the outer coil. 
 
 The overcoils of watch springs are usually turned with 
 curved nosed steel tweezers like Fig. 37. These curves 
 
 Fig. 37- 
 
 should be of suitable shape for the particular spring, and 
 be well polished. Beginners are not sufficiently careful in 
 forming the curved noses, When a bit of .spring wire is 
 gripped between them they must show an exactly circular 
 outline. If defective they would twist the spring, even in the 
 hands of an experienced operator. In forming the Phillips' 
 curves, some watchmakers use hot pliers of the requisite 
 shape to set the curve to the required form, but this is 
 exceptional treatment, and does not appear to be necessary.
 
 Fig. 36 is a useful little clamp of brass, by Mr. T. D. 
 Wright, for holding a spring while turning 
 the overcoil, the jaws of which are curved to 
 the outer coil of a spring of average size. 
 The upper screw passes through a clearing 
 hole in the loose jaw and is tapped into the 
 body of the tool ; the pin below is a steady 
 pin ; the bottom screw tapped into the loose jaw serves to 
 keep the opening between the jaws parallel to suit the 
 thickness of the spring ; a is a screw, and b a steady pin for 
 attaching it to an old watch plate or other table. 
 
 With this tool, and care, the turning of the overcoil will 
 be comparatively easy. Having decided on the length and 
 form the overcoil is to take, and of which you will have 
 either a pattern or a photograph, or a drawing on paper or 
 metal, break off a piece of the waste outer coil and bend it 
 to the shape required ; then pull it out again to coincide 
 with the outer coil, and from the point where the spring 
 enters the stud, or if there is to be an index, the middle 
 position of the curb pins, mark off the length of the curve. 
 Then fix the outer coil in the clamp just short of the 
 point where the overcoil should start, and place underneath 
 it a sketch of the overcoil for guidance. There are now 
 two distinct operations first, to raise the overcoil above 
 the plane of the spring, and then to form the curve. If the 
 two are attempted at once, it is most likely that in unskilled 
 hands the spring will be twisted and practically spoiled. 
 Therefore with strong tweezers raise the spring sufficiently, 
 and then with the curved-nosed tweezers bend the curve 
 very gradually without twisting the wire, taking care not to 
 overdo it in any part, for bending to and fro is sure to 
 deteriorate the spring. 
 
 Presuming that the copy be a theoretically isochronal 
 curve, let the arch of the overcoil, if it deviate at all, be 
 rather nearer the centre than the other way, so as to be, if 
 anything, quicker in the short arcs. When the curve is 
 correct it may be set by heating the spring to a bluing 
 temperature. Grandjean suggests gently heating the spring
 
 59 
 
 on a white enamel dial till a drop of oil placed alongside the 
 spring begins to smoke. 
 
 If the spring is to be free, without an index, pinning to 
 the stud is a simple affair, requiring no instruction beyond 
 what has already been given when dealing with a flat 
 spring ; but with an index there are two methods of pro- 
 cedure, one of which is to be recommended. The plan 
 formerly most generally followed was to carefully shape 
 that part of the outer end of the curve where the index pins 
 would operate into a circular form of a radius equal to the 
 distance of the curve from the centre of the index ; that is, 
 from the centre of the cock, and then proceed to pin to the 
 stud. But in almost every case this last operation distorted 
 the circular part on which pains had been bestowed, and 
 the shaping of this had accordingly to be done over again. 
 The best plan is to pin to the stud first, and then shape the 
 circular part to suit the path of the index pins. This way 
 is more expeditious, and the spring is subjected to less 
 bending. 
 
 If the balance spring has been carefully selected the 
 watch will not show a considerable variation from mean 
 time. L/eaving the final adjustment to be made by screwing 
 the mean time screws in or out, it may now be brought to 
 within about two minutes a day by, if necessary, adding to 
 or taking weight from the balance. Usually two opposite 
 screws are removed and heavier or lighter ones substituted, 
 as may be required. Platinum screws are used when extra 
 heavy ones are needed. The overcoil of the spring should 
 not be touched for mean time adjustment. 
 
 Possible tests and corrections now to be made may be 
 classed under the heads of: 
 
 Isochronism. 
 
 Compensation for temperature. 
 
 Position errors. 
 
 Mean time adjustment. 
 There is no absolute rule as to the sequence of these
 
 6o 
 
 operations. Some adjusters proceed in one way while 
 others, equally successful, pursue a somewhat different 
 course. Isochronism is, as a rule, dealt with first, because 
 it is possible more easily to carry it to completion in 
 presence of errors coming under the other heads enume- 
 rated. Much, however, depends on the character of the 
 watch, and the performance required from it. Between the 
 straightforward procedure adopted with a medium class 
 piece and a fine watch which has to be subjected to the 
 exacting ordeal of an Observatory test there is a wide 
 difference. In the latter case the margin of error allowed 
 *is so narrow that perplexing constructional inaccuracies, 
 quite distinct from the balance and spring, have to be 
 located and rectified. 
 
 We will first take, as an example, a going barrel watch 
 not required to be adjusted closely for positions. 
 
 Rack or Holder for Watches on Edge. The 
 primitive and most usual rest for watches placed on edge 
 for positional adjustment consists 
 of a rack or rest of wood having 
 V-shaped notches to suit various 
 sized watches. 
 
 A convenient movement holder 
 sometimes used is shown in the 
 annexed cut (Fig. 38). Itis adjust- 
 able, and capable of being turned 
 in any desired way for position 
 timing. 
 
 Timing Box. A brass box should 
 be provided for the reception of an 
 uncased watch movement while it 
 is being timed. The edge of the 
 pillar plate rests in a rebate, and a jjjj 
 cover with a glass let into the top 
 screws on to the box, and keeps 
 the movement in position.
 
 6i 
 
 ISOCHRONISM ADJUSTMENT. 
 
 Tests for isochronism may extend over different periods, 
 but variations are always reduced to 24 hours for compari- 
 son. Although approximations may be first obtained by 
 observations at short intervals, 24 hour trials should, if 
 possible, be allowed as the rates become close. Some 
 adjusters check the watches just when they happen to be at 
 leisure, and record the duration of each trial in minutes, 
 reducing the variation to 24 hour periods by the aid of 
 logarithms, or a slide rule to save extended calculation. 
 Aliquot parts of 24 hours can, however, generally be 
 managed by systematic procedure, and there is then less 
 liability to error from miscalculation. Trials of less than four 
 hours' duration can scarcely be regarded as a reliable guide. 
 Dial up is general!}' taken to represent the long vibrations, 
 and the mean of two opposite vertical positions the short 
 vibrations. With short trials of going barrel watches the 
 same turns of the mainspring should be used for each test 
 of the same kind. If we begin with four hour trials it will 
 be convenient to start in the morning, to wind the main- 
 spring but one turn so as to get the shortest vibration the 
 balance will have in ordinary use, and to set the watch to 
 mean time or in agreement with a reliable regulator. After 
 four hours pendant up the watch is found to have lost three 
 seconds, and its variation from mean time 3 is noted. The 
 mainspring should then be wound about half-a-turn, and 
 the watch placed pendant down. After four hours running 
 it is found to be 8, having lost 5 sees., and its variation 
 from mean time 8 is noted. The mainspring may then 
 be fully wound to get the longest vibration, the watch 
 placed dial up and allowed to run for four hours ; it is then 
 found to be 4, having gained 4 sees. The mean variation 
 of the two positions representing the short vibrations is 
 represented by 4, and the variation in the long vibra- 
 tions by + 4 ; as 4 hours, the duration of each trial, are 
 one-sixth of 24 hours each of these results is multiplied by 
 6 to obtain the variation in 24 hours, giving respectively 
 24 and -\- 24, which added together 48 ; the short arcs 
 are said to be 48 sees. slow.
 
 02 
 
 If the variations had been plus in each case or minus in 
 each case one total would have been subtracted from the 
 other,and the remainder would have represented the daily rate. 
 
 The most usual way of making the long and short 
 vibrations isochronous is by altering the form of the over- 
 coil. When the short arcs are slow, 
 as in this case, the arch of the over- 
 coil is closed slightly, and a little 
 more of the body of the spring added 
 to the overcoil. When the short arcs 
 are fast, the arch of the overcoil is 
 made a trifle flatter, and a little of 
 the overcoil taken back into the body 
 of the spring. Fig. 39 shows clearly Fig. 39. 
 
 what is meant. If the full line represents the original form 
 of curve, it would be altered in the direction of b to quicken 
 the short, and in the direction of a to quicken the long arcs. 
 To alter the shape of the overcoil the spring is held with 
 a pair of tweezers with brass faces curved exactly to suit it, 
 while the alteration is made with a similar pair. Alterations 
 must be made very gradually, for if bent too much the spring 
 is likely to be spoilt in bending it back. 
 
 After the requisite alteration is made, the watch may be 
 again set to mean time in the morning with the mainspring 
 fully wound, and after 24 hours running, dial up, it is found 
 to have gained 35-5 sees., this variation -f- 35-5 sees, having 
 been noted, the watch with about two turns of the main- 
 spring wound may be tried 12 hours pendant up ; it is then 
 found to be + 50*5 sees, fast, having gained 15 sees ; this is 
 noted ; the mainspring wound about a turn and a half and 
 the watch placed pendant down for 12 hours ; it is then 
 found to be 59-5 sees, fast, having gained 9 sees. The short 
 arcs are now slow, at the rate of 11-5 sees, in 24 hours ; the 
 balance spring therefore requires a slight alteration in the 
 same direction as before. 
 
 After this has been done the watch is fully wound, again 
 set to mean time and after 24 hours' trial, dial up, is found 
 to be, say, 36 sees. fast. The watch is noted + 368. With
 
 63 
 
 the pendant up for 12 hours, it gains 21 s., making it + 575. 
 And with the pendant down 12 hours, it gains 15 s., and is 
 then + 72 s. So that the variation in the long vibrations is 
 + 36, and the sum of the variations in the short vibrations 
 is -f 36 also. The watch is now said to be isochronous, but 
 there is still a position error between pendant up and pen- 
 dant down of 6 seconds in 12 hours. 
 
 Twelve hours' trial with pendant to the right shows a 
 gain of 275., and 12 hours' trial with pendant to the left a 
 gain of 13 s., making a position error of 14 seconds in 12 
 hours in the quarter positions. 
 
 The mean time screws may be drawn out sufficient to 
 give a loss of 38 seconds in 24 hours, thus dividing the 
 error. The piece would then show a variation from mean 
 time in 24 hours running in each position : Dial up 2 s. ; 
 pendant up +48.; pendant down 8s.; pendant right -f- 
 1 6 s. ; Pendant left 12 s. 
 
 This is not at all a bad result, and the watch being a 
 little faster pendant up than dial up is likely to approach 
 still nearer in these positions as the oil thickens. We may, 
 therefore, leave it as sufficiently near. 
 
 Position errors, which are due to escapement faults, 
 bad jewelling, and other constructional inaccuracies, are 
 often confounded with a want of isochronism ; but a watch 
 may be perfectly isochronous, and yet have very large 
 position errors. 
 
 If the short vibrations are not more than a turn in 
 extent, the vertical positions could be equalized by drawing 
 out the quarter screws of the balance nearest the fast 
 position, and setting in the ones nearest the slow 7 position, 
 and this is the remedy generally adopted. If the short 
 vibrations are a turn and a quarter, altering the screws will 
 be useless. Should the short vibrations be a turn and a 
 half, which is not very likely, the opposite method would 
 have to be resorted to, viz., the screws opposite the fast 
 positions must be set in, and those opposite the slow 
 position drawn out. This is called "timing in reverse." 
 The best authorities strenuously object to tampering with 
 the balance, and prefer to leave the position errors alone.
 
 6 4 
 
 No doubt it is a grave fault to set the balance out of poise, 
 and as the extent of the vibration falls off from thickening 
 of oil and dirt the correction is destroyed and an additional 
 error introduced. Of course, with a first-rate watch, where 
 everything isof the best, the position errors will be very small. 
 
 There is another method employed by some to lessen 
 the position errors, but it is not less objectionable, and is 
 certainly not sufficiently efficacious to be worth considera- 
 tion if the position errors are large. It consists of bending 
 the balance spring (not the eye), so that the eye, instead of 
 being true with the cock jewel, presses against the staff 
 in the fast positions ; in the preceding example, the 
 spring would be moved from the pendant, and to the right 
 of it. The effect of this is to cause more friction on the 
 balance staff pivots at those points. 
 
 A watch out in positions may often be brought nearer 
 by turning the roller of the escapement half-way round on 
 the balance staff, slight defects in the pivots and holes 
 being thus neutralized. 
 
 There should not be much difference whether the short 
 vibrations are caused by decrease of motive power or by 
 change of position from lying to the rack, yet the results 
 will not be absolutely identical. If the discrepancy is very 
 great, it would probably indicate defective jewelling, holes 
 too long or too large, or not sufficiently polished. The 
 length of the straight part of the jewel hole should be 
 about equal to its diameter. 
 
 In adjusting marine chronometers and chronometer 
 clocks for isochronism, the long and short arcs are not 
 obtained by changing the position of the instrument, but 
 the arc of vibration is reduced about one-quarter by letting 
 down the mainspring, or reducing the power in any other 
 convenient manner. For instance, a stiffish spring may be 
 brought to bear on the fourth wheel pinion with sufficient 
 pressure to absorb as much of the force as may be desired. 
 
 The method of recording observations varies with 
 different adjusters. The subjoined arrangement appears to 
 be convenient. Symbols are used to indicate the the positions 
 follows : Dial up = Q 5 dial down = (") ; Pendant up = "J"
 
 pendant right = \ ; pendant left= \ ; pendant down = _L- 
 The time is set down in hours and quarter hours. 
 
 
 o 
 
 'C 
 
 
 | ; 
 
 aJ 
 
 
 _o 
 
 aJ 
 
 $"S 
 
 ^ c 
 
 .5 
 
 I l^ : I-s 
 
 . 
 
 efl 
 
 ii 2 
 
 "rt 
 
 I 
 
 2-a 
 
 X 
 
 
 ^ 
 
 >i 
 
 > 
 
 Q 
 
 E< P 
 
 D 
 
 
 jo O S ' ** +3 
 
 "e> 
 
 ijj 
 
 OJ CS 
 
 
 '3 
 
 SJ 
 
 s 
 
 "S 
 
 CH O S 
 
 ed 
 
 cS 
 
 q 
 
 J 
 
 Oct. 
 
 2 
 
 i turn 
 
 Sa 
 
 R 
 
 T 
 
 12 
 
 4 
 
 -31 
 
 
 Shorts 48 sees. 
 
 ,, 
 
 ,, 
 
 12 
 
 3 
 
 
 
 4P 
 
 4 
 
 8) 
 
 24 
 
 slow. Spring set. 
 
 
 
 Full 
 
 4 
 
 8 
 
 O 
 
 8p 
 
 4 
 
 4 
 
 +-24 
 
 
 3 
 
 Full 
 
 7fa 
 
 R 
 
 o 
 
 7|a 
 
 24 
 
 -35'5 
 
 +-35 '5 
 
 
 4 
 
 2 turns 
 
 7|a 
 
 +-35;5 
 
 T 
 1 
 
 7|P 
 
 12 
 
 12 
 
 *-59'5 f 
 
 4-24 
 
 Shorts 1 1 -5 sees, 
 slow. Spring set. 
 
 5 
 
 Full 
 
 8a 
 
 R 
 
 
 
 8a 
 
 24 
 
 ^36 
 
 +-36 
 
 Set to mean time. 
 
 6 
 
 3 turns 
 
 8a 
 8p 
 
 +-36 
 +-57 
 
 T 
 1 
 
 8p 
 Sa 
 
 12 
 12 
 
 4-57} 
 
 4-72 ) 
 
 +-36 
 
 
 Adjustment of a Fine Watch. Where the best 
 possible performance is to be obtained from a watch in 
 every position and in varying temperatures an exact 
 standard of time for comparison is essential. Without this 
 it would be foolish to undertake the task. Even with an 
 Observatory time signal at frequent intervals there should be 
 two good regulators to check a doubtful signal before 
 making any alteration to the watch as it is approaching 
 completion. 
 
 However, assuming all the necessary apparatus to be at 
 hand the whole of the mechanism of the watch should 
 undergo a rigorous scrutiny prior to proceeding with the 
 timing and adjustment. It is often the next to nothing, the 
 very slight errors of construction which, if overlooked, 
 give trouble at the end. Taking, as before, a going barrel 
 watch with a lever escapement we will suppose that an over- 
 coil balance spring has been applied as already directed. 
 The escapement should then be examined in accordance 
 with the procedure given in Chapter IX., p. 132. 
 
 The escapement having been passed as correct, and the 
 
 E
 
 66 
 
 mainspring full} 7 wound the. balance should have a vibration 
 of not less than a turn and a half nor more than a turn and 
 three-quarters with the dial up, and not less than a turn 
 and one eighth with the pendant up. 
 
 If with the pendant up the arcs are only a turn in extent 
 a stronger mainspring should be tried, and this remedy will 
 probably prove effective ; of course, if a balance too large 
 or heavy has been selected a fresh balance must be obtained 
 and the watch resprung. 
 
 Assuming the vibration to be satisfactory, the watch may 
 be brought somewhere near to mean time say, to within 2 
 minutes a day by, if necessary, adding to or taking weight 
 from the balance. 
 
 Bonniksen recommends springing the watch so that 
 there is at this stage a loss of four minutes a day to allow 
 for weight to be taken from the balance in poising it, but 
 this is not the usual practice. 
 
 Then the compensation should be proceeded with and 
 almost perfected. If tests for isochronism were attempted 
 first variations due to changes of temperature might easily 
 be mistaken for want of isochronism. The compensation 
 cannot be quite concluded at this stage, because, subse- 
 quently poising the balance may slightly derange it. On 
 the other hand it is useless to think of finally poising the 
 balance now, because re-arrangement of the screws, and a 
 possible set of the balance during the compensation adjust- 
 ment will probably affect the poise. But the fact of the 
 balance being untested for poise render? it necessary to 
 provide perfectly level surfaces in the refrigerator and oven 
 on which to lay the watch so that want of poise will not 
 affect its rate. 
 
 With the compensation correct, or nearly so, try the 
 balance for truth in the callipers or on the poising tool as 
 directed in the chapter devoted to the subject ; get it as 
 perfect as possible and then proceed finally to test the poise 
 of the balance and spring in the watch. The mainspring 
 should be wound but little so as to give only a full turn or 
 rather less when pendant up. It is very rarely that the 
 vibration cannot be sufficiently reduced if the stop work is
 
 67 
 
 removed. In an extreme case it may be necessary to 
 substitute a weaker mainspring. 
 
 Suppose the watch when horizontal gains at the rate of 
 30 seconds a day. Try it pendant up ; its rate is then -f 36. 
 Pendant down it is -f *6. 
 
 It may be assumed that the balance is too heavy at the 
 bottom when the watch is pendant up and as a ready im- 
 provement for a watch which is not to be adjusted in positions 
 the timing screws nearest this position are proportionately 
 screwed in, and the opposite ones similarly withdrawn. 
 
 But clearly the heaviest part of the balance ma.y not be 
 exactly opposite the pendant, therefore let us try the 
 quarters : 
 
 We find with pendant right the daily rate is + 32. 
 ,, ,, pendant left ,, ,, -f- 21. 
 
 Probably the quarters would not be so close on account 
 of constructional errors and the possible presence of these 
 renders it unsafe to accept the deviation as due wholly to 
 want of poise in the balance. 
 
 Therefore wind the watch fully so that the vibration is 
 quite a turn and a half, and try round the vertical positions 
 again. 
 
 Pendant up, its daily rate is + 40. 
 ,, down, ,, ,, + 20. 
 
 ., right, + 35. 
 
 left, + 25. 
 
 It is again hardly likely there would be no other disturb- 
 ing elements, but accepting these figures as an example we 
 may ascertain if it is an open face watch that the balance 
 when at rest is too heavy between the IV. and the V. This 
 may be adjusted by screwing in a timing screw or nut at 
 this pointer by proportionally screwing in the adjacent ones. 
 
 ABSENCE OF ISOCHRONISM, AND VARIATIONS BETWEEN 
 HANGING AND IKYING. SUMMARY OF CAUSES AND 
 
 REMEDIES. 
 
 The natural tendency of the escapement which causes a 
 loss in the short arcs. 
 
