Most collectors seem to have a favorite company, and my favorite pocket watches were made by Illinois. For irrational reasons, I like the 12 and 16 size Illinois watches more than most, but perhaps rationality is beside the point when we’re talking about trying to adjust archaic timekeeping machines to high levels of accuracy.
Here’s a 94-year-old Illinois 12 size watch—a 17 jewel, 405 grade from 1922. I got it in a friendly local coin shop that has been family run for quite some time. They focus on numismatics but occasionally get watches that appeal to curmudgeonly adjusters. The dial was an old redial in terrible shape, so I thought I’d tune up the mechanism while the dial was being professionally restored.
Before trying to adjust a watch, you first need to decide if the watch can be (and should be) adjusted to close accuracy limits. For this watch, the omens are good:
- It is clean, intact, and in good shape. It doesn’t look like it has been mishandled or fiddled with by incompetent hands. (An exception—the balance bridge was probably swapped, given the different decorative pattern.)
- It has features that make fine adjusting easier: 17 jewels, a swan-neck regulator, and mean-time screws.
- The beat traces on the timing machine are clean and stable.
- Illinois watches from this period are excellent, even the humbler grades like this 405.
As a prelude to adjusting, the watch was cleaned, oiled, and demagnetized. It got a new white alloy mainspring and a serious jewel scrubbing. After letting it run for a day to let the lubricants and mainspring bed in, I fully wound it and set it on the timing machine.
Normally I would take baseline rates at this point, but this watch showed some ghastly knocking. The amplitude was ridiculously high: over 340 degrees in pendant up (PU) position, and high enough to knock in both the dial up and down positions.
If you haven’t come across knocking, we’ll dig into it in a later post. Basically, if the balance wheel starts to rotate almost a full circle, the roller jewel will bang against the other side of the pallet fork, causing a knocking sound, terrible timing, and the occasional cracked roller jewel. Bad.
So first we need to tackle the excessive amplitude. In this case, a previous owner had adjusted the banking pins to reduce the lock (see Fried’s The Watch Escapement (pages 83-85), one of our blog’s “textbooks,” if this is unfamiliar). Resetting the drop to banking eliminated the knocking and brought the amplitude back into a desirable range (around 290 horizontal, 270 vertical).
Aligning the Dial Up and Dial Down Positions
Now we’re ready to align the 6 positions. As Kleinlein teaches, the first step is to align the dial up (DU) and dial down (DD) positions.
Our Illinois was +5 DU and +14 DD. That’s more of a disparity than I’m happy with. Many, many things can make the DU and DD rates differ. The most likely ones—unequal oil or jewel cleanliness—seemed unlikely in light of how I cleaned this watch.
A good place to start is to clean the balance pivots. As a first, easy step, we can clean the pivots with pithwood. That sometimes works, but it didn’t here.
Next, I broke out a cleaning suggestion from Fried, who in The Watch Repairer’s Manual describes how to clean pivots with wood and jeweler’s rouge. This cleans and shines up the pivots without removing material or changing their shape.
While the balance wheel was off, I adjusted the beat error to a .4 ms, a great outcome that was part skill, part luck, and perhaps part good karma from not marking-up shipping on eBay.
That worked: afterward, the rates were -4 DU, -2 DD, with great amplitude (over 290) for each. Time for the next step.
Here are the baseline rates for evaluating our progress. Not bad, but not great—typical, I think, for old pocket watches.
We’ll use “straight limits” of 7 seconds—each position can be no more than +/- 7 seconds from zero—as our adjusting goal.
The mean absolute deviation from zero is 11.8 seconds, with a standard deviation of 13.1 seconds; the handy calculator is at the end of this post.
Round 1: Finding Poise Errors
Recall that the heaviest spot on the wheel is the point below the balance staff when the watch is in its fastest vertical position. Checking the rates at a low amplitude exaggerates the effect of a heavy spot, so it is easier to pinpoint where it is.
Here are the vertical rates when the watch runs at around 150 degrees of amplitude:
The watch is fastest when it is pendant up (PU), which is this position:
This means that the heavy spot on the balance wheel is below the staff when the watch is PU.
We can zoom in to see exactly where the heavy spot is. The red circle, the heaviest spot, is conveniently over a screw. The green circle, the lightest spot, is over the screw’s twin on the other side of the wheel.
What adjustment should we make here? When a big change is called for, it’s better to add than remove weight. Timing washers are reversible, unlike carving off a big chunk from a screw.
I found the smallest timing washer that would fit. It didn’t actually fit—the hole was too small—so I broached it to enlarge the hole and put it beneath the screw at the lightest spot (the green dot).
I brought DU and DD near zero and timed it in all 6 positions, and the results?
Clearly, we’re on to something here. There’s a reason why people admire these Illinois pocket watches.
The new average difference from zero is 4 seconds, and the standard deviation was 3.5 seconds. Much better!
Rinse and Repeat
These are pretty good rates, but this blog takes an obsessive and quixotic approach to adjusting, so we’ll try to squeeze some seconds out of those positions, especially PR.
And so we do it all again. Dynamic poising is iterative: we let down the mainspring until the amplitude is around 150 degrees, measure the rates in the 8 vertical positions, and then decide whether to add or remove weight based on the watch’s overall rate.
For all of the remaining rounds, I first slowed the watch slightly, to around -2 seconds when dial up, and then adjusted by lightly filing or undercutting the heavy spot with the balance screw file.
A few light rubs with the file is all we’re looking to do for the final couple rounds. It is easier to remove a tiny amount of weight than add a tiny amount. As a result, setting the watch slightly slow and removing a tiny amount brings the vertical rates closer to the horizontal rates.
The End is Nigh
Eventually, we need to call it quits. I probably went through 8 rounds of adjusting, although the last few involved very minor filing of the same screw.
The final rates are excellent. The mean absolute difference from zero was 2.2 seconds, with a tiny standard deviation (.8 seconds). This, people, is why I love Illinois pocket watches.
And for the doubters—because this is the Internet, and anyone can claim anything on a blog—the Timegrapher pictures are at the end. (For people interested in the details, the watch ran for at least a minute in each position, and the Timegrapher computed the rate based on a rolling 20-second window. At least 60 seconds of beat traces are displayed.)
And here’s the finished watch, with a shined-up case and a dial nicely restored by International Dial, ticking away and keeping great time.
This 94-year-old watch proved to have good potential for adjustment. After a rough start—heavy knocking caused by excessive amplitude, and some DU/DD differences—this watch was easily adjusted.
Our Illinois is a good example of a common theme in watch adjusting: if you can get the amplitude right, and if you can get the DU and DD positions to agree, then things will usually end well.