[LEAPSECS] the first TAI

[Michael.Deckers]

    On 2018-02-05 08:11, Steve Allen wrote:

> I have been diving through the library volumes with the contemporary
> records of the early days of atomic chronometers.  One of the things
> that stands out is in this image
> https://www.ucolick.org/~sla/temporary/tai1960.jpg

    Great find!

> ...But I digress from this first publication of TAI.  I have not seen any
> historical synopsis that mentions this first use of a time scale with
> the initials TAI.  Has anyone seen a reference to this use?
> If not, that begs the question of why not?

    I have seen "temps atomique integré" in [Audoin 1998, p 53..55], 
where the
    authors explain it in detail, and say that it was used from 1960 until
    1969-01-01 as a local atomic time scale at the BIH. Their point is that
    comparisons of the readings of distant atomic clocks (first done via VLF
    time signals) did provide good accuracy for the frequency but 
insufficient
    accuracy for the phase (even if done every so often). Hence the BIH was
    forced to integrate (over a time scale apparently determined with quartz
    clocks!) a mean value from atomic frequency observations to obtain a
    consistent time scale with the rate determined by Markowitz et al. The
    advent of LORAN-C reduced the uncertainty of long distance 
comparisons to
    the 1 µs level and the BIH then formed a mean reading of atomic clocks
    (modern TAI, or, rather, EAL) -- as opposed to the integral of a mean of
    the observed rates of these clocks (integrated atomic time). The name,
    TAI, is used (proleptically) to denote both scales, and more, since 
1955.

    HTH

    Michael Deckers.

    [Audoin 1998] Claude Adoin, Bernard Guinot: "Les fondements de la mesure
    du temps. Comment les fréquences atomiques règlent le monde".
    Masson. 1998 Paris. ISBN 2-225-83261-7.
    There seems to be an English translation for those who have not
    perused scores of volumes of the BIH.