On Tue 2003-07-01T21:51:09 +0100, Markus Kuhn hath writ: > Is the Simon et al. formula considered good enough to predict the mean > longitude of the sun within half a day for the next few tenthousand > years? The authors make only a statement on the quality of ephimerides > derived for the years 1000-3000.
Simon et al. write that they have attempted to make the expressions valid, in the sense of the mean, from 4000 BC to 8000 AD. The best other reference I know is Laskar in Astron. Astrophys. 157, 59-70 (1986) where the mean elements of the planets were studied over 10000 years. The mean longitude of the inner planets are given as 10th order polynomials. I believe that in this paper the elements are more mean of a mean than in Simon, by which I mean that the cyclical variations of relatively short term are not present. (Apologies for that last sentence :-) Nevertheless, half a day is half a degree, and all these expressions are probably that good over those time spans. It is very hard to be more sure than that. The only valid checks that reach 4000 years before present are a few records of solar eclipses and the alignment of megalithic structures. These checks are not much better than half a degree. Extrapolating as far into the future is just as dangerous; it is really better to wait and see. Getting more directly back to the purposes of leap seconds... The current scheme of UTC with leap seconds should work for well over 1000 years. Around 1700 years from now it would become necessary to have a leap second at least once a month, and then the current scheme requires modification. In the absence of leap seconds, an atomic timescale will have deviated from UT by an hour in 1000 years. As Markus pointed out, the situation gets worse quadratically. Night becomes day between 3000 and 4000 years from now, and in around 5000 years today is tomorrow. For the purposes of broadcast time signals, it should be sufficient that any change which may be adopted will last for 1000 years without forcing posterity to implement a large discontinuity. As for 10000 years -- there can be point in laying down specifications for activities to be carried out over 300 generations from now. IMHO, even planning for 500 years is hubris, but knowingly creating a problem which must be solved by somebody else in 500 years is criminal. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E 49 89 0E FE 26 B4 14 93