[time-nuts] Re: UTC is the tail that wags the dog

[Jürgen Appel]

Dear Steve,

On Wednesday, 1 February 2023 18:19:31 CET Steve Allen via time-nuts wrote:

> On Tue 2023-01-31T10:20:13-0800 Lux, Jim via time-nuts hath writ:

> > TAI please, not UTC with ickyness like leapseconds. And, yes, there are
> > people who use GMT (although they're really using UTC, they just call it
> > GMT).
> 
> But UTC is the primary.  UTC is the tail that wags the dog.
> TAI has no legal existence.  UTC is the legal time scale most places.

> Only BIPM can make TAI, no national laboratory can make TAI.
> Most national labs are required by statue to make UTC, and from
> those BIPM constructs TAI.

It's actually  bit more complicated: The time laboratories do make their own 
UTC(k) indeed and it's also correct that the time differences of UTC(k) and 
e.g. the GPS satellites are reported to BIPM. But what's also reported is the 
time difference of the individual atomic clocks (mostly H-masers nowadays) 
with respect to UTC(k) and all the deliberate tuning done to those clocks.

So from the UTC(K)-UTC(GPS) data together with these clock-tunings and clock 
measurements the 'bare' times of each clock is calculated. Based on that data, 
clocks are evaluated for their stability and assigned a weight factor 
determining how much they contribute to the atomic time scale "EAL". This time 
scale is stable, but perhaps not accurate.
EAL-seconds are then rescaled based on a second set of atomic clock that excel 
in accuracy (Optical lattice clocks, fountain clocks, ion clocks) but are 
limited in number, not always operational and have worse stability. This 
rescaling then gives TAI-seconds. From the TAI timescale UTC is then derived 
based on the eventual addition or subtraction of leap seconds and UTC-UTC(k) 
is then published along with the other data as Circular-T bulletin.

So in summary: I agree with every one of your statements above, except TAI is 
not calculated from UTC nor UTC(k) but rather from the individual clocks 
directly. How each time-lab creates their UTC(k) from their clocks has no 
influence on TAI nor UTC.

Cheers,
	Jürgen