On 19/11/2017 10:34, Mikko wrote:
In article ,
Gerald Kelleher wrote:

The basic experiment involves either a sand or water clock. The period of one
noon to the next full cycle would generate different volumes of water and
sand reflecting the variation in the total length of each individual cycle .

Sand and water clocks are too inaccurate to detect the variation in
duration from one noon to the next.

Actually they were just about good enough to do it a long time ago.

https://www.researchgate.net/publica...lack_Sea_Coast

They would need to be good to better than 10 minutes in 24 hours to get
a 3 sigma detection of shortest day vs longest day variation. They are
more accurate than that. I built one for a lecture (an equal interval
one constructed as a parabola of revolution). It was surprisingly
reproducible a larger scaled one probably good to

That is how the ancients knew that the suns motion across the heavens
was at a variable speed. Unclear how many times the water clock in its
various guises was invented. There are obvious systematic errors that
vary with temperature and humidity but it wasn't a bad start.

A Chinese inventor even did one with mercury as the working fluid which
avoided problems of freezing up in continental winters.

--
Regards,
Martin Brown