Watch-geeks and the Zen of Accuracy

On accuracy

Despite the fact that most watch collectors understand a mechanical watch is inherently less accurate than a quartz watch, the hobby tends to attract people with a certain penchant for paying close attention to minute details, and as such, it’s not uncommon to see a discussion regarding mechanical watch accuracy take place among watch-geeks.

Some of those people may purchase a timegrapher machine, like the one I have in my office, or the ones any good watchmaker would have. I got mine from Alibaba.com for a little over $100, and if you want to be certain about the accuracy of your mechanical watches, it’s a good investment, especially if you have a large collection, timekeeping accuracy is important to you, or if you plan to adjust or regulate your watches.

However, most people do not have a timegrapher, and will instead rely on a combination of mobile apps, web-based tools, and their own diligence in observation when determining the accuracy of their timepieces.

Although many of those tools are quite good for what they are, and I don’t want to assume anyone is making mistakes in observation, whenever I read online discussions about timekeeping accuracy, I invariably wonder about the methodology employed, particularly when the discussion involves any degree of complaint regarding the performance of a watch, even when it’s “within spec”.

As such, I thought it might be worth delving into the topic a bit, from the perspective of both the enthusiast, and the manufacturer. I’ll start with the often-used and much-dreaded car analogy.

The dreaded car analogy

When you buy a car, you likely pay some attention to the fuel efficiency ratings. These are provided by the manufacturer, typically in a standardized format: “20mpg City / 30mpg Highway”, or something to that effect.

Right off the bat, it should raise the question – in what city, or on what highway?

More than likely, the numbers were not the results of actual, real-world driving experience in any city or on any highway, but rather the results of the calculations done by the manufacturer’s team of engineers, making certain assumptions about road and weather conditions, car condition, and a host of other variables.

Many of those assumptions are likely to represent the ideal scenario - a new car, with new tires, on a perfectly level, well-maintained road, at sea level, mild temps, etc, etc, etc. As such, it’s not uncommon to see a car magazine review in which the reviewers state that they actually experienced different fuel efficiency during their testing.

It’s much the same way with horsepower ratings (at the driveshaft, or at the wheels?), and 0-60 or quarter-mile times. The figures likely result from testing under ideal conditions, and you may not be able to replicate the numbers provided by the manufacturer out in the real world.

Accuracy specs for a movement are similar, inasmuch as individuals may find that their actual, real-world experience differs from the manufacturer’s specs, and it raises questions, both about the methodology the manufacturer uses in determining the numbers they provide, and what can or should be done if your real-world experience isn’t as good.

Accuracy is more than just one number

While most people tend to boil timekeeping accuracy down to a single number – “-10 seconds per day”, “+6 seconds per day”, etc – the manufacturers’ specs are actually / typically expressed as a range of numbers, provided with a set of conditions for testing, and use terms which are worth understanding, at least conceptually, if you’re going to concern yourself with the accuracy of your watch.

Let’s take a specific example of a manufacturer’s accuracy specs, using the Miyota 9015.

Miyota’s spec sheet for the 9015 states that the movement should have an average daily rate of -10 to +30 seconds per day, with under 40 seconds/day of posture difference, tested in the four positions (or postures) of dial up, 6 (o’clock) up, crown down, and crown up, and tested between 10 minutes and 60 minutes of being fully wound (at full power), to avoid the effects of isochronism (a metric for which, Miyota does not provide in their technical documentation).

Obviously, that’s a lot more detailed than just saying your watch runs “-10 seconds per day”, or even, simply “-10 to +30 seconds/day”, but it’s important to understand the conditions and terms at work behind the numbers which are so often quoted, especially when we're discussing our real-world experience, and comparing them to the specs.

What those specs say is very specific – the watch is tested in 4 specific positions, and regulated/adjusted so that the average daily rate of those 4 specific positions is between -10 and +30 seconds per day, when measured at or very near to full power, with less than 40 seconds of difference between the readings in those positions.

Balance matters

One of the key factors in a mechanical watch’s accuracy is the balance, which can be affected by the position of the watch, due to gravity. As such, the timekeeping on the wrist can and often will differ from the manufacturer’s specs, which are the results of testing the watch at rest, albeit in multiple positions, or postures.

There are a total of 6 postures – dial up, dial down, crown up, crown down, and 6 (or 12) up, and 6 (or 12) down. The posture difference number provided by the manufacturer establishes the boundaries of the readings for each position. In the case of the 9015, the specs state that there has to be less than 40 seconds difference between the readings in the four positions tested.

