Chronometer testing in Switzerland took place at observatories in Neuchtel (1866 - 1975) and Geneva (1873 - 1967).
Each observatory maintained its own testing standards. Generally speaking, the observatories did not test large
numbers of movements meant for public sale. Typically, watches tested at the observatories were specially made
and regulated pieces. A manufacturer might submit a few watches, then report the best results in advertising.
The few watches that were actually tested were rarely sold to the public.
Testing of larger numbers of watches intended for public sale was conducted by official testing agencies. The testing
agencies were originally called "Bureaux officiels de contrle de la marche des montres", abbreviated B.O. These
agencies were founded between 1877 and 1956, and each operated independently of the others.
In 1973, the official testing agencies were placed under central administration and the name was changed to
"Contrle Officiel Suisse des Chronomtres" [Official Swiss Chronometer Control], or COSC. The director of the COSC
was located in La Chaux-de-Fonds, and offices were located in Le Locle, Bienne and Geneva.
By way of summary, watch manufacturers send batches of uncased movements (not watches) to the COSC for testing.
The COSC charges manufacturers a fee for each movement tested. Movements are fitted with dial blanks and hands,
and groups of movements are mounted on racks. The test lasts 15 days. During the test, the racks of movements are
placed in five different positions and kept at three different temperatures. Results are tracked with cameras and lasers
and reported to a computer. The passing movements receive a Bulletin de marche or watch rate certificate. These
certificates are usually provided to the ultimate consumer with the watch, though some companies now charge an extra
fee for the certificate.
For purposes of this article, I am using an actual COSC certificate posted by MJ as TZ Classic # 55. . For convenience,
I am including the relevant portions of the certificate in this article. We will work our way from the top down. The top of
the certificate appears below.
The COSC certificate states the results in French, German, Spanish and English. The certificate number appears in the
upper left portion of the certificate. In the example above, this number is 5523048. Next to that on the right is the serial
number of the movement tested. In the example, 102472.
Below the certificate and movement numbers, reading left to right, is a description of the movement by category, functions,
diameter and thickness. In the example, the movement is Category I/1, referring to a spring balance oscillator movement that
is more than 20 mm in diameter and which has a surface area greater than 314 square millimeters. This categorization
scheme is explained on the back of the certificate (in French). The movement in question is an automatic chronograph,
30.00 millimeters in diameter and 7.55 millimeters thick.
The next section of the certificate - the center - shows certain of the raw test results, as shown below:
Looking at the column from left to right, the first column is the days, 1 through 15, over which the test was conducted.
The second column shows the temperature in degrees Celsius at which the movement was being tested. Converting to
Fahrenheit, 8 = 46.4, 23 = 73.4, and 38 = 100.4.
Column three shows the position in which the movement was tested. Vertical means the movement is hanging
perpendicular to the ground, and horizontal means the movement is lying flat or is parallel with the ground. Dial up and
down means the dial is horizontal and either above the movement or below the movement. The references to 3, 6 and
9 o'clock up in the vertical position mean the number 3, 6 or 9 is highest when looking at the dial straight on
(or where 12 o'clock normally would be).
The next column shows the daily rates. The capital letter "M" with the small subscript numbers 1 - 15 represent the
days of the test. The results in this column are not aggregate. In other words, the results are the gain or loss in seconds
per day, without regard for the results of the previous day. In this column, we see the absolute gain or loss in seconds
per day as measured against an accurate, consistent reference such as an atomic radio signal clock. For example, in
the sample certificate shown above, the reading for M8 is -0.7. Reading across all the columns, that means that on
Day 8 of the test, the watch was in the horizontal, dial down position at 23 degrees Celsius (73.4 F), and over 24 hours
the rate was -0.7 seconds, or slow by 7/10 of a second compared with the reference.
The fifth and final column shows the difference between the first day measurement and the second day measurement
in seconds per day. In the example above, M1 is -0.5 and M2 is -0.4, so the difference reported in column 5 is 0.1. M1
and M2 are used because they are two 24 hour testing periods under identical circumstances (position and temperature).
Similarly, M3 is 1.9 and M4 is 2.8, so the difference reported in Column 5 is 0.9.
Just below the chart, on the right, is the date on which the COSC testing was completed. In the sample certificate, that
date is July 25, 1997. If you're buying a watch that comes with a COSC certificate, you can compare the date testing
ended with the present date to get an idea how long ago the test occurred. The test completion date won't necessarily
tell you how old the watch is or how long its been in the dealer's case, however. As discussed below, manufacturers
sometimes place movements in a vault upon return from the COSC, to be cased at a later date.
The important numbers are found at the bottom of the certificate, shown below. These are the Results or Summary.
This is where the raw numbers are plugged into various formulae and the results of the testing are shown. Note that
not all of the figures needed to make the final calculations appear in the center portion of the certificate. The Results or
Summary reflect seven specific types of error deemed significant by the COSC. Each of the seven results must fall
within COSC guidelines for the movement to pass. If even one result is outside the guidelines, the movement does not
pass. Manufacturers are free to re-submit movements that fail.
The bottom section of our sample certificate is set forth below. A description of the tests and calculations follows.
