the beauty and curse of leap seconds

From: Steve Allen <>
Date: Thu, 10 Jul 2003 22:49:27 -0700

On Thu 2003-07-10T16:26:52 -0700, Rob Seaman hath writ:
> However, I doubt whether any complete inventory has been taken of the
> dependencies of aviation - or of any other technical or civil field -
> on *accurate* synchronization with time of day.

I have opined that the UTC of the 1960s truly was chaos. The roughly
annual steps of 5 to 200 ms, and the concomitant changes of the
carrier frequency and pulse duration of radio broadcast time signals
by several parts in 1E8 were intolerable. The unilateral action of
the CCIR to switch UTC from chaos into leap seconds with a lead time
of less than two years was uncomfortable, and the failure of the CCIR
even to send a letter to the IAU, thus depriving the 1970 IAU GA of
any official basis to act on the change until after it would be in
effect, seems deplorable. But there were clearly some very suave
diplomats working behind the scenes at the IAU GA in 1970 who deserve
applause for smoothing things out and keeping impolitic remarks out of
the official record.

So we now have radio broadcasts with carriers and ticks in atomic
seconds, and that is definitely much better than the clumsy attempts
to match UTC with UT2 that preceded 1972. But there are almost no
clocks on the planet which actually keep UTC. A UTC clock (and by
extension of seconds into days, a UTC calendar) must have a table of
all historic leap seconds, and warning about upcoming leap seconds.
This is annoying at best, and impractical in most cases.

Wristwatches, grandfather clocks, battery clocks, line-current clocks,
and almost every other clock that is used for civil time all ignore
the existence of the occasional 61st second. Most of them are of a
precision that prevents them from claiming that they provide UTC; at
best they can be described as regularly being re-set to UTC. As noted
by Seago and Seidelmann (in their paper that was available but not
presented in Torino), all of these all better described as "generic UT
clocks"; they are not really UTC clocks.

Even the POSIX standard and the majority of Unix systems that obey it
are self-inconsistent in the presence of leap seconds. In a sentence
the standard says that POSIX time should be taken to be UTC, except
when it should not.

The invention of leap seconds for UTC was a brilliant compromise.
Unfortunately, the effect of that compromise was to obscure the
distinction between two concepts already known to be irreconcilable.
Time-of-day (earth rotation) is not time interval (atomic time).

UTC with leap seconds has permitted not only the general public, but
also clock builders, systems designers, standards writers, and even
the ITU-R and IERS to avoid contemplation of adding complexity to
their devices and practices for 33 years. All this continues to
obscure the distinction between universal time and atomic time.

If 33 years ago the CCIR had decided that time-of-day and time
interval were irreconcilable, then today we might have that
distinction built into most of our operational systems. Sooner or
later it must happen. Now we are again faced with figuring out
whether, how, and when to educate, motivate, and pay the cost of
trying to retrofit that distinction into everything.

Steve Allen          UCO/Lick Observatory       Santa Cruz, CA 95064      Voice: +1 831 459 3046
PGP: 1024/E46978C5   F6 78 D1 10 62 94 8F 2E    49 89 0E FE 26 B4 14 93
Received on Thu Jul 10 2003 - 22:51:23 PDT

This archive was generated by hypermail 2.3.0 : Sat Sep 04 2010 - 09:44:54 PDT