 Various escapement faults ; excess of shake in pivot 
 
 E 2
 
 68 
 
 holes, or insufficient freedom in pivot holes ; escape wheel 
 too heavy ; bad train depths ; escapement out of beat. 
 
 Variations between dial up and dial down. 
 Balance spring badly pinned to collet. Any twist in the 
 spring here is almost sure to cause variation in these posi- 
 tions. Probably the hole in the collet is not exactly in the 
 proper direction, and the best plan is to have a new collet. 
 
 Want of uniformity in escapement end shakes, causing 
 the end of the impulse pin to graze the lever of the safety 
 finger ; or the balance to graze the escape cock. 
 
 If the impulse pin is not perfectly upright it may in one 
 of these positions be free as it passes in and out of the 
 horns, and in the other positions may graze. 
 
 A short balance-staff pivot may in one position just rest 
 on its conical shoulder. 
 
 A defective or unevenly set end stone. 
 
 Excessive or careless oiling may cause a drop of oil on 
 some part of the escapement to spread and create sufficient 
 adhesion in passing an adjacent surface to retard the 
 balance and diminish its vibration. See especially that 
 contact between the lever and banking pins is free and not 
 sticky from oil. 
 
 Variations in the Quarters, Positional Errors. 
 Want ot poise in the balance. 
 
 Want of poise of the lever and pallets. 
 
 Too little draw, causing the safety pin or finger to rub 
 on the roller ; this may occur only in one position, owing 
 to want of poise in the lever and pallets. 
 
 Extra Long Arcs defective. If the extra long 
 arcs are fast, the fault may be that the impulse pin knocks 
 the outside of the lever. To correct this a weaker main- 
 spring may be substituted for the existing one. 
 
 If the balance spring is pressed out of its proper position 
 by one curb pin, though there is play between the two, it 
 may be that the spring leaves the curb pin and becomes 
 free in the extra long arcs, causing a loss. 
 
 The balance when expanded to its fullest extent through 
 natural action in low temperature and by centrifugal force 
 may just touch some part of the mechanism.
 
 69 
 
 Cylinder watches and carriage clocks with this escapement 
 after cleaning, will often so increase in vibration that a bank- 
 ing error is introduced. A weaker mainspring is the remedy. 
 
 In a cylinder watch which has been cleaned a sticky 
 banking will sometimes give trouble ; brushing is often not 
 sufficient to remove the holding tendency. Both pin and 
 block should be scraped. 
 
 Isochronism. The short arcs may be quickened by 
 rectifying faults in the escapement. 
 
 By increasing the angle of draw on the pallets. 
 
 By altering the overcoil of the balance spring. 
 
 By altering the point of attachment of the spring to the 
 collet. 
 
 By centrifugal force ; e.g. 
 
 A small hole is drilled through each of the half rims of 
 the balance close to their fixed ends, and, after trial, 
 broached out till the desired equality is obtained. This 
 remedy was, I know, successfully adopted by the late 
 B. Dennison. The greatest care must be taken to broach 
 the holes alike, for if one half of the rim is weakened more 
 than the other the balance will be out of poise when moving 
 quickly. This method is open to the objection that the 
 balance is rendered less rigid and more likely to be bent. 
 
 By using a shorter balance spring. 
 
 A shorter balance spring to give the same number of 
 vibrations will, of course, be thinner, and to reach the hole 
 in the stud will have more open coils. 
 
 If there is play between the curb pins ; close them till the 
 spring is but just free ; there should be no apparent motion 
 of the spring when its action is viewed through a glass. 
 
 By using a stronger mainspring. 
 
 As already explained (p. 63), the short arcs may be 
 quickened by putting the balance out of poise, but this 
 must be regarded as an illegitimate method, because the 
 remedy is but ephemeral. 
 
 So far I have endeavoured to summarize the practice of 
 various craftsmen. On request, Mr. W. N. Barber has 
 favoured me with a method of procedure, which, from his
 
 yo 
 
 high position as a theoretical and practical horologist, 
 cannot fail to be a reliable guide. 
 
 " To find the size of the balance spring make a small 
 prick in the board paper, put the balance pivot in this hole 
 and make another prick against the end of a screw-head, 
 the distance between the two pricks gives the diameter of 
 the spring. 
 
 The balance being cut, trued and poised with the roller 
 on in its permanent position, and each quarter screw turned 
 out about one turn and a half, the spring may be selected, 
 as described in pp. 49-50. 
 
 If no attention is going to be paid to the position of the 
 internal point of attachment, and no interior terminal curve 
 is to be used, break the eye of the spring out by putting it 
 on a flat plate of metal and cutting the spring with a sharp 
 graver, the eye must be broken out to such an extent that 
 when pinned on the collet there will be a space between 
 the collet and the first coil equal to the space between two 
 coils of spring ; hence, in breaking out, the bit of spring 
 which will go in the collet hole must be allowed for. The 
 pin having been fitted, as in p. 52, before pushing the 
 pin home, set the spring as nearly flat as possible, judging 
 of this by twisting the broach round in the fingers ; then 
 push the pin in tight by means of a very strong pair of 
 Uveezers, or small pliers ; this must be a perfectly solid job 
 or the spring cannot be got true and flat. When pushed in, 
 the pin must not project at either end if the collet be of the 
 usual round slit variety, because the coils of a close spring 
 are so near that there would be danger of touching the pin 
 end ; if the collet be of the flat-sided type the small end of 
 the pin should just peep through the collet, to facilitate a 
 possibly needed pushing out. 
 
 If an interior terminal curve is to be used, break out the 
 eye of the spring so as to leave a space equal to that which 
 was occupied by about four coils ; but an interior curve is 
 not to be recommended since its practical disadvantages 
 outweigh its theoretical attractions. 
 
 To get the spring true and flat, take a pivot-broach, blue 
 it, file and stone round, and having a nearly straight part
 
 about -05 mm. to 'i mm. in diameter. Put this between the 
 collet and spring, close up to the eye, and holding both 
 broaches together in the left hand the spring can be bent 
 inwards by pressing on it with a pair of tweezers beyond 
 the broach ; the broach being between the point where the 
 spring emerges from the collet and the part of the spring 
 being bent in ; if it is required to bend the spring out, 
 reverse the process, taking the broach a little farther along 
 and pressing between eye and broach ; the eye must be 
 bent very carefully round by this or some similar method, 
 and when approximately true and flat take the collet off the 
 broach and put it on an arbor in the callipers or turns, a fine 
 pointed pair of tweezers will now be required to complete 
 the truing process ; a very small eyed needle fitted to a 
 handle and the end of the eye stoned off makes a useful tool 
 for this purpose ; if the collet be a flat-sided one, the spring 
 can be easily got at, and can be bent by the aid of a very 
 small pair of curved -jaw tweezers. Getting it true will 
 probably send it out of flat, and so on, this must be got 
 over by observing which way the spring goes out of flat 
 when bent in for truth and then giving the spring such a 
 twist out of flat in the opposite direction that the bending 
 in for truth will bring it also flat. Unless the pivots are 
 unusally fine there will be no danger in using the staff as 
 the arbor to true the spring on. As a refinement it is 
 desirable that the collet, with its pin and a little bit of 
 spring in place, be itself in poise ; with a flat-sided collet 
 this will be nearly the case, but the slit in the round collet 
 needs compensating for; this may be done by putting the 
 slit at the back as usual and filing out with a seconds file, 
 and polishing with a rotating rod of metal, a small hollow 
 just in front of where the spring comes out of the collet ; 
 this hollow also allows the tweezers to get in to bend the 
 spring. Poising the balance with the collet on is useless 
 if this latter is not in poise and is going to be turned 
 into a different position. 
 
 Next see that the stud hole is clear of shellac, .soft 
 bread, etc., and is broached out sound and parallel to the 
 cock, then place the collet and spring on its seat on the
 
 72 
 
 balance, using a bone punch to press it down with, hold the 
 spring with a pair of tweezers and count up the vibrations as 
 explained at p. 49. The tweezers may have a brass slide 
 put on as in a pair of sliding tongs, if thought advantageous. 
 When to time, notice which screw-head, point or hole the 
 tweezers holding the spring are opposite or nearest to, then 
 put the spring between the curb-pins and pull through the 
 stud hole until the above desired point is between the pins, 
 the index being in the middle, then hold the whole affair 
 up by the cock and count up exactly. By counting for two 
 or three minutes as a final trial the average error should 
 not exceed -5 to i minute per day. When correct for time 
 nip off the waste spring at the back, close up to the stud and 
 remove balance and spring. 
 
 After seeing that the curb-pins are in good order, upright 
 and just far enough apart to fit the spring, yet not grip it 
 (if the pins require bending they must be bent with a knee 
 bend, so that they are parallel to each other and do not 
 form a taper slit), measure the radius of the spring and the 
 radial distance from the centre of the cock hole to midway 
 between the two curb-pins, multiply this latter by 100 and 
 divide the result by the radius of the spring, the quotient 
 will be the number of the terminal curve to be imitated. 
 The measurement may be done with a slide-gauge having 
 points. 
 
 The next step is to find how much spring will be 
 employed in making this terminal curve, plus the little bit 
 of spring used for elbow, plus the distance between the 
 stud and curb-pins when the index is in the centre. If the 
 length of the terminal curve itself is not known it must be 
 found by bending a bit of waste spring up into a similar 
 curve to the one going to be used, and then unbending this 
 again into the radius of the exterior coil of the spring, add 
 to this the two bits before mentioned and the point at which 
 to start bending up the spring for the elbow is determined, 
 Sets of copies of terminal curves can now be obtained, in 
 which all the curves are made out of a half-turn of the 
 exterior coil, this does away with all trouble in finding the 
 length. The total length will be found to be generally
 
 73 
 
 about 4 of a turn. Next to find the height to which the 
 overcoil has to be raised above the plane of the spring. See 
 that the spring collet is down to its place on the staff, place 
 the balance and spring in the watch, screw the cock on, see 
 that the staff end-shake is about right, and then look 
 through the watch and notice what position the stud -hole 
 occupies with reference to the top of the collet perhaps 
 the middle of the hole may be level with the top of the 
 collet or the bottom of the hole level witn the turning-off. 
 
 Next take the radius of the curb-pins, preferably with a 
 small pair of beam compasses having a pump-centre, or a 
 depthing tool ; put one centre in cock-hole, adjust for 
 upright and open until the other centre is midway between 
 the pins ; this distance may also be judged off by noticing 
 how many coils of spring, counting from the outside, the 
 centre of the pins is just over, and afterwards using this 
 coil as a guide or template to form the circular arc of the 
 overcoil along which the curb-pins travel, by looking 
 squarely on top of the spring. 
 
 The next step is to raise the overcoil ; at this stage we 
 shall have to decide whether the bend upwards is to be one 
 gradual slope or whether we shall have two bends so as to 
 bring the overcoil parallel to the body of the spring. Both 
 methods appear to give about the same practical results ; 
 two bends require more bending about of the spring than 
 one, this may be a disadvantage, but if a large model 
 Breguet spring be examined it will be at once evident that 
 the two bend variety gives a much squarer motion of the 
 spring, and that the single long slope causes an up and 
 down movement to the body of the spring; on a small 
 scale, as in a watch, this is not, perhaps, perceptible, but if 
 it exists we ought to try and avoid it. 
 
 Almost every springer has his own way of raising the 
 overcoil, there is no royal road, and however it is done 
 much will depend on the judgment of the operator. 
 
 Supposing the single long slope is going to be made, an 
 easy way of doing so is as follows : Put the spring on a 
 piece of brass or bone, having a small depression in it, hold 
 the spring down when the overcoil is to start with a pair of
 
 74 
 
 tweezers or with a bit of brass wire having a nick in its end 
 to go over the coil at this point ; now lift up the spring 
 with a peg, putting the point of the peg under the spring in 
 the hollow in the plate. Or it maybe done simply with two 
 pairs of tweezers, or in other ways. Many ways which are 
 perfectly successful with large springs are often quite 
 useless for small ones, or for open-coiled thin springs. The 
 spring will now need twisting with two pairs of tweezers in 
 about two places, one a little way up the elbow and the 
 other, perhaps, halfway to the end; the object is to bring 
 the overcoil as nearly as possible parallel with the plane of 
 the body of the spring, and also to raise it up exactly the 
 right height for the stud. To judge of this latter hold the 
 spring up sideways so that its plane is vertical, and see 
 whether the stud end occupies that relative height with 
 reference to the collet which was previously ascertained it 
 should do. It is admissible, and, perhaps, advisable, even 
 with the t\vo-bend variety of overcoil, that the overcoil 
 makes a very slight slope up to the stud, but on no account 
 must it have an up and down shape, or the spring will 
 wobble ; if the stud hole is drilled very close to the brass 
 work more slope must be allowed than would be otherwise 
 desirable in order that the spring may free the cock when 
 turned upside down. Making an overcoil with a knee-bend 
 is not quite so simple, and a tool of some kind is almost 
 absolutely necessary. A good and quick tool is shown in 
 Fig. 39A. 
 
 It is made entirely of steel, with the jaws hardened, the 
 jaws are about 3 mm. in breadth, and have four steel steady 
 pins about '5 mm. in diameter, fixed tight in one and a loose 
 fit in the other jaw, these pins must be as far apart as the 
 deepest spring to be used is broad, the jaw A is ground out 
 with a radius of curvature about 4-5111111., the jaw B with a 
 radius of about 4-3 mm., this makes the jaws fit at their centre 
 line and so grip the spring, while the slight divergence 
 towards their edges allows the spring to be bent yet
 
 75 
 
 prevents it from buckling ; the pins should be allowed to 
 show at the back of A, they then form a guide to judge 
 where the spring will bend. To use the tool put the spring 
 in between the four pins at the place where the first bend is 
 to be made ; shut the tweezers together, and pull the spring 
 up or down gently with a pair of roundish faced brass 
 tweezers, or push it with a peg ; to make the second bend 
 pull the spring back through the jaws a little and make 
 the spring butt against the opposite set of two pins. Now 
 look at the back of A, and the end of a steady pin will show 
 where the bend will come when made ; if this would not be 
 right height for the stud shift the spring either backwards 
 or forwards a little until it is right, then bend down as 
 before. These bends should be as round as possible and 
 not abrupt angular ones. If a superlatively nice round 
 curve is desired do not make the curve all at once, but bend 
 a little and then shift the spring between the jaws back- 
 wards or forwards by a miscroscopic quantity and then 
 finish the bending. Before taking the spring out of the 
 tool it had better be bent to get the overcoil parallel with 
 the body ot the spring ; to do this twist the overcoil up by 
 aid of a pair of tweezers, taking hold of it close to the jaw. 
 
 Take the spring out and place it on some flat white sur- 
 face, an enamel dial by preference, and look squarely down 
 on to the spring and see whether the outside coil is bent in 
 or out of place where it starts upwards for the elbow ; if 
 necessary correct this by means of a pair of tweezers and a peg. 
 
 The next step is to make the terminal curve itself, to do 
 this place the spring on top of the photograph chosen to be 
 copied, and gradually bend round to shape ; get in the 
 general outlines first and then form the exact copy, the 
 part along which the curb pins are to work must be made 
 to agree with the radius already measured with the beam 
 compass, the pump-centre of this tool being placed in the 
 collet-hole, adjusted for upright, and the other centre 
 moved round just clear of the top of the overcoil. The 
 curved-jaw tweezers used to effect this bending must not be 
 of too sharp a curvature or the spring will show a series of 
 distinct bends ; the tweezers must previously have been
 
 7 6 
 
 carefully polished, especially along the highest portion of 
 the convex jaw and the two edges of the concave jaw ; these 
 edges must be stoned nicely round and then polished by 
 means of a piece of brass with a groove in it, and finally 
 with a peg and diamantine ; the tweezers should be held 
 exactly upright, if inclined to one side they will bend the 
 overcoil up and vice versa, this property is often used to 
 raise or lower a part of the overcoil. 
 
 Now prepare a flattened pin for the stud-hole in the same 
 way as for the collet, and pin the spring in without the 
 balance, the pin must be level with the front of the stud and 
 just come through a little way at the back ; set the spring 
 about flat before pushing the pin home ; now push the 
 index along from end to end and see whether it moves the 
 spring out of place, if it does it arises from the stud-hole not 
 being drilled right for the pins ; find out whether the pins 
 are too near the centre or too far out by noticing which way 
 the spring is moved, then push the index right to the fast, 
 and with a pair of tweezers bend the spring either in or out 
 with a knee-bend close up to the stud ; finally the spring 
 may be set quite square to the plane of the cock, and a look 
 taken to see whether the cock-hole comes in the middle of 
 the staff-hole through the collet, and also as regards the 
 height of the overcoil. 
 
 Unscrew the stud, put the stud-screw up safe, put the 
 balance in the watch and notice where, say, the edge of the 
 stud will stand with reference to some screw in the balance 
 rim when the watch is in beat ; remove the balance and put 
 the spring on in this position ; oil the jewel-holes, screw 
 the stud and spring on, and put the balance in its place in 
 the watch. 
 
 The watch must be put exactly in beat by the drop of the 
 escapement, noticingwhere the twodrops occur and bisecting 
 the interval bet\veen them. It will now be generally found 
 that despite all care the spring is not exactly central, the 
 deviation will, however, be very slight and may be corrected 
 by pushing the overcoil in with a peg or pulling it out with a 
 bit of fine wire filed very small, burnished, flattened and turned 
 up at the end into a little hook, before doing this push the
 
 77 
 
 index right to the fast. The centrality of the spring is 
 judged of by seeing whether the space between the two out- 
 side coils is the same all round. The body of the spring 
 must, of course, be quite parallel to the cock, neither saucer- 
 shaped nor like an umbrella; there must be no motion of 
 the spring between the pins, the spring must not move 
 when the index is pushed from end to end, the overcoil at 
 at the elbow must not touch the second coil, and, finally, 
 it must expand and contract concentrically ; turn the 
 balance round and so wind up the spring, and notice 
 whether the coils touch on one side first, if not the spring 
 should be fairly isochronal. If well sprung there will be 
 one coil, or portion of a coil near the middle part of the 
 spring which will neither move in or out when the watch is 
 going, this isthe" dead coil, "and should always belooked for. 
 
 The watch should now be tried for mean time, and, if 
 necessary, corrected by altering the timing-screws, taking 
 a bit off two opposite screw-heads, putting in two heavier 
 screws, or by fitting balance washers ; there is no need, 
 however, to get the watch exactly to time at this stage. 
 
 Two courses are now open to us we may first adjust for 
 compensation and then for isochronism and positions ; or 
 adjust for isochronism first, try the \vatch for positions to 
 see if there be any serious structural faults, then adjust for 
 temperature, and finally for positions. If the latter course 
 be adopted the watch must be kept in a uniform tempera- 
 ture, during the other adjustments, if this cannot be done 
 it is probably better to compensate first, since the isoch- 
 ronism should never be so much in error as to materially 
 affect the compensation, then adjust for isochronism and 
 positions, taking care only to alter in weight those balance 
 screws which are quite close to the uncut end of the balance 
 rim ; if any material alteration is made either to the length 
 of the spring (as by cutting the eye out), or to the weight of 
 the balance it will only be prudent to try the compensation 
 again, and, as the balance may take a fresh form by the 
 heating, the positions will now again require trying. On 
 the other hand the arrangement of the screws for compen- 
 sation will affect the isochronism slightly, owing to centi-
 
 78 
 
 fugal force, but as the balance screws are put very nearly in 
 their right places when the balance is made or cut this dis- 
 advantage is very small. For these reasons the general 
 practice is to follow the second method. The watch may 
 now be tried for isochronism as directed in p. 61, always 
 taking the time at one place on the seconds dial of the 
 watch. If the isochronism is defective it may be altered 
 according to the following rules : 
 
 To QUICKEN THE SHORT ARCS. Looking at the point 
 where the spring emerges from the collet notice where a 
 radial line 90 from this point, counted along the spring the 
 same way it is coiled up, cuts the terminal curve ; bend the 
 terminal curve IN towards the centre an equal amount on 
 each side this imaginary line so that the line stands in the 
 middle of the alteration in shape. Or if this alteration 
 would come in an inconvenient place count 270 from the 
 the point of attachment as before and bring the curve OUT 
 towards the circumference along this line. 
 
 To SLOW DOWN THE SHORT ARCS reverse the above 
 process. 
 
