Re: Feb. 29, 2000

[wier@xxxxxxxxxxxxxxx (Bob Wier)]

|>Just as a side note - this is the last "normal"
|>Feb 29th until after the turn of the century -
|>the *next* one after this year doesn't occur but
|>once every 400 years - and will likely break a bunch
|>of code (electronic cameras?) - Of course, my circa
|>1955 Realist could NOT care less!
|
|My computer's calendar insists that there will be a Feb. 29 in the year 2000,
|but as a loyal "Jeopardy" follower, I know better. I'll bet this will create
|havoc among computerists, though!

Actually, it's working correctly (although it may just be coincidence) -
here's the algorithm...

If a year is evenly divisible by 4, then it's a leap year, and you add
     Feb 29th,

UNLESS the year is also evenly divisible by 100 (ie, 1900) in which case
    there is NOT a Feb 29th


UNLESS (this is the big one) the year is also evenly divisible by 400
    (ie, 1600, 2000) in which case you stick in Feb 29th again.

Thus, in 2000 there will be a Feb 29th, which normally there would not
be, given that it's a "century" year.

There is another cycle out in the 3600 year range, when you take it back
out again, but I'm not worrying about that :-)

All of this is due, of course, because the year is not exactly 365 days
long but is in the range of 365.24xxx. The .24 requires a correction
every 4 years, but since it's not QUITE .25, then that error accumulates
and requires a correction every 100 years, etc etc etc. Adding to the
confusion at this point is the fact that the year length is not constant
due to geophysical factors, and so "leap seconds" are periodically added
(usually once or twice a year) giving 61 second minutes. This however,
doesn't affect the basic calendar algorithm (although it does affect
the time readings for GPS units - satellite time doesn't exactly
equal "clock time").

There is a quite active mailing list (The Y2K - Year "2000") concerning
these issues. The bigger issue is the potential breakage of programs which
represent date codes only by the last two digits of the year number.
The first time this occurred was in 1970, when 30 year mortgage calculations
were being done. When the mortgage termination date rolled over to be
2000 ("00") and you calculated how many years you had left for interest
purposes, you got very strange results (ie, in 1971 you had 00 - 71 years
left, or NEGATIVE 71). Banks tended to get rather upset about this
because the calculations indicated that they owed you a refund on 71
years of overpayments on your mortgage (plus interest!).

There is some thought that it might be best to stay out of elevators
and so forth just after midnight on 2000, since a bad security
program might lockout elevator usage, power down HVAC systems, etc
(such as are found in large office buildings sometimes). An interesting
exercise is to set your PC clock to 11:58 PM, 1999 and see what
happens when it rolls over. Two methods to do this - set the time, power
off-wait 5 minutes, power back on. Also try just watching it roll over. Then
ask it the date/time. About 60 percent of the current PCs malfunction,
showing 1970, 1900, 1984, etc (depening on how the date/time is
calculated).

Lastly (a bit of trivia)  - the new millenia doesn't start on 2000, it
starts on Jan 1st, 2001 (since most calendars used in the West don't
recognize a year zero). It's said that's why Arthur Clarke named his
novel that, instead of "2000" :-)

For me, this is an interesting topic since it makes for some real
head-scratching programming assignments for students!

Sorry for the digression, Now, back to Photo-3d! (anyone needing
further info can e-mail me directly). A timely subject.
(Ouch!)

Thanks!

   ====== wier@xxxxxxxxxxxxxxx ======
   keeper of the Photo-3d, Motorola
 MC68HC11, Overland-Trails, LDS State
Research Outline Guides and other stuff
     (currently in Ouray, Colorado)



------------------------------