In case of over compensation a spring of the same make, one coil smaller, will permit of shifting one or two pairs of screws toward the free ends of rims.

In a series of tests it was demonstrated that by duplicating or changing springs of the same make and size, on balances that had previously been compensated, there was very slight difference in the temperature variation of the watch. Also by changing pinning points or breaking out one-fourth to one-half of the coil around collet and adding weight to the balances to correct the mean time the difference in the variation was almost negligible.

On the other hand it was found that by replacing the springs with others of larger or smaller size, variations of from three to ten seconds were noted in all instances.

In selecting and fitting a spring that will be one coil larger or smaller, it should be noted that the inner coil of the original spring and that of the new spring are approximately the same distance from the collet. For if there was considerable s.p.a.ce between the collet and inner coil of the original spring, and the new spring was colleted quite close, there might be the addition of an extra coil in the inside only. This was found to produce only a very slight correction, compared to that obtained by the addition of a complete outer coil.

These tests indicate that the proportion of strength of the spring in the temperatures varies with any appreciable change in length while slight changes make practically no difference.

20. _Example Demonstrating that Temperature Variation is not Always Due to the Balance and Spring._

Fig. 7

+--------------------------------------------------+

No. .................... Make...................

+--------+-----+-----+-----+-----+-----+-----+-----+

HEAT

-10

-10

+ 4

+ 4

+ 1

+ 1

+--------+-----+-----+-----+-----+-----+-----+-----+

NORMAL

- 6

+ 4

+ 5

+ 1

+ 4

+ 3

+--------+-----+-----+-----+-----+-----+-----+-----+

COLD

+12

+18

+ 1

- 4

+ 7

+ 3

+--------+-----+-----+-----+-----+-----+-----+-----+ 28 8 2

The following example is submitted to show that temperature variation is not always due to the balance and spring, and that the general condition of the watch may be responsible. The second column of Fig.

7, indicates an error of twenty-eight seconds slow in heat with all screws a.s.sembled in the holes nearest the free ends of the rims.

Examination proved that the motion of the balance in cold was reduced to about one-fourth of a turn. In heat the arc of motion was at least one full turn. This difference in motion was sufficient to prove that there was some binding in the train.

A very close fitting of the escape pivots was found and this undoubtedly caused binding of the pivots in heat due to slight expansion. Expansion of the stone would also tend to close the hole, and while the degree of temperature would hardly have any bearing on this point it is sufficient to show in what direction the tendency would be. The fourth wheel end shake was very close and probably caused binding of the wheel in cold, due to greater contraction of the bridge than of the fourth pinion. Furthermore the mainspring was only 0.02 of a millimeter narrower than the s.p.a.ce in the barrel box. This no doubt also caused binding through greater contraction of the barrel than occurred in the mainspring.

The above defects were remedied and the rate was found to be eight seconds plus in heat as per third and fourth columns Fig. 7.

This made it necessary to shift several of the screws away from the cut, in almost the same position in which they were before the alteration which caused the close a.s.sembling of the screws was made.

The final rate was two seconds slow in heat as shown in fifth and sixth columns.

The variation of thirty-six seconds between the second and fourth columns was entirely erroneous, and was due to condition of the watch irrespective of the balance and hairspring. Should the variation with the screws a.s.sembled have been by chance within the limits of allowance the watch would undoubtedly have been a very unreliable timepiece. The errors in the watch would no doubt have been corrected during the position adjustment later, but the large error in temperature which would have been introduced by wrongly moving the screws, would have prevented reliable timing until possibly at some future period a test in temperature would have been made and the screws replaced in the proper positions.

CHAPTER V

THE MIDDLE TEMPERATURE ERROR

21. _Why This Error Exists and What it Consists Of._

In adjusting watches to temperature it is not always possible nor expected to obtain a perfect rate between the two extremes, manufacturers generally allowing from two to ten seconds variation according to the grade.

Even when the rate obtained is perfect it will only be so at the two extremes and there will always be a few seconds variation in the middle or normal temperature.

This variation will always be a gain of from two to four seconds in the higher grades of steel bra.s.s balances and usually more in cheaper balances.

As there is no possible correction for this irregularity in ordinary balances it has long been known as the middle temperature error and for many years was one of the most perplexing problems that the manufacturer of specially fine timepieces had to deal with.