 These alterations will only need to be very slight indeed, 
 and they must be effected with the curved-jaw tweezers as 
 used for originally bending the overcoil, the spring being 
 unpinned from the stud and placed exactly on top of the 
 photograph from which it was copied. 
 
 THE POSITION OF THE INTERNAL POINT OF ATTACH- 
 MENT. Both theory and practice show that the position of 
 the point of attachment of the spring to its collet with 
 relation to the pendant has a great effect on the position 
 errors of the watch (especially if the spring has no interior 
 terminal curve). If the displacement of the centre of 
 gravity of the interior coils of a spring alone entered into the 
 problem it would be easy to fix the most advantageous 
 position of attachment, but when it is remembered that the 
 staff pivots must have some shake in their holes, and that 
 the lift of the escapement is not given as a pure couple, it 
 will be evident that the virtual axis of vibration of the 
 balance often does not coincide with its geometrical axis, 
 and this in conjunction with the weight of the balance will
 
 79 
 
 cause the latter to have varying moments of inertia in 
 different vertical positions ; if to this cause of error we add 
 the peculiarities of each escapement it is clear that the 
 proper position of attachment cannot be predicated with 
 certainty, and it is not to be wondered at that some timers 
 doubt whether it is worth the trouble of putting it in the 
 position indicated by the displacement effect alone, and so 
 pin the spring on the collet just where it happens to come ; 
 but since we know in what direction the errors caused by 
 the displacement tend, or in other words, we know the 
 effect on the rates in the four vertical positions which are 
 due to an}- given placing of the position of attachment, 
 there is no reason why we should not avail ourselves of this 
 knowledge, and make one error balance another by cutting 
 out the eye of the spring so as to bring the point of attach- 
 ment into the most suitable position for this particular 
 watch, the selection of the position being made according 
 to the observed rates of the watch itself. In a general way 
 the position depends on the calliper of the watch, and it 
 must be a matter of personal experiment to determine 
 whether, for any particular make of watch, it pays to go to 
 the initial trouble of getting the attachment into one 
 particular place. It may be thought that the effect of the 
 displacement of the centre of gravity of the spring can be 
 got over by poising the balance and spring, considered as a 
 system, in the watch according to the observed rates ; this 
 does not however by any means appear to meet the case, the 
 explanation probably lying in some balancing effect pro- 
 duced by the pull of the spring on the collet and the lift of 
 the escapement. 
 
 As regards poise what we really require is that the 
 balance, with its rollers, collet and spring, shall be and 
 remain in poise during the expansion and contraction of 
 the spring. 
 
 It is safe to say that this is quite impossible. Even if 
 the balance, etc., be perfectly in poise, and the spring is 
 furnished with two theoretical terminal curves, so that the 
 whole system is in poise when tried in the poising-tool, it 
 will no longer be so when in the watch, because a certain
 
 8o 
 
 proportion of the weight of the exterior coils is borne by 
 the stud ; much more will this be the case when there is 
 only one curve. It follows from this that although the 
 balance may have been poised in the tool it is not altogether 
 illegetimate to alter the screws a moderate amount accord- 
 ing to the rates shown, this is the universal way of correct- 
 ing small positional errors, and is fully explained on p. 63. 
 
 Although this is the best practical way under the cir- 
 cumstances it is by no means satisfactoty, for two reasons 
 the amount the balance spring is out of poise is a varying 
 quantity as the spring opens and closes, and secondly, the 
 virtual axis of the staff does not always coincide with its 
 geometrical axis. If the positional errors are large, and 
 faults in the machine likely to cause them cannot be 
 detected or remedied, it is, therefore, better to cut out the 
 eye of the spring according to the table on p. 33 or p. 35. 
 
 If it is decided to place the point of attachment in its most 
 favourable position when first springing the watch the follow- 
 ing method shown in Figs. 20 and 21 (p. 34) will be found 
 advantageous, as it avoids a large spring eye. The spring 
 being pinned on the collet as usual, we now require to know 
 how many more degrees the overcoil will subtend when 
 bent up than it does now when untouched. This is easily 
 found, as before explained (a divided plate is here useful). 
 Now place the collet central over o (Fig. 21), making the 
 point of attachment lie on the line oc if the spring be a 
 right-handed one, and on the line on if it be left-handed, 
 and cut off the spring the determined number of degrees in 
 front of the line OG, say along OM for example. 
 
 The weight of the balance must now be altered to bring 
 the watch to time. In this operation it is often convenient 
 to alter the weight of two opposite screws, and to be able to 
 replace them, feeling certain that the poise of the balance 
 has not been materially altered. This can be readily done 
 by using a large, light compensation balance, all the screws 
 except the quarter-screws having been removed. Notice 
 which side each screw taken out of the balance of the 
 watch belongs to, put them in the poising balance, adjust 
 for poise in the poising- tool and remove metal at discretion."
 
 Si 
 
 COMPENSATION ADJUSTMENT. 
 
 A hot and a cold chamber are usually required for the 
 temperature adjustment. A stove or oven for high tem- 
 peratures must certainly be provided. An ice box, essential 
 when marine chronometers are being tested during the 
 the summer, may, as a rule, be dispensed with for watches. 
 A Norwegian refrigerator or box padded with a non- 
 conductor can generally be so arranged in a cellar or other 
 place of fairly uniform temperature so as to provide a 
 receptacle with a sufficiently cold extreme, except in the 
 very hottest weather. The adjustment for temperature is 
 made after observations of the alteration of rate in the two 
 extremes at which it is decided 10 expose the piece. A 
 short exposure to the temperature, or a single observation, 
 cannot be taken as a reliable indication of the effect, for the 
 unnatural connection of the metals composing the rim of 
 the balance requires time to settle. 
 
 The usual course is to place the piece to be tested, after 
 its rate has been careiuliy noted, into the oveii. Alter 
 twenty-four hours, the rate is again noted, fciay it has 
 gained on its rate 8 seconds, It is then removed to the 
 refrigerator, and subjected to the other extreme ol tempera- 
 ture lor 24 hours. At the end of that period a comparison 
 shows that it lias lost on its rate 7 seconds. Although the 
 alteration in the two extremes is not equal, there is 
 sufficient evidence that the balance is over-compensated. 
 In the oven, the rims bending too far inwards reduced the 
 effective diameter of the balance too much, and caused a 
 consequent gain. In the refrigerator the rims expanded too 
 much, and as a consequence the piece went slower. If it is 
 a marine chronometer under trial, the weights have to be 
 shifted a little towards the fixed ends of the rim. They 
 must be shifted equally, or the balance will be thrown out 
 of poise, and it is well to see that the slots in the weights 
 arc easy and do not grip the rim. If a watch is being- 
 tested, two opposite screws must be shifted towards the 
 fixed end, care being taken not to screw them too tight to 
 the rim.
 
 82 
 
 To avoid mistakes impress on the memory the fact that 
 a minus sign ( ), that is losing on its rate in the oven. 
 always involves shifting weights towards the FREE ends of 
 the rim, aud plus ( + ), or gaining, shifting weights towards 
 the FIXED ends. 
 
 The piece is then again subjected to the extremes of 
 temperature, and as the compensation adjustment gets 
 closer the piece is taken from the refrigerator and placed a 
 second time in the oven for verification before the alteration 
 is made. As the trial proceeds the piece is allowed to 
 remain more than 24 hours in each extreme, oftentimes a 
 week. 
 
 Oven. An oven for marine chronometers is generally 
 a square box of sheet copper or stout zinc with one or two 
 shelves. There is a jacket creasing round the top, bottom 
 and three of the sides filled with water so as to keep the 
 temperature uniform. The jacket may with advantage be 
 covered with flannel or felt, except the bottom, which is 
 slightly domed and heated by gas. The remaining side is 
 formed into a door having double panes of glass let in so as 
 to leave an air space between them. The oven is furnished 
 with a self- registering maximum and minimum thermometer. 
 Hearson's Oven, shown in Fig. 40, is a good example 
 tastefully arranged as a fitting for a watchmaker's shop. 
 A is the water jacket ; R and D plugs for filling and empty- 
 ing ; B gas tap ; F shelves of coarsely woven wire on which 
 the watches E are placed. The "thermostat" c is a very 
 sensitive device on the principle of the aneroid, for keeping 
 the temperature constant. It is a brass capsule, two inches 
 square and quite flat, suspended from a tube running 
 through the water jacket; this is filled with ether: when 
 the ether boils the sides of the capsule bulge out, and by 
 means of a rod (o), passing through the tube the gas is 
 turned off at the tap. In this way a constant temperature 
 of about 85 F. is maintained. There is a small bye 
 pass for enough gas to keep the flame from being entirely 
 extinguished. x is a thermometer. The water jacket 
 though an excellent aid for ensuring an even temperature 
 is not absolutely essential. The oven at Greenwich
 
 Observatory is jacketted with glass wool, the waste from glass 
 spinning, which is a good non-con- 
 ductor. Many watch adjusters have 
 for an oven but a wooden box with a 
 sheet of metal under the bottom 
 which is heated by a small lamp. 
 
 Mr. School" describes a cheap way 
 of making an oven. To the middle 
 of a thin iron plate (A Fig. 41) he 
 fastens an iron tube (B), long enough 
 to go right through a wooden box of 
 the height the oven is to be, the iron 
 plate being under the bottom of the 
 box. The front of the box is formed 
 into a door with double panes of glass 
 as just described. A very small flame 
 from a gas or oil lamp suffices to keep 
 the highest temperature required for "1 
 
 compensation adjustment. The one [. 
 
 objection to the use of iron in the Fig. 40. 
 
 construction of an oven, is its liability, after hammering 
 or rolling, to retain magnetism. 
 
 A very handy form of oven large enough for watches is 
 shown in Fig. 42. There is an outer 
 ring of thin sheet copper or zinc, 18 
 inches high and 12 inches diameter, with 
 
 Fig- 41- Fig. 42. 
 
 a horizontal partition 6 inches from the bottom, and with 
 
 F 2
 
 three legs to raise it from the floor. Inside is a cylindrical 
 copper or zinc box as shown by the dotted lines, 9 
 inches in diameter and 9 inches deep with a rim to prevent it 
 going too far down, and a hinged bezil with double panes 
 of glass with an air .space between them. A disc of wood 
 is placed on the bottom to rest the watches upon. The oven 
 is heated by a paraffin lamp, and there are a number of 
 holes in the outer ring just below the partition as shown. 
 Round the sides above the holes are several layers of flannel, 
 with a loose cover of the same for the top. 
 
 There is one advantage of a square oven over the cylin- 
 drical shape, and that is the facility with which the 
 thermometer can be .seen through the glazed door, but 
 many good timers use a round one substantially as 
 described, which is cheaper to construct. 
 
 Gas Governor. Fig. 43 is Mr. Kullberg's effective 
 and ingenious attachment to an oven heated by gas. The 
 gas enters by the right-hand pipe into the glass tube, and 
 from there through a hole in the left-hand pipe to the 
 burner that heats the oven. It will be observed that the 
 left-hand pipe is prolonged to nearly the bottom of the glass 
 tube to furnish an attachment for a laminated arm. This 
 arm, made of a thick piece of mainspring and rather thicker 
 brass outside, carries at its free end a weight, and just 
 beyond a conical tit that acts as a stopper for the hole in the 
 exit pipe. A nick is made in this hole to allow of the 
 passage of sufficient gas to keep the burner just alight even 
 when the stopper is pressed home. As the temperature 
 rises the bending of the laminated arm causes the tit to 
 approach the hole, and the reverse action takes place with 
 a fall of temperature. By tilting the instrument to the right 
 or left the weight retards or assists the action of the lami- 
 nated arm, so that any adjustment needed maybe made. A 
 screw with a large head nips the two pipes when the proper 
 inclination has been obtained. There are also a pointer 
 and circular scale for denoting the angle of inclination. 
 The two pipes are of brass, but they may have flexible 
 tube connections to the gas supply and burner respectively,
 
 to allow of the tilting of the governor, the lower part of 
 which is, of course, immersed in the oven. 
 
 Fig. 44 is another form of a regulator, consisting of a 
 glass tube with a reservoir of mercury (c) at the bottom. A 
 hollow tapered glass plug is ground 
 into the top of the tube. The gas 
 enters this plug at A, and passes out of 
 the bottom of the plug into the tube, 
 and through B to the burner. The 
 volume of gas allowed to pass depends 
 upon the distance between the bottom 
 of the plug and the top of the mercu- 
 rial column. The screw allows the 
 the temperature to be adjusted by 
 enlarging or restricting the space for 
 the mercury. A minute hole a in the side 
 of the plug allows enough gas to flow to 
 just keep the flame alight irrespective of 
 
 regulation. 
 
 There is another 
 kind of gas governor 
 in which a sensitive 
 valve of Persian sheep- 
 skin or similar mate- 
 rial is interposed in 
 the supply pipe, and 
 weighted so as to keep 
 the supply of gas at a 
 desired pressure. This 
 principle is not so satis- 
 factory for watch- 
 makers' ovens as those 
 which deal directly 
 with the temperature. 
 
 The Ice Box is a 
 metal receptacle, very 
 similar to the oven, for 
 the chronometers or watches, 
 
 44. 
 surrounded
 
 86 
 
 by a space for ice ; the outer vessel has a tap to drain off 
 the water, and is covered with some non-conducting 
 substance. 
 
 Great caution is needed in using an ice box, to preserve 
 the steel work from rust caused by condensation of the air. 
 Mr. Arthur Webb places his chronometers in a brass box 
 with a cover screwing against a waxed joint to exclude the 
 air. Whenever this box is taken out of the ice box, it is 
 allowed to remain unopened for about three hours, and as a 
 further precaution the balance and spring are washed in 
 benzine after a chronometer has been subjected to ice. If 
 this is not done the chronometer, immediately after 
 removal from the low temperature, should be placed for a 
 period in a moderately warm receptacle, with an open 
 vessel containing chloride of calcium, which absorbs 
 moisture. 
 
 Balance out of truth. To a balance maker setting 
 a balance true is but a small matter. The following is 
 related as the procedure of the late Mr. Earnshaw : "He 
 first took out the screws which he arranged in holes in his 
 work board, and then cut the rim open. This caused it to 
 spring outward, and he then placed it over a hollow filed in 
 a piece of boxwood screwed in the vice, and rubbed the 
 inside with an oval metal polisher. There \vere holes of 
 different length and depth in the boxwood, and he also used 
 polishers for rubbing according to the size and thickness of 
 the rim. The first rubbing sent the rim in too much, and 
 he pulled it outwards with his fingernail from the cutting. 
 Of course the calipers \vere in constant requisition, and 
 when the rim was quite circular, he subjected it to a heat 
 test in order to 'set it.' This was done by putting the 
 balance (with the staff) in an escapement box and holding 
 it over the gas until the rim sensibly curved inwards. He 
 remarked while doing this, that it was equal in effect to a 
 six weeks' trial in the oven. After this he again set it true 
 in the manner previously described, and then replaced the 
 screws with due regard to proper compensation, at the same 
 time getting the balance fairly in poise."
 
 87 
 
 Mr. Plose has given the following excellent directions 
 for truing a balance : 
 
 " Before attempting to true a balance that is in bad 
 condition, remove all the screws. Place them in holes made 
 in the lid of a scape or pill box, in the same position as they 
 occupy in the balance blank holes to represent those in the 
 the balance being also blank in the lid they will then not 
 become mixed. A rough sketch of the balance on the lid, 
 with the holes made perpendicularly instead of horizontally, 
 will be a help to identity. Be careful when unscrewing or 
 screwing them in to hold the rim firmly in the fingers as 
 near to the screw } r ou are working at as possible, or it may 
 bend, by this operation, out of shape. 
 
 Put the balance in the callipers, and test it for round and 
 flat with a toucher of brass or German silver, screwed to 
 one limb and having a long hole to allow of adjustment. Be 
 sure the calliper sinks are not worn, or the job of making 
 flat will become more difficult. 
 
 By looking down the edge of the toucher which is best 
 about the thickness of a sixpence, and perfectly flat and 
 smooth for this job on to the rim, you will be best able to 
 see whether it is necessary to bend in or out. Before 
 making any alteration examine your job closely, for a slight 
 crack almost imperceptible, mostly at a screw hole is 
 often a bar to a true balance. It is, in fact, almost safest 
 not to do anything to such a thing. 
 
 Be prepared with brass-lined pliers, the noses tapering 
 to about one-eighth of an inch, the linings hollow one nose 
 and rounding the other, to fit the circular shape of the rim 
 and to prevent it becoming marked. As these brass cover- 
 ings cannot be filed to the right shape after they are on the 
 insides of the noses, it is necessary to prepare them prior to 
 fixing them. To do this, soften the pliers sufficiently to be 
 able to drill them where the linings come, and steady pin a 
 piece of brass, with a pin top and bottom, to each nose. 
 The required shape can then be easily obtained by filing 
 them when off, and testing their fit one to the other after 
 being placed in position through the pins. Finally screw 
 on with brass screws.
 
 88 
 
 A gradual inclination to the staff, from the arm to the 
 cut end of the rim, may sometimes be altered, by pulling 
 this part gently out with the fingers; but any projections 
 or indentations between these parts are best managed with 
 the tool mentioned. A slight pressure, when the balance is 
 gripped between the pliers at the spot where a sudden 
 inclination takes place, and the rim from this spot goes in 
 toward the staff, will bring the rim out to its proper place. 
 The reverse mav be obtained bv holding the balance steadily 
 and tightly with one hand, while the other uses the pliers 
 to bend inward the part required. Take care, rather to 
 do too little than too much bending in one operation. 
 All alterations to be made with the balance out of the 
 callipers. The toucher fixed by the screw in the position 
 required, and a little red stuff applied to the rim, on the top, 
 at the part to be corrected, will be a guide as to whether it 
 has been bent enough, too much, or too little. After one 
 part has been brought in. another close to it mav require 
 the reverse, or vice versa. Should the balance be mounted 
 out of round, it will be useless to attempt to true it In- 
 bending ; but close examination will show if this is so, and 
 a new staff that fits the hole in the balance is the best 
 remedy. Manv balances are spoilt through being badlv 
 mounted, becoming therefrom out of round or out of flat. 
 sometimes both. The seat for the balance should be turned 
 quite flat, otherwise there will be great uncertaintv about 
 riveting it flat; and the staff, as mentioned, should fit the 
 hole, or true running will be out of the question. 
 
 The job of flatting may require an arm to be bent or 
 bumped, or the rim to be bent or hammered. In most cases 
 when the arm is bumped or bent, the rim will be found to 
 have altered its position, but increasingly towards the cut 
 end. This can generally be set right with the fingers, by 
 holding the part at the root, near the arm, tightly between 
 the thumb and finger of one hand while the other adjusts. 
 A brass hammer or a brass punch will not deface the arm if 
 it must be driven up or down with a blow, but the finger- 
 nails often do the work as well. A sharp dent in the bar 
 causing the balance to run out of flat is best removed with
 
 8 9 
 
 a brass punch like a blunt screw-driver at the face. Some 
 smooth lead stakes of different sizes to go inside the rim, 
 and one large one for any large size balance to rest on com- 
 fortably, with a hole for the staff, will lie found very useful 
 when it is necessary to recover its lost shape by impact. 
 An ivory or boxwood hammer, ' loaded ' at the opposite 
 end to the pane this part being used to make the rim flat 
 when lying on the lead and brass, ivory, and boxwood 
 punches are very handy. 
 
 The flatting process mostly causes error in truth, and 
 truing generally throws the balance out of flat; inconse- 
 quence a repetition of the work though to a less extent if 
 care is taken is often necessary. When satisfied with flat- 
 ness and truth in the callipers, put the balance in the frame, 
 without the roller, spin it gently round, and any little error 
 still existing will show itself more perceptibly by watching 
 a certain screw, or screw end, or anything close to the inside 
 or outside of the rim as a guide to its being true. For the 
 flat, observe the daylight above and below the balance when 
 looking along the plate. I should have mentioned that, 
 before anything is done that will raise or lower any part 
 of the balance, it should be tried in the frame, and when 
 running observe which part it will be best to operate on to 
 also procure needful freedom ; for time may be wasted in 
 making flat in the wrong direction, when there is very little 
 freedom above and below. The rim may be improved in 
 appearance, if any slight scratches or marks are on it 
 by polishing it with a boxwood slip and diamantine, the 
 balance resting the while on a sound flat cork secured in 
 the vice. 
 
