Dear dialers,
Tom Semadeni <[EMAIL PROTECTED]> asks for clarification of a few points in
my earlier mail.
Firstly, I stated that an EoT table would have most error in 1903 and 2096.
The only factor I was considering was indeed the difference between the
calendar year and the mean tropical year. The only really easy way to
demonstrate this is with a graph of the solar longitude on, say, July 1st
each year. The graph steps by the equivalent of about quarter of a day every
year until a leap year, when it steps back by about three quarters of a day.
This sawtooth pattern is repeated every four years, but drifts by 3/4 of a
day per century, except of course that the years divisible by 100 but not
400 give it a kick in the opposite direction.
I'll try a picture. Each line represents 25 years, starting from 1600, the
distance along the line being the solar longitude on, say, 1st July. The ***
represent the range of longitudes during those 25 years:
***
***
***
***
***
***
***
***
***
***
***
***
*** 1903 is at the left of this bunch - over 7 years since a leap
day.
***
***
*** 2000 is right in the middle here.
***
***
***
*** 2095 is at the right of this bunch - the last leap year for
eight years.
***
***
***
***
***
***
***
***
***
***
***
***
Hope that helped.
Tom also asked why one didn't simply have an analemma (figure of eight) at
noon, with single straight hour lines otherwise. Well, you could, but
converting from sun time to mean time would be a mental operation: the
shadow says 3:45, I follow the declination line back to the analemma,
estimate (how?) that it says the sun is 8 minutes fast (or is that slow -
easy to get confused), so the mean time is 3:37.
Too much work for most people, I would expect. Look at the URL (
http://www.uwrf.edu/sundial/ ) pointed to by John Shepherd's email for a
much more elegant solution.
Regards
Chris Lusby Taylor
=======================================================
Email: [EMAIL PROTECTED]
(Formerly [EMAIL PROTECTED])
=======================================================
