Notes

Rolex Watch Timing Data Explained: Witschi X1 G2 in use.
replica rolex

Tolerances Mechanical watches are not exact, they keep time within a certain tolerance.
The tolerance of a watch is given as the maximum amount of seconds per day it may lose or win. For example, if the specification of a model says it has an accuracy of –4/+6 seconds per day, that means it is going to deviate at most that much. That does not mean your own timepiece will reach the extremes, however, could be the case that in practice it just gains around a couple of seconds. This post explains how to measure that.
According to Frederique Constant, a tolerance –20/+20 is considered to be generally acceptable, and –7/+7 is considered to be excellent.
About 3% of the Swiss production earns the COSC chronometer certification, which means their movements are certified to operate within –4/+6. Movements are tested uncased and under certain standardized conditions.
Some watches have even better accuracy. For example, the Caliber 9S of Grand Seikos works at –3/+5, also tested uncased and under some controlled conditions. When you wear the watch, though, the expected tolerance changes to –1/+8.
Rolex, on the other hand, goes an extra mile: their movements are sent to COSC, and afterwards they individually undergo in-house testing that certifies every single unit that leaves the factory keeps time within –2/+2, cased, and in simulated real-life conditions.
These tolerances do not mean that the watch is going to operate within those margins under any circumstance, we are going to see that the deviation of a watch depends on a number of factors.
Factors that influence accuracy Mechanical watches are driven by an engine made of delicate springs, tiny wheels, pinions, etc. This is a cool didactic 360° live schematic model.
Therefore, first and foremost, the design, manufacture, materials, and quality control of the movements, set a baseline for how well they may perform.
Additionally, winding, shakes, temperature, gravity, magnetism, the state of inner lubricants, etc., also influence how they function in real conditions.
Regarding gravity, it is interesting to note that a watch typically presents different deviations in different positions. That is, the deviation if you leave the watch dial up, or leaning on the side crown up, might be different.
Fine testing If you go to your watch repair shop to get a time test done, they are going to put your timepiece in a device that listens to the tick-tock of the escapement, in different positions.
The resulting metrics, extrapolated, provide expected deviations per day. For example, a test may yield something like this:
Dial Up: +1.2 s/dDial Down: +2.7 s/dCrown Up: +2.0 s/dCrown Right: +1.5 s/dCrown Down: –0.3 s/dCrown Left: +1.2 s/dX: +1.1 s/d The first row means that if you leave that particular watch flat, with the dial facing up, it is going to gain around 1.2 seconds after 24h. Whereas if you leave it leaning on its 9H (crown up) for a day, it is going to gain around 2 seconds. The X metric at the bottom is a sort of estimated average behavior, calculated following some built-in heuristics.
This is a great video of a watchmaker testing a high-end model with a Witschi Chronoscope X1.
Professional equipment like that one may cost thousands of dollars, but there are personal options like Lepsi’s Watch Scope and Watch Analyzer, ONEOF Accuracy², or the more affordable Frederique Constant’s Analytics.
As explained in the video linked above, these devices give additional information that tells experts about the health of the movement. At the bottom of Frederique Constant’s product page there’s also a FAQ that covers some related concepts like rate, beat error, amplitude, or lift angle.
Mobile applications Some mobile applications implement the same idea using the mic of your phone or earphones for input. I have tried Hairspring and Watch Tuner Lite, but with unconvincing results. It was hard to get consistent measures. I guess the applications work correctly software-wise, but the audio seems to be just too difficult to precisely control, even if you try to run the test in a really silent room. I have seen good reviews of them, though, so YMMV.
There is an interesting app called Twixt Time that instead of listening to the escapement, applies visual recognition to detect the hands of the watch and calculate drift compared to a reference clock. This app has also a feature that is a must for me, which is the ability to record a series. More on that later.
Toolwatch is based on a different approach: You tap when the second hand is at twelve o’clock. Then do that again hours later, say 24h for instance, and the application calculates the deviation with respect to a reference clock.
Indeed, the personal system I’m following implements a similar idea, only I use a spreadsheet and keep a track record.
My personal systemObservations If you wear your watch regularly, your arm is oriented and accelerates in all directions throughout the day, so the estimated drifts in the reported fixed positions of a test don’t tell you enough to know how it behaves in practice. The X metric is practical, but still, it’s an estimation.
On the other hand, single data points do not extrapolate well. For example, if you measure the deviation in a 24h interval and it yields +0.32 s/d, can you multiply by 30 to conclude that in a month the watch is going to be at +9.60? My experience says this is not necessarily so.
A simple solution For me, a good compromise is to keep a spreadsheet where you write down the deviation in a series of consecutive days.
Since the precision manually looking at the second hand is not going to be extremely good, it is important that you track the total deviation time, that is, the deviation compared to a reference clock, ignoring the measure of the previous day. Daily deltas and averages are a function of the observed absolute drifts, which is something a spreadsheet can automate for you.
For example, this is a record leaving my watch with the crown down at night:
There, only the “Total Drift” column is filled. “24h Drift” is a calculated cell that shows the difference with respect to the previous day, and “Average” is the total drift divided by the number of days, both in the same row.
The rationale for filling only the total accumulated deviation is that you are a human and will introduce measure errors trying to hit a button at exactly twelve o’clock. It might be off by 0.2 seconds, right? We are not so precise. When you stopped the second hand to set your watch at the beginning, maybe it was not perfectly vertical either and the series started 0.12 seconds off?
In your manually tracked reports, tiny differences of a few tenths are not significative due to these natural and small errors. By focusing on the total drift, the individual daily errors will at least be gone and won’t propagate.
Point is, if your watch is regularly gaining about 2 seconds every day, in a week your total drift reading will be around 14 seconds. That’s it, you’ll see the trend, and good enough figures for personal use in the calculated cells.
Manual procedure A manual way to do this is:
Prepare a stopwatch, for example the one in your phone.Prepare a reference clock like Emerald Time, or the one included in the HODINKEE app. Ideally one that plays a sound each second, it helps. With that ready, then:
When the second hand of your watch is exactly at twelve o’clock, start the stopwatch.Now look at the reference clock.When the reference clock reaches the next minute, stop the stopwatch. If your watch was too close or past real twelve o’clock don’t stress, just let the reference clock run for a minute.Look at the stopwatch reading, and do the math.Programmatic aid I am a computer programmer, and while I followed that procedure for a while, nowadays I run a program that I wrote that tells me the drift right away:
That is written in the C language because it is blazingly fast, that runs in about 0.004s, so it won’t affect the measure by any significant amount. If you know what this means, you can compile it with
gcc -O3 -o wdrift wdrift.c Sample execution from today:
$ wdrift+0.21Magnification Tapping when the second hand is exactly at twelve o’clock may be challenging, but it becomes significantly easier if you augment the dial with the camera zoom of your phone or a magnifying glass.
In my case, I can see the coronet with a crisp center, which is easy to target.
By the way, my watch is awesome To finish, let me share the table with the metrics for my watch leaving it with the crown up at night:
As you see, this watch keeps correct time during days and days. Turns out my watch is not only gorgeous and versatile, but also ridiculously accurate.

Here's my website: https://replicauhren.ru