 All being satisfactory, it only remains to replace the 
 screws very carefully. Use a nice light screw-driver, with 
 thin level blade, that will go well into the slit in the 
 head." 
 
 Figs. 45 and 46 are two tools described by Mr. Robert 
 Gardner. Fig. 45 is for truing the rim. The pin that is 
 to be used on the brass side of the rim is of ivory, and the 
 other of brass ; the brass pin being used as a fulcrum for 
 bending the rim in either direction by means of the ivory
 
 9 o 
 
 one. The ivory pin should be rather less than -, 1 .,- inch 
 diameter, and there should be a space of rather more than 
 
 Fi-. 45- Fijf. 46. 
 
 twice the thickness of the rim between the pins. By 
 means of this tool a balance can be very easily brought 
 true in circle. To get a t balance flat which has been dis- 
 torted is sometimes troublesome, but it is greatly facilitated 
 by using a piece of brass with a notch in it (Fig. 46) for 
 slipping over the bar, so that the bar can be slightly twisted 
 at the root without disturbing the position of the staff. 
 
 Poising the Balance. To ascertain if a balance is 
 correctly poised, or that the metal composing it is evenly 
 balanced around its axis, the balace staff is centred between 
 the points of callipers and one limb of the callipers gently 
 tapped, when it the balance is not in poise the heaviest part 
 is drawn to the lowest position. The sinks in the limbs 
 must not be too large or worn. In Fig. 47 is .shown an 
 adjustable link connecting the upper and lower portions, 
 By means of this, afterthe limbs have been opened to suit any 
 particular balance staff, the link can be set so that if the calli- 
 pers are opened to remove the balance they can afterwards 
 be closed to clasp the staff again without fear of damaging 
 the pivots, because the link stops them from going beyond 
 their former position. As a poising tool for fine work,
 
 jewelled callipers are preferable. The ends of the limbs of 
 an ordinary pair of brass callipers may be straightened as 
 in Fig. 48, and a hole drilled and tapped in each ; to these 
 holes brass screws without slits in the heads are fitted ; the 
 
 Fig. 48. 
 Fig- 47- 
 
 screw heads jewelled with good endstones and holes large 
 enough to take any ordinary pivot. Some prefer V-shaped 
 jewels instead of holes and endstones. One outside edge 
 of the callipers is roughened, and when the tool is in use it 
 is slightly rubbed with a screw-driver or other tool that 
 may be handy, to cause the heav} r part of the balance to 
 travel to the lowest point. Plose's pattern is a very 
 superior kind. It has ruby pins, with polished sinks and an 
 adjusting screw to prevent the callipers closing too far and 
 causing injury to the pivot ends from careless handling. 
 The "Excelsior" callipers, having the limbs crossed as 
 shown in Fig. 49, embody a distinct improvement over the 
 ordinary kind. When held in the left hand, with the 
 thumb and forefinger placed at opposite sides of the upper 
 part, a slight pressure will close the callipers, while a pres- 
 sure with the ball of the hand below the crossing of the 
 limbs will open them as easily. That the old-fashioned 
 callipers can only be opened or closed by the use of two 
 hands is a disagreeable experience to every watchmaker, 
 especially when the joint of the callipers works stiffly. 
 Therefore this advantage in the " Excelsior." 
 
 The plyer form of callipers shown in Fig. 50 are excel- 
 lent for poising. They are furnished with male and female
 
 92 
 
 centres and a spring for keeping the object to be trued in 
 position without undue pressure. 
 
 Fig. 49. "Excelsior' 
 Callipers. 
 
 Fig. 50. Flyer-shaped 
 
 Poising Callipers. 
 
 In the opinion of many, testing in the callipers does not 
 afford a sufficiently exact indication of the truth of a com- 
 pensation balance for fine work, and recourse is had to a 
 special tool, of which the one in the drawing (Fig. 51) by 
 Mr. R. Bridgman is a very fine example. The movable 
 standard is kept in contact with the bed of the tool by 
 springs, so as to be independent of the truth or otherwise of 
 the traversing screw. The pivots of the balance rest on
 
 93 
 
 knife edges formed of sapphire, which may be adjusted to 
 the desired distance by means of a screw. Screws on each 
 
 Fig- 5 1. 
 
 side of the standard serve to level the knife edges. P'ig. 52, 
 
 the "Grossman," is a less costly 
 instrument more generally used. 
 As to the correction of the 
 poise, if the divergence is not 
 great it may be adjusted by 
 screwing in a timing screw or 
 nut at the heaviest point, or by 
 proportionately screwing in the 
 Fig. 52. adjacent ones. For larger errors 
 
 it may be desirable to substitute a heavier screw of 
 platinum for one at the lightest point, or to lighten a com- 
 pensation screw at the heaviest point, or instead or in 
 addition to interpose a thin collet or washer underneath 
 the head of the screw at the lightest point. All these 
 methods are adopted, and a choice will rest with the 
 discretion of the operator. If the balance has been adjusted
 
 94 
 
 for compensation, screws approaching the free ends should 
 be interfered with as little as possible in order that the 
 compensation may not be unnecessarily deranged. 
 
 The Stud. The stud for the attachment of the outer 
 encl of the spring has generally been a weak point. \Yith 
 the primitive English full plate stud screwed into the 
 plate the spring had to be unpinned every time the balance 
 was removed, and was as a consequence often distorted or 
 broken at that point. In old verge watches it was nothing 
 uncommon to find the spring attenuated and much shortened, 
 for it was the practice of the repairer each time the end of the 
 spring was broken off to correspondingly reduce the 
 strength of the spring by rubbing it on an oil-stone. The 
 more modern stud fastened to the cock by means of a screw 
 is costly and not perfect. A careful workman before 
 unscrewing the stud will move the index as far as possible 
 to the fast position and then release the spring from the 
 curb pins, but in unskilful hands the spring is often bent 
 through disregarding this precautionary measure. Then, 
 as a rule, the balance spring stud, when screwed to the 
 plate or cock, admits of no alteration in its position, and 
 though it presents the advantage that if removed it may be 
 replaced with the certainty that it is unaltered with relation 
 to the spring, it is undoubtedly a convenience to have a 
 
 53- 
 
 stud which allows of adjustment, either in circle to bring 
 the stud hole right for the spring, or to and from the centre 
 of the balance, to suit the diameter of the spring. The 
 adjustable stud shown in Fig. 53 answers the latter con- 
 dition, and is found mostly in Swiss watches. The wing of 
 the cock is formed into a slot, into which the body of the 
 stud fits. The head of the stud rests on the top of the 
 wing, and is kept in position by a cover plate and two 
 screws.
 
 95 
 
 Walker and Barber's stud, shown in sectional elevation 
 (Fig. 54) and plan (Fig. 55) allows of adjustment in circle. 
 
 Fig- 54- 
 
 1'iR- 55- 
 
 Fig. i = Longitudinal section through balance-cock. 
 
 Fig. 2 Plan of balance-cock, inverted, showing stud and part of 
 spring. 
 
 A A Balance-cock. 
 
 J> J> - Steel ring sprung into balance-cock. 
 
 B 1 B 1 = Projections from steel ring, one of which is drilled to take 
 outer end of spring, the other to enable the ring to be contracted 
 when it is desired to turn round or remove the stud. 
 
 S S S = Terminal curve of spring. 
 
 Guye's stud also allows of adjustment in circle, 
 Fig. 56 is a plan of the stud fixed under the balance-cock ; 
 and Fig. 57 a section through a b of the stud attached to 
 the cock.
 
 9 6 
 
 In this case when the position of the stud is decided on 
 Fig- 56. 
 
 57- 
 
 a screw at a serves to fix it to the cock. 
 
 Fig. 58 and 59 illustrate a stud described in Dcnischc 
 Uhrmacher Zeitung, and which appears to be especially 
 suited for cheap watches. 
 
 As seen in Fig. 58 it is made of one piece of thin steel, the 
 balance spring being clasped spring tight by the jaws, b b. 
 The jaw b is a tongue punched from the body k and turned 
 up b' being brought to it by forming the loop // and passing 
 the end throught the slot s. The slot s from which b is cut 
 serves for the screw e to attach the stud to the cock B, as 
 seen in the sectional view Fig. 59. ^>" is the balance spring ; 
 R the collet, and z the regulator. 
 
 Fig. 59-
 
 97 
 
 This stud allows the spring to be clasped at am- height 
 to suit the collet, and also permits adjustment to the 
 diameter of the spring. By its use the buckling of the 
 spring by pinning into a round hole is avoided. 
 
 Stud Remover. Mr. Daldorph has devised a useful 
 table with four slits as shown in the sketch (Fig. 60) for 
 
 Fi^. 60. 
 
 supporting a Swiss cock while pushing out the stud. The 
 holes in the table, and also the slits, allow the tool to be 
 used as a support when pressing a balance spring collet to 
 position. 
 
 The engraving (Fig. 61) shows a pair of tweezers for 
 removing studs such as are pivoted to the balance cock. 
 
 By placing the tweezers so that the slit embraces the stud 
 and the projecting finger is pressed on the pivot, a tight- 
 fitting stud may be readily removed without fear of 
 damaging the balance spring. 
 
 CHAPTER V. 
 
 THE MANUFACTURE OF BALANCE SPRINGS. 
 So long as the verge escapement was in use watch 
 balance springs were usually coiled from soft or hard-drawn 
 wire with the aid of curved nosed tweezers, or a coiling pin, 
 which latter was very often simply a round broach. The 
 manner in which the coiling pin was used to bend to a
 
 9 8 
 
 volute the wire as it lay on the thumb of the operator will 
 be understood from the sketch (Fig. 62). Starting with 
 the eye of the spring each coil was finished before the next 
 was begun, and many of the springs so made were marvel- 
 lously true. Inaccuracies in flatness were remedied in 
 bluing. For this purpose the spring was placed in the 
 centre of a copper pan and held down by a skeleton disc 
 on the free end of a springy arm, the other extremity of 
 which was attached to the handle of the pan. 
 
 Fi^. 62. 
 
 Soft balance springs soon get out of shape and are there- 
 fore not reliable, but their strength is not materially 
 increased by hardening. If a watch is brought to time 
 with a soft spring, and the spring is afterwards hardened, 
 the rate of the watch would be accelerated but a few seconds 
 in twenty-four hours. 
 
 For many years certain Swiss spring manufacturers have 
 prepared springs of very hard drawn wire, which, being 
 comparatively inexpensive, serve a useful purpose. They 
 are much more lasting than soft springs, but more brittle, 
 and not so good as the best fire hardened springs. At one 
 time springs made from hard wire were supposed to be 
 hardened after being made by some chemical process, and 
 the secret of their manufacture was well kept. 
 
 Arnold invented the helical spring for chronometers, and
 
 99 
 
 patented it in 1776. To form the spring he wound the 
 wire on a spirally grooved cylindrical block and hardened 
 it, practically just as such springs were prepared till quite 
 recently. 
 
 From this time springs for cylinder and duplex watches 
 though turned up by hand were sometimes hardened. 
 
 The method of hardening a flat spring without distort- 
 ing it after it had been applied was at one time kept a secret 
 by the few who practised it. It is however exceedingly 
 simple. The spring is placed on a round, PERFECTLY FLAT 
 plate of German silver ; the space between the coils being 
 filled in with black lead, the spring is covered with a plate 
 similar to the first. These plates, with the spring between 
 them, are then laid on a small press of German silver very 
 much resembling in form the presses used for copying 
 letters. (See Fig. 63.) The plates are kept together by 
 
 Fig. 63. 
 
 means of a screw. The spring, press and all, are then 
 heated to a cherry red and plunged into water. When cool 
 a thin slip of bright steel is laid on the press, which is 
 heated till the slip of steel is brought to a blue colour, when 
 it is plunged into oil. It is not imperative that the press 
 should be of German silver, but it must not be of steel or 
 wrought iron, or the springs will be spoiled. A press of 
 this sort is also useful for restoring to shape a spring dis- 
 torted from flatness. This may be placed between the flat 
 plates and heated to a blue, as already described. 
 
 Modern Hardened and Tempered Springs. 
 
 In making balance springs the first requirement is good 
 wire. After being drawn to suitable sizes, the round wire 
 is flattened between rollers and then drawn through jewelled 
 clams or rollers to ensure uniformity of substance. The 
 
 G 2
 
 100 
 
 most general fault is that, during the frequent annealing 
 necessary in drawing it so small, the steel has lost too much 
 of its carbon, and is difficult to harden except at an 
 excessive temperature. Before making balance springs it 
 would be prudent to test a piece of the wire for hardening 
 and tempering. If the wire is suitable it should harden 
 well after heating to what is called a cherry red, that is 
 before it attains a brilliant red hue. Flat spiral springs 
 are coiled up in a circular box, like a small watch barrel, 
 and cover (see Fig. 64), made preferably of aluminium- 
 bronze or of platinum. Copper and German silver are 
 sometimes used, but platinum and aluminium-bronze are 
 found to best retain their form after heating. Obviously 
 the top and the bottom of the box must be perfectly 
 flat if flat springs are to be produced. Though rigid the 
 box and cover must be as thin as possible, because if a con- 
 siderable body of metal surrounds the spring it has to be 
 raised to a greater heat to harden. Either two, three or 
 four springs are coiled at a time, generally three ; if coiled 
 up four at a time the coils would be more open than usual ; 
 springs for overcoils are generally closer and would be 
 coiled two at once. 
 
 The following is substantially the process described in 
 the Horological Journal by Mr. Glasgow: "The winder 
 
 . 64. I'itf. 65. 
 
 (Fig. 65) is of steel, with a pivot formed on the end 
 to pass through a hole in the bottom of the box, and 
 project into the box just the width of the wire of which 
 the springs are to be made. Across the pivot of the 
 winder are slits to receive the ends of the wire. For 
 springs, coiled up two together, there is one straight 
 slit across the centre of the pivot. If three springs are 
 to be coiled up together, three equidistant radial slits are 
 cut (as at a, Fig. 65), and for open springs coiled up four
 
 101 
 
 together, two slits at right angles across the centre are 
 made. The pivot should be snailed from the slits according 
 to the wire to be used, so as to ensure the truth of the 
 eyes of the springs. A hole is drilled and tapped in the 
 centre of the pivot to receive a small screw, the head of 
 which passes through a hole in the cover. Holes not radial, 
 but almost tangential, are drilled through the side of the 
 box to thread the wire through. The ends of the wire 
 passed into the slits are secured by the screw, the cover is 
 gently pressed on, and the winder rotated till the box is 
 full. The ends of the wire are cut off, the screw and 
 winder removed, and the cover bound tightly with wire. 
 The holes are then stopped, usually with a mixture of soap 
 and animal charcoal to exclude the air, and the box put in 
 an open iron vessel containing charcoal, which has been 
 placed in the fire of an ordinary grate." Animal charcoal is 
 usually prepared from burnt bones or burnt leather reduced 
 to powder. The object of using it in this connection is to 
 further enrich the steel, but some discretion is necessary, 
 for if the steel is already sufficiently rich in carbon it may 
 be rendered too brittle. Difficulty would then be experi- 
 enced, especially in bending overcoils, if such were needed. 
 If the steel is found to harden readily at a proper tempera- 
 ture vegetable charcoal may be used to exclude the air. If 
 soap is used it should not be wetted, but softened by heat- 
 ing. Kither beeswax or oil appears to be a more suitable 
 vehicle than soap. When the spring box is cherry red, it is 
 dashed into a vessel containing plenty of cold water. The 
 spring box is then placed in a boiling out pan or an old 
 metal spoon filled with oil, and held over a flame till the 
 flame just flickers over the oil. 
 
 After cooling, the springs may be turned out of the box. 
 If they do ~>ot come apart quite readily they should not be 
 forced, as tney would be likely to be permanently distorted 
 thereby. If thrown into a pill box and rattled for a minute 
 the slight concussion against the sides of the box is usually 
 enough to induce separation unless the springs have been 
 burnt. Water or oil is the medium generally selected for 
 plunging steel in to cool it when it has been heated to the
 
 1 02 
 
 requisite redness for hardening. Either mercury or petro- 
 leum may be recommended if extra hardness is desired. 
 Salt water will give great hardness, but the steel is rendered 
 brittle. Oil and mercury are considered the two best media 
 for hardening steel if toughness is desired. If the steel 
 has been protected from the air, and not overheated during 
 the process of hardening, its surface will not be scaled nor 
 materially injured, and its brightness may be restored by 
 polishing, but if the surface is intricate, as is the case with 
 balance springs, a quicker method of cleaning may be 
 employed. The steel may be washed in a strong solution 
 of hydrochloric acid (about one-third of pure acid to two- 
 thirds water is recommended), and immediately afterwards 
 rinsed in a saturated solution of cyanide of potassium. 
 In one respect springs finished with a high polish are 
 inferior to others left grey, for they are more liable to rust. 
 
 To avoid oxidation of the surface during hardening and 
 tempering the springs melted crystals of potassium cyanide 
 may with advantage be employed as a heating medium. 
 The cyanide having been placed in a wrought iron pot on a 
 stove and raised to a red heat, the spring box is immersed 
 and allowed to remain till it becomes of the temperature of 
 the bath, when it is removed and plunged into cold water. 
 A film of cyanide clings to the exposed surface when the 
 box is removed from the bath, and so the air is entirely 
 excluded from contact with any part of the springs. This 
 protecting film leaves the surface when the box is plunged 
 in the water. 
 
 To temper the springs, the box on being removed from 
 the water may be again momentarily plunged in the cyanide 
 bath and then subjected to an even heat in a small stove or 
 iron box. When the film of cyanide peels off the box it is 
 considered the desired temper is obtained. 
 
 This is, I believe, the process adopted by some of the 
 Swiss spring manufacturers, and in America. Of course, 
 wire of uniform quality must be selected to be sure of 
 always attaining the desired temper. The difficulty generally 
 is to ensure uniformity of quality, and for this reason no exact
 
 103 
 
 tempering temperature for balance springs can be given, 
 because the steel varies in quality. Continual care is neces- 
 sary to ensure success. Mr. Arthur Webb, on taking a 
 fresh bobbin of wire, hardens a piece, and then finds the 
 temperature at which that particular wire yields the desired 
 temper, by placing it in the oil bath with a thermometer. 
 When satisfied with the temper he notes the temperature 
 which he assumes will be the best for tempering so long as 
 that wire lasts. 
 
 Still, the following table, giving the characteristics of a 
 wide range of temperature may be useful : 
 
 
 
 Alloy, whose 
 
 
 Colour. 
 
 Purpose, 
 
 Tempera- fusing point is 
 ture. of the same 
 
 Effect on Tallow. 
 
 
 
 temperature. 
 
 
 Pale straw - 
 
 Lancets and Tools for Cut- 
 
 
 
 
 tiny Iron - 
 
 420Fah. 7 lead 4 tin 
 
 Vapourizes. 
 
 Straw - 
 
 Watchmakers' Tools - 
 
 450" ,, 4 
 
 Smokes. 
 
 Straw yellow 
 
 Pen Knives and Razors 
 
 48o 8J ,, 4 
 
 More smoke. 
 
 Nut brown 
 
 Small Pinions and Arbors - 
 
 500-* 14 4 
 
 Dense smoke. 
 
 1'urple - 
 
 Large Pinions and .Arbors 
 
 S3o 19 4 
 
 Black smok". 
 
 Bright blue - 
 
 Swords and Watch Springs 
 
 580" 48 4 
 
 Flashes if light is 
 
 
 
 
 applied. 
 
 Deep blue - 
 
 Watch Balance Springs 
 
 490 ,, 50 ,, 2 
 
 Continuous burn- 
 
 
 
 f All lead or 1. 
 
 ing. 
 
 Blackish blue 
 
 Chronometer Balance 
 Springs 
 
 640 ^ boiling ^ 
 (, linseed oilj 
 
 All burns away. 
 
 Steel is less oxidized by tempering in an alloy than if 
 tempered in the air, and the required temperature is 
 obtained with much greater certainty. Olive oil, which 
 boils at 600, is now most generally used as a bath for 
 tempering balance springs. But great care must betaken to 
 ensure the purity of the oil. Olive oil adulterated with cotton 
 seed oil, when used as a tempering bath, has been found 
 to rust the springs, possibly from the presence of some 
 chemical employed in the preparation of the oil. Pure 
 olive oil may be used with confidence. It may be heated 
 in a wrought iron pot, which must not be tinned. The 
 cover may have a hole in it for the insertion of a ther- 
 mometer, and hooks on the under side from which the 
 springs may be hung. 
 
