Do you shake your watch? - Better watch this! -- Video -- January 2002

Dear All

Like so many interesting things we discover, it started with a coincidence. I was going to explain to my 3 years 8 months old daughter how a mechanical watch receives energy by shaking it. At the time I was conducting a Seiko 7S26 timing experiment and many times she had helped to shake them. But she might not have known, why. So I tried to explain...

I had just wound the Omega Seamaster GMT a couple of times and I had hacked it and synchronized it with my Seiko Perpetual Calendar (left in video).

Then I swayed (shook, you decide after seeing the video) for about 30 seconds, I looked down at the watch, the Seamaster was 10 second fast! Mysterious?!

Well, I could not explain, had to continue playing with my daughter. But that strange incident stuck in my mind.

I posted the incident on the TZ Omega Forum. One guitar player made some interesting comments about his watch gaining seconds on the Omega forum too. Another fellow WIS asked, whether I could reproduce the effect. I tried, and yes, it was quite easy. I suspected at that time, that it had something to do with that SPECIFIC Omega or else with the 2892. So I tried a Dunhill Automatic with the same movement. Same effect, 7 seconds gained with ease. The same evening I asked my neighbour at the Christmas dinner to lend me her Rolex DateJust, 6 seconds gain with no problem. Now I was really curious.

I went back home and took the Seiko divers out of the watchbox that I had been using for the timing experiment and yes, also these movements showed the same effect, 3 to 5 seconds gain after short shaking. Side note: I had to cancel the timing experiment that I had been running for 4 weeks in 4 positions...

In the mean time I had some conversation with John (ei8htohms) and we discussed the probable causes. Finally I remembered an article from Walt Arnstein Does Faster Mean More Accurate? that talked about the influence of wrist movements near half the escapement frequency (1/2 bps) that could trip off the swinging of the escapement wheel and thus create disturbances. 

My daughter holding the watch I discovered the effect with first - Omega Seamaster GMT. Note: All other brands I tested: Dunhill (2892), Rolex DJ, Seiko (7S26) showed the same effect, I add this note, because I do not want to start a rumour that Omega watches are especially sensitive to shaking. All mechanical watches featuring a balance wheel seem to show the same effect.

The funny thing is, after reading that article again, that if I tried to "shake" the watches near those 1/2 bps, that is 4 moves/second for the 2892, and 3 moves/second for the 7S26, it was even easier to repeat that phenomenon.

I guess most of the times, those deviations go unnoticed. It was really a coincident, that I noticed it.

Conclusion and most feasible Explanation: Be careful how you sway/shake/wind your automatic watches! Do not shake them near the resonance frequency of the balance wheel. The resonance of the balance assortment is simply f = 1/2 bps (beats per second). I trust you will not damage the watch if you shake it so fast. The jolts on the escapement or watch mechanism are not as violent as e.g. when playing sports (tennis).
Obviously shaking the watch near its 1/2 bps resonance results in forces on the balance wheel that might advance the escape wheel two cycles instead of one. I have never noticed a loss from shaking, the watches always gained a couple of seconds. I noticed also that an almost unwound main spring helped showing this phenomenon.
Automatic watches without winding mechanism (like many Seikos featuring the Cal. 7S26 and many Orient movements) are best wound by swaying them in ca. 8 inch (20 cm) circles. Just think of swaying a glass of Cognac.

Best regards


Notes that help when watching the video:

*The second hands are quite hard to see, they are at about 26, 26 when the movie starts,
*The seconds are at about 45 (Seiko) and 50 (Omega) after the shaking
*I shook the watch ca. 30 seconds, I did it twice, because it is tiring, but you can check the elapsed time from the hands.
*You can pause the video anytime if you have problems seeing details, Windows Media Player: just click into the movie window, click again to continue.

Shaking a Omega Seamaster GMT
Format MPEG-1, Length 12 Seconds, Size 1569 KB, No Sound

(Just realized today May 12th, 2004 that the link to the Video was not working because it pointed to an old hard coded IP address, sorry for this! It works now!)