[time-nuts] 15 Seconds error...??

Tom Van Baak tvb at LeapSecond.com
Wed Jan 18 18:15:43 EST 2012


> On 1/18/2012 4:36 PM, Magnus Danielson wrote:
>> Use of TAI is fine, as long as you won't tell anyone near the timelords,
>> as obvious TAI is supposed to be "hands off".
> You said something similar a couple of times. Where does that come from? 
> Who says you're not supposed to use TAI, and why not? Why would there be 
> a well defined time scale which isn't meant to be used?

The beauty is that it is *defined* very well. But strictly speaking
TAI cannot be *used* because there is no clock that is TAI. It's
a timescale, a paper clock. One with no 1 PPS or 10 MHz output.

What a nation uses for their legal time is their own realization
of TAI/UTC, specifically the UTC(k) as maintained by their own
national metrology institute.

OK, it's a fine line, I know. But it's convenient to think of "TAI" as
the abstract paper clock and "UTC(k)" as the physical clock that
makes the ticks you hear.

>> For physical realizations we have all the UTC(k) clocks, one or
>> sometimes more than one per country.
> Well, yes and no. From the perspective of being real-time exact, even 
> the UTC(k) clocks are only good for those with direct access, and even 
> that is debatable, since any measurement is subject to propagation 
> delay. (which may be variable at some scale, and there's phase noise, 
> and now we're picking nits)

Correct. There's no end to the nits here; piconits!
> I think the proper comparison is between UTC and TAI, as they are 
> available to average users in real time. Aren't they the same, except 
> for how they're enumerated? i.e. don't they differ only in the number of 
> integer seconds they are apart? Aren't both based on the same ensemble 
> of atomic clocks, and aren't both only really only known "after the 
> fact," when comparisons and calculations have been made?
> So, I'd still like to know who says "don't use TAI," and for what reason.

My recollection is this came out of the BIPM back when people
were abusing the UTC(k) nomenclature, as in "UTC(GPS)". The
BIPM wanted to make it very clear that there was no such thing
as UTC(GPS). Yes, there is UTC(SU) and UTC(USNO), and yes
there is a GLONASS and GPS time transfer system, but there is
no such thing as UTC(GLONASS) and UTC(GPS). Similarly you
tend not to see phrases like TAI(USNO) or TAI(NIST) used.

I don't know if there is a "style guide" to timescales. There's no
one to enforce it anyway. I'm just calling it like I see it used in
the literature. The acronyms TAI, and especially UTC, can have
many different connotations, so you have to gauge their use
against your audience to minimize misunderstanding.


More information about the time-nuts mailing list