I have lately become fascinated with Hamilton Railroad pocket watches, and have obtained several. Curious about their accuracy and the “railroad” timekeeping standard, I purchased a cheap iPad app that utilizes the earbuds microphone to record beats and to indicate gain/loss in seconds per day. Probably not what a watchmaker would use, but good enough for my hobby purposes. Here are the results of tests on four watches. There are two 992’s, and two 992B’s, ranging from 1909 to 1954.

Loss/gain Seconds per Day (spd)
992B. 4992B. 992. 992
#1. #2. #3. #4
Dial up 7.4 -29.7 -7.2 13.9
Dial down 12.2 -29.5 -6.7 36.6
Stem up 36.2 -37 -5.6 -18.8
Stem down-31.7. -46.5 14 -31.8
Stem 9 -23.4 -49.7 -11.2 -13.9
Stem 3 20.5 -32.6 -63.2 -20

1. My first impression was, “WOW!”, how surprising the variation by position is for a given watch. #1 looks really good Dial up until you look at the other positions. #1 has a range of 67spd between stem up and stem down. #4 has about the same range between dial down and stem down.

2. How would a watchmaker analyze this data? Can anything specific be concluded by, say, for #3, the extreme loss (63.2 spd) at position Stem 3 compared to the more moderate losses at other positions? Perhaps the pin on one side if the pallet is broken? I am guessing not. Especially considering how truly enmeshed the gear train is, it seems to me one symptom cannot definitely identify an ailment. Nor can looking at them in combination tell you much about what’s going on inside .... except that they are not up to specified standards.

3. Watch #2 appears to have the worst performance, but might in fact be in the best condition since the variation between positions is by far the least.

4. If the only practical adjustment is by adjusting the regulator, how do the large variances get addressed with a single adjustment? I’m thinking that the don’t, and the first thing to do would be cleaning and oiling, which might eliminate or reduce the variation range by cleaning out gunk that might contribute to a variation in a particular position, and bring variations for each position closer in line with each other. THEN a particular variation might be addressed with the regulator (which MIGHT by coincidence also improve performance in another position. Finally, if that doesn’t work, replace parts.

5. Finally, this little exercise has left me in awe of those great craftsmen/artists/mechanics who could ever manufacture, assemble and adjust even a new timepiece to conform to such strict standards over a variety of conditions (several seconds per day for railroad standards, and only 1 second per day for the 4992B!). Utterly amazing.

Steve
There are whole books on the subject of adjusting the timing of watches. The most straight forward and understandable explanation I have found is on the web site "Adjustingvintagewatches.com". The writer goes through the process of adjusting the positional timing of a number of vintage pocket and wrist watches. It will give you a good idea of what's involved when you are ready to try it.