Re: [LEAPSECS] what should a time standard encompass?

From: Markus Kuhn <>
Date: Mon, 27 Jan 2003 14:44:14 +0000

It has been pointed out here before, but might be worth reiterating:

If a leap-second-free atomic time such as TAI or some TAI+fixed_delta_t
became the future basis for the definition of civilian time, then all
that would happen is:

  - The Royal Greenwich Observatory would loose its distinguished
    role as being a point whose locale time is of any particular

      -> greater fairness, as eventually every spot on the
         planet will be like Greenwich in a number of millenia
         for a few years. As the universal-time meridian slowly moves
         eastwards, Paris, Berlin, Cairo, Moscow, etc. will soon
         (well...) enjoy that role, too.

  - After considerable time (every several millenia), the full-hour
    offsets between local civilian time and the broadcast international
    standard atomic time will have to be adjusted.

      -> No big deal, as we are already used to adjusting this full-hour
         offset twice per year anyway. We would just skip the end adjustment
         of daylight saving time every couple of millenia. Existing
         software for converting UTC to local time can easily handle this.
         Historically, it would vanish almost completely as a weak long-term
         trend in the noise that politically motivated civilian time zone
         adjustments generate.

      -> As a result of the continuous adjustment of the UTC offset of the
         civilian time zones east and west of the International Date Line,
         the IDL itself would not have to move.

      -> Greater fairness, as the points where local time and civilian
         time coincide are now not any more static, but move slowly
         from west to east within each civilian time zone between two
         civilian time zone adjustments.

      -> Potential problem: At the moment the difference between UTC and
         any local civilian time is limited to around +/- 13 hours. This would
         now grow without bound. Date and time encoding standards such as
         ISO 8601 need to be updated once the difference between any
         civilian time zone and UTC starts to move outside the range
         +/- 99 hours and 59 minutes, though that's quite some time off
         (In the order of 0.1 million years, and I would hope very much
         that by then people, if they still exist, got fed up with the
         only 0.005 million years old Babylonian base-60 time notation,
         which still cripples our common units of measurements)

So apart from a day overflow in a future far longer away than any
recorded history, basing civilian local time on TAI would not really
cause any noticeable practical or even philosophical problems.


Markus Kuhn, Computer Lab, Univ of Cambridge, GB | __oo_O..O_oo__
Received on Mon Jan 27 2003 - 06:44:35 PST

This archive was generated by hypermail 2.3.0 : Sat Sep 04 2010 - 09:44:54 PDT