On Fri 2003-07-04T10:32:51 +0100, Markus Kuhn hath writ:
> > It also deserves to be pointed out that the Julian Date would not
> > match in the two different calendrical schemes. With that it becomes
> > tricky to calculate elapsed time.
>
> "Elapsed time" ("time" in the sense of ephemeris time or atomic time)
> would actually be easier to calculate, at least if you used a Julian
> Date derived from real time, not UT.
Julian Date is a count of days. Historically "day" has been
understood as a count of earth rotations minus earth revolutions, but
we should agree to admit that a "SI day" would be 86400 SI seconds. A
calendar is a count of days modulo, but also keeping track of, months
and/or years.
By the current reckoning, the date 7001-01-01T12:00:00 (Thursday)
corresponds to JD 4278123 in the Gregorian calendar. But by about
that time a hypothetical atomic timescale (TI) will have incremented
by one full day more than an earth rotation timescale (UT).
So when TI may have the value
7001-01-01T12:00:00 TI, JD(TI) 4278123, Thursday
it is completely understandable that UT might have the value
7000-12-31T12:00:00 UT, JD(UT) 4278122, Wednesday
Trying to reconcile those two in any way is not a good thing to do.
The difference is itself the simplest way to acknowledge what will
have actually happened.
Eventually the difference might become large enough to be considered
awkward (i.e., when TI and UT differ by a significant fraction of a
year), but we are silly to propose or plan any remedies for that.
> As someone from the BIPM stressed at the Torino meeting, we shouldn't
> forget that UT isn't really a measure of time. It is merely the angle of
> a mass in the solar system that was historically used as one popular
> practical representation of time.
UT is time-of-day. TI will be time interval counted from its epoch.
In general, people do not have a clue that time-of-day differs from
time interval. They do not understand and will not care that
differencing two values of UT provides a value which cannot be used as
time interval without a footnote.
The first concept is not just popular, it's primeval. Stromatolites
record it at least half a billion years into the past. Protozoa still
react to it. Dinosaurs understood it. Cockroaches understand it.
Cows understand it. People use it to count the span and plan the
interactions of their lives.
The second concept is (almost *) our best measure of the independent
variable underlying the evolution of physical processes on this
planet.
(* For the purposes of this discourse ignore the fact that TAI is a
statistical average of clocks globally, and as such its rate does not
vary diurnally. But the proper time of any particular clock on the
surface of the earth ticks faster at night -- integrating to a
picosecond level. This will surely result in a Nobel prize when it is
detected, and then it will have to be modelled as a correction to the
meaning of TAI.)
Both concepts are valid. Both concepts must be explicitly recognized
by broadcast time signals.
Aside from the status quo, the only plan that has been put forward by
the SRG has, in the sketchy form that was available, seemed tantamount
to declaring that time-of-day is no longer relevant to anyone.
Yes, we must have a broadcast time that gives readily available
access, and possibly even primary access, to atomic time. Yes, it
must be stressed that time-of-day is not time interval. That
difference must be taught in schools. It must be made evident in
timestamping schemes and interface definitions that are used by
information systems. But taking away readily available access to
time-of-day it not best the way to teach it. The users of broadcast
time are humans, and the primeval notion of time cannot be discarded
by fiat.
I hope this was part of the message that was communicated in Torino.
But it means that the simplest possible scheme for changing to
broadcasting atomic time is unacceptable, and that means that the
change will be incompatible with some aspect of the existing systems.
--
Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064
sla_at_ucolick.org Voice: +1 831 459 3046 http://www.ucolick.org/~sla
PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E 49 89 0E FE 26 B4 14 93
Received on Sun Jul 06 2003 - 00:01:41 PDT