After reading the earlier posts in this series, you should understand the basic reasons why watches can have different rates across positions. But like everything else in life, you really get dynamic poising when you do it.

This post brings it all home with an example of adjusting a watch to 5 positions. You should “adjust along at home” with a watch of your own that is clean, oiled, and ready for tweaking. We’ll use the method described by Jendritzki in *Watch Adjustment*.

## Picking a Standard & Tracking Progress

For our Hamilton 747 Fulton, we’re going to use “straight limits” of 10 seconds: I’m aiming for accuracy of at least +/- 10 seconds in all 5 positions. The 747 is a great mechanism, but with its common regulator (vs a micrometer regulator), tighter limits are harder to achieve.

Dynamic poising is an iterative, rinse-and-repeat process, so we need a way to track our progress and know when to stop. When using an absolute-deviation standard, a good way to chart your progress is to compute the **mean** and **standard deviation** of the 5 rates. The mean is the average, or central tendency, of the 5 rates. The standard deviation is their dispersion. We want a mean close to zero as well as a small dispersion, which shows that the scores are clustering closer together.

With each round of dynamic poising, the mean should get closer to 0, and the standard deviation should get smaller. A handy calculator for computing means and SDs is embedded at the bottom of this page.

## 1. Center the Regulator & Wind Fully

We first center the regulator: it should be in the exact middle of the fast/slow index region. Then we’ll wind the watch fully and measure the rates in the 5 positions. This gives us our baseline readings. Here’s what Fulton looks like.

Our mean of the absolute values for our 5 positions is 138.8, and our SD = 16.3. (Keep in mind that we’re averaging absolute rates [deviations from 0, regardless of fast or slow] and that PL and PR are defined from the dial side.)

## 2. Find the Heavy and Light Spots

Now we need to find the heavy spot. To do this, we take advantage of the fact that poise errors are exaggerated at low amplitudes. If we measure the watch’s vertical rates at a low amplitude, poise errors are more apparent.

So, we’ll let down the mainspring and wind it up a few times. We want the amplitude in the vertical positions to be around 140-160 degrees, but no more than that.

Then we measure the rates in all 8 vertical positions. For our Fulton, they look like this:

Now we need to identify the heavy spot. We’ll simply look for fastest position—the one with the largest positive rate, or the least-slow position if they are all slow. **The heavy spot is below the balance staff when the watch is in this position.** For our Hamilton 747, the fastest position is PR, in red. We find the heavy spot by visualizing the bottom on the balance wheel. For this watch, the heavy spot is in the red circle when the wheel is at rest.

## 3. Adjust

We now know where our heavy spot is. What next? Do we remove weight from the heavy spot, or do we add weight to the spot exactly opposite it? Let’s think this through.

Remember that we want to align our horizontal rates and vertical rates. In this example, what would happen if we added a small timing washer to the lightest spot, in green?

- the wheel’s poise would improve, so the 4 vertical positions would start to align
- the watch would get even slower in the DU/DD positions

Likewise, what would happen if we removed weight from the heavy spot, in red?

- the wheel’s poise would improve, so the 4 vertical positions would start to align
- the watch would get faster in the DU/DD positions

So, in both cases, improving poise will help the vertical positions. But our Fulton is already running very slow DU/DD, so adding weight would cause the horizontal and vertical rates to diverge.

Removing weight at the heavy spot, in contrast, would improve poise while also speeding up the watch in the DU/DD positions. Removing weight will thus have the side effect of making the horizontal and vertical positions more similar.

**This point is crucial: we always want to improve poise in a way that aligns the horizontal and vertical positions:**

- if the watch is running slow DU/DD, remove weight
- if the watch is running fast DU/DD, add weight

## 4. Rinse & Repeat

I let down the mainspring, removed a screw that was conveniently exactly at the heavy spot, and did a bit of undercutting.

Then Round 2 begins: we fully wind the watch and measure the rates in the 5 positions. As before:

- we evaluate if the watch is running fast or slow DU/DD at full wind
- we let down the mainspring to an amplitude of around 140 to 160 degrees
- we measure the rates in the 8 vertical positions to find the heavy and light spots
- we add or remove weight as necessary
- we calculate the mean and SD of the absolute deviations from zero to chart our progress

I did a couple rounds of adjusting, and I got to this point:

Not too shabby. Our mean of the absolute values for our 5 positions is 5.2, and our SD = 4.7.

## 5. The Final Rounds

Up to this point, we haven’t moved the regulator from the center. When you get to the final few rounds, however, you’ll likely need to nudge it slightly in one direction.

Why? At the final rounds, you’ll need to add or remove tiny, almost imperceptible amounts of weight. It is easy to remove a tiny amount of weight: a light swipe with an escapement file will do it. But it is hard to add a tiny amount of weight: even a tiny timing washer might change your rate by 15 seconds, thus making your horizontal positions differ from your vertical positions.

For the final round, then, you’ll often nudge the regulator to make the full-wind DU/DD rate **a little slow**. Why a little slow and not a little fast? Let’s think it through:

- if we made it a bit
**fast**and removed a tiny amount of weight, poise would improve, but the watch would get even faster, so the horizontal and vertical rates would diverge - if we made it a bit
**slow**and removed a tiny bit of weight, poise would improve, and the watch would get a little faster, returning to a near-zero rate. The horizontal and vertical rates would thus get more similar.

I nudged the watch to be a couple seconds slow DU and did one more round. A tiny bit of weight was sanded off a screw. Here’s the final timing chart:

At this point, I’m going to call it. I could probably squeeze out a couple more seconds from the PU position, but sometimes discretion is the better part of positional-timing valor. For the watch’s grade, this is a good outcome:

- the mean of the absolute values for our 5 positions is now 3 seconds, and our SD is 3.2, our lowest values yet.
- all the values are under our target limit of +/-10 seconds
- the biggest difference between any position is only 8 seconds
- the most important wristwatch positions—DU, DD, and PD—are essentially the same

Here’s how it looks with a polished case, new crystal, ostrich strap, and restored dial. Not bad for a humble Hamilton 747 from the early 1950s!

## A Handy Calculator

Here’s a useful calculator for computing the mean and standard deviation of your rates. Remember to enter the absolute deviations from zero: all values are positive.

We use absolute deviations because they are less likely to mislead. The average of +20 and -20 is zero, after all, which implies that all is well when in fact the positions are far apart. But the average of their absolute values is [20+20]/2 = 20, which accurately reflects the need for more adjustment.