So summarizing what I find in Steve's paper, and only concerning the pointing of the telescope and not the reporting results from the scope, large telescopes with guidance systems from the 1970s and 1980s have a guide camera field of view of 3 arcminutes, or a bit more. So if the telescope was aimed to bring a certain object to the center of view, and the time were off by the time equivalent of 1.5 arcminutes, the object would be out of view unless other fortuitous errors brought it back into view. If the object is on the celestial equator and the telescope has an equitorial mount, this would be 6 seconds of time. In this event, the situation is salvageable but it requires intervention by the on-site observing technicians.
In the case of a large fully-automated 21st century telescope, the APF, the specified pointing accuracy is 10 arc seconds. If it were an equitorial mount (which it isn't) this would correspond to a time error of up to 0.7 s. If we consider that a measurement standard should have a tolerance 1/3 as great as that which is being measured, and that in some directions the time equivalent of 10 arc seconds, we might estimate that a time base tolerance in the 10s of milliseconds would be negligible compared to other pointing errors. The same telescope will develop serious problems within a year of the discontinuance of leap seconds. I take this to mean that an error of 1 second in time is intolerable, which corresponds to an angle of, at most, 15 arc seconds. This would suggest that when converting UTC to sidereal time, the older telescopes could ignore the distinction between mean and apparent Greenwich Sidereal Time. Probably even for the APF the subtle differences in the definition of Greenwich Sidereal Time that have been made in the last 30 years or so could be neglected, right? (That is, just for pointing, not for reporting results.) Gerry Ashton _______________________________________________ LEAPSECS mailing list [email protected] http://six.pairlist.net/mailman/listinfo/leapsecs
