[ntp:questions] Re: GMT: UT1 vs UTC

Rob Seaman seaman at noao.edu
Thu Oct 30 18:44:45 UTC 2003


Steve Allen at Lick Observatory has an excellent UTC resource page:

    http://www.ucolick.org/~sla/leapsecs

His bibliography, http://www.ucolick.org/~sla/leapsecs/onlinebib.html,
is especially useful.

Ian G. Batten says:

> Come the winter, for us in the UK date(1) reports:
>
> Thu Oct 30 11:01:15 GMT 2003
>
> when in fact for almost everyone's computers it's UTC, which is +/- 0.9s
> from GMT.  And the `Greenwich Time Signal' on the radio is UTC(GPS)
> these days, too.

I would hardly refer to the current Unix/Posix/Linux date command as any
sort of fundamental source of time :-)  What is really happening here,
of course, is that "GMT" is simply your local time zone.  If I ask for
the date in Tucson, I get:

    Thu Oct 30 10:38:03 MST 2003

If I ask for universal time (date -u), I get various answers:

    Thu Oct 30 17:39:56 UTC 2003 (from Linux), but
    Thu Oct 30 17:40:10 GMT 2003 (MacOS X)
    Thu Oct 30 17:40:27 GMT 2003 (Solaris)
    Thu Oct 30 17:40:15 GMT 2003 (SunOS)

In those cases that both GMT and UTC are mentioned in the web page for
date, they are described as synonyms:

    "Display the date in GMT (universal time)."

    "Coordinated Universal Time is often called "Greenwich Mean Time"
    (GMT) for historical reasons."

Often the approximation is implicitly assumed - the man page refers to
Universal Time and issuing the "date -u" command reports GMT.  The fact
is that Universal Time (of all flavors, including UT1 and UTC) has always
been taken to refer to an approximation to Greenwich Mean Time.  This is
a very useful approximation that should not be lightly discarded.

Ian further wonders:

> Does anyone have a clock which actually ticks UT1?

UT1 is only known after the fact when observations can be consulted to
derive the precise Earth orientation at a given epoch (and also remove
the higher order terms that go into UT2).  That said, the precision
timing community's predictions of UT1 have become exceedingly accurate.
Take a look at Figure 2 from http://iraf.noao.edu/~seaman/leap.  (In the
interest of full disclosure, you'll note that I have an axe to grind :-)

Astronomical observatories sometimes do present clocks to the personnel
in their control rooms that display UTC corrected by the current DUT
value to report UT1 to within 0.1s.  More often, however, the clocks
simply present UTC while internal operations for pointing the telescope
and tracking objects and other utility chores include the DUT correction.

Can anyone comment on whether commercial clocks are available that
include the DUT correction?  This would seem a prerequisite before any
action might be taken to decommission the leap second and retire UTC as
the international civil time standard.


Terje Mathisen says:

> Why would you want one?

Good question.  For many purposes raw UTC suffices.  An approximation
of the Earth's orientation to one second of time is often adequate.
For some purposes, it is not.  On the other hand, why would anybody
want any precision source of time?  Is your statement that precision
time is only useful for marking relative intervals?  I would suggest
that absolute standards of time are important for many purposes.

> Even in England/Great Britain GMT is just a local name for UTC these days.

Legally this is not true, as Ian points out:

> Legally I think GMT is still UT1 --- the bill to redefine it as UTC
> didn't pass in 1997.

Imagine reintroducing such bills in countries around the world - with
the express (or hidden?) intent being to get each country to switch
from GMT to UTC simply such that UTC can be later stripped of any
underlying connection to the spinning Earth.  Should make great
theater on C-SPAN - at 3:00 am :-(

Terje continues:

> UT1 is for astronomers, as well as a way to figure out when/if to
> add/subtract leap seconds.

UT1, and Universal Time in general, is for everybody - at some level
of approximation.  There is indeed an initiative among the precision
timing community to eliminate leap seconds.  Such a move would
eventually turn day into night (literally).  Everybody on the planet -
for all activities - cares at some level.  The current arguments for
such a (rather silly, in my estimation) change key on suggestions that -
for instance - nobody cares about daylight saving time so nobody will
care about the elimination of leap seconds.  The flaw here - a flaw
that has been repeated many times during the Kabuki-like leap second
debates - is to confuse periodic effects (not just Spring Forward, but
ALSO Fall Back) with secular effects (UTC being allowed to diverge
indefinitely from GMT).  (The latter could even be interpreted as the
Prime Meridian being allowed to drift out to sea.)  Folks most certainly
would care if clocks were set forward another hour every spring and then
never set back.

It's also just plain weird for the precision timing community to rally
around a proposal to make the civil time standard completely imprecise.

> BTW, anyone want to bet how much (supposedly leap-second handling) gear
> will fall down if we actually have to subtract a leap second?

And how much "gear" will fail if we don't?

    http://www.motorola.com/ies/GPS/docs_pdf/notification_oncore.pdf

(As Steve Allen points out, "It is rumored that these [GPS] receivers
may be built into some JDAM smart bombs and other munitions.")

We live on a rotating Earth that is spinning down.  That is a reality
that our institutions and infrastructure can't ignore.


David L. Mills says:

> If we ever had to give back a leapsecond, it would be a disaster. The
> WWV/WWVB timecode has only a single bit to signal leap insertion and no
> way to signal deletion. The GPS timecode has the UTC-GPS offset in
> seconds, but the GPS receivers I have indicate only leap insertion. From
> the ERTS data the rate of increase has flattened out over the past thre
> years, which certainly gives me the willies, but my professional friends
> tell me not to worry about it.

We don't have leap seconds because the Earth's rotation is slowing down -
we have leap seconds because it has already slowed down over the century
since the equatorial epoch of 1900.  The figure labeled "Figure 1" from
http://iraf.noao.edu/~seaman/leap shows this trend.  It would take much
more than the Earth's deceleration flattening out temporarily (a third
or fourth order effect) to overcome such an accumulated lever arm.

This would, however, be a good opportunity to redesign WWV and GPS and NTP 
to be actually responsive to the technical requirements of the underlying
time standards.

Rob Seaman
National Optical Astronomy Observatory
Tucson, Arizona  USA



More information about the questions mailing list