Various devices were originated from time to time for the purpose of counteracting the error but they were always too infinitely complicated to be of commercial or scientific value, and none of them were ever adopted as a solution of the problem.

In chapter I, No. 3, will be found a description of the distortions of compensation balances in the extremes of temperature and the cause of the middle error is due entirely to the fact that these distortions are not exactly equal in both directions. The free ends of the rims are drawn outward from the concentric form to a slightly greater proportional degree as the temperature decreases from normal and they are not forced inward at an even proportional degree with increase of temperature.

22. _How Nickel Steel Balances Overcome the Middle Temperature Error._

Through extensive experiment in the foreign laboratories balances containing nickel steel have been found to almost eliminate the middle error, which is reduced to one second or less, making it possible to obtain perfect adjustment in various temperatures.

All highest prize watches pa.s.sing through the Geneva Observatory are equipped with these balances and they have been adopted for commercial use to a large extent by the manufacturers of the finer grades of watches.

From the same source success has recently been attained in applying this metal to hairsprings and using them in connection with uncut balances, but owing to the necessary high cost of production, their general use may be delayed for some years to come. Their general use however would revolutionize the present-day methods of adjusting to temperature as there would be practically no expansion or contraction to deal with.

Nickel steel balances will always be found to have the cuts about one eighth of the circle distant from the arms instead of close to the arms. This is made necessary by the fact that the coefficient of nickel steel is about ten times less than that of ordinary steel, and if the cuts were made close to the arms the bra.s.s in expansion would force the free end of the rims to curve inward to such an extent that it would cause an abnormally fast rate in heat.

By making the cuts more central the length of the segments are reduced, thereby causing less curvature of the extreme ends and more nearly equalizing the extent of curvature both ways from the concentric form. This equalization is what causes the reduction in the middle error and its absence in ordinary balances is what causes the larger error.

Non-magnetic or palladium balances are also credited with a smaller middle temperature error than the ordinary steel bra.s.s balance, but owing to the unstable nature of the metal they have not proved to be as reliable in other respects and are not used to any large extent.

The middle temperature error is of course a small factor in the larger sense of obtaining time from commercial watches but its influence is apparent in timing and it will therefore be considered further in the section devoted to Final Regulation, Chapter XV, No. 77.

PART II

THE ADJUSTMENT TO ISOCHRONISM AND POSITIONS

CHAPTER VI

GENERAL CONSIDERATION

23. _Optional Allowances for Variation._

The phrase "Adjusted to Isochronism and Positions" does not always indicate the same high quality or the expense a.s.sumed in obtaining close rating in different kinds of watches.

One particular model may be stamped "Adjusted to Five Positions" and this may indicate that the manufacturer of this model has tested all watches of this grade for twenty-four hours in each of five positions and that the extreme extent of variation from one position to any other, among any of these watches, did not exceed six seconds. Another model may be stamped in exactly the same way and it may indicate that all watches of that particular grade have been tested in exactly the same way and that the extreme extent of variation from one position to any other, did not exceed twenty-five seconds.

The statement regarding the number of positions to which the watch has been adjusted is just as legitimate in the latter instance as it is in the former, for the watches are really tested in five positions and required to perform within specified allowances.

The important difference is in the established limits of requirement, one demanding an extreme of only six seconds variation and the other allowing twenty-five seconds. Both watches may have the same number of jewels and there is no way to discern the actual variation except through a test in positions.

Technically it would be just as legitimate to stamp and advertise watches as above and have an allowance of fifty or more seconds, providing that they were actually tested and not allowed to pa.s.s with a variation greater than this limit.

Close limits of allowance require adjusters of greater skill and material of a finer degree of accuracy, however, than do greater allowances, but the dealer and consumer are generally not informed in regard to this particular point. Some watchmakers also do not understand this feature clearly and the limits of variation to which watches have been adjusted are seldom considered.

Should the difference in allowances and identical advertising be interpreted as an injustice to the manufacturer who maintains close limits for his various grades of watches, it must be remembered that they speak for themselves after pa.s.sing over the counter and into the hands of satisfied customers. His reputation after a period of years will be more firmly established than will that of his less particular compet.i.tor in the high grade field. A similar situation prevails in the repair shop, and the fact that many of the leading dealers and railroad watch inspectors require at least a three position adjustment in the repairing of high grade watches, is convincing evidence that position rating demonstrates its importance in actual service when applied to repair work, as surely as it does when applied to new watches.