 Springs are more liable to distortion when perfectly hard 
 than when tempered. No explanation appears to have been 
 given of this curious fact, but, mention of it may prevent
 
 IO4 
 
 a beginner attempting to polish watch springs before 
 tempering. 
 
 The most usual method of mechanical polishing is as 
 follows : 
 
 The flat faces are polished by gently pressing the tip of 
 the finger on the spring, and moving it in a circular direc- 
 tion on a piece of writing paper on which red stuff has been 
 rubbed. 
 
 The spring is drawn down over a piece of wood with a 
 conical end like Fig. 66, with a pin at the apex for the eye 
 to go over, and the outsides are polished with a well-worn 
 brush charged with red stuff. 
 
 The inner sides are much more difficult to polish ; but 
 if the wire is good, and has not been raised to an excessive 
 heat in hardening, and if the air has been properly excluded 
 during that operation, but little polish will be needed. 
 The spring is placed on a flat piece of cork, and the inner 
 sides rubbed to and fro and from side to side with a finely- 
 
 Fig. 67. 
 
 . 66. 
 
 pointed peg charged with red stuff, the spring taking the 
 form of a cone the while, as shown in Fig. 67.
 
 105 
 
 After polishing, the spring is washed in PERFECTLY 
 CLEAN benzine, and blued on a flat plate. The plate should 
 be heated before the spring is brought into contact with it 
 and care taken that no dust is allowed to settle on the 
 spring. A glass tube may be placed over it during the 
 operation. 
 
 An alternative method of polishing flat springs on the 
 surface and between the coils at the same time is to place 
 the spring on an india-rubber pad under two plates with 
 radial sides terminating in a point, as in the subjoined 
 sketch, Fig. 68. Over the spring is a holder on a universal 
 
 Fig. 68. 
 
 joint. Through a hole in the holder passes freely the 
 polisher spindle, carrying at its lower end a thin slip of 
 willow wood cut almost to a knife edge where it bears upon 
 the spring. A little pressure on the knob at the top of the 
 polisher holder will cause the coils of the spring to make 
 serrations in the edge of the willow. Charged with a little 
 fine emery the polisher is now pressed on the spring and 
 rotated as far as the radial edges of the plates will allow 
 for, say, six times backward and forward ; the position of 
 the spring is then altered so as to bring the part at first 
 under the plates into contact with the polisher and treated
 
 io6 
 
 in like manner. The spring is then turned over and the 
 other side operated on. The india-rubber pad is on a 
 circular metal table, and underneath the tool is a bridge ; 
 one end a helical spring around the stalk of the table 
 presses against the underneath surface of the table and the 
 other end against the bridge, and so keeps the spring on 
 the rubber pad sufficiently tight against the plates. Care 
 must be taken not to rub the spring more than is necessary 
 and to subject each part of the surface to the same number 
 of rubs. Excessive and indiscriminate rubbing will spoil 
 the spring. 
 
 Fig. 69. 
 
 Fig. 69 shows a different arrangement for winding 
 springs, for which I am indebted to Mr. R. B. North. Here 
 the winding machine stands on the bench, and the centre 
 arbor is actuated by turning a handle at the side with which 
 it is connected by a pair of mitre wheels. The spring box 
 without the cover is placed on the table of the machine and 
 prevented from turning by a notch at the edge which 
 engages with a pin in the table. At the end of a hinged 
 arm is pivotted a little bar which may be pressed on the 
 wire while it is being wound in. In this way the winding 
 is under perfect control, and can be watched as it proceeds.
 
 107 
 
 When the box is full the cover, with a hole to pass over the 
 centre arbor, is placed in position and fastened with three 
 screws. There is a special feature in the box which is 
 worth}' of mention. It has a loose disc or false bottom 
 which can be readily trued or a fresh one substituted. The 
 cover is flat, and does not enter the box, therefore narrow 
 w T ire would not fill the space between the disc and the 
 cover, but the central hole in the bottom of the box is 
 larger than the hole in the disc and tapped, so that when 
 the cover is screwed on and the centre arbor removed a 
 screw can be inserted to press the disc till the whole surface 
 of the wire is in contact with both the disc and the cover. 
 
 J. F. Cole used instead of a spring box a square plate, 
 rising from which were four half round pins instead of a rim, 
 keeping the wire within the required limit. Helical springs 
 are formed by winding the wire on a thin solid drawn tube of 
 brass or aluminium-bronze. If the same tube is to be used 
 for tempering it has shallow grooves cut to the pitch the 
 spring is to be, and in depth about half the thickness of the 
 wire. The ends of the spring are fastened to the tube 
 under the heads of brass screws which must be made left- 
 handed, so as to draw the wire tighter on the tube as they 
 are screwed up. The tube is mounted on an arbor in the 
 turns; one end of the wire is fixed by one of the screws, the 
 remainder of the wire hangs freely in the air, a weight of 
 about 12 ounces being fastened to the lower end to keep the 
 wire taut. A piece of thin sheet platinum or sheet copper 
 is wrapped round the spring and tube and kept close with 
 binding wire prior to hardening. It is hardened by being 
 heated to redness and plunged into cold water. Immedi- 
 ately it is withdrawn from the water one of the screws is 
 slackened, the spring drawn tight on the tube and held 
 again in position by the screw. It is then tempered in 
 boiling oil. 
 
 A grooved tube is not essential for coiling the spring. 
 Mr. Mercer uses a plain thin tube of solid drawn brass or 
 aluminium -bronze. The wire is coiled closely, covered with 
 a thin sheet copper wrapper on the inner side of which 
 has been smeared either beeswax or a little lard ; the ends
 
 io8 
 
 of the copper are tucked over the ends of the tube 
 which is bound round with wire. A light wire holder, 
 sufficientl)' long to protect the operator from the heat, is 
 attached, and the tube is subjected to a charcoal fire heated 
 by gas. The operator turns the tube, watching till it 
 attains a cherry red, when it is dashed into mercury, which 
 Mr. Mercer prefers as a hardening medium. During hot 
 weather the pot containing mercury should be surrounded 
 with cold water or the steel may not harden satisfactorily. 
 After hardening, the spring is tempered on a grooved tube. 
 The lard should be melted fresh pork. L,ard, as purchased, 
 is liable to contain salt or other impurity which might 
 engender rust. 
 
 The outsides and edges are polished on a block a little 
 larger than the tube used for coiling the springs. For 
 polishing the inside, the spring is mounted on a piece of 
 wood charged with red stuff and rotated in the turns, the 
 spring being held between the thumb and finger the while. 
 Before bluing the spring is thoroughly cleaned with soap 
 and water, and afterwards with benzine. Unless the spring 
 is perfectly clean it will not blue evenly. For bluing, the 
 springs are secured by screws to a block with very shallow 
 grooves very similar to the tube used for tempering, except 
 that it is solid and a shade larger in diameter. The block 
 is placed end upwards on a bluing pan, which is heated 
 over a spirit lamp. The spring may be encircled with an 
 open glass tube to keep the temperature uniform, and keep 
 off dust which might make the spring speckv. 
 
 Mr. T. Hewitt's procedure in coiling and hardening is a 
 little different. He turns down his piece of solid drawn 
 brass tubing, leaving the surface plain without grooves. 
 On this he winds the wire, the coils following each other 
 close together. He smears a mixture of oil and vegetable 
 charcoal on the coiled wire, and over all slips another thin 
 brass tube, stopping the ends of the two with a luting of 
 the oil and charcoal. It is then hardened. After harden- 
 ing, the spri'ig is wound on to another brass tube having 
 shallow grooves of the desired pitch and of exactly the 
 diameter at the bottom of the grooves as the hardening tube
 
 log 
 
 was, fixed with left-handed screws and tempered. Temper- 
 ing sets the spring satisfactorily to the pitch decided on. 
 
 The late F. Knudsen, a very successful chronometer 
 springer, wound his springs on a plain tube, and heated 
 them for hardening by hanging the tube on a pin in a piece 
 of charcoal and directing a flame through the tube with a 
 blow-pipe. After plunging in water he heated the tube to 
 nearly the ultimate tempering heat to avoid breaking the 
 spring in loosening the screw. He set the springs by 
 tempering them on a grooved tube, and after polishing 
 finally blued them on a plain block, the spring being secured 
 at one end by a screw. 
 
 After helical springs are tempered the greatest care is 
 taken to avoid rust and stain. I have heard of springers 
 conducting subsequent operations necessitating handling 
 with stalls of oiled silk on their fingers to ensure that when 
 the terminals of a spring are formed, and it is primed on, 
 the even colour remains without blemish. This, though, is 
 not the usual practice, and is, I should say, not necessary 
 unless the operator has very damp hands. Indeed, rust is 
 generally the result of dirt or impurity met with in course 
 of manufacture. 
 
 CHAPTER VI. 
 
 NON- MAGNETIC MATERIAL AND MATERIAL INSENSIBLE 
 TO CHANGES OF TEMPERATURE. 
 
 Watches with quickly moving parts of .steel are rendered 
 unreliable if exposed to the influence of magnetic or electric 
 currents, and as the number of electric appliances for light- 
 ing, traction, and other purposes is continually increasing 
 it appears to be evident that some other material than steel 
 must be found for such parts of watches as the balance, 
 balance spring, and escapement. 
 
 Aluminium - bronze, which combines strength with 
 lightness, is particularly suited for the lever and pallets. 
 The steel balance staff, pallet staff, and escape pinion 
 may be retained, their circumferential velocity being
 
 no 
 
 small. A steel seconds hand must not be used, and steel 
 hands had better be altogether avoided. For the ordinary 
 run of watches, a plain gold or brass balance may be used. 
 Many attempts have been made to devise a compensation 
 balance, in which the use of steel is dispensed with. 
 
 Frederic Houriet, in the early part of the century, pro- 
 duced a non-magnetic watch using a compensation balance 
 of platinum and gold. J. G. Ulrich took out a patent in 
 1828, and two subsequently, for non-magnetic balances. 
 His claim seems not to have been based on exhaustive 
 experiment, though it embraces nearly every possible metal. 
 Messrs. Arnold and Dent made many experiments between 
 1830 and 1840, and they used balance springs of glass, of gold, 
 and of palladium. Briefly, their greatest success with com- 
 pensation balances was obtained with a construction of 
 platinum and silver, which compensated fairly well, but 
 were lacking in rigidity. 
 
 C. A. Paillard, in some instances, appears to have used 
 a palladium alloy for the inner part and brass for the 
 outer part of the rim, and in others to have formed both 
 laminae of different alloys of palladium. Professor Houston, 
 in an exhaustive paper to the Franklin Institute, gave 
 the following interesting particulars respecting Paillard's 
 alloys : 
 
 " The alloy described in the first U.S. specification 
 contains the following metals, viz. : 
 
 Palladium 60 to 75 parts. 
 
 Copper 15-25 
 
 Iron i ,, 5 
 
 "We will for convenience call this alloy No. i. 
 
 " The specification states that the preceding proportions, 
 or percentage may be somewhat varied, without appreci- 
 ably affecting the essential characteristics or properties of 
 the alloy. 
 
 "The process for the production of this alloy is as 
 follows : About half of the palladium to be used is placed 
 with the other metallic ingredients, and with a small quan- 
 tity of borax and powdered charcoal, in a clay crucible and
 
 Ill 
 
 heated until melted. The remaining part of the palladium 
 is then added, and when the whole is melted the molten 
 mass is poured into a suitable mould, and when cooled is 
 ready for use. 
 
 "U.S. Patent, No. 367,159, describes an alloy cheaper 
 in its production and suitable for a lower grade of watches, 
 or for certain parts of watches. 
 
 "The inventor makes the following statement in the 
 specification : ' I have found by experiment that the alloy 
 hereinafter described and claimed can be successfully 
 employed in the manufacture of watches and timepieces 
 for all the parts not required to be non-magnetic and hardly 
 dilatable, as it is not perceptibly affected by ordinary mag- 
 netic or other disturbing causes or influences. The cost of 
 its production is also such that it can be profitably and 
 economically used for all the ordinary parts of watches and 
 for the mechanism of the cheaper grade of watches, where 
 perfect uniformity and regularity of movement under all 
 circumstances is not required.' 
 
 "The composition of the above alloy (No. 2) is as 
 follows : 
 
 Palladium 50 to 75 parts. 
 
 Copper 20 ,, 30 
 
 Iron 5 ,, 20 ,, 
 
 "The method adopted for its manufacture is the same 
 
 as that described for the first alloy. 
 
 " U.S. Patent, No. 367,160, describes an alloy possessing 
 
 in the highest degree the properties desired in parts of 
 
 high-grade watches liable to change of rate. 
 
 "The following is the composition of the alloy (No. 3) 
 
 that is described and claimed in this patent : 
 
 Palladium 65 to 75 parts. 
 
 Copper 15 ,,25 
 
 Nickel i 5 ,, 
 
 Gold i ,, 2i ,, 
 
 Platinum ] , ,, 2 ,, 
 
 Silver 3 ,, 10 
 
 Steel ... 15
 
 112 
 
 " The same method is used in the manufacture of this 
 alloy as in the preceding alloys. 
 
 "U.S. Patent, No. 367,161, describes an alloy which, 
 besides possessing the properties of the preceding, possesses 
 the additional property of being raised by tempering to a 
 very high degree of hardness. He uses the alloy as follows : 
 ' In order that the best result may be obtained in watches 
 and chronometers, I have found that it is necessary that 
 other parts of the mechanism employed therein, viz., the 
 escape-wheel, escape-lever, guard-pin, and regulator-index 
 shall possess the same characteristics and properties, and 
 in addition thereto be capable of being tempered to a high 
 degree of hardness to prevent wear and abrasion.' 
 
 "This alloy (No. 4) has the following composition, 
 viz. : 
 
 Palladium 45 to 50 parts. 
 
 Silver 20 ,, 25 
 
 Copper 15 ,, 25 ,, 
 
 Gold 2 5 
 
 Platinum 2 ,, 5 ,, 
 
 Nickel 2 ,, 5 ,, 
 
 Steel 2 5 ,, 
 
 " The process of manufacture of this alloy is the same 
 as that used in the other alloys. 
 
 " The following experiments were made to test the non- 
 magnetic character of the palladium alloys and the extent 
 of the protection afforded to the watches before alluded to, 
 when placed in various magnetic fields. 
 
 "Two palladium balance springs of different alloys were 
 placed in a uniform field, the direction of the lines of force 
 in which was determined by a very thin layer of iron filings. 
 On placing the springs in various positions in the field no 
 change in the grouping of the lines was observable. They 
 were neither concentrated on the palladium alloys nor 
 repelled from them ; that is to say, they were neither 
 appreciably paramagnetic nor diamagnetic. 
 
 According to the experiments of Faraday and the later 
 investigations of Pliicker, palladium is a paramagnetic sub-
 
 stance ; that is to say, it concentrates the lines of force upon 
 it after the manner of iron. Its paramagnetic properties 
 are, however, comparatively feeble. 
 
 " Bearing in mind the paramagnetic character of many 
 of the components of the preceding alloys, their failure to 
 exhibit any of the properties ordinarily recogni/ed as mag- 
 netic is interesting from a scientific standpoint. Thus in 
 alloys Nos. i and 2 the copper is the only component that 
 is diamagnetic. In alloy No. 2, the steel and nickel are 
 powerfully paramagnetic, and the palladium and platinum 
 are also paramagnetic. The copper, gold, and silver only 
 are diamagnetic. In alloy No. 4, all the components are 
 paramagnetic except the copper, silver, and gold. 
 
 " The masking of the paramagnetic properties of some 
 of the components of an alloy would seem to point to the 
 probability of such alloys being formed by true atomic or 
 chemical combination. 
 
 " In order to observe the effect of an intense field on the 
 springs they were placed in the very powerful field of an 
 electro-magnet, the massive pole-pieces of which were 
 but a few inches apart. No deflection of the light springs 
 was observed when suspended in this field. When allowed 
 to fall through the narrow gap between the pole-pieces, they 
 fell quite freely. The intense field failed to produce any 
 appreciable magnetism in them. 
 
 "Similar experiments were tried with the compensating 
 balance with the same result." 
 
 Professor Houston says the segments of Paillard's balances 
 consist of two laminae of different palladium alloys, the co- 
 efficients of expansion of which are so proportioned as to 
 permit them to act as in the compound balance of brass and 
 steel. This does not agree with the prescription in 
 Paillard's English patent, which clearly defines the inner 
 part of the rim as palladium alloy and the outer part as 
 silver or brass. It adds : 
 
 " After melting the alloy, the molten metal is poured 
 into a mould to form a plate. This plate, when cold, is 
 forged or hammered, being frequently heated to red heat, if 
 becoming too hard, until the desired thickness is obtained. 
 
 H
 
 There is then stamped out, by means of a die, a disc, in the 
 centre of which a hole is pierced, and the disc is turned 
 true. The hole and the one side of the disc are then filled 
 up and covered with a fire-proof clay, or any such suitable 
 material (plumbago, for instance), which is allowed to dry 
 on the disc, so that the latter, in all its parts so covered, 
 will not be touched by the second metal in the subsequent 
 operation. The so prepared disc is now inserted in a 
 crucible, in which pieces of brass or silver and borax are 
 placed and melted, the crucible being introduced into a 
 muffle furnace. The molten metal will form a rim around 
 the disc. If preferred, the silver rim may be produced by 
 depositing this metal by a galvanic process. 
 
 " The melting of the metal around the alloy disc will 
 require some skill or experience, inasmuch as care must 
 be taken that the metal is not heated more than re- 
 quired for melting the same, so as to avoid melting of the 
 alloy disc, or even to avoid the molten metal attacking the 
 alloy disc. If the rim or ring is made of brass, the melting 
 of the latter may be facilitated by an ample addition of 
 zinc, which may form 40 to 50 per cent, of the brass. In 
 using silver, about 20 per cent, of zinc may be added. The 
 ample addition of zinc offers this advantage, that the metal 
 becomes more expansible, and thereby increases the 
 function of the bimetallic balance. If silver is deposited by 
 the galvanic process, this will require the employment of 
 screws or heavy masses, for the compensation to be a 
 sufficient one. 
 
 "The rest of the work for making the balance is entirely 
 mechanical, and does not differ materially from the method 
 heretofore in use. 
 
 " First, the excess of metal is removed by means of a 
 file or a hand-tool lathe, and then the centre part of the 
 disc, either from one side, or from both sides, is hollowed 
 out, so that in this centre part there remains a thickness 
 only, corresponding to the thickness of the middle arm, 
 which is to be cut out, care being taken to preserve a 
 bimetallic rim of such width, that it is composed by ^ of 
 palladium alloy and by 3 of brass and silver. By partially
 
 "5 
 
 removing the remaining central part of the disc (by filing 
 or otherwise), the central arm or arms for the central pin 
 are termed, so as to complete a balance. 
 
 " The boring of the screw holes and the cutting of the 
 thread in them, is made by means of a dividing engine. 
 
 " The arm or arms of the balance, as well as the interior 
 of the bimetallic rim are generally tempered, whereas the 
 exterior, and the faces of the said rim, may be polished. 
 
 " It is desirable, in order to give hardness, to heat the 
 balance in the same way as for tempering balance springs. 
 The bimetallic rim of the balance is then cut towards the 
 arms, and provided with its screws or set weights." 
 
 M. Rambal, describes a balance with the inside of 
 pure malleable nickel and the outside of brass : good 
 timing was obtained, but the balance was soft. 
 
 In 1888 E. Golay patented a non-magnetic balance ; the 
 interior ring being formed of an alloy composed of about 40 
 per cent, platinum, 35 per cent, copper, and 25 per cent, 
 nickel, the external ring consisting of about 55 per cent, 
 silver, 35 per cent, zinc, and 10 per cent, copper. 
 
 Mr. T. D. Wright, who found constructional difficulty 
 with brass and zinc, has used with success balances having 
 the arm an inner rim of platinum and the outer part 
 of the rim of silver. He describes them as compensating 
 well, but they required careful handling by reason of their 
 softness. 
 
 Good results have also been obtained by the Waltham 
 Watch Company, but the nature of their alloy is not 
 divulged. Messrs. Nicole, Nielsen and Co. have introduced 
 very excellent but costly balances in which the inner part 
 is an alloy of platinum and iridium. Mr. Daniel Buckney 
 patented the use of steel alloyed with about 24 per cent, of 
 manganese for the inner ring. Both the platinum-iridium 
 and the manganese-steel alloys are, I believe, very intractable 
 when subjected to cutters in course of manufacture. 
 