The more positions a watch is tested in, and the lower the posture difference number, generally the more reliable the manufacturer’s stated accuracy specs are going to be out in the real world.

Isochronism, when the term is used in watchmaking, refers to how the amplitude of a watch, which is the measure of the balance’s swing, decreases as the watch winds down, affecting its accuracy. 

Generally, a watch will become less accurate as it nears the end of its power reserve, which is why Miyota tests the watch within a narrow window of time, between 10 minutes and 60 minutes of full power, since an automatic is generally assumed to always be at or near full power, at least while being worn.

Here again, when manufacturers provide a metric for isochronism, the lower that number is, the less the movement’s accuracy is affected by isochronism, and thus the more reliable the accuracy specs are going to be out in the real world.

It’s unfortunate that Miyota’s specs do not provide a specific number, only a bit of fine-print verbiage stating that the daily rate will be different by a “maximum of several ten[s of] seconds”. We can’t know precisely how much the timekeeping accuracy will change as the watch winds down, but can reasonably assume isochronism could as much as double the daily rate, from -10 to +30 at or near full power to -20 to +60 at or near the end of its power reserve. 

As such, testing a watch over 24 hours, with some of that time being worn, in various positions, and unknown power reserve, and some being stationary, with the watch steadily losing power, is not likely to result in numbers identical to what I'd get if I put the watch on my timegrapher, at full power, and measured it in 4 to 6 positions, getting an instant reading for the daily rate in each position.

Hands off the thermostat

Believe it or not, environmental conditions, such as temperature and altitude can also affect accuracy, and yet, few manufacturers provide any guidelines for test conditions. 

I generally assume they test at or close to sea level, and at “room temps” of 68-72 degrees Fahrenheit, or otherwise adjust their numbers to take altitude and temperature into account, much the same way automotive engineers likely assume ideal road conditions and brand new tires when calculating fuel efficiency.

Where the rubber meets the road

Knowing all the above, it’s not surprising when anyone finds their watch’s performance differs from manufacturer’s specs.

Usually, people find that a watch runs better than spec, and that’s probably to be expected, particularly if we believe, as I do, that most manufacturers are trying to strike the right balance with the numbers they provide. They want to be able to brag about how accurate their movements are, but they do not want to have to deal with a large number of warranty claims for watches running outside of spec.

As such, it seems that most movement manufacturer’s accuracy ratings are likely to be near the ideal conditions, best-case-scenario numbers, but with a little margin for error built in for an acceptable range of variation and environmental conditions, the same way auto engineers will build a bit of “fudge factor” into their numbers.

So, if the manufacturer’s specs say your watch should be within -10 to +30 per day, and you believe that your watch is not running within that range, what do you do?

First off, relax. If you didn’t already know it, mechanical watches are not as accurate as quartz watches, so if you require the utmost in accuracy, consider getting a quartz watch.

If it helps put things in perspective, there are 86,400 seconds in a 24-hour day (60 seconds x 60 minutes x 24 hours). If your watch gains or loses a full minute each day, it’s still accurately tracking within 99.93% of “perfect” ([1-{60 / 86,400}] X 100).

Before contacting the manufacturer about it, take a moment to look up the exact specs provided by the manufacturer, but make sure you are also testing the same way they do, in the same positions, under the same conditions, or as close as you can get to them, and doing the math just like they do.

But, if accuracy is important to you, even in a mechanical watch, and yours isn’t running within spec, without any good explanation why, then, by all means, contact the manufacturer, assuming the watch is still under warranty, or if not, talk to a watchmaker about what can be done to make it more accurate.

One final note – it’s quite common for watches to become magnetized, due to our frequent proximity to magnetic fields, unbeknownst to us. Magnets are everywhere, from speakers to mobile devices. Magnetized watches will generally run fast by several minutes per day. Before trying to determine if your watch is running accurately or not, make sure it’s not magnetized. Holding a compass near your watch is a quick and easy way to tell.

If it is magnetized, getting it demagnetized is cheap and easy to do, either using a degausser machine (several available online for as little as $10), having a watchmaker do it (typically a free, or at least a cheap service), or laying it down for a few seconds on the anti-theft-device de-activation pad on the counter at many store’s checkouts.

Chris Vail is the founder, majority owner and lead designer for NTH and Lew & Huey. He owns more than two dozen mechanical watches, none of which get worn for more than a few hours at a time, and so he'd never notice, nor does he obsess over the accuracy or lack of it in any of them. Completely unrelated - he's usually running late.