1. Mean daily rate in the  different positions. The first figure is mean daily rate in the different positions. This is the
average rate over the first ten testing days. Simply add the rates for the first 10 days (in our example, the total is 18),
then divide by 10 to get the result (1.8 in the example). To pass, the mean daily rate must be -4 to +6 seconds per day.
COSC determines the mean daily rates by subtracting the time indicated by the movement 24 hours earlier from the
time indicated on the day of observation. The -4 to +6 figures are the most often quoted when someone asks "what
are the COSC guidelines". As we'll see, these two figures are only the tip of the iceberg.
The mean daily rate figure is said by some to be the most important to the owner of the watch. This "daily rate" tends
to indicate how the watch will perform on the wrist in actual use. In his book "Wristwatch Chronometers", Fritz von
Osterhausen states at page 59: "the lower the daily rate, the more accurate time a watch gives."
2. Mean Variation, or mean daily rate variation in 5 positions. This test is sometimes referred to as rate deviation.
The COSC observes the movement's rate in the 5 positions each day over 10 days for a total of 50 rates. To pass,
the mean variation in rates can be no more than 2 seconds per day. Generally speaking, this figure shows if a watch
tends to run fast, slow or correct, or perhaps both fast and slow; in other words, the continuous rate of the movement.
It is said this measurement shows the quality and reliability of a watch - a good quality watch will be consistent,
whether it is fast, slow or correct. The disadvantage of a consistent error is that it accumulates. If the watch is 10 seconds
fast every day, it will be over a minute fast at the end of a week. However, a watch that is consistent is more easily
regulated, and so the error is more easily eliminated.
3. Maximum Variation. This is the largest difference between any two day's readings in a single position. In the sample
certificate, we see that M9 (9th test day) was 1.3, and M10 (10th test day) was 4.3. On both of these days, the watch
was in the horizontal position, and the difference was 3.0. This was the largest difference or variation between readings
in any position. To pass, the greatest variation in any position cannot exceed 5 seconds.
4. Difference between flat and hanging positions. The figure is calculated by subtracting the average of the rates in the
vertical position (first and second days) from the average of the rates in the horizontal position (ninth and tenth days).
To pass, the difference must be no more than -6 to +8 seconds. Some experts feel this test shows the greatest difference
between mass produced and hand adjusted movements. This figure tends to be lower for better quality movements.
5. Greatest difference between the mean daily rate and any individual rate. This is the largest difference between the
mean daily rate (the figure reported in test number 1) and any individual rate during the first 10 days of positional testing.
In our sample certificate, we see that on M6 (6th day of testing), the rate was 5.3. The mean daily rate was calculated
at 1.8. The difference between these two is 3.5. The figure reported for test 5. To pass, the difference between the
greatest daily rate and the mean daily rate can be no more than 10 seconds per day
6. Variation of Rate per 1 Centigrade. This is the temperature error of the movement, expressed in seconds per degree.
The COSC subtracts the cold temperature rate (8 degrees) from the hot temperature rate (38 degrees) and divides by 30.
In the sample certificate, the rate at 8 degrees was 7.1, and at 38 degrees it was -3.2. So -3.2 - 7.1 = -10.3 30 = -0.34.
To pass, the variation cannot exceed 0.6 seconds per day.
7. Rate Resumption. This figure is obtained by subtracting the average mean daily rate of the first two days of testing
from the mean daily rate of the last test day. To pass, the resumption of rate can be no more than 5 seconds.
By way of summary, the COSC passing guidelines for all seven tests for medium and large wristwatches (movements
greater than 20 mm in diameter) are currently as follows:
1. a mean daily rate over 5 positions of a minimum of -4 seconds and a maximum of +6 seconds;
2. a mean variation between the days in each position of 2 seconds maximum;
3. a maximum variation between days in each position of 5 seconds;
4. a difference between horizontal and vertical positions of -6 seconds minimum and +8 seconds maximum;
5. a difference between mean daily rate and any individual rate of 10 seconds maximum;
6. a variation rate per degree centigrade or -0.6 seconds minimum to +0.6 seconds maximum
7. rate resumption of -5 seconds minimum to +5 seconds maximum.
The COSC certificate is not a precise indicator of how well a watch will perform on the owner's wrist. Rather, the test
results show how well a given movement performed over a given period of time under certain conditions. Several factors
can affect the actual performance the owner observes. First, The COSC test is only an artificial approximation of
actual use. Second, the COSC does not test "watches" but only uncased movements. Following the test (assuming a
movement passes), the movement returns to the manufacturer, where it is either cased or placed in a vault for casing
at a later date. Movements can spend significant time in the vault. When they do, they are usually disassembled, cleaned,
re-assembled, oiled and regulated, then cased. This can certainly affect the movement's performance, yet re-testing is not required.
Next, the watch may be affected by shipping in the distribution chain, where it might be exposed to shocks
or extreme temperatures. Conditions at the dealer's shop also affect performance. If the watch sits unused for many months,
if it sits under hot lights or in the window in the sun, if customers drop it while trying it on, this will affect how the watch performs
on the owner's wrist.
Finally, the owner's personal habits can affect performance. Is the owner active or sedentary? Is the watch kept near full
wind or allowed to run down? Is the watch in one position most of the time? Is the owner in cold or hot temperatures, or
moving between the two? Is the watch placed on a winder? How is the winder set? Each of these also affects performance.