 Dealing with the metals and alloys enumerated, it seems 
 that so far as a material for balance springs is concerned, 
 Paillard's palladium alloy appears to best meet all require- 
 ments. It is not liable to rust. It is non -magnetic. It is 
 
 H2
 
 Ili 
 
 lasting. With it the troublesome " middle temperature 
 error" is materially lessened. Special treatment is, how- 
 ever, required. It is heavier than steel, and therefore 
 smaller springs should be used foi watches ; the overcoil 
 should be raised more above the body of the spring and 
 brought in nearer to the centre. It is comparatively soft, 
 and when its form is to be altered ivory-tipped or other 
 suitable tools should be used, and it must be bent with care. 
 Bending to and fro will quickly spoil a palladium spring. 
 For balances, there is no particular style or material so far as 
 my observation and enquiry go that can be indicated as 
 being absolutely non-magnetic, yet certain in its action and 
 free from objection. 
 
 Steel-Nickel Alloys. The researches of Dr. C. K. 
 Guillaume, at the Sevres Office of Weights and Measures, 
 have revealed the fact that steel liberally alloyed with 
 nickel produces a compound with a very small co-efficient 
 of expansion. Dr. Guillaume found the co-efficient of 
 expansion of steel alloyed with 36^2 per cent, of nickel 
 (named Sevres alloy) to be but 8, in comparison with brass 
 189; steel or iron, 108 to 122; glass, 86: wood (Yw'.f <lc 
 sapttt), 44. 
 
 Professor M. Thury conducted some experiments with a 
 view of determining the variations in the elasticity of 
 Sevres alloy when subjected to changes of temperature : 
 and he discovered that in a range of temperature of 22 
 Centigrade (4- 15 to 4- 37 Cent.), the elasticity actually 
 increased with a rise in temperature, thus reversing the 
 behaviour of steel under similar circumstances. Such a 
 startling result leads at once to conjectures of a revolution 
 in the methods of compensating watches. Accepting the 
 Sevres allo)- with its low co-efficient of expansion as a suit- 
 able material for the balance of a watch, it is easy to 
 suppose that by some slight variation in the proportion of 
 nickel employed a substance may be obtained for the 
 balance spring with a ratio of expansion calculated to 
 entirely neutralise the temperature error in timekeepers 
 furnished with a balance. Though watches with nickel - 
 steel balances and springs have been, I believe, tested with
 
 H7 
 
 encouraging results, further experimentsnn the preparation 
 of the alloy, and much labour in the production and appli- 
 cation of springs, are of course necessary before one can say 
 positively the position steel-nickel alloys will take in the 
 construction of horological mechanism. Though the steel- 
 nickel alloys cannot be fire-hardened, they are readily 
 compressed and stiffened by hammering or rolling. They 
 are less affected by magnetism than steel, but can hardly be 
 classed under the head of non-magnetic. 
 
 CHAPTER VII. 
 GAUGES. 
 
 Hewitt's Balance Spring Tester. This gauge, 
 designed by Mr. T. Hewitt, is invaluable to the springer of 
 
 I'ig. 70. 
 
 high-class timekeepers. Dr. Hooke, the inventor of the 
 balance spring, in explanation of its isochronous property
 
 gave utterance to the now well-worn maxim, it/ tcnsio sic ris. 
 (As the tension is, so is the force.) And in theory this is 
 true of springs perfect in construction, though it sometimes 
 happens that a spring apparently faultless contains some 
 latent defect that renders its action not regularly progressive. 
 \Yith the tool shown in the engraving, the behaviour of a 
 spring under different degrees of tension may be tested 
 before it is applied to the timekeeper, and if it prove to be 
 bad it may be at once rejected, and much labour saved that 
 would be otherwise thrown away. The eye of the spring is 
 held in a spring collet, and its outer end caught in a split 
 nose at the end of the arm in front. This arm is held in 
 a split bearing, and is capable of adjustment to suit springs 
 of various kinds. The circular scale is divided into degrees 
 and into radians, and turns friction-tight on its axis so that 
 it may be readily set to zero. By means of the knurled 
 knob at the back of the tool the spring may be wound up, 
 weights being gradually added to the scale pan to balance 
 it. A fair average of the extremes of winding a balance 
 spring would be subjected to in the long and short vibra- 
 tions of a watch, would be from three to four radians, and 
 if for every tenth of a radian between these points the same 
 increment of weight were required, the progression of the 
 force of the spring under trial would be regular. The 
 bending moment of any spring may be ascertained, and in 
 the case of substituting one balance for another in order to 
 obtain increased vibration, the relative moments of inertia 
 of the old and the new balance may be readily found. The 
 central staff holding the eye of the spring is very nicely 
 fitted into jewelled holes. It carries the lever to which the 
 scale pan is hung at exactly i inch from the centre of 
 motion, and then the bending moments, etc., are very 
 readily calculated in inch-grains. 
 
 Daldorph's Gauge. A gauge by Mr. Daldorph which 
 is shown in the accompanying engravings, is a useful form 
 when it is necessary to substitute for a spring another of a 
 different size or strength. 
 
 Fig. 71 is a view of the back of the gauge. Upon a plate 
 of brass is mounted an arbor, with a balance spring of
 
 medium strength pinned in as shown. The projecting nib 
 
 A is carried round from its point of rest (that is, bearing 
 
 Fig. 72. 
 against the stop, as shown by the dotted lines) nearly a
 
 I2O 
 
 whole turn, when it is prevented from flying back by the 
 catch B. Referring to Fig, 72 it will be seen that the arbor 
 is carried through the plate, and has a pointer, resembling 
 a long seconds hand, attached to it. A word of explanation 
 is now required respecting the little holder (D), represented 
 in Fig. 73. The twee/.er-like points are kept together by 
 
 - 73- 
 
 the spring forming one of the legs, with just sufficient force 
 to grasp a balance-spring tightly. But if both legs are 
 pressed together somewhere about c the points will open. 
 With this tool take hold of the eye or inner coil of the 
 balance-spring that has to be discarded. The stout leg of 
 the tool has a hole drilled up it to fit the end of the arbor, 
 and, holding the outer coil with an ordinary pair of tweezers 
 (see Fig. 72), push the knob E. The nib A being released 
 the pointer flies back some distance according to the 
 strength of the spring being tried. 
 
 The spot indicated by the pointer being noted, the rela- 
 tive strength of any spring which it may be proposed to 
 substitute for the one discarded may be ascertained with 
 great exactness. The whole operation occupies but a 
 remarkably short space of time, and, in addition to the dial 
 being sub-divided into small divisions, it will be a great 
 practical help if the spot indicated by the pointer for every 
 number of springs in general use be marked by correspond- 
 ing numbers on the dial. 
 
 In factories where thousands of watches are produced, 
 each one the counterpart of the others, a more systematic 
 procedure may be adopted than is otherwise possible. The 
 balance sprr, : .l fitted to collets and studs, and their 
 
 power of resistance measured in the following way : The 
 spring to be tested is held on a centre by a collet, while its 
 stud end is caused to rotate one turn. l]ut in moving from 
 the zero point it had to overcome the resistance of a strong
 
 121 
 
 balance spring fixed to the underside of a table, at the edge 
 of which is a finely graduated and numbered scale, and the 
 number of degrees or graduations which the spring under 
 trial is able to move the stronger one attached to the 
 machine is thus registered. The spring is then turned back 
 two rotations, thus testing its resistance to one turn of 
 unbending. The mean of the two readings is then taken, 
 and the girl who presides at the machine takes off the 
 spring tested, and drops it into a compartment marked with 
 a corresponding number. All the balances used are weighed 
 with great exactness, and placed in marked compartments 
 in the same way as the springs. As it is known that a 
 balance of a certain size and weight takes a certain number 
 of spring, the selection of either is thus reduced to a mere 
 perfunctory operation. 
 
 CHAPTER VIII. 
 
 OBSERVATORY TESTS. 
 
 Kew Observatory. At the Observatory, Richmond, 
 watches are tested, and if the performance is satisfactory, 
 certificates are issued. The trial for Class A certificates is 
 divided into eight periods of five days each : ist, pendant 
 up ; 2nd, pendant right ; 3rd, pendant left ; 4th, dial up in 
 refrigerator (about 40 Fahr.) ; 5th, dial up in room (about 
 65 Fahr.) ; 6th, dial up in oven (about 90 Fahr.) ; yth, dial 
 down in room ; 8th, pendant up. All but periods 4 and 6 
 are at the ordinary temperature of the room. The 4th, 5th, 
 6th and yth periods are each extended one da)-, and on the 
 first day of each the watch is not rated. To obtain the 
 certificate, the daily rate must not exceed 10 sees. ; the 
 mean difference of daily rate during each period must not 
 exceed 2 sees. ; the difference of mean daily rate between 
 pendant up and dial up must not exceed 5 sees., and between 
 pendant up and any other position, 10 sees. ; change of 
 temperature must affect the daily rate by less than one-third 
 of a second per degree Fahrenheit. The behaviour of a 
 watch keeping just within the limits throughout the trial is 
 appended.
 
 122 
 
 Mean daily rate gaining 9-9 sees. 
 
 variation of daily rate .. .. .. .. .. 1*9 ,, 
 
 ,, for i F 0-3 
 
 Difference of mean rate between pendant up and dial up. . 4.9 ,, 
 
 ,, ,, ,, pendant right. . 9*9 ,, 
 
 ,, ,, ,, ,, pendant left . . 9^9 ,, 
 
 ,, ,, dial up and dial down. . 9^9 ,, 
 
 It will be observed that the MEAN variation is taken into 
 account. For instance, between pendant up and dial up 
 the piece may vary more than 5 sees, between one day and 
 the next, provided the MEAN daily variation of the pendant 
 up period does not differ from the MEAN daily rate of the 
 dial up period by more than 5 sees. Marks, showing the 
 excellence of the watch, are awarded to A certificates : 100 
 marks representing absolute perfection, that is, 40 for no 
 variation of rate, 40 for no change of rate with change of 
 position, and 20 for no temperature error. A watch just 
 touching the limit of variation allowed would get no marks, 
 and for every mark obtained the 2 sees, limit would have 
 to be reduced -05, the 10 sees, limit '25, and the temperature 
 error '015. The words " especially good" are added to an A 
 certificate when the watch obtains not less than 80 marks. 
 
 For B certificates the watches are tried 14 days pendant 
 up, 14 days dial up, one day in the oven, one day at 
 temperature of the room, and one day in the refrigerator; 
 variation limit in each position 2 sees., and between hang- 
 ing and lying 10 sees. " Especially good" when the 2 sees. 
 limit is reduced to 075 sec., the 10 sees, to 5 sees., and the 
 temperature limit o - 2 per sec. Fahr. With the sanction of 
 the Kew Committee, Dr. Charles Chree, superintendent of 
 the Observatory, has favoured me with the following 
 detailed account of the course pursued, and, as an example, 
 the performance of a watch throughout the trial for an A 
 certificate. 
 
 On arrival at the Observatory, the watches are wound, 
 divided into classes A and B in accordance with the 
 instructions given by the sender on the " Application 
 Form," and full particulars of the description and escape- 
 ment, &c., of each watch are entered in the Observatory 
 Kegister. Each movement then receives a " K.O. register
 
 123 
 
 number," by which number it is known and distinguished 
 throughout the trial. 
 
 Watches that come to hand a few days previous to the 
 commencement of the trial are kept dial up and wound 
 daily until the test begins. The following short description 
 of the procedure followed refers to the watches entered for 
 the Class A trial, as they form by far the largest proportion 
 of the movements sent for rating : 
 
 When the movements are classified and numbered they 
 are placed upright, with the pendant up, in mahogany 
 racks with semi-circular recesses, specially made for the 
 purpose. The racks are of various sizes so as to accommo- 
 date all ordinary watches, and wood or paper pluggets are 
 used to fix any watches not quite secure in their place. 
 
 These racks fit into mahogany trays, which run in 
 grooves cut in the sides of the interior casing of a large 
 Milner safe, and are arranged so that each tray can be 
 removed for examination without disturbing neighbouring 
 watches. The interior of the safe is kept as closely as 
 possible at a temperature of 65 P., and to allow this to be 
 done during the colder months of the year, the safe is built 
 up so as to enable a Bunsen gas burner to be fitted beneath 
 it. This burner is provided with a guard shield between 
 the flame and the bottom of the safe, and enables the mean 
 temperature to be regularly maintained. The extremes 
 of temperature variation are registered by maximum and 
 minimum thermometers. 
 
 The separate sets of watches are carefully compared at 
 the same time each day [immediately after the noon obser- 
 vations] with the standard mean -time clock, and, as soon 
 as the comparison is made and the reading entered in the 
 Register, the watch is wound up and returned to its place. 
 The error on G.M.T. of the standard clock (always very 
 small) is determined mainly by direct time current from 
 Greenwich, and also should it be required by sun and 
 star transits, and by inter-comparisons with two other 
 regulators. 
 
 After being observed for six days with the pendant zip, 
 the watches are placed with the pendant right and com-
 
 124 
 
 pared for five days, and then changed to pendant left, and 
 the same procedure gone through. 
 
 The daily rates on the "quarters" being thus deter- 
 mined, the movements are next tested for temperature 
 compensation, for which purpose they are kept, dial up, for 
 periods of six days each at temperatures of about 40", 65, 
 and oxr F. 
 
 The "cold" chamber used for the test at 40" consists 
 essentially of a small safe, fitted with a sheet-zinc covering. 
 Built round this a yellow deal casing, surrounded with four 
 inches of non-conducting packing. 
 
 The low temperature is maintained by means of blocks of 
 ice, and proper provision is made for draining away the water 
 resulting from the melting of the ice, so that practically the 
 chamber is surrounded by cold air only. The interior is 
 furnished with sliding lattice shelves to hold the watches, 
 dial up, with dishes containing potassic chloride for drying 
 the air, and with thermometers to register the daily range 
 of temperature. To avoid the necessity of taking out the 
 watches into a warmer air during their comparison, the 
 cold chamber is built as near as possible to the clock dial, 
 so that the observer can read and wind the watches without 
 removal. 
 
 Experience with some thousands of movements has 
 shown that no danger from moisture condensation during 
 the temperature trials need be apprehended. 
 
 The "oven" or hot chamber is designed on similar 
 lines to the " cold" safe, with zinc casing, non-conducting 
 packing, &c., and hot water is employed to keep it at about 
 90 F., the temperature of the water being maintained by 
 means of a Fletcher tubular boiler. 
 
 On the completion of the temperature tests, the watches 
 are placed in the large safe (kept at 65), dial clown, in 
 which position five daily rates are secured, and then they 
 are changed to the final position, which is the same as the 
 first, vi/.., pendant up. 
 
 Having now obtained five daily rates in each of the 
 positions pendant up, right, and left, dial up and dial 
 down at 65, and also with dial up at 40 and 90 F. we
 
 125 
 
 have the necessary data to give us the marks for positional 
 adjustment, for temperature compensation, and for varia- 
 tion of daily rate. 
 
 The accompanying " Rate-record" gives the full history 
 of the performance of a watch entered for the Class A trial, 
 with the average daily rates, variation of rates, and 
 temperatures. 
 
 In the " Abstract of Results," supplied with each certifi- 
 cate, these averages are all given, and in addition " Marks' 
 are added for superior performance, based upon the scale 
 O 100, where O represents the marks which would be 
 awarded to the watch which only just managed to pass 
 each separate part of the tests and obtain the bare certifi- 
 cate, while loo would be the marks gained by the ideal 
 and absolutely perfect watch. 
 
 The total, which of course is only theoretically possible, 
 is made up as follows : 
 For complete absence of variation of rate = 40 
 
 ,, Absolute freedom from change of rate. j 
 
 } 40 
 with change of position * 
 
 ,, Perfect temperature compensation = 20 
 
 100 
 
 Copy of the "Rate-Records" of gold keyless crystal 
 watch, No. 10001, by X.Y.Z., rated from May i to June 14, 
 1898, and awarded a Class A certificate, especially good : 
 
 In awarding the marks the following rules are followed : 
 
 I. [ 'a nation of daily rate. 
 
 Number of marks possible =40 
 
 Limit of variation of rate allowed . . 2 seconds. 
 
 For the number .v, of marks awarded we have the 
 
 J 
 
 formula x where ['is the mean variation of rate, 
 
 0-05, 
 
 in seconds, for the watch.
 
 126 
 
 Davs 
 of 
 Trial. 
 
 Posit'n. 
 
 Mean 
 Temp- 
 erature 
 for 
 period. 
 
 Daily 
 Rate. 
 
 Mean 
 Daily 
 
 Rate 
 for 
 period. 
 
 Varia- 
 tion 
 from 
 Mean 
 Rate. 
 
 Mean 
 Varia- 
 tion 
 of Rate. 
 
 i6 
 
 I 
 Pend'nt 
 up. 
 
 Degrees 
 
 65 
 64 
 66 
 64 
 66 
 
 SITS. ' 
 
 *-i*S 
 
 -1-0-75 
 4-1-25 
 
 4-2-0 
 4-2-0 
 
 Sees. 
 4-1-50 
 
 Sees. 
 
 O'OO 
 
 75 
 25 
 50 
 
 0-50 
 
 Sees. 
 
 4- 
 0-40 
 
 7 ii 
 
 II 
 Pend'nt 
 right. 
 
 67 
 66 
 66 
 64 
 63 
 
 *-rs 
 
 4-0-75 
 
 *-o-5 
 
 4-0-75 
 
 4-1-0 
 
 4-0-90 
 
 0-60 
 
 '5 
 40 
 
 15 
 
 O'lO 
 
 4- 
 0-28 
 
 12 16 
 
 Pend'nt 
 left. 
 
 Ill 
 
 63 
 64 
 64 
 65 
 65 
 
 4-2-25 
 4-2'25 
 
 -i*5 
 
 -i'75 
 
 4- I '25 
 
 4-i-8o 
 
 0'45 
 45 
 30 
 'OS 
 
 0'55 
 
 4- 
 0-36 
 
 INTERMEDIATE DAY. 
 
 
 Dial up 
 
 4i 
 
 4-0-25 
 
 
 0'35 
 
 
 in 
 
 40 
 
 4-0-25 
 
 
 '35 
 
 17 22 Refrige- 39 
 
 4- 1-0 
 
 4-0-60 
 
 40 
 
 rator. 39 
 
 4- ro 
 
 
 40 
 
 IV. 41 
 
 4-0-5 
 
 O'lO 
 
 INTERMEDIATE DAY. 
 
 
 V 
 
 66 
 
 -t-i'75 
 
 
 0-15 
 
 
 Dial up 
 
 64 
 
 t-i'75 
 
 
 15 
 
 2328 
 
 at Or- 
 
 64 
 
 *-i*5 
 
 4-1-60 
 
 io 
 
 
 dinary 
 
 65 
 
 4-1-25 
 
 
 35 
 
 
 Temp. 
 
 65 
 
 +-I-75 
 
 
 0-15 
 
 INTERMEDIATE DAY. 
 
 
 VI 
 
 90 
 
 4-2-75 
 
 
 0-60 | 
 
 29-34 
 
 Dial up 
 in 
 
 9 
 92 
 
 4-2-25 
 
 4-2-O 
 
 4-2-15 
 
 io 
 
 15 
 
 
 Oven. 
 
 90 
 
 4- 2-0 
 
 
 15 
 
 
 
 89 
 
 -f-i'75 
 
 0-40 ! 
 
 INTERMEDIATE DAY. 
 
 0-32 
 
 0-18 
 
 0-28 
 
 
 
 6 4 
 
 4-1-0 
 
 
 0-30 
 
 
 
 VII 
 
 65 
 
 4 ro 
 
 
 3 
 
 4- 
 
 3540 
 
 Dial 
 
 66 
 
 4-0-25 
 
 4-0-70 
 
 '45 
 
 0-26 
 
 
 Down. 
 
