Daniel Wegner ([EMAIL PROTECTED]) is only partly correct in
saying that an analemma must have an error due to leap years. The error can
be avoided.

It is true that tables of the Equation of Time are slightly inaccurate
because they take a mean value for the solar longitude on a named date (such
as February 17th), whereas the 4 year and 400 year cycles should be
allowed-for to be totally accurate. Fortunately for us, the peak error is
less in the next few years than at any other time in the 400 year cycle. How
convenient. The worst case is in 1903+400n and 2096+400n, when the longitude
is 7/8 of a day different from its mean value. But even 7/8 of a day
accounts for less than 30 seconds of EoT, so still allows a sundial to be
less than a minute out. Around the year 2000, the worst case is half this -
about 14 seconds.
If an EoT table is drawn graphically to allow a sundial reading to be
converted to mean time, then this too must have an error with the same 4 and
400 year cycles.

But if the sundial is marked with figure-of-eight hour lines, then there
need be no such error, since the sun's declination and longitude are related
by geometry, not by what we call the date. Even if we lost another 11 days
in a calendar reform (I am from England), such a sundial would continue to
read correct mean time. Therefore, I suggest that this is a purer and
altogether more satisfactory solution than an EoT table or figure. Except
for the little point that the EoT changes rather a lot, and the longitude
does not, at the solstices. Pity.

By the way, if you are ever making a circular date scale - to calibrate a
declination scale, for instance - you should divide it into 365.25 and make
February 29th be just the .25. This is the best simple way to allow for one
February 29th every four years.


Chris Lusby Taylor

======================================================= 
Email:  [EMAIL PROTECTED]
 (Formerly [EMAIL PROTECTED])           
======================================================= 

Reply via email to