I HAVE BEEN experimenting for some time with methods of timing watches by ear. The results have been very gratifying.
My employer uses a watch accurately timed, synchronizing with it the beat of the watch he wishes to regulate. He is able to tell within a very short time whether the watch in question is gaining or losing. I could tell the extent of the error by this method but could not tell whether the watch gained or lost.
Another method was referred to in the February, 1937, issue of HOROLOGY. This method was to use a watch that gained a beat in a minute or 4 minutes and 48 seconds a day. If it takes this watch more than a minute to gain a beat on another, the other is gaining; if it takes less than a minute to gain a beat, the other is losing. The difficulty with this method is that it is hard to tell just when the beats coincide, as they appear to be exactly together for some six or seven seconds.
Therefore, a watch that gained two beats in a minute, or 9 minutes and 36 seconds a day, was tried. It was found to be much easier to determine just when they came together, and the accuracy of the results over a minute's time was the same. A watch can be timed very closely by this method. When it takes 5 seconds too long or too short a time to gain a beat, the error is 24 seconds a day, and it is easy to determine much smaller variations.
For best results, a watch with a fairly loud beat should be used. Then the weaker beat disappears completely for about three seconds when they coincide, and the exact seconds on which this happens can be noted.