 67 
 
 4-0-75 
 
 
 05 
 
 
 
 
 65 
 
 4- 0-5 
 
 O'2O 
 
 
 
 
 64 
 
 4-I'O [ O-^O 
 
 
 
 Pend'nt 
 
 63 
 
 4-0-75 
 
 '55 +- 
 
 4145 
 
 up. 
 VIII 
 
 63 
 65 
 
 4-1-25 4-I-30 
 
 t-i'5 
 
 'OS 0-36 
 2O 
 
 
 64 4-2'0 0-70 
 
 Signifies rate gaining. - Rate losing.
 
 I2 7 
 
 In the example of the performance of a watch given on 
 page 126, the average variation of rate in each of the 8 
 periods of the trial is 0-40, 0-28, 0-36, 0-32, o-iS, 0-28, 0*26, 
 
 0*36 second, and the mean of the 8 values= 0-305, which 
 
 corresponds to ]' in the above formula. Hence, in this 
 example, the marks awarded in respect of variation of daily 
 rate wonld be 
 
 r _ 2-0 0-305 _ 1-695 _ , rq 
 0-05 " 0-05 ^^' 
 
 II. POSITIONAL ADJUSTMENT. 
 The number of marks possible . . =40 
 
 ,, limit allowed for positional error 10 seconds. 
 For the number x* of marks awarded we have the 
 
 formula x. l( ~- where p is the mean change of rate 
 0-25 , 
 
 in seconds due to changes of position. 
 
 Referring again to the record of watch 10001, we see 
 that the rates in the various positions are : 
 
 I. Pendant up : + i -4 sees, (mean of periods iand8.) 
 
 II. ,, right = + 0-9 
 
 III. left : + 1-8 
 
 V. Dial up (ord.) = + r6 
 
 VII. Dial down = + 07 ,, 
 
 The average rate was thus + 1-28 sees., and the rates in 
 the various positions differed from the average as follows: 
 
 I. o.i 2 sec. 
 
 II. 0-38 
 
 III. 0-52 
 
 V. 0.32 ,, 
 
 VII. 0-58 ,, givingamean difference
 
 128 
 
 T "O "J 
 
 of - = 0-38 sec., (to the nearest - oi). Hence, applying 
 
 
 
 the formula, we have 
 
 io'o o"?8 9 - 62 , 
 
 .r. = 38-5 marks. 
 
 0-25 0-25 
 
 III. TEMPERATURE COMPENSATION. 
 
 In the case of the temperature compensation, we have 
 two factors to deal with (i) the three different mean tem- 
 peratures in the trials in the " dial up " position, and (2) the 
 three corresponding mean rates. We first take the mean of 
 the above three mean temperatures, and the mean of the 
 above three mean rates. We next find the departures of 
 the three mean temperatures from their mean, and the 
 departures of the three mean rates from their mean. We 
 then find the value of a quantity, 0, such that 
 
 _Sum of departures of the 3 mean rates from their mean 
 Sum of departures of the- 3 mean temps, from their mean' 
 
 and deduce the marks, .r :1 , from the formula 
 
 0.30 - 
 
 -*-a = 
 
 OI 5 
 
 This gives a total possible of 20 marks, and allows of an 
 extreme variation of 0-30 seconds for i F. 
 
 In the case of watch [10001] the three mean tempera- 
 tures (to the nearest whole degree) are 40, 65, and 90, and 
 the corresponding mean rates are -f o'6o, -f- r6o, and + 2-15 
 seconds. The mean of the three mean temperatures is 65, 
 and the departures from this are 25, o and 25, whose sum is 
 50. The mean of the three mean rates is -f- 1-45 sees., and 
 the departures from this are 0-85, 0-15, and 075, whose sum 
 (irrespective of sign) is 175. Hence in this case 
 
 0= ?= -035; 
 
 50 
 
 and so bv the formula x, = '3-'35 __ '265 = 
 
 015 .015
 
 129 
 
 Hence the total marks awarded to this watch would be 
 
 In respect of variation of daily rate 33*9 
 
 ,, ,, ,, change of rate with change of position 38.5 
 
 ,, ,, ,, temperature compensation = 177 
 
 90' i 
 
 In the case of chronograph watches, at the conclusion of 
 the ordinary Class A trial just described, the running and 
 adjustment of the chronograph work is examined. 
 
 The chronographs arc run for periods of 24 hours each 
 with the fly-back hands in action (both ordinary and split- 
 seconds), and the effect upon the ordinary dail}' rate of the 
 watch of this additional work is noted, a change of + 5 
 seconds daily being allowed. 
 
 Examination is also made of the general behaviour with 
 respect to stopping and starting, action of minute recorder, 
 and freedom of the "fifths" hands from a tendency to 
 "hold-up" with improper use of the plunger, and should 
 any defects be exhibited (now happily much less rare than 
 formerly) the watch is not granted a certificate until the 
 action has been corrected. 
 
 " NON-MAGNETIC " TEST. 
 
 Watches are subjected to a "non-magnetic" test only 
 when a special request is made to that effect, and when the 
 test has been applied the fact is specifically stated on the 
 certificate. 
 
 The examination is conducted with the aid of a powerful 
 electro-magnet, possessing an iron core and moveable pole- 
 pieces, each 2f inches in diameter. 
 
 i
 
 130 
 
 The watch is placed in the air-gap between the core and 
 pole-piece, and its behaviour examined with the lines of 
 force passing from front to back, and from side to side. In 
 both positions, after being held stationary for some time, 
 the watch is slowl)' rotated through a complete revolution, 
 first in one direction and then in the other. The rate is 
 noted before and after the magnetic trial, and any serious 
 change of rate is made a cause of rejection. 
 
 Greenwich Observatory. Here the tests are organ- 
 ized for a different purpose. Watches and chronometers 
 required for the British Navy are purchased after estimation 
 of their value from their performance at the Greenwich 
 trials. Instead of giving marks for good going, a "trial 
 number " is deduced b} 7 deviations from exactness. Trials 
 of " Deck Watches " submitted are held annually, beginning 
 in October and lasting 16 weeks. Any manufacturer is 
 accorded permission to deposit suitable timekeepers on 
 written application to the Hydrographer to the Xavy at 
 Whitehall. The watches are used principally for noting the 
 time when observations of the sun or stars are being taken. 
 The chronometers not being on deck their error is ascer- 
 tained by comparison with the deck watches. Watches for 
 this purpose are preferred to be of about No. 20 size, with 
 seconds dial not less than -,-v of an inch diameter, accurately 
 and distinctly divided. The hour and minute hands and 
 the hour figures should not be so heavy as to interfere with 
 the visibility of the seconds. 
 
 The trial is conducted in the following way : 
 
 Watch horizontal, dial tip, in room for 6 weeks 
 
 ,, ,, ,, in oven ,, i week 
 
 ,, vertical, pendant up ,, ,, 4 days 
 
 right 3 
 
 left 3 
 
 up ,, 4 
 
 ,, horizontal, dial up ,, ,, i week 
 
 ,, ,, in room ,, ,, 6 weeks 
 
 Total duration of trial . . 16 weeks
 
 The mean temperature in the oven is from 80 to 90 
 Fahrenheit. 
 
 The watches are wound and noted every morning but to 
 obtain the trial number the rating is summarized in this way : 
 
 ^f: p. 
 
 T3 
 
 
 
 
 ha 
 
 ^ ^ - 
 
 Difference of Weekly Rates. ^-j- 
 
 !* . 
 
 b '>2 
 
 
 
 i 4 
 
 '- i 5 S a .0 
 
 
 Pendant 
 
 Pendant 
 
 
 
 *j a x 
 w 
 
 Pendant up 
 Dial up. 
 
 left. 
 Dial up. 
 
 left. 
 Dial up. 
 
 1 4 
 
 sif 
 
 
 e. 
 
 d. 
 
 * 
 
 i^ 
 
 s. 
 
 S. S. S. 
 
 S. 
 
 
 14-8 
 
 4-4 + 1-2 4*0 
 
 4-5 
 
 25-8 
 
 The figures given are those relating to the first watch 
 on the 1898 list. It will be observed that in fomulating the 
 trial number steady going is regarded as of primary impor- 
 tance, and errors between any week and the next carry 
 double the penalty of deviations spread over the 16 weeks of 
 trial. The trial number obtained in the case quoted (25-8) 
 is small. The greater the variations the higher the trial 
 number, and in preparing the list watches with a higher 
 trial number than 100 are unnoticed. 
 
 Marine chronometer trials are also conducted annually 
 at Greenwich. They begin in July and extend over 29 
 weeks. They are tested in the oven for two periods of four 
 weeks each, at temperatures ranging from 75 to 100 
 Fahrenheit, and also at the ordinary summer and winter 
 temperatures of the room. 
 
 To obtain the trial number the rates are summarized as 
 below. The figures given represent the performance of the 
 chronometer occupying the premier position in the 1898 
 Report. 
 
 Difference between 
 greatest 
 and least weekly rates. 
 a. 
 
 Greatest difference 
 between one week I 
 and the next. 
 b. 
 
 Trial Number. 
 a plus zb. 
 
 s. 
 6-0 
 
 s. 
 2-8 
 
 u-6 
 
 12
 
 132 
 
 Note on Timing Repeating Carriage Clocks. 
 
 The quickest method is to listen to the first blow of the 
 hammer on the bell (or gong) at each hour and half- 
 hour, noting at the same time the position of the seconds 
 hands on the regulator. Thus : supposing the blow is given 
 exactly at 12 h. o m. o s. make a note of it, and compare 
 again at 12 h. 30 m. o s., when the half-hour blow is given 
 at, say, 12 h. 30 m. 5 s., make a note of this also, but do not 
 alter the index, as this difference may arise from the imper- 
 fect position of the half-hour pin in the cannon wheel; at 
 i h. o. m. o. s. compare as before, and as it is the same pin 
 coming into action as at 12 h. om.os., any difference in the 
 time of the first b!o\v will be indicative of a gain or loss, 
 as the case may be, thus : it may be given at i h. om. 4 sec., 
 which will show a loss of 4 s. in the last hour, and the clock 
 may now l>c regulated accordingly. The blow may next be 
 given at i h. 30 in. 8 s., showing the effect the moving of the 
 index has had in making it gain i s. in the preceding half- 
 hour ; regulate and compare again at 2 h. o m. o s., and if 
 the first blow is given at 2 h. o in. 3 s. it will show that the 
 clock has been keeping correct time for the last half-hour. 
 
 CHAPTER IX. 
 EXAMINATION OF ESCAPEMENTS. 
 
 Examination of the Lever Escapement. See 
 
 that the balance staff is perfectly upright. See that the 
 escape wheel is perfectly true on edge and on face, and that 
 the teeth are equally divided and smooth; also, by gently 
 turning the wheel backwards, see that the pallets free the 
 backs of the teeth. If the wheel is out of truth it must be 
 set up in the lathe and re-bored. It can be fixed either 
 with shellac or in a brass sink bored out the exact size to 
 receive it. If the divisions are unequal, or the wheel has
 
 133 
 
 some thick teeth, it should be discarded. It is useless to 
 attempt to make the wheel right, and to reduce the corners 
 of the pallet to free the wheel is simply to spoil the escape- 
 ment for the sake of the wheel. At the same time, it must 
 be left to the operator to judge whether the amount of the 
 inaccuracy is serious. The whole affair is so minute that 
 no rule can be given. 
 
 Is the wheel the right size? If the lockings are too 
 light, and the greater part of the shake INSIDK, the wheel is 
 too small, and should be replaced by one larger. Before 
 removing the wheel, gently draw the balance round till the 
 point of the tooth is exactly on the locking corner, and see 
 if there is sufficient shake. If not, it will be prudent to 
 have the new wheel with the teeth a little 'straighter than 
 tne old ones. If the lockings are too deep and most of the 
 drop OUTSIDE, the wheel is too large and should be topped.* 
 
 The wheel is so fragile that care is required in topping, 
 which is done by revolving it in the turns against a diamond 
 or sapphire file. A brass collet is broached to fit friction- 
 tight on one of the runners of a depth tool : one side of this 
 collet is then filed away, leaving sufficient substance to 
 avoid bursting into the hole. On this flat a small piece of 
 sapphire file is attached with shellac, taking care that the 
 face of the file is parallel to the centre of the runner. The 
 escape wheel on its pinion with the ferrule attached is 
 placed in the centres of the depth tool furthest from the 
 adjusting screw, and the collet and file on one of the oppo- 
 site centres, and that centre fixed firmly by its clamping 
 screw. A very light hair bow is used to rotate the pinion, 
 and the depth tool laid on its side on the work-board, the 
 tool being closed by its screw until the teeth of the wheel 
 nearly touch the surface of the file ; now if a slight pressure 
 is made by the fingers on the uppermost limb of the tool, 
 at the same time rotating the wheel by the bow, the spring 
 
 * In planting the wheel and pallets it is always best to err, if at 
 all, by making them too deep rather than too light. If they are a 
 shade deep, topping the wheel soon puts matters right.
 
 134 
 
 of the tool will allow the teeth to be brought into contact 
 very slightly and without fear of bending the teeth ; the 
 wheel can be reduced as much as is necessary. 
 
 If the wheel is the right size and there is no shake 
 (which try, as before directed), the discharging corner 
 of the pallets may be rounded off by means of a 
 diamond file if they are of garnet. If they are of 
 ruby, the} 7 may be held against an ivory mill charged 
 with diamond powder. If the lockings are too light 
 and there is but little shake, they may be made safe 
 by polishing away the locking face a sufficient quan- 
 tity. If one locking is right and one is too light, the 
 one that is too light may be made safe by polishing 
 away the locking face as before, or the pallet may be 
 warmed and the stone brought out a bit. The locking 
 faces of the pallets should be sufficiently undercut to 
 draw the lever to the banking pins without hesitation. 
 If they require alteration in this respect, polish away 
 the upper part of the locking faces so as to give more 
 draw, leaving the locking CORNER quite untouched. But 
 proceed with great care, lest in curing this fault the 
 watch sets on the locking, as small watches with light 
 balances are very liable to do. If a watch sets on the 
 lockings, or on one of them, the locking face or faces 
 may be polished away so as to give less draw i.e., 
 have most taken off the CORNER of the locking. If 
 the watch sets on the impulse, the impulse face may 
 be polished to a less angle if the locking is sufficiently 
 deep to allow it, for it must be remembered that in 
 reducing the impulse the locking of the opposite pallet 
 will also be reduced. In fact, the greatest caution 
 should be exercised in making any alteration in the 
 pallets. 
 
 Sometimes, in new escapements, the oil at the escape 
 wheel teeth will be found to thicken rapidly through 
 the pallet cutting the wheel, showing that one or both 
 corners of the pallet are too sharp. If ruby, the corner 
 may be polished off with a peg cut to the shape of a
 
 135 
 
 pivot polisher, and with a little of the finest diamond 
 powder in oil ; if garnet, diamantine on a peg will do 
 it very well. Great care should be taken to remove 
 every trace of the polishing material, or the wheel may 
 become charged with it. 
 
 See that the pivots are well polished, of proper length 
 to come through the holes, and neither bull-headed nor 
 taper. A conical pivot should be conical only as far as 
 the shoulder ; the part that runs in the hole must be 
 perfectly cylindrical. They must have perceptible and 
 equal side shake, or if any difference be made the pallet 
 pivots should fit the closest. Both balance staff pivots 
 should be of exactly the same size. The end shakes 
 should all be equal. Bad pivots, bad uprighting, exces- 
 sive and unequal shake of the pivots are responsible for 
 much of the trouble experienced in position timing. 
 With unequal end shakes the pallet depth is liable to 
 be altered owing to the curved form of the pallet faces. 
 The action of the escapement will also be affected if the 
 end shakes are not equal, by a banking pin slightly 
 bent, a slight inaccuracy in uprighting, and other minute 
 faults. The infinitesimal quantity necessary to derange 
 the wheel and pallet action may be gathered from the 
 fact that a difference of '002 of an inch is quite enough 
 to make a tripping pallet depth safe or correct depth 
 quite unsound. 
 
 When the wheel and pallets are right see that the 
 impulse pin is in a line with an arm of the balance, 
 and proceed to try if the lever is fixed in the correct 
 position with relation to the pallets. Gently move the 
 balance round till the tooth drops off the pallet. Ob- 
 serve the position of the balance arm, and see if it 
 comes the same distance on the other side of the pallet 
 hole when the other pallet falls off. If not, the pins 
 connecting pallet and lever are generally light enough 
 to allow of the lever being twisted. To do this success- 
 fully a clamp to grasp the back and belly of the pallets, 
 as shown in Fig. 74, or some similar tool is necessary.
 
 I 3 6 
 
 Mr. R. Bridgman has devised a very superior clamping 
 tool for pallets, which, with some modification by Mr. 
 
 Fig, 74- 
 
 C. Curzon, is shown in Fig. 75. There is a spring for 
 keeping the slide in contact with the pallets when they 
 are placed in position, and until they are gripped tight 
 by the screw, and a lever under the body of the tool 
 by means of which the spring is overcome and the 
 
 pallets released when the operation is completed. It 
 will be seen that the screw passes through a split nut, 
 which may be set up as it wears. The jaws of the 
 tool should be faced with tin to avoid marking the 
 pallets. When the lever is right with relation to the 
 pallets, see that the pallets are quite firmly fixed to the 
 lever, and that the lever and pallets are perfectly in poise. 
 This latter is an essential point in a fine watch to be 
 timed in positions, but it is often neglected. 
 
 See that the escapement is in beat. When the 
 balance spring is at rest, the impulse pin should be on
 
 137 
 
 the line of centres, that is in the middle of its motion. 
 If this is not so, the spring should be drawn through 
 or let out from the stud if the position of the index 
 allows ; if it does not, the roller may be twisted round 
 on the staff in the direction required. 
 
 Is the roller depth right? If the safety pin has 
 insufficient freedom while there is enough run, the 
 roller is probably planted too deep. On the other hand, 
 if it is found that while the safety pin has plenty of 
 freedom there is no shake between the bankings the 
 roller depth is probably too shallow. When the impulse 
 pin is led round there should be an equal clearance 
 all round the inside of the horn, and the pin must 
 fall safely into the notch. If it binds in the horn and 
 bottoms in the notch it is too deep, and, on the other 
 hand, if with excessive clearance in the horn the pin 
 when it falls does not pass well into the notch it is 
 too shallow. The readiest method of altering is to warm 
 the roller, remove the impulse pin, and using a to-and- 
 fro motion with a wire and oilstone dust, draw the 
 hole in the required direction. If the pin is deep in 
 the notch and too tight in the roller to give a little, 
 it should be removed and flattened off a trifle more. 
 If too shallow, a triangular pin, or one of some other 
 shape with the point of contact more forward, can 
 generally be substituted by polishing out the hole to- 
 wards the crescent. If not, the staff hole in the lever 
 may be drawn to allow of shifting the lever sufficiently ; 
 or the recesses for the jewel settings of the balance 
 staff pivots may be scraped away on one side and rubbed 
 over on the other to suit. See as it passes round that 
 the impulse pin is free when in the notch. 
 
 Just as the safety pin is about to enter the crescent, 
 the impulse pin must be well inside of the horn. In 
 the single roller escapement a very little horn is re- 
 quired, unless the crescent has been made of an un- 
 necessary width. In very common work one occasionally 
 sees a flat filed on the edge of the roller instead of
 
 138 
 
 a crescent. There is no excuse for such a piece of 
 bungling. 
 
 A fault occasionally met with is that the impulse 
 pin after leaving the notch just touches on some part 
 of the inside of the horn in passing out. If a wedge 
 of cork is placed under the lever, so that the lever 
 moves stiffly, it can be readily seen whether or not the 
 impulse pin is free to leave the notch, and is free all 
 round the horn when the wheel tooth drops on the 
 locking. 
 
 See to the safety action. When the tooth drops 
 on to the locking, the safety pin should be just clear 
 of the roller. If it is not clear, the edge of the roller 
 should be polished down till it is right. If there is 
 more than clearance, the safety pin must be brought 
 closer to the roller. See upon pressing the safety pin 
 against the roller that the tooth does not leave the 
 locking, and that the impulse pin is free to enter the 
 notch without butting on the horn of the lever : also 
 that the safety action is sound, so that the pin is in 
 no danger of passing the roller. If the action is not 
 sound the diameter of the roller should be reduced 
 and the safety pin brought towards it sufficiently to get 
 a sound action if it can be done, but if the escape- 
 ment has been so badly proportioned as not to allow 
 of a sound action being obtained in this way, the pin 
 must be shifted forward and the bankings opened to 
 allow more run. 
 
