[Backdrop] A. Lange & Söhne Sax-O-Mat Perpetual Calendar, Adjusted To Positions, Temperature, And Isochronism
When it comes to performance in watches, the single most important element to the owner is generally “accuracy,” and by that, is usually meant the ability of a watch to keep time as measured against a certain reliable external standard. Internet time signals are widely available and provide an easy, fast time reference for setting a watch. If a watch matches the time signal reasonably closely, the owner concludes the watch is “accurate” and is satisfied.
However, to a watchmaker, accuracy is a by-product of precision, and relatively trivial. A watch is considered by professionals to be precise not if it shows little drift from a particular time reference, but rather if it has a stable rate – that is to say, if the frequency of its oscillator, whether a balance wheel or quartz resonator, changes very little over time.
That a watch is accurate or precise may seem to be the same thing but they’re not. A watch that is relatively imprecise may have a rather unstable rate. Its daily rate variation on one day may be -10 seconds. The next day, it may be +8, then +10, then -5, then -7, then +4. At the end of five days the watch appears to be exactly synchronized to the time standard and the owner feels it’s accurate, but this is merely a matter of having been lucky and having looked at the watch at the right moment.
Though perceived to be accurate, “luck” is a problem because the rate is inconsistent, drifting under the influence of factors like temperature variation and changes in position. Indeed, the watch may be dependent on the habits of the owner who tends to spend time in positions with less variation in rate, or to be exposed to an ambient temperature most apt to produce minimal rate variation.
If, on the other hand a watch consistently loses 5 seconds per day over the same period, at the end of that period it will have lost 30 seconds. Now (for the sake of argument) this may well be a watch with zero daily variation in rate across all positions, and temperature fluctuations but the owner will still feel his watch is not precise, when it is in fact extremely precise and can easily be regulated to compensate for the variation in rate.
This is why it’s so difficult to objectively evaluate the performance of even a single watch, let alone compare multiple watches. First of all, the watches need to be tested under as nearly identical conditions as possible in order to eliminate random environmental variations that will skew results. Without standardized testing procedures results are obviously unreliable and no conclusions can be drawn.
Simply wearing one watch on the right wrist and one on the left can make a difference in performance – a watch on the right vs. left wrist is exposed to a different range of positions as well as potentially different temperature variations, different types and frequency of shocks, and in the case of self-winding watches, different amounts of torque delivered to the balance depending on the difference in degree of movement from the right vs. the left hand.
In order to eliminate these variables and to evaluate performance objectively, testing agencies from the COSC, the Concours de Chronométrie, to the observatories at Kew, Besançon, Glashütte, Geneva, and Neuchatel all developed standardized testing procedures to eliminate what are called confounding variables that can lead to bad data.
The observatory trials were especially strict, as in general they were testing, not series produced watches, but specially made, finely tuned watches that were not particularly suitable for daily wear. Procedures varied, but the observatory at Geneva tested watches for a total of 44 days; there were nine testing periods (two of which were a six-day period at low temperature, and an eight-day period at high temperature) and 11 different types of errors were assessed, including:
Patek Philippe, calibre 324 SC, adjusted to heat, cold, isochronism and positions
Points were assigned for watches that performed better than certain threshold values. The COSC certification is a less complicated process as the purpose is different – the observatory trials were competitions between thoroughbred one-off movements for bragging rights; COSC exists to certify that mass produced movements meet certain minimum performance criteria.
The COSC standards are sometimes criticized for their laxity relative to the stringency of the observatory trials but in fact they’re reasonably rigorous for a process that must evaluate many thousands of movements per year, and which is not intended to establish a “best” movement but merely to ensure that the criteria of the ISO for chronometers is met. (For one thing, to test watches to observatory competition standards and issue a separate bulletin for each movement would be prohibitively expensive.)
Given the number of variables involved, it should be obvious that merely wearing a watch for a week tells us surprisingly little – indeed, it tells us essentially nothing – about the precision of a watch. Watchmakers, watch manufacturers, the COSC, the Concours de Chronométrie, and observatories undertake the elaborate procedures they do for a reason – to even have a hope of evaluating actual precision (not accuracy, which is the hoped-for by product of precision) it’s necessary to eliminate the host of variables (everything from state of wind to variations across test samples in position to variations across test samples in temperature) that can cloud the results.
If you’re comparing two watches even storing them on a nightstand in different positions can make a visible (and confusing) difference in results. Your best chance of getting good accuracy out of a watch is still what it’s always been – precision in manufacture, craft in lubrication, care in assembly, adjustment, and regulation, and reasonable care in daily use.
Understanding precision and performance is crucial in appreciating one of the most genuinely interesting features of mechanical horology: the incredible complexity of the problem of precision, and the incredible ingenuity exerted by some of the best scientific and mechanical minds in the world, over five centuries of watch and clock design, in solving them.
More immediately, how do you know if you have an accurate watch? Most of us don’t have the time or inclination to fiddle with professional tools like Witschi timing machines, or the leisure to time our watches over a 44-day period to observatory standards; we want to wear them. The simplest answer is to trust the manufacturer’s commitment to accuracy – if they certify their watches through COSC, if they have known high internal standards for chronometry (as do Patek Philippe and A Lange & Söhne) and if they show care in the quality of finish and level of craft visible in their watches and movements, you will probably do well.
PS: When you get a good watch, if it’s not running up to a reasonable standard (say, minimum COSC standards, or the manufacturer’s stated internal standards) after you’ve given the poor thing a chance to run in (remember, God knows how long it’s been sitting in a case at the dealer) just take it in and have it regulated. If you own good watches, remember, your watchmaker should be as good as your watch. Precision in manufacturing is a necessary but not sufficient condition for accuracy; good adjustment and regulation is necessary too. Of course, you’ll never make a Trabant into an F1 car and you’ll never make a cheap watch into a precision chronometer, so drive in the reality lane.
Recommended reading: Fritz von Osterhausen, Wristwatch Chronometers
On Timezone.com: Tweaking the Mark XII, by Walt Odets
For an understanding of basic principles of precision timekeeping: Marine & Pocket Chronometers, Hans von Bertele
Also: Precision Pendulum Clocks, Derek Roberts