IIRC the H1 prototype was tested aboard ship to confirm his theory of using oscillation. 25 years later he not only produced a practical instrument but also a beautiful work of art.
2 Likes
H1 through H3 were all too sensitive to external motion. During the several years of his work on H3 he designed a watch for his own use built for him by John Jefferys using a different type of movement than the sea clocks. Its accuracy proved astonishing, and realizing that higher beat frequency (implying smaller size) was an important factor he abandoned work on H3 and started H4.
Later, however, he built a terrestrial pendulum clock for which he claimed a rate of one second in 100 days. He was laughed and sneered at as delusional. However this clock was recently tested at the Royal Observator, sealed in a plastic box so it could not be touched except for winding it, and in a 100 day test it lost 5/8 of a second.
1 Like
It should be noted that a three dollar crystal-controlled watch employing a tuning fork crystal oscillating at 32,768 Hz has a specified rate over its working temperature range of +/- fifteen seconds a month.
Slightly fancier ones using a 4 MHz crystal are good for about +/- five seconds a month under the same conditions.
Worn on a wrist with smaller temperature variations each type will do a good deal better than the spec.
Until the battery runs out 
Cheers,
Earl
1 Like
I expect you could generate a few microwatts from almost any cyclic or stream phenomenon.
Not for three dollars
Just kidding. I’m very happy with my self-wind Marathon with manual wind backup. You used to be able to get quartz watches whose batteries you could change yourself but they seem to have disappeared, especially if you are a fan of tritium dials.
Cheers,
Earl