 See if the banking pins are so placed as to allow 
 of an equal run on each side. If not, they should 
 not be bent, for with bent banking pins a difference 
 in the end shakes of the pivots will cause a difference 
 in the run. The banking p'.n allowing of the most 
 run should be removed, and the hole broached out to 
 receive a larger pin. 
 
 A fault rather difficult to detect, which is sometimes 
 met with in the double roller escapement, is the end
 
 139 
 
 of the impulse pin slightly touching the safety finger, 
 caused by excessive end-shake, or from a longer im- 
 pulse pin than was originally intended having been 
 put in. 
 
 To Examine the Chronometer Escapement. 
 
 See that the wheel is true and the teeth smooth and 
 perfect, and that the rollers properly fit the staff. See 
 that the end shakes and side shakes are correct. See 
 that that the "lights" between the wheel teeth and the 
 edge of the roller are equal on both sides when the 
 wheel is locked. If they are not, the foot of the detent 
 must be knocked a trifle to or from the centre of the 
 roller till the lights are equal. If the light is more than 
 sufficient for clearance, the roller must be warmed to 
 soften the shellac, and the impulse pallet moved out a 
 little. If the light is excessive there will be too much 
 drop on to the locking after the wheel tooth leaves the 
 impulse pallet, and with a large drop there is danger 
 of tripping. 
 
 To ensure safe locking the detent should be set on 
 so that when the banking screw is removed, and the 
 locking pallet is free of the wheel teeth, it will just 
 spring in as far as the rim of the wheel. 
 
 In pocket chronometer escapements it is especially 
 necessary to see that the face of the locking stone is 
 angled so as to give perceptible draw. Many pocket 
 chronometers fail for want of it. 
 
 The gold spring should point to the centre of the 
 roller. Bring the balance round till the discharging 
 pallet touches the gold spring preparatory to unlocking, 
 and notice how far from that point the balance moves 
 before the gold spring drops off the face of the pallet. 
 Then reverse the motion of the balance, and see if the 
 same arc is travelled through from the time the back of 
 the pallet touches the gold spring till it releases it. If 
 not, the horn of the detent must be bent to make the 
 action equal.
 
 140 
 
 Bring the discharging pallet on to the gold spring, 
 and let it bend the detent so that the locking stone is 
 as much outside the wheel as it was within when the 
 wheel was locked. The gold spring should then drop 
 off the discharging pallet. Make it to length, sloping off 
 the end from the side on which the pallet falls to 
 unlock, and finish it with great care. The gold spring 
 should be thinned near its fixed end as much as pos- 
 sible, and the detent spring thinned if it is needed. 
 The judgment of the operator must determine the proper 
 strength in both cases. The nose of the detent horn 
 should be nicely flattened and the corners rounded off. 
 
 The locking pallet should not be perfectly upright. 
 It should lean a little from the centre of the wheel, and 
 a little towards the foot of the detent, so that the lock- 
 ing takes place at the root of the stone, and then the 
 action of locking and unlocking does not tend so much 
 to buckle the detent. The face of the impulse pallet, 
 too, should be slightly inclined, so that it bears on the 
 upper part of the wheel teeth. By this means the 
 impulse pallet will not mark the wheel in the same spot 
 as the locking pallet. 
 
 Try if the escape wheel teeth drop safely on the 
 impulse pallet by letting each tooth in succession drop 
 on, and, after it has dropped, turn the balance gently 
 backwards ; you can then judge if it is safe by the 
 amount the balance has to be turned back before the 
 tooth leaves the pallet. If some teeth do not get a safe 
 hold, the impulse roller must be twisted round on the 
 arbor to give more drop. 
 
 If the escapement is in beat, the balance, when the 
 balance spring is at rest, will have to be turned round 
 an equal distance each way to start the escapement. 
 When the balance spring is in repose, the back of the 
 discharging pallet will be near the gold spring, and if 
 the balance is moved round till the gold spring falls off 
 the back of the pallet and then released, the escapt-ment 
 should start of itself: and in the other direction also, if
 
 the balance is released directly the wheel tooth leaves 
 the face of the impulse pallet, the escapement should go 
 on of itself. 
 
 Examination of the Cylinder Escapement. 
 
 See that cylinder and wheel are perfectly upright. Put 
 the cylinder and cock in their places. Then, with a 
 little power on, and a wedge of cork under the balance 
 to check its motion, try if all the escape wheel teeth 
 have sufficient drop, both inside and out. If, with the 
 cylinder planted the correct depth, there is sufficient 
 drop inside the cylinder and none without, the cylinder 
 is too large ; if the reverse fault is apparent, the cylinder 
 is too small. If some of the teeth only are without 
 necessary freedom, make a hole in thin sheet brass of 
 such a size that one of the teeth that has proper shake 
 will just enter. Use this as a gauge to shorten the 
 full teeth by. For this purpose use either steel and 
 oilstone dust or a sapphire file, polish well with bell 
 metal and red stuff, and finish with a burnisher. Be 
 careful to operate on the noses of the teeth only, and 
 round them both ways so that a mere point is in con- 
 tact with the cylinder. If the inside drop is right, and 
 there is no outside drop with any of the teeth, the 
 cylinder may be changed for one a little smaller or for 
 one of the same inside diameter, but thinner. Or the 
 wheel may be changed for one a little LARGER, but in 
 this case be sure the larger wheel will clear the fourth 
 pinion. And with insufficient drop inside changing the 
 wheel for one a little smaller will often be more expe- 
 ditious than removing the cylinder. 
 
 If the teeth of the escape wheel are too high or too 
 low in passing the opening of the cylinder, the wheel 
 should be placed on a cylinder of soft brass or zinc 
 small enough to go inside the teeth, with a hole through 
 it and with a slightly concave face. A hollow punch 
 is placed over the middle of the w r heel while it is rest- 
 ing on the concave face of the brass or zinc cylinder, 
 and one or two light taps with a hammer will bend the
 
 wheel sufficiently. In fact, care must be taken not to 
 overdo it. It rarely happens that the wheel is free 
 neither of the top nor bottom plug, but should this be 
 the case sufficient clearance may be obtained by deepen- 
 ing the opening with a steel polisher and oilstone dust 
 or with a sapphire file. A cylinder with too high an 
 opening is bad, for the oil is drawn away from the teeth 
 of the escape wheel. 
 
 If a cylinder pivot is bent, it may very readily be 
 straightened if a bouchon of a proper size is placed over 
 it to get a leverage. 
 
 When the balance spring is at rest, the balance should 
 have to be moved an equal amount each way before 
 a tooth escapes. By gently pressing against the fourth 
 wheel with a peg this may be tried. There is a dot on the 
 balance and three dots on the plate to assist in esti- 
 mating the amount of lift. When the balance spring is 
 at rest, the dot on the balance should be opposite to 
 the centre dot on the plate. The escapement will then 
 be in beat that is, provided the dots are properly 
 placed, which should be tested. Turn the balance from 
 its point of rest till a tooth just drops, and note the 
 position of the dot on the balance with reference to 
 one of the outer dots on the plate. Turn the balance 
 in the opposite direction till a tooth drops again, and 
 if the dot on the balance is then in the same position 
 with reference to the other outer dot the escapement 
 will be in beat. The two outer dots should mark the 
 extent of the lifting, and the dot on the balance would 
 then be coincident with them as the teeth dropped when 
 tried this way: but the dots may be a little too wide 
 or too close, and it will therefore be sufficient if the 
 dot on the balance bears the same relative position to 
 them as just explained ; but if it is found that the lift 
 is unequal from the point of rest, the balance spring 
 collet must be shifted in the direction of the least lift 
 till the lift is equal. A new mark should then be made 
 on the balance opposite to the central dot on the plate.
 
 H3 
 
 When the balance is at rest, the banking pin in the 
 balance should be opposite to the banking stud in the 
 cock, so as to give equal vibration on both sides. 
 This is important for the following reason : The bank- 
 ing pin allows nearly a turn of vibration, and the shell 
 of the cylinder is but little over half a turn, so that 
 as the outside of the shell gets round towards the centre 
 of the escape wheel, when a tooth is at rest outside of 
 the cylinder, the point of a tooth may escape over the 
 exit lip and jamb the cylinder unless the vibration is 
 pretty equally divided. When the banking is properly 
 adjusted, and a tooth is at rest inside the cylinder, 
 bring the balance round till the banking pin is against 
 the stud ; there should then be perceptible shake between 
 the cylinder and the plane of the escape wheel. If there 
 is no shake the wheel may be freed by taking a little 
 off the edge of the passage of the cylinder where it 
 fouls the wheel by means of a sapphire file, or a larger 
 banking pin may be substituted at the judgment of the 
 operator. See that the banking pin and stud are per- 
 fectly dry and clean before leaving them : a sticky 
 banking often stops a watch. Cylinder watches and 
 timepieces, after going for a few months, sometimes 
 increase their vibration so much as to persistently bank. 
 To meet this fault a weaker mainspring may be used, 
 or a larger balance, or a wheel with a smaller angle of 
 impulse. By far the quickest and best way is to very 
 slightly top the wheel by holding a piece of Arkansas 
 stone against the teeth, afterwards polishing with box- 
 wood and red stuff. So little taken off the wheel in this 
 way as to be hardly perceptible will have great effect. 
 
 Revolving Escapements. The tourbillon, invented 
 by A. I,. Breguet, is a revolving carriage in which 
 the escapement is placed, the object of the revolution 
 being to eliminate the errors due to varying positions, 
 and particularly the quarter positions, which present 
 the greatest difficulty to the adjuster. In its original 
 form, it is scarcely ever applied except to watches sub- 
 jected to Observatory or other competitive trials.
 
 144 
 
 B. Bonniksen has invented a more compact arrange- 
 ment which he calls a karrusel, in which the carriage 
 driven by the third pinion rotates once in 52^] minutes, 
 which gives sufficiently quick change of position for 
 all practical purposes. 
 
 Breguet's and Bonniksen s devices are both illus- 
 trated in the IVatcJi and Clockinakcrs Handbook, and I 
 do not think anything need be said here in reference 
 thereto, except a warning that the large pivot carrying 
 the karrusel carriage should not be oiled. The rotation 
 is so very slow that lubrication is unnecessary, and 
 would be mischievous.
 
 INDEX. 
 
 Acceleration - - . - 16 
 
 Adjustable stud - - 95 
 
 Adjusting for positions - . 32 
 
 ,, ,, temperature . 36 
 
 Airy- - . . 38 
 
 Aluminium bronze . IQO, 107, 109 
 
 Annealing ... Ioo 
 
 Applying a balance spring - . 48 
 
 Arnold, John - - . -37 
 
 Attachment with relation to pendant - - 32 
 
 Auxiliary - . . -AT. 
 
 Balance - 4> 86 
 
 " arc ' - 3. 3i 
 
 cock 48) 9? 
 
 ,, compensation . 30 g z 
 
 spring - 4 
 
 applying - 48 
 
 ,, hardening - 1OI 
 
 ,, spring holder - - ' 56 
 
 ,, spring making - . 97 
 
 ,, spring tester . uy 
 
 staff 
 
 135 
 
 Balances, size of - - - . 5, 1 1 
 
 Banking - 138, X43 
 
 pins ... I3 8 
 
 Barber, \V. N. -18, 25, 35, 71, 134 
 
 Barrow's Regulator - - 10 
 
 Beat I3 6 
 
 Benzine 
 Berthoud 
 
 105 
 37 
 
 Bickley 55 
 
 Bird cage - - 7 
 
 Bluing ... I05 
 
 Boiling-out pan . . g,S 
 
 Bonniksen - . . . 67, 144 
 
 K
 
 146 
 
 Bosley I0 
 
 Breguet X 5. H3 
 
 ,, drawing of curve 15 
 
 ,, spring - J 5 
 
 Bridgman, R. 92, 136 
 
 Buckney, D. n 5 
 
 Callipers - 9 
 
 Carriage clock - - J 3 2 
 
 Centre of gravity 3 2 , 33 
 
 gyration 
 
 Centrifugal tendency J 3 
 Charles II. - 
 
 Chree, Dr. Charles 122 
 
 Chronometer escapement 3 2 , I4 1 
 
 Clamp for lever pallets - 1 3& 
 
 Cock 48, 97 
 
 Cole, J. F. - J 4, 107 
 
 Collet 49. 52 
 
 ,, adjuster 55 
 
 lifter 55 
 
 table for 97 
 
 Compensation adjustment - 3 6 , 8l 
 
 ,, balance - 3$ 
 
 curb 3 6 
 
 Counting vibrations 4 8 
 
 Crescent 
 
 Gumming, A. 37 
 Curb pins 
 
 Curzon, C. - r 3 6 
 
 Cyanide 97 
 
 Cylinder escapement - 3, J 4 T 
 Cylindrical springs 
 
 Daldorph's stud and collet table 97 
 
 gauge 
 
 118 
 
 Dead point - 
 Deck watches 
 Dennison, B. 
 
 Dent, E. J. 46 
 
 Depthing - 73 
 
 Derham 7 
 
 Detached escapement - 3 2 , 139 
 
 Detent ... 139
 
 Dial - 71 
 
 Draw 134 
 
 Drilling - 74 
 
 Drop 31 
 
 Duo in uno - 6 
 
 Duplex escapement 12 
 
 Earnshaw, T. 37 
 
 Elasticity 5, 6 
 
 Emery 37 
 
 End-stone 68 
 
 Escapement, examination of 132 
 
 ,, chronometer 139 
 
 ,, cylinder - 141 
 
 ,, lever- 132 
 
 Escape pinion 48 
 
 wheel - 133, 140, 141 
 
 Escaping arc 30, 31 
 
 Escape wheel too heavy 68 
 
 Excelsior callipers 92 
 
 Expansion of metals 37 
 
 Flat spring - 6, 48 
 
 Fourth wheel - 14 
 
 ,, pinion 48 
 
 Free spring 59 
 
 Friction - - 68 
 
 Frictional escapement 30, 141 
 
 Gardner, R.- 6, 14, 89 
 
 Gas governor - 84 
 
 Kullberg's - 85 
 
 ,, ,, Mercurial 85 
 
 Glasgow ... loo 
 
 Going barrel - - 60 
 
 Golay . 115 
 
 Gravity - - 32, 33 
 
 Greenwich - 3 6 > 47> J 3o 
 
 Grossmann Jules 18, 33 
 
 Guard pin - 138 
 
 Guildhall Museum 43 
 
 Guillaume, C. E. - n6 
 
 Guye's adjustable stud - 95
 
 148 
 
 Hammersley - 6, 17 
 
 Hands - no, 130 
 
 Hardening - 101 
 
 Hardy, W. - 45 
 
 Harrison, J. - 36 
 
 Hartnup - 4 2 
 
 Heavy escape wheel - 68 
 
 Helical springs - 13 
 
 Huygens - - 8 
 
 Hog's bristle - - 9 
 
 Hooke, R. - -7, 17, 108, 117 
 
 Horizontal escapement - - 30, 141 
 
 Houriet, F. 7, no 
 
 Houston - - - - 113 
 
 Hydrochloric - 102 
 
 Ice box .... 85 
 
 Impulse pin ... 137 
 
 Index - 10, 137 
 
 Inertia - 5 
 
 Isochronism - 17, 61 
 
 Isochronous - - 18, 64 
 
 Jewel hole .... 50 
 
 Joint pusher 53 
 
 Journal Suisse d'Horlogerie - 18 
 
 Kew Observatory - - - 121 
 
 Karrusel - - - - M4 
 
 Knudsen . - 109 
 
 Kullberg, V. - - - 15. y 4 
 
 Kullberg's balance - - 46 
 
 Leroux, John ... 43 
 
 Locking pallet - - - 14 
 
 Lever escapement, examination of !3 2 
 
 Line of centres - - - ?8 
 
 Long arcs, extra defective - 68 
 
 Loseby, E. T. 46 
 
 Lossier, L. - - 18, 21, 22, 36 
 
 Macartney's collet adjuster 
 
 Magnetism - 
 
 Mainspring, stronger to quicken short arcs
 
 149 
 
 Marine chronometer - . 14, 4^ j^j 
 
 Mass- r 
 
 Massey, E. - - 46 
 
 Mean time - 
 
 77 
 
 107, 108 
 
 Method of forming terminal curves - - 25 
 
 Middle temperature error - - 40 
 
 Modulus of elasticity ... 6 
 
 Molyneux, R. - . . _ _ 44 
 
 Measurement 
 
 72 
 
 Moment of elasticity . 5 
 
 Moray .... 7 
 
 Movement holder - - 60 
 
 III, 112 
 
 Nickel .... 
 
 Nickel-steel - ... 
 
 Nicole, Nielsen & Co. - - - 115 
 
 Non-magnetizable watch . 109 
 
 North, R. B. . . . IO 6 
 
 Norwegian refrigerator - . - 81 
 
 Oil- .... 42 
 
 Oven . . 82 
 
 ,, Hearson's - 82 
 
 ,, Schoof's - - - 83 
 
 Overcoil - 6, 11, 19 
 
 Paillard - . . - lio 
 
 Palladium springs - - 33, 42, no, 115 
 
 Pallet . . . 30,31 
 
 ,, staff - 30 
 
 Passing spring (gold spring) - . 139 
 
 Point of attachment - - 78 
 
 Phillip's spring - - - 18, 20 
 
 Pinning in, under or over full turns - 12, 35 
 
 Pivot 135 
 
 Plate . .50 
 
 Platinum - 112 
 
 Pliers 55 
 
 Plose . 55, 87 
 
 Plose's collet lifter 55 
 
 Pocket chronometer 14, 32 
 
 Point of attachment to collet 33 
 
 Poising '90
 
 150 
 
 Poising tool - - 9 1 
 
 Polishing 104 
 
 Position errors 45 
 
 Poole, J. 45 
 
 Position errors 63 
 
 Quarter screws - 63 
 
 Rack 60 
 
 Rambal n 5 
 
 Red-stuff - S3. 104 
 
 Refrigerator Si, 85 
 
 Regulator - 49 
 
 Roberts 9 
 
 Remedies - 6? 
 
 Run - ! 37 
 
 Rust I0 9 
 
 Safety pin 
 Sapphire 
 Sch loss 
 Schoof - 
 Screw 
 
 ,, driver 
 ,, left-handed 
 Seconds hand 
 Sevres alloy 
 
 Short arcs, to quicken - 
 Silver 
 
 Single beat escapement - 
 Spherical - 
 
 Springs, making 
 Staif- 
 Steel 
 
 Steel nickel 
 Stud 
 
 ,, adjustable 
 
 ,, remover 
 
 Table roller 
 Tempering - 
 Terminal curves 
 Thermometer 
 Theoretical curves
 
 Thury n6 
 
 Timing 64 
 
 ,, carriage clocks - 132 
 
 ,, in positions . 64 
 
 ,, in reverse 64 
 
 ,, screws - 39 
 
 Tompion, T. 9 
 
 Tourbillon 143 
 
 Train - - 48 
 
 Trial number, deck watches and marine chronometers - 131 
 
 Timing box- - 60 
 
 Turns . 71 
 
 Tweezers - 97 
 
 ,, for balance springs - 57 
 
 ,, ,, studs 97 
 
 Ulrich - no 
 
 Unlocking resistance, - 30, 134 
 
 Vibrations - - 48 
 
 Vibrator - - - - 51 
 
 Villarceau - - 38 
 
 Walker, George - - 18, 25, 35 
 Walker and Barber's method ot forming and testing ter- 
 minal curves 25 
 Walker and Barber's adjustable stud 95 
 Walsh - - - 40 
 Waltham Watch Co. 115 
 Watch tests - - - - 121 
 ,, balances - 4 
 ,, size of ii 
 Webb, A. - 40, 103 
 Wilkins 7 
 Woerd, C. V. 46 
 Wright, T. D. 38, 115 
 
 Young - - 6
 
 152 
 
 [While these sheets are passing through the press, I 
 learn, with deep grief, the Mr. George Walker, whose name 
 frequently occurs therein, is no more. He was a thorough 
 horologist, devoting attention particularly to the laws which 
 govern the motion of the balance and balance spring, and 
 an expert springer. But two months ago he made the 
 sketches for Walker and Barber's tester, described on pp. 
 26 29, and had he lived would have contributed an article 
 dealing with the turning of overcoils and other points on 
 which he was admittedly an authority.] 
 
 October, 1898. F. J. B.
 
 ""'in inn MIII mil mil mi Illl 
 
 A 000093132 9