Stan writes in response to my question about whether UNIX handles dates
correctly prior to November, 1582 (the first full month following the
adoption of the Gregorian calendar):

>The basic answer is:
>
>   *no* Julian/Gregorian calendar makes a lot of sense prior to
>1752-09-14.
>   ...and possibly even later than that.   ...  forget about 1582!
>
>Why?
>
>   Many western countries finally adopted the Gregorian Calendar in
>   September, 1752 ... dropping  almost two weeks:
>
>      cal 9 1752
>
>         September 1752
>       S  M Tu  W Th  F  S
>             1  2 14 15 16
>      17 18 19 20 21 22 23
>      24 25 26 27 28 29 30
>
>I seem to recall that a few countries (including Russia?) waited even
>longer to adopt the Gregorian calendar ... yes, found it on the net:
>
>   From: http://genealogy.org/~scottlee/cal-overview.html  ...
>
>   The Gregorian calendar was not instituted until October 15, 1582
>   (or October 5, 1582 in the Julian calendar). Some countries did not
>   accept it until much later. For example, Britain converted in 1752,
>   The USSR in 1918 and Greece in 1923. Most European countries used
>   the Julian calendar prior to the Gregorian.
>
>So the bottom line is that unless you define the adoption point of the
>Gregorian (and earlier) calendars for your particular pre-1923 calendar
>date,

The story's actually a little more complicated than that referenced
above in the web excerpt. The points at time when the Gregorian (or "New
Style") calendar was adopted in preference to the Julian ("Old Style")
calendar was based greatly on religious and political reasons --
but that's exactly as you would expect. Britain was, by 1582, staunchly
anti-Catholic (Henry VIII ruled 1509-47), and the Anglicans would be
damned if they would do anything that the Pope in Rome thought was a
good idea. Thus, it took until 1752 for the British Empire (which was
then worldwide and growing) to finally capitulate and admit the
correctness of the calculations of the astronomers that Gregory had
hired.

At the time of Gregory, the accumulated error in sidereal time (the time
it takes the Earth to return to same point on the star background, as
referenced from the Sun) from the initiation of the Julian calendar
to 1582 was 10 days; by the time the English adopted the NS calendar,
that error had grown to slightly more than 11.5 days.

As it occurs, the specific time of the changeover still has
reverberating effects on our own lives (if you're an American). George
Washington, who was native-born to the American colonies, was born on
March 5, 1732 (under the Julian calendar). When England and the colonies
changed over, George Washington personally changed his birthday to
February 22 to keep his birthday on the same sidereal day. Abraham
Lincoln, of course, was born much later, in 1809, when the NS calendar
was well in place. But if Washington hadn't elected to rewrite his own
history, the US Congress probably would not have been able later to
argue the case that the birthdays of the two most prestigous presidents
in American history should be celebrated on the same day -- and thus we
could have well had an additional holiday in March.

The reason for Greece's reluctance to change was essentially the same as
England's. The only difference was that this time it was the Eastern
Orthodox Christian Church that refused to go along with the Holy Roman
See -- and it took until 1923 for them to see the futility in trying to
fool Mother Nature.

Russia was only slightly more progressive -- by a few years -- but for
different reasons: complete social upheaval. The Communists came to
power in 1917-1918 and were pre-wired to overthrow anything that smacked
of the previous regime, especially that of one of the overwhelming
symbols of establishmentarism in Tsarist Russia, the Greek Orthodox
Church. Immediately changing the calendar was just one more way of
poking a sharp stick in the eye of the oppressor elite in Russia.

Thus, Stan's right in one sense about this confused space-time continuum
(where you stood on the Earth's surface determined what day it was --
and still does!) as to when to switch the computed calendar from
Gregorian to Julian depends on which country you believe the machine
will be primarily used (and its predominant religion). It could be well
argued on these grounds that I made a mistake in running QueryCalc's
calendar back to 1583 rather than 1753.

But the question that I was originally asking doesn't depend on the
particular date of the changeover. The more important point was that the
leap-year counting algorithms of the Julian and Gregorian calendars are
different -- and if these differences aren't taken into account, the
reported days of the weeks will be in error.

There is an easy test for to see if UNIX has it at least basically
right, regardless of the adopted changeover date: Are the years 1000,
1200, and 1204 all leap years? They are under the Julian calendar -- and
must be in the computed calendar if the reported days are to be correct.
If all three dates are not leap years, the Gregorian algorithm is being
erroneously applied beyond its bounds.

Wirt Atmar