[fitsbits] leap second alert

[Mark Calabretta]

On Thu 1999/12/16 10:10:14 BST, Paul Schlyter wrote
in a message to: fitsbits at cv.nrao.edu

>> In any case, my feeling is that anti-leap-secondists should be lobbying
>> for civil time to be changed from UTC to TAI (+ offset?).  The proposal
>> to change the definition of UTC appears to me to be a back-door route to
>> achieving this.
> 
>The anti-leap-secondists should leave civil time alone, and instead
>adopt some time scale without leap seconds!  For instance TT, or TAI,

Yes, I didn't mean that I'd support that either, only that they should be
upfront about what they want.

>or TAI+some_offset.  Astronomers have done this for decades already:
>their ET (Ephemeris Time) since 1960, and TT (Terrestial Time) since
>1984 serves this purpose well -- however astronomers need to keep
>track of both ET/TT and UT, since UT *is* a measure of the Earth's
>rotation, which matters astronomically.

Astronomers usually keep TAI (perhaps as BAT) using atomic clocks because
that's the most convenient method.  Conversion to UT1 is via tables
published in IERS-A.  (While I'm on that, wouldn't it much more convenient
if IERS-A tabulated UT1-TAI rather than UT1-UTC?  Then the table wouldn't
jump and we could drop UTC out of the calculation altogether.)

>  Computer folks have it
>easier: they need not worry much about the Earth's rotation, so they
>could forget about UT and simply adopt TAI.

The time reported by your computer (perhaps synchronized via NTP) has to
agree with the time reported by the radio announcer, train timetable, TV,
etc. so the NTP community have to conform to civil time.  The argument is
about changing civil timekeeping.

It occured to me later that the proposal to drop leap seconds from civil
timekeeping is like lobbying to bring back the Julian calendar to save
having to keep track of those irritating leap years (although admittedly
they're much more predictable).  After all, it took over a millenium for
the Julian calendar to get noticably out-of-whack with respect to the
vernal equinox (and therefore Easter).  If we drop leap seconds from
civil time it will take a millenium for it to deviate from the sun by
30min, safely someone else's problem - just like Y2K was!

And what about that irritating daylight saving?!  If we dropped that we
wouldn't have to bother keeping track of when it starts and stops in
different countries/states.

The more I think about the proposal the sillier it seems - unless, that
is, you go the whole way and drop the notion that civil time keeps track
of the sun.  If everyone kept TAI as civil time (with no time zone
offsets) then instead of knowing that Sydney is 10:00 ahead of Greenwich
you'd know that midnight occurs there at approximately 10:00 TAI - at
least in this millenium.  Daylight saving in Sydney would consist of a
general agreement to start work at (say) 18:00 TAI rather than 19:00 TAI.
I think this would be a workable system, but I don't imagine for a
leap-second that you'd persuade the general public of that.
 
>Don't you think the French would want us to use the Paris meridian instead? 
:-)

While they succeeded admirably with the metric system, the French missed a
golden opportunity for calendar reform.  Instead of trying (and failing) to
introduce a metric calendar, a much better and more achievable goal would
have been to divide the year into 13 x 28 day "lunars" + 1-day, or in leap
years 2-day, residual.  In this system, in a particular year the Nth day of
each month would correspond to a particular day of the week and, to a
reasonable approximation, a particular phase of the moon.

If we're going for big civil timekeeping changes then I'd say that one
should have a higher priority.  (Just thought I'd throw that red herring
in for you!)

Mark Calabretta
ATNF