At first this project seemed straight forward. All the parts are here, nothing is broken.
The movement cleaned up well and ran. Good end-shake and side-shake through-out, good pallet, everything was fine. It timed well on the timing machine also, reading at less than +/- 30 seconds per day at 18,000 BPH (beats per hour).
My last step in watch testing is to run each watch for several days in various orientations; dial up, dial down, hanging, etc. I didn't expect any issues with this watch. But it turned out to be one of my more difficult and puzzling projects.
Finally in desperation, I counted the teeth on the wheels, and the leaves on the pinions.
For a bit of background here, it may be helpful to refer back to this post on counting the train. Watches operate by regulating, slowly releasing at a certain beat rate, a power source through a set of gears. These gears, or wheels, have ratios such that hands attached to the right gears rotate at a rate we recognize in seconds, minutes and hours per day. It isn't magic, it's mechanics.
These are the count values for this movement.
Great Pinion: 10
Great Wheel: 65
3rd Pinion: 8
3rd Wheel: 60
4th Pinion: 8
4th Wheel: 70
Esc Pinion: 8
Esc Wheel: 15
The 4th wheel's revolutions per hour is therefore:
( 18,000 * 8 ) / ( 70 * 14 + 2 ) = 68.57
The 4th wheel is the seconds hand. This should be 60 revolutions per hour, one per minute!
But is gets worse. If we calculate the rate of the center wheel:
(8 * 8 * 68.57 ) / ( 65 * 60 ) = 1.57
At 18,000 BPH, this leaves an erroneous rate of the center wheel, which is to say the minute hand, of 1.57 revolutions per hour. That's +180.4 minutes per 24 hours on the hour and minute hands! But that's only the beginning. The error on the second hand is not relatively the same as that of the minute and hour hands. In other words even if we did awful things to the balance wheel to shift the rate away from 18,000 BPH to the point where the main hands read OK, the seconds hand would go around a bit more than one time per minute.
This watch's train contains incorrect parts.
In fact, it contains more than one incorrect part. The strange thing though is that this watch "ticks" fine - quite well in fact. Because of the geometry of the way gears work, all possible combinations of tooth and pinion counts will not all run smoothly in practice. They have to mesh well. This is why any 18,000 BPH movement, with a seconds hand, will tend to have the same counts. This movement's combination also runs fine at 18,000 BPH, but it is physically impossible for the hands to read correctly.
For comparison, here are the values of the train counts for a typical 18 size Elgin pocketwatch, non-slow beat.
Great Pinion: 12
Great Wheel: 80
3rd Pinion: 10
3rd Wheel: 75
4th Pinion: 10
4th Wheel: 80
Esc Pinion: 8
Esc Wheel: 15
Fortunately, this watch's train layout is very similar to other American watches or this size and era. The height, the distance between the plates, is a bit atypical in places, but the layout is common. It was a bit of work but I was able to swap to wheels with replacements having different counts. The wheels only had to be altered a small amount for a good pivot size and for the height of the movement.
I replaced three wheels for an atypical count, but one that worked out to give the correct time readings at 18,000 beats per hour.
These additional images show the lever-setting mechanism.