[tz] Was there a 1971-12-31 23:59:60?

Levine, Judah Dr. (Fed) judah.levine at nist.gov
Tue Nov 13 15:17:29 UTC 2018

Applying a time stamp to an event in an end-user application with nanosecond accuracy or precision is very difficult because of the problems of measuring the transmission delay from the source to the end-user and the latency in the end-user application. These problems are generally at least as large as knowing the accuracy of the source with respect to UTC or UTC(lab). Although all of these things can be characterized statistically, that is not of much use for a single event which cannot benefit from averaging. Note that post-processing may not be of much help in determining the jitter in the processing latency.

The data in  Circular T are what they are because of the administrative and processing delays in computing UTC. The UTC(lab) is now and will probably always be an extrapolation, which is limited by the flicker and random-walk processes of the laboratory time scale. Although the publication of UTCr has decreased the processing delay, it is a provisional scale and is not of much use at the nanosecond level.

Judah Levine
Time and Frequency Division
NIST Boulder

On: 10 November 2018 22:24, "Steve Allen" <sla at ucolick.org<mailto:sla at ucolick.org>> wrote:
On Sat 2018-11-10T19:54:11+0000 Levine, Judah Dr. (Fed) hath writ:
> It is quite challenging to compute time intervals in the pre-1972
> era.

That depends on the required precision.

It is still challenging to compute intervals to nanoseconds.
That requires grabbing all the relevant issues of BIPM Circular T
and similar publications of the particular source of time if they
are not members of the BIPM atomic clock club.

One hundred years ago it was challenging to compute intervals to
centiseconds, and that was the reason for the creation of BIH.

When Anna Stoyko retired from the BIH in 1965 she published one final
analysis of all the BIH data from 1922 to 1963.  She produced plots
showing how the time services of observatories had improved.

Those plots are reproduced here

In order to gather these data the BIH had been monitoring the arrival
times of radio broadcast time signals since 1922.  The volumes of
Bulletin Horaire have the numbers that show many ways that various
national observatories and time services messed up (and messed around
with) the time they were providing.

The data in the issues of Bulletin Horaire are the nightmare of the
tzdb because taken to the absurd limit they would require that tzdb
encode the millisecond-scale differences between the legal civil times
of every nation whose transmitters were monitored by BIH.

> Different timing laboratories had slightly different values for this
> final time step.  The goal was to bring all of the real-time
> realizations of UTC to the same time.

Well, much closer than before, but if they were the same then the
subsequent issues of BIH Bulletin Horaire and now BIPM Circular T
would be much terser publications.

Steve Allen                    <sla at ucolick.org<mailto:sla at ucolick.org>>              WGS-84 (GPS)
UCO/Lick Observatory--ISB 260  Natural Sciences II, Room 165  Lat  +36.99855
1156 High Street               Voice: +1 831 459 3046         Lng -122.06015
Santa Cruz, CA 95064           https://na01.safelinks.protection.outlook.com/?url=http:%2F%2Fwww.ucolick.org%2F~sla%2F&data=02%7C01%7Cjudah.levine%40nist.gov%7C45aee56fa8e74e78eac908d64795e7ed%7C2ab5d82fd8fa4797a93e054655c61dec%7C1%7C0%7C636775106502069769&sdata=xJUyk%2BhmFsoH%2Fpya3BaZT4sDDhp1XEQUTzU12V1CH2I%3D&reserved=0   Hgt +250 m
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mm.icann.org/pipermail/tz/attachments/20181113/35b27ba4/attachment.html>

More information about the tz mailing list