We often repeat Jendritzki’s definition of adjustment
to alter the period of the balance and spring so that the daily rate of the watch comes as near to zero as possible.
But how “near” is “near zero”? Certainly, one wouldn’t quibble if your pocket watch ran with a rate of zero in all 5 positions, but that’s not realistic. When are we supposed to stop tinkering with the watch and declare it good?
This post describes some ways of thinking about accuracy standards. You could use many kinds of standards to decide if your watch keeps what Kleinlein calls “close time.”
One approach is to use informal guidelines developed from an adjuster’s experience. For example, in Watch Adjustment (p. 88) Jendritzki suggests these guidelines for “precision watches”:
Mean daily rate in the five positions: -3 to +12 s.
Greatest difference between mean daily rate and one of the rates in the five positions: +/-22 s.
This sort of standard specifies an average rate—within 3 seconds slow to 12 seconds fast per day—as well as a maximum deviation. In this case, no one position can vary from the average by more than 22 seconds. Such conditions keep adjusters honest. After all, the average of -45, 0, and +45 is zero, but we wouldn’t be happy with a 3-position watch with those results.
A simple approach is to define absolute variances from zero. For example, the goal might be to get each of the 5 positions running within +/-10 seconds per day. Kleinlein calls these “straight limits.”
If you picked +/- 7 seconds, for example, no one position could be more than 7 seconds slow or fast. As a result, the average rate across the positions could not exceed 7 and would usually be much less. (The average rate would be +/-7 only in the weird case that all 5 positions were +7 or -7 and you refused to tweak the regulator). Also, the largest difference between any 2 positions could not exceed 14 (the limit doubled).
For vintage watches, in my experience, you could consider straight limits of 10 as good, 7 as great, and 5 as excellent.
Another approach is to research factory standards, either for your watch or from high-grade watches from the same era. The idea is to call it a win if you can meet or exceed the standards represented by the best production watches of the time. This is an interesting historical standard for vintage watches.
6 seconds maximum variation between dial up and dial down
10 seconds maximum variation between dial up, dial down, 12 up
13 seconds maximum variation between dial up, dial down, 12 up, 3 up, 9 up
20 seconds maximum variation between dial up, dial down, 12 up, 3 up, 9 up, 6 up
A fun approach is to pick a particular watch as a standard. For example, you might try to beat the timing results for a watch that you had professionally adjusted or that you saw adjusted on the Internet.
As a case in point, have a look at this interesting vintage 1950s Rolex 6074 restoration that was carried out by professional watchmaking instructors. (Their final image shows the timing results.) Can you adjust one of your 1950s wristwatches to run at least as well?
Finally, you can use standards used by formal certification centers, such as COSC, METAS, or TIMELAB. The Swiss COSC standards, for example, have their infamous -4/+6 average rate standard.
Keep in mind that these are collections of standards that include tests for isochronism and temperature. Some of the newer tests include resistance to magnetism and require testing the cased watch.
So if you pick something like COSC, stick to it. Don’t declare it “COSC chronometer accurate” on a watch forum or eBay unless you also carried out and met the isochronism and temperature tests.