karn@JUPITER.BELLCORE.COM (Phil R. Karn) (01/05/88)
I just had a most enjoyable chat with Mr. Kerry Kingham of the US Naval Observatory. Kerry provided me with the following list of leap seconds. Each took place on the very last minute of the last day of the indicated month: June 1972 December 1972 December 1973 December 1974 December 1975 December 1976 December 1977 December 1978 December 1979 June 1981 June 1982 June 1983 June 1985 December 1987 It turns out that there is a very amusing reason for the obvious patterns you see here. The original BIH policy was to declare leap seconds at the end of the year whenever possible, using the June date only when necessary to stay within the +/- 0.8 second UTC-UT1 limit. However, it seems that the French eventually rebelled at having to come in to work, thereby missing their New Year's Eve parties year after year, so in 1981 they decided to "anticipate" the need for a year-end leap second by doing it in June. That one took UTC as far behind UT1 as it has ever been (-0.78 seconds), just barely within the limit. (Fortunately WWV UT1 corrections (which only go to +/- 0.7 sec) could still handle this because their absolute values are truncated down to the next lower 100 ms multiple.) I mentioned the griping we've heard from various parties about what leap seconds do to radio clocks, and he says "Oh, that's nothing". He then told me about how several very upset people called him up once to ask why there was no leap second in that year. It seems that they had systems with a hardwired "positive leap second every December 31st" rule -- and some of those systems were in orbit! He then took great pleasure in letting them know that it is also possible to have NEGATIVE leap seconds... Kerry also improved my understanding of UT1. It is based on raw observations of the earth's rotation angle (defined as UT0) corrected for the effects of polar wandering. The earth's rotational poles wander over a roughly circular path every 14 months or so. This is caused by changes in the earth's mass distribution (solar tides, seasonal atmosphere mass shifts, melting snow caps, etc). The amplitude of the polar variation is roughly the size of a baseball diamond, and it causes small changes in the observer's true latitude and longitude -- therefore affecting the observer's astronomical observations on the order of +/- 30 ms. Phil